Lipid modifying and glucose therapies in clinical...
102
Linköping University Medical Dissertations No. 1285 Lipid‐modifying and glucose‐ lowering therapies in clinical practice: The impact of guidelines and changing reimbursement schemes Billie Pettersson Division of Health Care Analysis Department of Medical and Health Sciences Linköping University, Sweden Linköping 2012
Transcript of Lipid modifying and glucose therapies in clinical...
Linköping University Medical Dissertations No. 1285
Lipid‐modifying and glucose‐lowering therapies in clinical
practice: The impact of guidelines and changing
reimbursement schemes
Billie Pettersson
Division of Health Care Analysis Department of Medical and Health Sciences
Linköping University, Sweden
Linköping 2012
©Billie Pettersson, 2012 Cover picture/illustration: Published articles are reprinted with the permission of the copyright holder. Printed in Sweden by LiU‐Tryck, Linköping, Sweden, 2012 ISBN 978‐91‐7519‐998‐6 ISSN 0345‐0082
To my late parents Naser and Aida Kanngo: I felt your support all the way
Thinking: the talking of the soul with itself. Plato (427‐347 BC)
Contents
CONTENTS
ABSTRACT .................................................................................................................. 1
LIST OF PAPERS ........................................................................................................ 3
ABBREVIATIONS...................................................................................................... 5
1. INTRODUCTION................................................................................................... 7
1.1. Health policy for prevention ...................................................................... 9 1.1.1. Economics of prevention ................................................................... 10 1.1.2. Prevention of cardiovascular risk factors........................................ 11
1.2. Health technology assessment ................................................................. 13 1.2.1. Economic evaluations in HTA.......................................................... 14 1.2.1.1. Capturing quality of life in economic evaluations.............................. 16
1.2.1.2. Guidelines in economic evaluations ................................................... 18
1.2.1.3. The theoretical foundation of economic evaluations........................... 19
1.3. Pharmaceutical policy ................................................................................ 21 1.3.1. Guidelines ............................................................................................ 24 1.3.1.1. Lipid‐modifying therapies................................................................... 25
1.3.1.2. Glucose‐lowering therapies................................................................. 26
1.3.2. Pricing and reimbursement............................................................... 28 1.3.2.1. Lipid‐modifying therapies................................................................... 30
1.3.2.2. Glucose‐lowering therapies................................................................. 30
1.4. Use of preventive medicines..................................................................... 31 1.4.1. Lipid‐modifying therapies ................................................................ 31 1.4.2. Glucose‐lowering therapies............................................................... 33
2. AIMS OF THE THESIS........................................................................................ 34
3. METHODS AND MATERIALS......................................................................... 35
3.1. General design............................................................................................. 35 3.1.1. Observational studies......................................................................... 35
Contents
3.1.2. Quasi‐experimental studies............................................................... 36 3.2. Papers I‐IV.................................................................................................... 37
3.2.1. Paper I................................................................................................... 38 3.2.2. Paper II ................................................................................................. 40 3.2.3. Paper III ................................................................................................ 42 3.2.4. Paper IV................................................................................................ 45
4. RESULTS ................................................................................................................ 47
4.1. Lipid‐modifying therapies........................................................................ 47 4.1.1. Use and costs of lipid‐modifying therapies .................................... 47 4.1.2. Quality aspects of lipid‐modifying therapies................................. 51
4.2. Glucose‐lowering therapies...................................................................... 54 4.2.1. Use and costs of glucose‐lowering therapies.................................. 54 4.2.2. Quality aspects of glucose‐lowering therapies............................... 58
5. DISCUSSION........................................................................................................ 62
5.1. Main findings .............................................................................................. 62 5.1.1. Lipid‐modifying therapies ................................................................ 62 5.1.2. Glucose‐lowering therapies............................................................... 67 5.1.3. Use of preventive medicines ............................................................. 69
5.2. Methodological considerations................................................................ 72 5.2.1. Papers I & III ........................................................................................ 72 5.2.2. Paper II ................................................................................................. 73 5.2.3. Paper IV................................................................................................ 74
6. CONCLUSIONS ................................................................................................... 77
7. IMPLICATIONS FOR THE FUTURE............................................................... 78
8. FUTURE RESEARCH........................................................................................... 79
9. ACKNOWLEDGEMENTS.................................................................................. 81
10. REFERENCES ...................................................................................................... 82
Abstracts
1
ABSTRACT
Preventive medicine has evolved in recent decades as an important way of reducing
the risk of cardiovascular disease, which is still a major cause of death that creates
large burdens to society in terms of costs and morbidity. Dyslipidemia and type 2
diabetes mellitus are the main risk factors for cardiovascular disease, and national
and international guidelines recommend lipid‐modifying and glucose‐lowering
treatments for prevention. In 2010, about 836,000 (9% of the population) and 372,000
patients respectively were treated with these therapies in Sweden.
Various pharmaceutical policies aimed at improving the efficiency of drug use have
been introduced over the years. Health technology assessment (HTA) was
introduced in Sweden in 2002 as a foundation for informing pricing and
reimbursement decisions by the Dental and Pharmaceutical Benefits Agency (TLV).
Following HTA reviews, new reimbursement schemes for lipid‐modifying and
glucose‐lowering therapies were introduced in 2009 and 2010 respectively. To assess
the impact of the changing reimbursement schemes on the use and costs of these
therapies, we analyzed data from the Swedish drug registry, using a quasi‐
experimental design and interrupted time series analyses.
Our results showed that the new reimbursement scheme for lipid‐modifying
treatment had a major effect on use; following the implementation of this scheme,
there was a substantial increase in both discontinuation and switching to higher
doses. Conversely, the new reimbursement scheme for glucose‐lowering therapies
had overall only a minor effect on use. Larger savings in the lipid market were
anticipated but not fully realized, while even the minor anticipated changes in costs
in the glucose‐lowering market were not realized due to increased costs for insulins.
We found that changes in reimbursement schemes might lead to unintended effects,
which should be considered before implementation. Softer demand‐side policies,
Abstract
2
such as recommendations and guidelines, might be a better option under some
circumstances.
Clinical and national guidelines are other policies aimed at improving quality of care
and drug use. We assessed the impact of guidelines on the quality of lipid‐modifying
therapies, defined as proportions of patients attaining goal/normal levels according
to guidelines for lipid management. A longitudinal retrospective observational study
was carried out, covering time periods before and after initiation of lipid‐modifying
treatment. The findings show that about 40% of the patients attained the
recommended low‐density lipoprotein cholesterol goals following treatment, but
only 18% attained goals/normal levels in all lipid parameters. Improvement in
triglycerides was moderate, and low levels of high‐density lipoprotein cholesterol
persisted, showing only modest improvement following therapy. Treatment patterns
were found to have a better degree of adherence to guidelines regarding low‐density
lipoprotein cholesterol as compared to other lipid parameters.
The overall objective of treatment of type 2 diabetes mellitus is to improve glycemic
control without negatively affecting quality of life. Hypoglycemia is a common side
effect of intensive blood glucose control, mostly seen in patients treated with
insulins. Earlier studies have suggested that hypoglycemia has a negative impact on
quality of life, even in patients treated with oral glucose‐lowering therapies. We
carried out a cross‐sectional retrospective study to assess the impact of self‐reported
experience of hypoglycemia on quality of life in Swedish adult patients with type 2
diabetes mellitus treated with a combination of metformin and sulfonylureas. The
results showed that about 40% of the patients achieved the goal of glycemic control.
About 19% reported experience of moderate or more severe hypoglycemia, and these
patients were found to have lower quality of life than those patients reporting no or
mild hypoglycemia, as measured by EQ‐5D, a generic quality of life instrument. This
could be important to consider in clinical practice.
List of Papers
3
LIST OF PAPERS
I Billie Pettersson, Mikael Hoffmann, Per Wändell, Lars‐Åke Levin: Utilization and Costs of Lipid‐Modifying Therapies following Health Technology Assessment for the new reimbursement scheme in Sweden. Health Policy. 2011. 104(1): p. 84‐91. II Billie Pettersson, Baishali Ambegaonkar, Vasilisa Sazonov, Mats Martinell, Jan Stålhammar, Per Wändell. Prevalence of lipid abnormalities before and after introduction of lipid modifying therapy among Swedish patients with dyslipidemia (PRIMULA). BMC Public Health, 2010. 10(1): p. 737. III Billie Pettersson, Mikael Hoffmann, Per Wändell, Lars‐Åke Levin: Utilization and Costs of Glucose‐ lowering Therapies following Health Technology Assessment for the new reimbursement scheme in Sweden. (Submitted to Health Policy). IV Billie Pettersson, Ulf Rosenqvist, Anna Deleskog, Gunilla Journath, Per Wändell. Self‐reported experience of hypoglycemia among adults with type 2 diabetes mellitus (Exhype). Diabetes Res Clin Pract. 2010. 92(1): p. 19‐25.
4
Abbreviations
5
ABBREVIATIONS
CVD Cardiovascular disease CHD Coronary heart disease DCCT Diabetes Control and Complications Trial DDD Defined daily doses EQ5D EuroQol GLT Glucose‐lowering therapies HbA1c Glycated hemoglobin HDL‐C High‐density lipoprotein cholesterol HTA Health technology assessment LDL‐C Low‐density lipoprotein cholesterol LMT Lipid‐modifying therapies NCD Non‐communicable disease P&R Pricing and reimbursement QoL Quality of life SU Sulphonylureas QALY Quality‐adjusted life year TGs Triglycerides TIM Thousand inhabitants per month TLV The Dental and Pharmaceutical Benefits Agency T2DM Type 2 diabetes mellitus
6
Introduction
7
1. INTRODUCTION
Preventive medicine has evolved in recent decades as an important way of reducing
the risk of cardiovascular disease (CVD). In the OECD countries, life expectancy has
increased by 10 years since 1960, and CVD mortality has decreased substantially but
is still the main cause of death in Europe [1]. Technological change has had a major
impact on both health care outcomes and the quality of care. The introduction of
lipid‐lowering drugs has contributed significantly to the 50% reduction in
cardiovascular mortality observed in many countries during the last two decades [2,
3]. However, CVD‐related morbidity still constitutes a major burden to society, in
both economic and human terms. In 2006, the costs related to CVD were estimated at
around 10% of total health care expenditures in Sweden [1].
Technological change has been a major driver of health care spending over the post‐
war period [4, 5]. Pharmaceuticals represent around 15% of overall health
expenditure in the OECD countries, and increasing expenditures have led to the
introduction of different policies aimed at improving the efficiency of drug use [6].
In recent decades in Sweden, such policies have been introduced at a regional and
national level [7], leading to major reformation and changes in the pharmaceutical
market. Pharmaco‐economic assessments in support of listings of publicly‐funded
benefits were initiated in Australia in 1993 [8], and have now been introduced in
many OECD countries, in one form or another [9]. In Sweden, health technology
assessment (HTA) has emerged as an important foundation for guiding decision‐
making and allocating resources in health care. HTA was introduced in 2002 as the
basis for pricing and reimbursement (P&R) of new drugs as well as older drugs,
within the framework of the reviews of the Swedish P&R agency, the Dental and
Pharmaceutical Benefits Agency (TLV). HTA is increasingly used for production of
clinical, national, and regional guidelines.
Introduction
8
In preventive medicine in particular, private demand could be influenced by market
failures such as asymmetric information and moral hazard, and so the use of
preventive medicines could be other than the socially desired level. While prevention
is widely recommended by public health professionals as a strategy for improving
health, there is an increasing recognition of resource constraints, which has led to the
introduction of various pharmaceutical policies to control costs. Pharmaceutical
policies such as P&R and guidelines are instruments employed by governments to
steer demand and use of medicines towards a desirable level.
A review by Green et al. (2010)[10] concluded that policy measures should be
carefully designed and should be based on research quantifying the harm and benefit
profiles of the target and alternative drugs; otherwise there may be unwanted health
system and health effects, particularly where drugs are not interchangeable.
Furthermore, the authors concluded that removing restrictions on drugs that prevent
complications of disease might remove barriers to access, resulting in a desired
increase in their use as well as cost savings [10]. It has also been shown that
guidelines accompanied by a change in reimbursement rules had a significant
influence on the prescribing of lipid‐lowering drugs [11].
It is therefore of particular interest to study the impact and effectiveness of emerging
P&R policies and guidelines to steer use of preventive medicines. This dissertation
focuses on lipid‐modifying and glucose‐lowering therapies, which are preventive
therapies for two of the most prevalent risk factors for CVD. About 9% of the
population in Sweden is prescribed a lipid‐modifying therapy, and the prevalence of
type 2 diabetes mellitus (T2DM) in Sweden has been estimated at around 4.5% [12,
13], affecting more than 400,000 individuals.
Introduction
9
1.1. Health policy for prevention
CVD mortality has decreased substantially, but is still the main cause of death in the
European region, causing over two million deaths per year [1]. In Sweden, CVD
mortality is about 40% of total mortality, though the rate of death form heart disease
and stroke is decreasing for both women and men[14], see figure 1.
Figure 1. Number of deaths from heart disease and stroke in women and men per 100 000 in Sweden 1952-2008.
Source: Socialstyrelsen.
The European division of the World Health Organization has a vision of a health‐
promoting Europe free of preventable non‐communicable diseases (NCD),
premature death, and avoidable disability; this vision is laid out in the European
Strategy for the Prevention and Control of NCD [15]. The goal of this strategy is to
avoid premature death and significantly reduce the disease burden from NCD by
taking integrated action, by improving quality of life, and by making healthy life
expectancy more equitable within and between Member States.
In 2003, a new national public health strategy for Sweden was presented, one of its
objectives being health and medical care that more actively promotes good health.
This strategy stated that: “A health‐promotion and disease‐prevention perspective
shall be an integral part of the whole health and medical care service and be a
Women
0
200
400
600
800
1952 59 66 73 80 87 94 01 2008
Heart disease
Men
0
200
400
600
800
1952 59 66 73 80 87 94 01 2008
Stroke
Introduction
10
palpable component of all care and treatment” [16]1. Improving preventive care and
using the potential of prevention was also highlighted in the objectives of the WHO
strategy: “To take integrated action on risk factors and their underlying determinants
across sectors” and “To strengthen health systems for improved prevention and
control of NCD” [15]2.
1.1.1. Economics of prevention
Prevention is a broad concept, and a standardized approach identifies three
categories of intervention: primary, secondary, and tertiary. Primary prevention
consist of actions that reduce the occurrence or incidence of disease, secondary
prevention consists of actions that reduce or eliminate the health consequences of a
disease given its occurrence, and tertiary prevention consists of actions that reduce
the disability associated with a chronic illness [17].
Use of preventive medicine depends on individual decisions and the functioning of
the private markets; it is also affected by relevant market failures, such as moral
hazard, asymmetric information, and other externalities on prevention decisions.
Moral hazard is one type of market failure related to asymmetric information. It
refers in general to those actions of the insured which alter the accident probability
but are not observable by the insurer [18]. One example in the field of health care is
that health insurance for curative care might reduce incentives for prevention [19],
mainly because the individuals subjective estimated risk of morbidity is in general
lower than an objective estimation done by experts[20, 21]. However, other factors
might create further private incentives for prevention, because in many cases the
uninsurable utility loss from health risks, for example pain and suffering, far exceeds
the insurable monetary loss; that is, the coverage is incomplete. Furthermore, despite
1 Objective 6. Page 7. 2 Page 17
Introduction
11
insurance for curative care, prevention remains attractive because in cases when
complete cure is not possible, the choice is between completely preventing disease or
incompletely curing it.
Another source of failure for consumers to make optimal prevention decisions arises
from lack of (correct) information on a wide variety of primary and secondary
prevention activities [18, 22]. When this occurs, the decision is heavily influenced by
the recommendation from the physician, who is in turn influenced by factors such as
clinical guidelines and recommendations. In a publicly‐financed health care system,
incentives for prevention are partly shifted away from the insured onto the providers
of insurance. The public sector has a general incentive to encourage prevention; the
challenge is then to internalize this incentive to relevant agents who can influence
consumer preventive behavior in order to steer the use of preventive drugs towards
an optimal and cost‐effective level. Removing restrictions for drugs that prevent
complications of disease has been suggested to result in a desired increase in their
use as well as cost savings [10].
Health science research and the development of new medical technologies are other
important factors in determining trends in health as in cost and quality of medical
care [23].
1.1.2. Prevention of cardiovascular risk factors
Cardiovascular disease – disease of the heart and blood vessels – has three major
manifestations: coronary heart disease (CHD), transient ischemic attack, stroke and
peripheral arterial disease [24]. The underlying pathology for CVD is atherosclerosis,
which develops over many years and is usually advanced by the time symptoms
occur, generally in middle age.
Prevention and modification of risk factors can reduce clinical events and premature
death in people with established CVD as well as in those who are at high
Introduction
12
cardiovascular risk due to one or more risk factors [25]. A cardiovascular risk factor
is a condition that is associated with an increased risk of developing CVD [26]. The
concept of risk factors has evolved over the past 45 years, and new factors are
periodically added to the list as comprehension of the disease process grows [26]. Box
1 lists the currently accepted cardiovascular risk factors classified as factors that
cannot be changed, factors that can be changed, and factors that are protective.
Cigarette smoking, diabetes, hyperlipidemia, and hypertension have been
established as independent risk factors for CHD and are often labeled as
“conventional” risk factors because of the strength of evidence supporting their role
in the pathogenesis of CHD [27]. There is clear evidence that the four conventional
risk factors and their resulting health risks are largely preventable by a healthy
lifestyle [28].
Total risk estimation is a crucial tool to guide patient management, and has been a
cornerstone of guidelines. Individual risk factors should be evaluated against total
cardiovascular risk, since the combined effects of several risk factors may
interact [29]. Targets for individual risk factors are problematic in that they will
always be open to debate, they are not always achievable, and they seem to promote
a single risk factor approach to prevention [29]. There is, however, a consensus that
the risk increases continuously as blood pressure rises from levels that are
considered to be within the normal range [29].
Introduction
13
Box 1. Cardiovascular risk factors Risk factors that cannot be changed Age Gender Heredity Risk factors that can be changed High blood pressure Elevated serum cholesterol Lipoprotein (a) Cigarette smoking Obesity Glucose intolerance Diabetes Fibrinogen Left ventricular hypertrophy Cocaine Behavioral factors (stress, type A) Protective factors HDL cholesterol Exercise Estrogen Moderate alcohol intake Adapted from Black, Yale University School of Medicine Heart Book [26].
1.2. Health technology assessment
HTA is increasingly used in many countries to assist decision‐making regarding the
optimal use of competing health technologies. It has been defined as “a multi‐
disciplinary field of policy analysis, studying the medical, economic, social and
ethical implications of development, diffusion and use of health technology” [30].
The declared purpose of HTA is to support the process of decision‐making in health
care at policy level by providing reliable information; in this respect, HTA can be
seen as a bridge between the world of research and the world of decision‐
making [31].
HTA originated from growing concerns in the 1970s about the expanding costs of
new medical technology and the ability to finance them [32‐35]. During the
subsequent decades, there has been substantial demand for well‐founded
information from HTA to support decisions on the development, uptake, and
Introduction
14
diffusion of health technologies. This has led to a massive growth and development
in HTA, with the subsequent establishment of HTA programs in almost all European
countries, either in new agencies or institutes or in established academic units [9, 35].
While European HTA agencies share many of the same basic objectives, their
structures and how they operate differ widely across countries [9, 36]. Decision‐
makers in most European countries have increasingly relied on the use of HTA to
support P&R decisions regarding existing and new pharmaceuticals, prioritization,
development of clinical guidelines, and the direction of resources to the most cost‐
effective treatments in health care [9, 37]. HTA can therefore play a major role in
various phases in the use and diffusion of a health technology, notably when the
decision on reimbursement of the technology is taken (or revised) and when
recommendations on its use are made to the professionals using the technology.
1.2.1. Economic evaluations in HTA
Economic evaluations are important components of HTA. Economic evaluations
have been defined as the comparative analysis of alternative courses of action in
terms of both their costs and consequences. The general approach in full economic
evaluations is to compare the consequences of health care programs with their costs,
while partial evaluation might compare only consequences or only costs. The basic
tasks of any economic evaluation are to identify, measure, value, and compare the
costs and consequences of the alternatives being considered.
The main forms of full economic evaluations are tabulated in Table 1. All forms of
economic evaluations analyze costs in the same way, but differ in the way that the
consequences of health care programs are measured and valued. A general rule
when assessing two programs, A and B, is that the difference in costs is compared
with the difference in consequences, in an incremental analysis. Incremental analysis
means that difference between the costs of the two treatments to reach the defined
outcome, is divided by the difference in their effectiveness:
Introduction
15
CA– CB = ΔC EA – EB ΔE
Table 1. The main forms of full economic evaluations.
Type of evaluation Measurement/valuat
ions of costs in both
alternatives
Identification of
consequences
Measurement/valuatio
n of consequences
1. Cost‐minimization
analysis
Monetary Identical in all relevant
aspects
None
2. Cost‐effectiveness
analysis
Monetary Single effect of interest
but achieved to different
degree
Natural units (e.g. life‐
years gained, points of
blood pressure
reduction)
3. Cost‐utility analysis Monetary Single or multiple
effects, not necessarily
common to both
alternatives
Health years or quality‐
adjusted life‐years
(QALYs)
4. Cost‐benefit analysis Monetary Single or multiple
effects, not necessarily
common to both
alternatives
Monetary
Adapted from Drummond et al. [38].
Cost‐minimization analysis deals only with costs and can therefore be regarded as a
partial form of economic evaluation. When the consequences of two or more
alternatives are considered to be equivalent, cost‐minimization can be used to
compare the costs; hence, this analysis is a special form of cost‐effectiveness analysis
(CEA). In CEA, the consequences are measured in the most appropriate natural
effects or physical units, such as life‐years gained or units in any efficacy surrogate
Introduction
16
parameter, such as blood pressure or lipid values. In cost‐utility analysis (CUA), the
consequences of programs are measured in physical units as in CEA, but adjusted by
health state preference scores or utility weights. This gives the possibility to assess
the gain in quality of life and gain in life‐years in one single outcome measure, the
quality‐adjusted life year (QALY).
One severe limitation of CEA and CUA is that these analyses cannot provide
information on whether a program is efficient or worthwhile; that is, whether the
benefits exceed the costs. It is possible only to compare the cost‐effectiveness ratios of
various options [22]. In cost‐benefit analysis (CBA), however, the consequences are
valued in monetary terms and so can be directly compared to the costs related to a
program [38]. CBA can therefore be used to evaluate whether the beneficial
consequences of a program justify its costs.
1.2.1.1. Capturing quality of life in economic evaluations
Quality of life (QoL) has been defined as the “value assigned to duration of life as
modified by the impairments; physical, social, and psychological functional states;
perceptions; and opportunities that are influenced by disease, injury, treatment, or
policy” [39]. QoL is a subjective, multidimensional, and dynamic concept, and
therefore it is argued that QoL should be reported by patients whenever relevant and
appropriate, under the premise that the best way to find out about the effectiveness
of a certain treatment is to ask the patient [40]. QoL weights are fundamental in
health economic evaluations aimed at estimating the cost of QALYs gained, and they
play a major role in valuing the benefit of drugs where QALYs are used as the basis
for P&R decisions [41].
The QALY methodology is preferred by the TLV and many other HTA agencies,
because the outcome measure combines the dimensions of quantity of life (mortality)
and quality of life (morbidity), thus allowing comparison between different disease
areas with different clinical outcomes. In the QALY approach, the quality adjustment
Introduction
17
is based on a set of quality weights that represent the health‐related quality of life of
the health state under consideration. The weights (utilities) are derived from
consumer preferences, and so the consumers play a crucial role in valuing outcomes
in QALY methodology.
The concept of the QALY is illustrated in Figure 2. QoL weights are shown on the
vertical axis, with 0 representing death and 1 representing perfect health, and the
quantity of life (mortality) is shown on the horizontal axis. In this example, the
QALY gained by a treatment is illustrated in area A and B and it can be calculated as
((0.8*1+0.7*1)‐ (0.4*1))=1.1 QALY compared to no treatment. Area A shows the gain in
QoL, and area B shows the gain in both QoL and life‐years from the treatment.
Figure 2. The concept of the QALY.
Utility and values are different types of preferences. Preferences can be measured
using direct or indirect methods/techniques. The three most widely used techniques
for directly measuring the preferences of individuals for health outcomes are the
rating scale, the standard gamble, and the time trade‐off [38]. A simpler alternative to
these direct methods is to use one of the existing pre‐scored multi‐attribute health
A
B No treatment
Years
Quality of life weights
1.0
0.4
0.8
1 2
0.7
0.0 Death
Perfect health
Introduction
18
status classification systems, for example Quality of Well Being, Health Utility Index,
or EuroQol (EQ‐5D), to measure QoL and weights related to health status. These
methods and instruments have become widely accepted in health economics, and are
therefore considered a standard [41].
1.2.1.2.Guidelines in economic evaluations
The results from economic evaluations and analyses are heavily influenced by the
fundamentals and methodological choices related to costs and consequences. For
instance, if the analysis takes a health care perspective, then only the costs and
consequences arising for the health care sector are considered and included in the
cost analysis, while a societal perspective would allow for all costs and consequences
to be considered in the analysis. The employment of economic evaluations differs
between countries [42]. In Sweden, the TLV takes the following preferred approach
to drawing up a health economic analysis [43]:
• The health economic analysis should be done from a social economic
perspective. Among other things, this means that all relevant costs and
revenues for treatment and ill health should be considered, irrespective of the
payee (county council, local authority, state, patient).
• The information must describe the situation in Sweden.
• The costs and health effects of using the drug in question should be compared
with the most appropriate alternative treatment in Sweden. This could be
drug treatment, another treatment, or no treatment at all.
• The analysis should include the whole patient population to which the
subsidy application refers. Separate calculations should be made for different
patient groups where the treatment is expected to have different cost‐
effectiveness.
Introduction
19
• An estimation of the number of persons in each patient group in Sweden
should be attached.
• All relevant costs associated with treatment and illness should be identified,
quantified, and evaluated. The production loss for treatment and sickness
should also be included (estimated using the human capital method).
• Cost‐effectiveness analysis is recommended, with QALYs as the measure of
effect. Cost‐effectiveness ratios should be calculated based on the differences
in costs and effects (QALYs) that exist between treatment alternatives
(incremental analysis).
• QALY weightings should be based either on direct methods such as the
standard gamble or time trade‐off methods or on indirect measurements
(where a health classification system such as EQ‐5D is linked to QALY
weightings).
1.2.1.3.The theoretical foundation of economic evaluations
The theoretical foundation of economic evaluations is rooted in welfare economics, a
branch of normative economics that analyzes the desirability of different changes or
policies. Welfare economics is concerned with providing criteria to rank different
alternative changes or policies, with the aim of defining the optimal allocation of
resources [18]. The most widely used criterion for evaluating resource allocation is
Pareto efficiency, which states that a change is desirable if it makes some
individual(s) better off without making any other individual(s) worse off. Hence, a
situation is Pareto optimal if it is impossible to improve the situation of any
individuals(s) without making at least one other individual worse off [18, 22].
Introduction
20
Two key assumptions of the Pareto principle and the welfare approach are that 1)
social welfare is made up from the welfare (or utilities) of each individual member of
society; and 2) individuals are the best judges of their own welfare [41].
A reinterpretation of the Pareto principle, a potential Pareto improvement, was
provided by Kaldor and Hicks in the compensation test [18, 22]. Potential Pareto
improvements or compensation tests (Kaldor‐Hicks criterion) refer to the situation of
a policy that creates gainers and losers in welfare; if gainers in that situation could
compensate the losers and remain better off themselves after the change, then society
as a whole has benefited [41]. The reinterpretation of the Pareto principle and the
compensation tests form a basis for CBA to be operationalized, with program
benefits being valued using a compensation test based on the principle of willingness
to pay.
The approach of extra‐welfarism has emerged from critics of the traditional welfare
economics, mainly due to the narrow focus on individual utilities for resource
allocation that is implied by the key assumptions of traditional welfarist view [44].
Extra‐welfarism is considered as a pragmatic approach, taking as its theoretical
framework the aim of optimizing health benefits from a given budget; the evaluation
is aimed at informing decision‐makers rather than prescribing what decisions should
be made [45]. The most prominent differences between extra‐welfarism and welfarist
economics are that extra‐welfarism allows elements other than individual utility to
be considered in the analysis, it allows other sources of valuation of the relevant
outcomes, and it allows for interpersonal comparisons [44]. The extra‐welfarist
approach has however been criticized for its lack of theoretical framework, as it is not
embedded in standard welfare economic theory [46]. While CBA has its theoretical
roots in welfare economics, CUA and CEA are frequently referred to as non‐
welfarist, decision‐maker, or extra‐welfarist approaches [45]. Hence the two latter
types of analysis are criticized for having weak theoretical foundations in
comparison to CBA [45, 46].
Introduction
21
1.3. Pharmaceutical policy
Pharmaceutical policies are the instruments used by governments to control the
development, distribution, subsidization, pricing, and use of drugs in the
communities they govern [47]. While each OECD country has a unique mix of
pharmaceutical policies, their policy environments share several common features
that have implications for the resulting market dynamics [6]. These common features
comprise supply‐side policies such as intellectual property rights and regulation for
market authorizations, and demand‐side policies aimed at promoting affordable
access to medicines through various models for range and scope of subsidies through
P&R policies. The net effect of intellectual property rights and market authorization
is to raise prices by limiting competition, while the net effect of demand‐side policies
is to lower prices to consumers for pharmaceuticals through reduction or elimination
of out‐of‐pocket costs paid by the consumer [6].
Pharmaceutical expenditures in Sweden have been rising since the early 1990s, as in many other countries [6, 48‐51], see Figure 3. Figure 3. Public expenditures on pharmaceuticals including medical devices and other in the pharmaceutical benefits in 2010 years prices.
0
5 000
10 000
15 000
20 000
25 000
30 000
1989
1990
1991
1992
1993
1994
1995
1996
1997
1998
1999
2000
2001
2002
2003
2004
2005
2006
2007
2008
2009
2010
Miljoner Swedish crowns
Source: Socialstyrelsen: Läkemedel ‐statistik för år 2010.
Introduction
22
The main factor behind the escalation in drug costs in Sweden between 1990 and
2000 has been suggested to be the change from old to new and more innovative and
expensive drug therapies [51]. Increasing pharmaceutical expenditures have led to
the introduction of a variety of mainly demand‐side policies aimed at restricting the
escalation [48, 49, 52]. These policies can be divided into those aimed mostly at
promoting cost‐effectiveness and those aimed mostly at containing costs. Figure 4
presents the most important reforms in the Swedish pharmaceutical market since
1997 and the extent to which their purpose has been primarily to contain costs or to
promote cost‐effectiveness.
Figure 4. Major pharmaceutical reforms introduced in Sweden 1997-2004.
Adapted from Anell and Persson 2005[48].
Increased user charges (co‐payments), parallel trade, and the introduction of generic
substitution have been oriented towards cost containment, while other reforms have
been oriented towards promoting a rational and cost‐effective use of
pharmaceuticals [48].
The reforms to encourage rational use of prescription medicines at a regional level
have gathered pace, with drug budgets devolved to the counties as of 1998. These
reforms include measures operated via regional Drug and Therapeutic Committees,
such as the production of regional guidelines, academic detailing, benchmarking,
Introduction
23
prescribing targets, and incentives [7]. A variety of such initiatives have been
introduced in Sweden [53‐55]. These can be categorized under one or more of the
following four “E”s [54]: Education (programs that influence prescribing through
dissemination of material, which can be passive or active), Engineering
(organizational or managerial interventions), Economics (changes in insurance and
reimbursement, patient contributory payment, financial interventions, etc.), and
Enforcement (regulations including those enforced by law).
The reforms aimed at promoting a rational and cost‐effective use of drugs at a
national level involved already‐established organizations in Sweden, but also led to
the establishment in October 2002 of the Pharmaceutical Benefits Board
(Läkemedelsförmånsnämnden/LFN), which was renamed in 2008 to the Dental and
Pharmaceutical Benefits Agency (Tandvårds‐ och
läkemedelsförmånsverket/TLV) [53]. The establishment of this agency markedly
changed the principles of P&R of drugs in Sweden.
Reforms involving increased co‐payments were found to have limited effects on
expenditure and utilization of prescribed pharmaceuticals in Sweden, while policies
using indirect pricing, reference‐based pricing, and generic substitution were
associated with decreased cost per volume; generic substitution was also associated
with a decrease of total pharmaceutical expenditure [49, 52].
A review [10] found that under some circumstances, reimbursement restriction
policies can ensure better use of medicines with reduced costs and without an
increase in the use of other health services; this occurs, for example, when the
relevant drugs are aimed at targeting symptoms and have cheaper yet still effective
alternatives. On the other hand, relaxing reimbursement rules for drugs used for
prevention might remove barriers to access and increase use towards a desirable
level [10]. Green et al. concluded that policy measures should be carefully designed
and should be based on research quantifying the harm and benefit profiles of target
Introduction
24
and alternative drugs. Otherwise, there is a risk of unwanted health system and
health effects, particularly where drugs are not interchangeable.
A review by Cheah et al. (1998) found that there is considerable uncertainty whether
clinical guidelines will improve or influence clinical practice [56]. Another review by
Worrall et al. (1997) found that there is little evidence that the use of clinical
guidelines produces significant changes in clinical outcomes in primary care [57].
There is only limited knowledge about the impact of guidelines on quality and the
impact of the new Swedish P&R environment on use of medicines. This dissertation
will focus on these two policy instruments and their effects on the use of preventive
treatment with lipid‐modifying and glucose‐lowering medicines.
1.3.1. Guidelines
Regional and national clinical practice guidelines are developed by medical
specialists, national authorities, and regional authorities as a guide to physician
decision‐making. HTA is an important cornerstone in the production of national
guidelines to steer health care decision‐making. The main aim and focus of the
national guidelines are to steer towards an equal and cost‐effective care across the
country [58].
There are four national organizations in Sweden involved in producing and issuing
guidelines and recommendations and making decisions that influence the use and
quality of pharmaceuticals [53]: the Medical Products Agency
(Läkemedelsverket/LMV), the Swedish Council on Health Technology Assessment
(Statens Beredning för Medicinsk Utvärdering/SBU), the National Board of Health
and Welfare (Socialstyrelsen/SoS), and the TLV. The LMV produces
recommendations for pharmaceutical treatments, the SBU produces HTA reports and
issues recommendations, the SoS produces and issues national guidelines on health
care, and the TLV makes P&R decisions.
Introduction
25
1.3.1.1.Lipid-modifying therapies
Dyslipidemia is one of the major risk factors for CVD and CHD [27, 59]. Low‐density
lipoprotein cholesterol (LDL‐C) has been established as a key causative factor in the
progression of CHD [60‐62]. Independently of levels of LDL‐C, there is an inverse
association between high‐density lipoprotein cholesterol (HDL‐C) and increased risk
for CHD [62‐65] and CVD [66]. T2DM and CVD are associated with increased risk of
metabolic syndrome, which includes dyslipidemia [67]. Dyslipidemia in metabolic
syndrome is characterized by hypertriglyceridemia and low levels of HDL‐C [68, 69].
Low levels of HDL‐C have been shown to be predictive of and an independent risk
factor for developing CHD [70‐72].
Recommended thresholds for LDL‐C and total cholesterol (TC) and normal levels of
HDL‐C and triglycerides (TGs) as per the Swedish guidelines are outlined in Box
1 [73].
Box 2. Swedish guidelines for treating dyslipidemia.
Recommended lipid levels:
TC <5.0 mmol/L (very high risk: <4.5 mmol/L)
LDL‐C <3.0 mmol/L (very high risk: <2.5 mmol/L)
Indications for increased risk:
TGs >1.7 mmol/L
HDL‐C <1.0 mmol/L (men), and <1.3 mmol/L (women)
The average cholesterol level in the population in Sweden has decreased in recent
decades, see Figure 5. A large cohort study in northern Sweden [74] showed that
mean TC lipid levels decreased significantly during 1986‐2004, from 6.4 to 5.8 mmol/l
in men and 6.3 to 5.5 mmol/l in women aged 25‐64 years, and from 6.4 to 5.5 mmol/l
in men and 7.1 to 6.2 mmol/l in women in men and women aged >65 [75, 76]. Data
Introduction
26
from the Västerbotten Intervention Programme [77] has shown declining lipid levels
in the population since the introduction of the new guidelines in 1999 [78]. However,
from 2004 onwards there has been a tendency to an increase in mean TC lipid levels
among both men and women [79].
Figure 5. Age-adjusted levels of total cholesterol in the population of northern Sweden, 25–64 years old, 1986–2004.
Adapted from Eliasson et al. 2006[75].
1.3.1.2.Glucose-lowering therapies
The risk of developing CVD is increased twofold to fourfold in patients with T2DM,
independently of other concomitant risk factors [80]. Glycemic control in these
patients reduces the risk of developing complications [81, 82]. International [83] and
Swedish national guidelines recommend that glycated hemoglobin (HbA1c) level is
used as a treatment target. The Swedish standard is a level of <6% on the Mono S
scale [84, 85], which is comparable to the <7% standard used in the Diabetes Control
and Complications Trial (DCCT) and the <52 mmol/mol threshold recommended by
the International Federation of Clinical Chemistry and Laboratory Medicine
(IFCC) [86]. Most patients require pharmacological treatment to reach these
Introduction
27
treatment goals. Even so, only 50% of all T2DM patients in Sweden reach the goal of
HbA1c <6% [87], with goal attainment varying widely between regions in the range
of 40–60% [88].
However, while intensive glycemic control reduces the risk of microvascular
complications and non‐fatal myocardial infarction, it increases the risk of
hypoglycemia [82]. Failure to achieve the treatment goals could, among other
reasons, be due to poor adherence because of the side effects of certain antidiabetic
treatments [89]. Hypoglycemia has been suggested to be the main limiting factor in
achieving adequate glycemic control [90]. Furthermore, hypoglycemia episodes
induce costs, the annual cost in Sweden of managing hypoglycemic events is
estimated at around 50 million SEK [91].
T2DM is a progressive disease [81], and so most patients will sooner or later become
candidates for an add‐on treatment [84]. In the new international and Swedish
national guidelines (introduced in 2010) [83‐85], sulfonylureas (SUs) are
recommended for use as a second‐line treatment in patients not adequately
controlled on metformin. Thus SUs are seen as an alternative to neutral protamine
Hagedorn (NPH) insulin treatment, and although they are known to induce
hypoglycemia [92, 93], they are recommended before newer treatments such as
thiazolidinediones (glitazones), dipeptidyl peptidase‐4 (DPP‐4) inhibitors, and
glucagon‐like peptide‐1 (GLP‐1) analogues [84, 85].
The overall objective of diabetes treatment is to maintain good QoL while avoiding
acute and long‐term complications [94]. A study in France showed that
hypoglycemia induced by oral antidiabetic agents has a negative effect on QoL [95].
A study of Swedish patients found similar results, but 43% of the patients were
treated with insulin [96], why the results might not be appropriate to generalize to
patients treated with oral therapies. There is, therefore, still limited knowledge of
how QoL is affected in Swedish patients with T2DM in general; this was confirmed
by a recent evaluation of the national guidelines [97]. In particular, there is a lack of
Introduction
28
knowledge of how Swedish patients with T2DM treated with oral antidiabetic agents
experience hypoglycemia.
This lack of knowledge could be due to a lack of clarity in the stated goal for QoL.
While the objective of glucose control is a fairly clear goal, the objective of avoiding
deteriorating QoL is somewhat vague and therefore difficult to evaluate. A TLV
review states: “Acute serious events in T2DM are unusual, however even less serious
symptoms caused by either hyper or hypoglycemia should be avoided to retain good
quality of life” (freely translated into English) [98]3. Swedish national guidelines also
emphasize QoL: “In the clinical situation one has to make individual assessment of
the remaining life expectancy and quality of life. In particular, the latter (QoL)
requires care and responsiveness vis‐à‐vis the individual it concerns” (freely
translated into English) [84]4.
1.3.2. Pricing and reimbursement
In 2002, the Swedish parliament passed a new pharmaceutical benefits reform aimed
at promoting cost‐effective use of publicly‐financed pharmaceuticals and ensuring
equal drug benefits throughout the country. Following the reform, HTA was
introduced as a foundation for P&R decision making and the LFN was established as
a new government body. As previously mentioned, the LFN was later transformed
and renamed to the TLV after its responsibility was extended to include
reimbursement decisions for dental care.
The establishment of the LFN/TLV produced significant changes in the P&R of
pharmaceuticals in Sweden. The assigned task of the P&R agency was to make
decisions on new prescription pharmaceuticals to be included in the public
pharmaceutical benefits scheme, and to review all pharmaceuticals already included
3 Page 16. 4 Pages 14‐15.
Introduction
29
in this scheme [7, 99, 100]. The primary purpose of the establishment of TLV was to
replace the former system that “automatically” subsidized pharmaceuticals with
market authorization by a system that employed HTA coupled with economic
efficiency criteria (i.e. marginal cost‐effectiveness) and ethical principles for
prioritization in health care as the basis for reimbursement [48], as stated in law.
These ethical principles are: 1. human dignity, 2. need and solidarity, and 3. cost‐
effectiveness. The aim was to improve the rational and cost‐effective use of
medicines.
Economic evaluations which provide evidence on the cost‐effectiveness of a product
have since been a central foundation for the decisions made by the TLV. The
guidelines and practices for these evaluations are of major importance for the whole
functioning of the market for pharmaceuticals (see section 1.2.1.2.Guidelines in
economic evaluations).
The universal coverage scheme in Sweden acts as a combined pharmaceutical
subsidy and de facto nationwide price regulation mechanism for the subsidized
products [6]. A company applies for reimbursement of a pharmaceutical at a given
price, and TLV’s evaluation of whether the product is cost‐effective is normally based
on documentation enclosed in this application. The same principle also applies for
TLV’s pharmaceutical reviews, which evaluate those pharmaceuticals that were
included in the pharmaceutical benefits according to the old system. From its
establishment until 2010, the TLV was following a priority plan for reviewing nearly
2,000 drugs included in the national reimbursement scheme prior to the new system,
employing the same principles as in the new reimbursement system [99, 101, 102].
This plan was recently abandoned for a more flexible process and sequence for the
reviews, where TLV can initiate reviews in any area at any time in order to improve
the efficiency of the reviews [101].
Introduction
30
1.3.2.1. Lipid-modifying therapies
The TLV review of the lipid‐modifying market was presented in a report published
in February 2009. Following this review, the new reimbursement scheme was
implemented on 1 June 2009 [103]. The review concluded that regardless of the statin
used, a decrease in LDL‐C is correlated to the risk of CVD. Consequently, the
reimbursement for each lipid‐modifying medicine was continued, restricted, or
halted on the basis of the medicine’s marginal cost‐effectiveness on documented
decrease in LDL‐C. TLV estimated that the new reimbursement scheme could result
in savings of 170 million SEK per annum (≈ €18 million at a rate of 1 SEK ≈ €0.104,
June 2011).
An evaluation of the initial effects of the review estimated that likely savings within
the reimbursement scheme amounted to 47 million SEK (≈ €5 million) for the first 6
months, a large part of which was related to lower prices for generic simvastatin
rather than changes in reimbursement status [104].
1.3.2.2. Glucose-lowering therapies
The TLV review of glucose‐lowering therapies (GLT) was presented in a report
published on 2 December 2009, which concluded that use of GLT in Sweden was
cost‐effective, with a few exceptions [101]. Consequently, all glucose‐lowering
therapies were assigned one of the following four reimbursement statuses in the new
reimbursement scheme: retained, restricted, no reimbursement, or no reimbursement
for new courses. This new reimbursement scheme was introduced on 1 March
2010 [105]. TLV estimated that the decisions made in the review could result in cost
savings of at least 12 million SEK per annum [105] (≈€ 1.3 million at a rate of 1 SEK ≈
€ 0.11, 5 November 2011).
Introduction
31
An independent evaluation of the initial effects of the new GLT reimbursement
scheme showed that savings at constant volumes based on data for the first six
months following the new scheme were in line with those estimated by TLV [106].
1.4. Use of preventive medicines
1.4.1. Lipid-modifying therapies
International and Swedish guidelines recommend control and management of
dyslipidemia for primary and secondary prevention in patients at risk of CVD [58,
73, 107, 108]. There is wide documentation on the protective effects of lipid‐
modifying therapies (LMT), mostly statins, for various patient groups in clinical trials
[61, 109], and on their cost‐effectiveness for primary as well as secondary
prevention [24, 110‐112].
Statins are widely and increasingly used in many countries, though their use varies
extensively. Use in Sweden is at an average level in comparison with other Nordic
and western European countries [113, 114]. About 8.7% (816,000) of the population in
Sweden in 2010 were dispensed a statin, an increase from 6.7% in 2006 [115]. The
total number of patients treated with lipid‐modifying drugs in 2010 was about
836,000 (98% receiving statins), an increase of about 32% since 2006, while the
increase in defined daily doses (DDD) was about 55% over the same period [115].
There have been considerable price differences between patented and off‐patent
statin substances since the Swedish patent for Zocord (simvastatin) expired in 2003.
Prices for the off‐patent versions can now be about 90% lower, as is the case in many
other countries [116], and since the patent expiration there has been a sharp increase
in use defined in DDD and substantial decrease in costs, see Figure 6. While over
610,000 (83%) patients were treated with simvastatin and 100,000 (14%) were treated
Introduction
32
with atorvastatin in 2008 in Sweden, these two drugs accounted for 30% and 60% of
total costs, respectively [103]. Statins have therefore been a particular target for
various types of cost containment measures [117‐121], mostly through demand‐side
mechanisms enforced by law [122, 123].
Figure 6. Defined daily dose (DDD) and costs (Swedish crowns: SEK) per thousand inhabitants/day 2000-2008.
0
10
20
30
40
50
60
70
2000 2001 2002 2003 2004 2005 2006 2007 2008
DDD
/TIN
D
0
20
40
60
80
100
120
140
160
Cost
/TIN
D (S
EK)
C10AA statins DDD C10AA statins costs
Source: Swedish drug registry (Läkemedelsregistret).
New restrictive regulations were introduced in Norway in 2005 and in Finland in
2006. Following these new regulations, about 40% of the users of the more expensive
statins in Norway switched to simvastatin [117]. In Finland, 58% of those using
atorvastatin and 49% of those using rosuvastatin before the restriction switched to a
less expensive statin [118]. In both countries, the policies were considered successful
in reducing the overall cost of statins [117, 118].
Therapeutic reference pricing strategies have not been proven conclusively
successful as a cost containment tool [124]. In Hungary, such strategies caused
increased use of higher doses of statins, which increased overall expenditure [119].
Including statins in the German reference pricing scheme resulted in total savings
ranging from €94.4 million to €108.7 million in 2005, but also led to higher
Introduction
33
contributory payments for patients, which might explain the higher discontinuation
rates for patients initially treated with atorvastatin [121].
1.4.2.Glucose-lowering therapies
Use of glucose‐lowering therapies has increased over time in many European
countries, but at different rates and levels. In Sweden, the number of patients treated
with GLT in 2010 was 372,000, an increase of 17% since 2006. Of these, 183,000 were
treated with insulins and 265,000 with oral GLT, representing an increase of 15% and
19% respectively since 2006 [115]. In total, 76,000 patients were treated with both
insulins and oral GLT during 2010, either at the same time or due to a change of
therapy.
Until 2000 the use of insulin was highest in Sweden, in a comparison of ten European
countries, while the use of oral GLT was on an average level [125]. The prevalence of
T2DM has been estimated at around 3.5%‐4.5% in Sweden [12, 126], and around 3%
on average among European countries, ranging from 1.7% in the Netherlands to 4.2%
in Germany [127]. Earlier findings suggest, however, that this variation is
overestimated and due more to variation in factors related to definitions, detection,
and registration, among others [128]. The variation in prevalence does not, therefore,
fully explain the variation in use of GLT. Another factor that has been suggested to
explain the variation is differences in reimbursement schemes [125]. Insulins are fully
reimbursed for individuals in Sweden, the county councils being responsible for
covering the co‐payments, while the costs of oral GLT are covered by individuals
within the regular co‐payment scheme [129].
Aims of the thesis
34
2. AIMS OF THE THESIS
The main purpose of this thesis was to analyze the impact of changes in
reimbursement schemes and guidelines on use, costs, and quality of preventive
treatment with lipid‐modifying and glucose‐lowering therapies.
The purposes and aims of the specific papers were:
Paper I: To compare use, costs, and switching behavior regarding LMT before and
after the implementation of the new reimbursement scheme in June 2009.
Paper II: To estimate the prevalence of dyslipidemia and attainment of goal/normal
lipid levels in patients treated with LMT.
Paper III: To compare use and costs of GLT before and after the implementation of
the new reimbursement scheme in March 2010.
Paper IV: To evaluate the experience of hypoglycemia in patients treated with
metformin in combination with SUs, and the impact on patients’ QoL and level of
worry about hypoglycemia.
Methods and materials
35
3.METHODS AND MATERIALS
3.1. General design
A quantitative, deductive approach using an observational or quasi‐experimental
design was adopted for all four studies.
3.1.1. Observational studies
A basic distinction in quantitative research is that between experimental and non‐
experimental (observational) research. Experimental studies involve some type of
intervention, while non‐experimental or observational studies do not. In an
observational study, the investigator observes and evaluates the results that occur
without intervention. Randomized controlled trials are considered the most
scientifically rigorous method for hypothesis testing, yielding high internal validity
for the association between exposure and outcome [130, 131], due to the highly
controlled settings in which they operate. This type of setting, however, is also a
limitation when it comes to generalizing the findings to reflect real‐life clinical
practice; this is in contrast to observational studies. The performance of medicines in
real‐life clinical practice is of crucial importance to inform decision‐makers about the
effectiveness of a treatment [41].
Methods and materials
36
3.1.2. Quasi-experimental studies
In quasi‐experimental or experimental study designs, the investigator allocates or
controls the exposure of interest in an attempt to isolate the effect of the exposure
only; in this way, causal associations can be better established [130]. Quasi‐
experimental studies, like true experiments, involve an intervention. However, the
quasi‐experimental design lacks the randomization that is the signature of a true
experiment [131]. Quasi‐experimental designs are useful in guideline
implementation research for evaluating the effects of interventions when it is difficult
to randomize or identify an appropriate control group [132].
The three most commonly used designs in guideline implementation studies are
uncontrolled before‐and‐after studies, time series designs, and controlled before‐and‐
after studies [132].
Interrupted time series design is the strongest quasi‐experimental approach for
evaluating the longitudinal effects of interventions. Segmented regression analyses of
interrupted time series data are often used to assess how an intervention changed an
outcome of interest [133]. Segments in a time series are defined when the sequence of
measures is divided into two or more portions at change points, with two parameters
defining each segment: the level and the trend (or slope) [133]. The level is the value
of the series at the beginning of a given time interval, and the trend is the rate of
change of a measure during a segment.
Methods and materials
37
3.2. Papers I-IV
Table 2. Overview of the design and methods of each study. Paper I Paper II Paper III Paper IV Title Utilization and costs
of lipid modifying therapies following health technology assessment for the new reimbursement scheme in Sweden
Prevalence of lipid abnormalities before and after introduction of lipid modifying therapy among Swedish patients with dyslipidemia (PRIMULA)
Utilization and costs of glucose lowering therapies following health technology assessment for the new reimbursement scheme in Sweden
Self‐reported experience of hypoglycemia among adults with type 2 diabetes mellitus (Exhype)
Aims The aim of this study was to compare utilization, costs and switching behavior in patients treated with LMT before and after the new reimbursement scheme.
To estimate the prevalence of dyslipidemia and attainment of goal/normal lipid levels in patients treated with lipid modifying therapy (LMT).
To compare utilization and costs of GLT for type 2 diabetes Mellitus (T2DM) before and after the implementation of the changed Reimbursement schemes
To evaluate the experience of hypoglycemia in patients treated with metformin in combination with sulphonylureas (SUs) and the impact on patients’ quality of life (QoL) and worry about hypoglycemia.
Methods and materials
A quasi‐experimental study using data on dispensed LMT and costs from a database on dispensed individual prescriptions in Sweden. Segmented regression analyses were used to assess utilization and costs of LMT.
Longitudinal retrospective observational study covering time periods before and after treatment. Data were collected from 1994‐2007 electronic patient records in public primary healthcare centers in Uppsala County, Sweden.
This was a quasi‐experimental study using data on dispensed GLT and costs from a database on dispensed individual based prescriptions in Sweden. Segmented regression analyses were used to assess utilization and costs.
A national, cross‐sectional, multicenter study. Patients with type 2 diabetes treated with metformin and SU dual therapy were recruited by 54 investigators between January 2009 and August 2009. The patients were asked to complete a QoL instrument, (EQ5D) and the Hypoglycemia Fear Survey (HFS‐II) questionnaire.
Methods and materials
38
3.2.1. Paper I
This was a quasi‐experimental study [132] using a segmented time series [133] design
to analyze the use and costs of LMT. The study consisted of two time periods: before
and after the new reimbursement scheme (the intervention) was implemented.
Under the new reimbursement scheme, each drug was assigned to one of the
following three reimbursement statuses: continued, restricted, or excluded, Table 3.
1. Continued full reimbursement: generic pravastatin and simvastatin.
2. Restricted reimbursement: atorvastatin (Lipitor) and rosuvastatin (Crestor) in
higher strengths (>10 mg and >5 mg respectively) are reimbursed as a new treatment
only if generic simvastatin has been tried and the patient has not reached the
treatment objectives. Patients who have previously used atorvastatin 10 mg or
rosuvastatin 5 mg should first have tried simvastatin before higher doses of
atorvastatin and rosuvastatin may be prescribed with reimbursement. Ezetimibe is
reimbursed if generic simvastatin has been tried and the patient has not achieved the
treatment objectives, or if it has been established that the patient does not tolerate
statins.
3. Excluded from the reimbursement scheme: atorvastatin 10 mg, rosuvastatin 5 mg,
fluvastatin, pravastatin, cholestyramine, and branded simvastatin (Zocor) for all
packages except 80 mg in packs of 49 tablets (this refers only to the patented Zocord
80 mg in 49 tablets pack, since this pack was deemed to have an acceptable
price/tablet [134]).
Methods and materials
39
Table 3. The new reimbursement status for respective products following the TLV review for LMT.
a Individual products and packages which contain any of these substances lose their reimbursement.
b These medicines are not reimbursed for the indication blood lipid disorder.
Drug dispensing data were collected from the Swedish prescribed drug register held
by the National Board of Health and Welfare [115]. The register is described
elsewhere [135], but in short, it contains data for all dispensed prescriptions covering
the whole population of Sweden (9.5 million inhabitants). It contains patient‐specific
data (personal identifier, age, gender, place of residence) as well as drug data such as
the Nordic article number (which provides the trade name, pharmaceutical form,
strength, and package size), number of packages, Anatomical Therapeutic Chemical
(ATC) code, amount in DDD [136], prescription category, reimbursement code,
prescribing date, dispensing date, and price. It covers all prescribed drugs purchased
at Swedish pharmacies, but not drugs used in hospitals or purchased over the
counter, and it does not contain clinical information on diagnoses/indications for
treatment. Use of LMT was defined in volumes, expressed in terms of thousand
inhabitants per month (TIM) as number of patients/TIM or DDD/TIM. Costs were
measured in SEK and all analyses were carried out for total costs (reimbursed
expenditure and patient contributory payment) and converted to Euros (SEK, 1 SEK
≈ € 0.104, June 2011).
Methods and materials
40
Linear segmented regression analyses were used to analyze the changes in the levels
and trends in use and costs before and after the intervention. Number of patients,
DDD, and costs were entered as dependent variables. A dichotomous indicator
variable for the intervention was entered as an independent variable in the
regression models. The regression models allowed for a slope for the time period
preceding the intervention and a slope and a level shift to account for the change
after the intervention. Separate models were constructed for the total group and for
each reimbursement category. Any shifts in level (intercept) or slope related to the
regulation with p<0.05 were considered statistically significant.
Switching behavior was assessed as the number of individuals initially treated with
atorvastatin 10 mg or rosuvastatin 5 mg (at least the two last consecutive
dispensations but dispensation of other statin before the two subsequent
dispensations possible) during a one year period prior to the intervention (May 2008‐
May 2009) and the first dispensation during a six‐month period following the
intervention. The same analyses were carried out for a reference period using a
corresponding time period of one year before the intervention (May 2007‐ May 2008).
All analyses were performed using version 16.0 of the SPSS Statistical Package for
Windows and version 9.1.3 of the SAS software package (SAS Institute Inc., North
Carolina, USA).
3.2.2. Paper II
This was a longitudinal retrospective observational study covering time periods
before and after LMT. The study consisted of a baseline period (15 months prior to
initiation of LMT) and a follow‐up period (12 months following LMT). Data were
collected retrospectively (1994‐2007) from electronic patient protocols using a search
engine to scan patient protocols in 26 out of 30 public primary health care centers
Methods and materials
41
serving 77% of the total population in the county of Uppsala, Sweden (total
population: 322,043 in 2007).
The study included patients >35 years of age whose lipid values indicated
dyslipidemia, who had initiated LMT (ATC code C10: Lipid modifying agents) [136]
between May 1994 and June 2006, and who had complete lipid profiles at baseline
and at follow‐up. Treatment gaps of up to 6 weeks were allowed during the follow‐
up period except for the first 6 weeks post index date (initiation of LMT). A total of
5,424 patients met the criteria and were included in the study.
Normal and goal lipid levels were defined according to Swedish guidelines (see
section 1.3.1.1.Lipid‐modifying therapies). Mixed dyslipidemia was defined as
abnormal levels of more than one lipid fraction. High‐risk groups were defined as
those with CHD, T2DM without CHD, and 10‐year CHD risk>20% without CHD or
T2DM. Patients with T2DM and those with CHD were identified from the
International Classification of Disease (ICD) diagnostic codes. Patients with 10‐year
CHD risk>20% were identified by calculating risk per Framingham Risk Score [137].
Descriptive analyses were performed to evaluate baseline patient characteristics and
the prevalence of dyslipidemia and goal attainment, mixed dyslipidemia, and
treatment patterns, using thresholds for dyslipidemia according to clinical
guidelines. These analyses were carried out for the total study population as well as
for subgroups. Chi‐squared tests were used to detect significance in differences in
proportions between groups at the level of p<0.05 (two‐tailed). Multivariate logistic
regressions were used to evaluate factors associated with attainment of goal/normal
lipid levels.
Methods and materials
42
3.2.3. Paper III
This was a quasi‐experimental study [132] using a segmented time series [133] design
to analyze the use and costs of GLT in Sweden following the new reimbursement
scheme. The study consisted of two time periods: before and after the new
reimbursement scheme (the intervention) was implemented. We studied 38 separate
months in the period between February 2008 and March 2011 (25 months before and
13 months after the intervention on March 1, 2010).
The TLV review of GLT included all drugs in group A10 of the ATC
classification [138]. Under the new reimbursement scheme, each drug was assigned
to one of the following four different reimbursement statuses: retained, restricted,
excluded, and excluded for new courses of treatment, see Table 4.
Methods and materials
43
Table 4. Reimbursement categories in the new reimbursement scheme for glucose-lowering therapies.
Retained
Insulins: Rapid‐acting human insulin, rapid‐acting insulin analogues, intermediate‐acting human
insulin (NPH), intermediate‐acting insulin with rapid onset (2‐step).
Oral: Biguanides (generic metformin), SUs (Mindiab, generic glimepiride, generic glibenclamide,
Amaryl).
Restricted
Insulins: For T2DM patients the coverage of long‐acting insulin analogues, insulin glargine, and
insulin detemir is restricted to patients for whom other insulin treatment is not sufficient to reach the
treatment objectives due to recurring hypoglycemic episodes. (The restrictions did not apply for
patients with type 1 diabetes.) Insulin detemir used to be restricted to patients with type 1 diabetes
from 2004 until June 19, 2007, when the restriction was lifted [139].
Oral: Acarbose, rosiglitazone, pioglitazone, sitagliptin, vildagliptin, repaglinide, exenatide, and
combinations of these with metformin, will only be reimbursed for patients who have first tried
metformin, SUs, or insulin, or if these treatments are not suitable.
Excluded
Oral: Nateglinide (Starlix) is no longer covered by the benefit scheme. Glibenclamide sold under the
trade name of Daonil is excluded while generic products with glibenclamide are retained. The
combination of rosiglitazone and glimepiride is excluded from the reimbursement scheme, while the
active substances as separate substances are retained.
Excluded for new courses of treatment
Oral: Glibenclamide (Glibenklamid Recip) is excluded for incident patients but retained for patients
already treated with the drug.
The analyses were carried out on total effects from the intervention for oral and for
insulin therapies respectively. Oral GLT were analyzed in total and for each of the
four reimbursement statuses. Insulin therapies were analyzed in total and for
reimbursement statuses ‘retained’ and ‘restricted’.
Methods and materials
44
As in Paper I, drug dispensing data were collected from the Swedish prescribed drug
register held by the National Board of Health and Welfare [115].
Use of GLT was defined as volumes expressed as number of patients/TIM. We used a
cut‐off age of >40 years at the time of dispensation as a proxy for patients with
T2DM, since indications are not available in the registry and the onset of T2DM
usually occurs in patients >40 years [126]. Costs were measured in SEK and all
analyses were carried out for total costs (reimbursed expenditure and patient co‐
payment). We converted to Euros using current exchange rates (1 SEK ≈ € 0.11, 5
November 2011.
Linear segmented regression analyses were used to analyze the changes in the levels
and trends in use and costs before and after the intervention. Number of patients and
costs were entered as dependent variables. A dichotomous indicator variable for the
intervention was entered as an independent variable in the regression models. The
regression models allowed for a slope for the time period preceding the intervention
and a slope and a level shift to account for the change after the intervention. Separate
models were fitted for oral GLT and for insulin‐based GLT in total and for each
reimbursement category. Shifts in level (intercept) or slope related to the intervention
with p<0.05 were considered statistically significant.
Total costs were computed one year before (March 2009‐February 2010) and one year
after (March 2010‐February 2011) the new reimbursement scheme was implemented,
and the results were used to analyze cost savings following the intervention.
All analyses were performed using version 16.0 of the SPSS Statistical Package for
Windows and version 9.1.3 of the SAS software package (SAS institute Inc., North
Carolina, USA).
Methods and materials
45
3.2.4. Paper IV
This was a cross‐sectional, multicenter study. Patients were recruited by their general
practitioner (GP). After consenting to participate in the study, patients received a
self‐administered questionnaire. Data on patient characteristics and medical record
data were entered into an online form by the physician or a research nurse. A sample
of GPs and diabetologists from 54 sites in Sweden recruited patients consecutively
during their usual GP visit between January 2009 and August 2009. Patients with
T2DM aged 35 years or older, male or female, on treatment with metformin and SU
for the last six months were enrolled. Medical data were collected from the patient’s
records.
The patients were asked to complete questionnaires covering QoL and worry about
hypoglycemia during the past 6 months, demographics, and experience of
hypoglycemia. The specific questionnaires used were Experience of Low Blood
Sugar, EuroQol‐5 Dimensions (EQ‐5D), and the Hypoglycemia Fear Survey (HFS‐II).
The form concerning experience of low blood sugar (hypoglycemia) has been used in
previous studies [95] and contains 10 items on the frequency and seriousness of
hypoglycemic events in the patient’s history. Mild symptoms of low blood sugar
were defined as causing “little or no interruption of your activities, and you didn’t
feel you needed assistance to manage symptoms”, moderate symptoms caused
“some interruption of your activities, but [you] didn’t feel you needed assistance to
manage symptoms”, severe symptoms were described as “[you] felt that you needed
the assistance of others to manage symptoms (for example, to bring you food or
drink)”, and very severe symptoms “needed medical attention (for example, called
an ambulance, visited an emergency room or hospital, or saw a doctor or nurse)”.
QoL was evaluated with EQ‐5D, a generic instrument for use as a measure of health
outcome covering the following dimensions: mobility, self‐care, usual activities,
pain/discomfort, and anxiety/depression [140]. The EQ‐5D consists of 5 items scored
on a 3‐point Likert scale plus a visual analogue scale (VAS). Summary score
Methods and materials
46
calculated from the responses ranges from 0 to 1, where 1 is perfect health and 0 is
death. Scores were weighted using published weights from the UK population [141].
Worry about hypoglycemic symptoms was quantified using the Worry subscale of
HFS‐II [142, 143]. This subscale consists of 18 items scored on a 5‐point Likert scale
ranging from 0 (never) to 5 (very often). Scores on the worry subscale range from 0 to
72, with 0 representing the least worry.
Descriptive statistics were computed for all quantitative and qualitative variables.
Hypoglycemia was dichotomized into none/mild or moderate/severe/very severe.
Between‐group comparisons on continuous variables were carried out using a t‐test
(or ANOVA if more than 2 groups) or the Mann‐Whitney‐Wilcoxon test (or Kruskal‐
Wallis if more than 2 groups) if the requirements for the t‐test were not met. For
categorical variables, the comparisons between groups were carried out using a Chi
squared test, or a Fisher exact test if the requirements for the Chi squared test were
not met. Age‐adjusted p‐values were calculated using ANCOVA for continuous
variables and Cochranʹs Mantel‐Haenszel for categorical variables. A p‐value <0.05
was considered statistically significant.
Results
47
4.RESULTS
4.1. Lipid-modifying therapies
4.1.1. Use and costs of lipid-modifying therapies
Results from the regression analyses on the number of patients/TIM treated with
LMT before and after the intervention are shown in Figure 7 A‐D. Looking at total
LMT, before the intervention there was a slightly increasing trend in number of
patients/TIM (p=0.0007) (Figure 7 A), with no statistically significant differences
following the intervention. In the “continued” category (Figure 7 B), the number of
patients/TIM was slightly increasing before the intervention (p=0.0002), with no
statistically significant differences in level or trend following the intervention. The
“restricted” category (Figure 7 C) also showed an increase prior to the intervention
(p=0.002), and there was a statistically significant increase in level following the
intervention (p=0.0336). Finally, in the “excluded” category (Figure 7 D), the trend
was slightly negative before intervention (p=0.0001) and there was a negative shift in
level following the intervention (p<0.0001).
Results
48
Figure 7 A-D. Segmented regression analyses of number of patients/1000 inhabitants treated with LMT in all and in respective reimbursement categories before and after implementation of the intervention.
Switching behavior for all patients and for a subgroup of patients with diabetes
(patients treated with glucose‐lowering drugs) initially treated with atorvastatin 10
mg during a reference period and the period after the intervention is shown in Table
5.
Table 5. Switching of all patients and patients with diabetes initially treated with atorvastatin 10 mg, reference period and period after introduction of the new reimbursement scheme.
Initial treatment 10 mg atorvastatinAll patients Diabetes coprescription
Alternative Period before Period after Period before Period aftern=38443 n=32751 n=9168 n=7834
Remain on initial treatment 87.3% (87.1 ‐ 87.6) 20.6% (20.2 ‐ 21.0) 87.1% (86.5 ‐ 87.6) 19.5% (18.7 ‐ 20.2)
Switch to higher dose of initial treatment 1.2% (1.1 ‐ 1.3) 19.8% (19.4 ‐ 20.2) 1.4% (1.2 ‐ 1.6) 18.2% (17.4 ‐ 18.9)
Switch to simvastatin 10 mg 0.3% (0.3 ‐ 0.4) 6.5% (6.3 ‐ 6.7) 0.2% (0.2 ‐ 0.4) 6.5% (6.1 ‐ 7.0)
Switch to simvastatin 20/40/80 mg 1.2% (1.1 ‐ 1.3) 29.3% (28.9 ‐ 29.8) 1.4% (1.2 ‐ 1.6) 32.0% (31.2 ‐ 32.9)
Switch to other 0.4% (0.4 ‐ 0.5) 3.5% (3.3 ‐ 3.7) 0.5% (0.4 ‐ 0.6) 3.3% (3.0 ‐ 3.7)
Discontinuation 9.4% (9.2 ‐ 9.7) 20.2% (19.9 ‐ 20.6) 9.4% (8.9 ‐ 10.0) 20.5% (19.8 ‐ 21.3)
Results
49
An alternative view is shown in Figure 8. The proportion of patients initially treated
with atorvastatin who remained on the same strength following the intervention was
21%, while the corresponding figure for the reference period was 87%. Of all patients,
20% switched to higher doses of atorvastatin following the intervention compared to
1% in the reference period. Switching to higher doses (20/40/80 mg) of simvastatin
from atorvastatin 10 mg was increased from 1% to 30% following the intervention. In
the reference period, 10% of patients discontinued treatment, while the
corresponding figure following the intervention increased to 20%.
Figure 8. Switching of patients initially treated with 10 mg atorvastatin before and after the introduction of the new reimbursement scheme.
0%
10%
20%
30%
40%
50%
60%
70%
80%
90%
100%
Period before (n=38443) Period after (n=32751)
Remain on initial treatment Switch to higher dose of initial treatmentSwitch to simvastatin 10 mg Switch to simvastatin 20/40/80 mgSwitch to other Discontinuation
Results
50
Costs
Results from the regression analyses are shown in Figure 9 A–D. In total, the
costs/TIM decreased significantly following the intervention by about 1,122 SEK/TIM
(≈ €0.12/TIM) (p<0.0001). This corresponds to a total saving of 125 million SEK/year (≈
€13 million/year) (figures not shown). This was predominately reflected in the “no
reimbursement” category (p<0.0001) (Figure 9 D). In the “restricted reimbursement”
category, neither level nor trend showed a statistically significant difference
following the intervention (Figure 9 C).
Figure 9 A-D. Segmented regression analyses of total cost/1000 inhabitants in all and in respective reimbursement categories before and after implementation of the intervention.
In summary, the new reimbursement scheme had a dramatic effect on the use of
LMT. Patients initially treated with low doses of statins that were eventually
excluded from the reimbursement scheme, switched to higher doses or discontinued
to a much greater extent compared to a reference period. However, the new
reimbursement scheme had only a moderate decreasing effect on costs, and the
expected savings were not realized.
Results
51
4.1.2. Quality aspects of lipid-modifying therapies
Among the 5,424 patients included, the prevalence of dyslipidemia (≥1 lipid
abnormality) was 100% by definition at baseline, and 82% at follow‐up. At baseline,
60% had elevated LDL‐C combined with low HDL‐C and/or elevated TGs, while the
corresponding figure at follow‐up was 36%. Low HDL‐C and/or elevated TGs at
follow‐up remained at 69% for patients with T2DM, 50% among patients with CHD,
and 66% among patients with 10‐year CHD risk >20%. At baseline, there were large
differences between different high risk groups in terms of prevalence of elevated
TGs, low HDL‐C, and mixed dyslipidemia, while no major differences were seen
concerning elevated TC or LDL‐C. The most prominent differences were in the
prevalence of dyslipidemia in the group with T2DM compared to those with CHD.
At follow‐up, differences in TC and LDL‐C were most prominent in the group with
10‐year CHD risk >20%, where they persisted in about 70%. The prevalence of
elevated TGs, low HDL‐C, and combinations decreased modestly and persisted to a
greater extent in the groups with T2DM and with 10‐year CHD risk >20%, Table 6.
Table 6. Prevalence of dyslipidemia at baseline (BL) and follow-up (FU) by riskgroups and p-values for comparison of proportions.
Results from multivariate regression models (Table 7) to determine attainment of
goal/normal levels in TC, LDL‐C, TGs, and HDL‐C suggested a slight positive
association between age and attainment of goal/normal levels in all lipid parameters.
Results
52
Female compared to male gender was associated with significantly lower odds of
attaining normal levels in HDL‐C (odds ratio [OR]: 0.07; 95% confidence interval [CI]:
0.05‐0.08). Patients with T2DM had significantly lower odds of attaining lipid
goals/normal levels in any lipid parameter than patients without T2DM. Patients
with a history of CHD had significantly lower odds of reaching goal/normal level in
TC or LDL‐C, compared with patients without history of CHD. Baseline lipid values
were strongly associated with attainment of goal/normal levels in all lipid
parameters. For each 0.1 mmol/l increase in TC or LDL‐C at baseline, the odds of
attaining goal levels decreased by about 7% for TC and for LDL‐C respectively (OR:
0.93; 95% CI: 0.92‐0.93) and (OR: 0.93; 95% CI: 0.92‐0.94). For each 0.1 mmol/l increase
in TG at baseline, the odds of attaining normal levels were about 15% lower. Baseline
HDL‐C values were strongly and positively associated with attainment of normal
levels, as the odds increased by more than 200% for each 0.1 mmol/l increase in
baseline value (OR: 2.15; 95% CI: 2.06‐2.25). Duration of statin therapy was associated
with lower odds of attaining goal/normal levels of TC, LDL‐C, and TG. For each year
on statin treatment, the odds of attaining TC or LDL‐C goal were about 15% lower
and the odds of attaining normal levels of TGs were 6% lower.
Results
53
Table 7. Logistic regressions on goal/normal lipid level attainment.
In summary, we found that the majority of patients (98%) were treated with statins,
and 40% of the patients attained goal levels of LDL‐C but only 18% attained
goal/normal levels on all three lipid parameters. Improvement in TGs was moderate.
Low HDL‐C persisted, showing only modest improvement following therapy, and
this was most notable in patients with T2DM. These findings show that treatment
patterns were fairly well in line with treatment guidelines concerning LDL‐C, but
less in line in treatment of other lipid parameters.
Results
54
4.2. Glucose-lowering therapies
4.2.1. Use and costs of glucose-lowering therapies
Use of glucose‐lowering therapies
Insulins
Results from the regression analyses of number of patients/TIM treated with insulin‐
based GLT before and after the intervention are shown in Figure 10 A–C. There was
an accelerated increasing trend in total number of patients/TIM treated with insulins
following the intervention (p=0.0027) (Figure 10 A); this was the case for insulins
with retained reimbursement as well as for restricted insulins (Figure 10 B–C).
Figure 10 A-C. Segmented regression analyses of number of patients/1000 inhabitants treated with insulin in total and in respective reimbursement categories before and after implementation of the intervention.
Insulin total reimbursement
Time of intervention
0
2
4
6
8
10
0 10 20 30 40
Month
Patien
ts/1000
Inhabitant
Patients Regress ion li ne
Level p=0.5474Trend p=0.0027
Insulin restricted reimbursement
Time of intervention
0
1
2
3
0 10 20 30 40
Patients/1000 inhabitants
Patients Regression l ine
Level p=0.0422Trend p=0.0007
Insulin, patients/1000 inhabitants
A
B
C
Insulin continued reimbursement
Time of intervention
0
3
6
9
0 10 20 30 40
Month
Patients/1000
Inhabitants
Patients Regression line
Leve l p=0.7100Trend p=0.0021
B
Insulin total reimbursement
Time of intervention
0
2
4
6
8
10
0 10 20 30 40
Month
Patien
ts/1000
Inhabitant
Patients Regress ion li ne
Level p=0.5474Trend p=0.0027
Insulin restricted reimbursement
Time of intervention
0
1
2
3
0 10 20 30 40
Patients/1000 inhabitants
Patients Regression l ine
Level p=0.0422Trend p=0.0007
Insulin, patients/1000 inhabitants
A
B
C
Insulin continued reimbursement
Time of intervention
0
3
6
9
0 10 20 30 40
Month
Patients/1000
Inhabitants
Patients Regression line
Leve l p=0.7100Trend p=0.0021
B
Results
55
Oral glucose‐lowering therapies
Results from the regression analyses of number of patients/TIM treated with oral
GLT before and after the intervention are shown in Figure 11 A‐E. There were no
significant differences when the total group of drugs was considered (Figure 11 A).
There was a negative trend in number of patients/TIM treated with products with
restricted reimbursement (p=0.0035) (Figure 11 C) and a negative shift in products
that were excluded from the reimbursement scheme following the intervention
(p<0.0001) (Figure 11 D).
Figure 11 A-E. Segmented regression analyses of number of patients/1000 inhabitants treated with oral glucose lowering therapies in total and in respective reimbursement categories before and after implementation of the intervention.
Oral total reimbursement
Time of intervention
0
5
10
15
0 10 20 30 40
Month
Patients/1000
Inhabitants
Patients Regression line
Level p=0.1610Trend p=0.6084
Oral no reimbursement
Time of intervention
0
0,01
0,02
0,03
0,04
0 10 20 30 40
Month
Patients/10
00 Inhabitants
Patients Regress ion line
Level p=<.0001Trend p=0.1603
Oral No reimbursement for new courses
Time of interventi on
0
0,5
1
1,5
2
2,5
0 10 20 30 40
Month
Patients/10
00 Inhabitants
Patients Regress ion line
Level p=0.5867Trend p=0.1880
Oral continued reimbursement
Time of intervention
0
4
8
12
0 10 20 30 40
Month
Patients/10
00 Inhabitants
Patients Regress ion line
Level p=0.1255Trend p=0.2625
Oral restricted reimbursement
Time of interventi on
0
0,4
0,8
1,2
1,6
2
0 10 20 30 40
Month
Patie
nts/1000
Inhabitants
Patients Regression line
Level p=0.1348Trend p=0.0035
Oral glucoseloweringtherapies, patients/1000 inhabitants
A
B C
D E
C
Oral total reimbursement
Time of intervention
0
5
10
15
0 10 20 30 40
Month
Patients/1000
Inhabitants
Patients Regression line
Level p=0.1610Trend p=0.6084
Oral no reimbursement
Time of intervention
0
0,01
0,02
0,03
0,04
0 10 20 30 40
Month
Patients/10
00 Inhabitants
Patients Regress ion line
Level p=<.0001Trend p=0.1603
Oral No reimbursement for new courses
Time of interventi on
0
0,5
1
1,5
2
2,5
0 10 20 30 40
Month
Patients/10
00 Inhabitants
Patients Regress ion line
Level p=0.5867Trend p=0.1880
Oral continued reimbursement
Time of intervention
0
4
8
12
0 10 20 30 40
Month
Patients/10
00 Inhabitants
Patients Regress ion line
Level p=0.1255Trend p=0.2625
Oral restricted reimbursement
Time of interventi on
0
0,4
0,8
1,2
1,6
2
0 10 20 30 40
Month
Patie
nts/1000
Inhabitants
Patients Regression line
Level p=0.1348Trend p=0.0035
Oral glucoseloweringtherapies, patients/1000 inhabitants
A
B C
D E
Oral total reimbursement
Time of intervention
0
5
10
15
0 10 20 30 40
Month
Patients/1000
Inhabitants
Patients Regression line
Level p=0.1610Trend p=0.6084
Oral no reimbursement
Time of intervention
0
0,01
0,02
0,03
0,04
0 10 20 30 40
Month
Patients/10
00 Inhabitants
Patients Regress ion line
Level p=<.0001Trend p=0.1603
Oral No reimbursement for new courses
Time of interventi on
0
0,5
1
1,5
2
2,5
0 10 20 30 40
Month
Patients/10
00 Inhabitants
Patients Regress ion line
Level p=0.5867Trend p=0.1880
Oral continued reimbursement
Time of intervention
0
4
8
12
0 10 20 30 40
Month
Patients/10
00 Inhabitants
Patients Regress ion line
Level p=0.1255Trend p=0.2625
Oral restricted reimbursement
Time of interventi on
0
0,4
0,8
1,2
1,6
2
0 10 20 30 40
Month
Patie
nts/1000
Inhabitants
Patients Regression line
Level p=0.1348Trend p=0.0035
Oral glucoseloweringtherapies, patients/1000 inhabitants
Oral total reimbursement
Time of intervention
0
5
10
15
0 10 20 30 40
Month
Patients/1000
Inhabitants
Patients Regression line
Level p=0.1610Trend p=0.6084
Oral no reimbursement
Time of intervention
0
0,01
0,02
0,03
0,04
0 10 20 30 40
Month
Patients/10
00 Inhabitants
Patients Regress ion line
Level p=<.0001Trend p=0.1603
Oral No reimbursement for new courses
Time of interventi on
0
0,5
1
1,5
2
2,5
0 10 20 30 40
Month
Patients/10
00 Inhabitants
Patients Regress ion line
Level p=0.5867Trend p=0.1880
Oral continued reimbursement
Time of intervention
0
4
8
12
0 10 20 30 40
Month
Patients/10
00 Inhabitants
Patients Regress ion line
Level p=0.1255Trend p=0.2625
Oral restricted reimbursement
Time of interventi on
0
0,4
0,8
1,2
1,6
2
0 10 20 30 40
Month
Patie
nts/1000
Inhabitants
Patients Regression line
Level p=0.1348Trend p=0.0035
Oral glucoseloweringtherapies, patients/1000 inhabitants
A
B C
D E
C
Results
56
Costs
Total costs in the year following the intervention (March 2010–February 2011)
amounted to 1.1 billion SEK (≈€ 121 million), an increase of 58 million SEK (≈€6.4
million) compared to a reference period one year before the intervention (March
2009–February 2010). Total costs for insulins increased by 60 million SEK (≈€ 6.6
million), while costs for oral GLT decreased by 2 million SEK (≈€ 0.22 million) (data
not shown).
Insulins
Regression analyses showed accelerated increasing trends in total costs (Figure 12 A),
both for insulins with retained reimbursement and for insulins with restricted
reimbursement, following the intervention. There was an increase of 100 SEK/TIM
(≈€ 11/TIM); of this, 60 SEK (≈€6.6) was for products with retained reimbursement
and 40 SEK (≈€4.4) was for restricted products (data not shown).
Oral glucose‐lowering therapies
There were negative trends in costs for oral GLT in total (Figure 12 B) and for drugs
which were retained or restricted in the new reimbursement scheme. There was a
negative level shift in products excluded from the reimbursement scheme (p<0.0001)
(data not shown).
Results
57
Figure 12 A-B. Segmented regression analyses of costs/1000 inhabitants for insulin and oral GLT on total before and after implementation of the intervention.
Following the changed reimbursement status, there was an accelerated increasing
trend in the number of patients treated with restricted or retained insulins, as well as
in costs for insulin‐based GLT. No impact was detected in the total number of
patients treated with oral GLT, but there was a slightly negative trend in total costs
for oral GLT following the intervention.
In summary, we found the new reimbursement scheme for GLT had a minor impact
on use and costs of oral GLT. Despite restricted reimbursement for patients with
T2DM, the use of insulin‐based GLT and related costs increased faster following the
intervention.
Oral total reimbursement
Time of intervention
0
1000
2000
3000
4000
0 10 20 30 40
Month
Cost/1000 Inhabitants
Cost Regression line
Level p=0.2738Trend p=0.0177
Insulin total reimbursement
Time of intervention
0
4000
8000
12000
0 10 20 30 40
Month
Cost/1
000 Inhabitants
Cost Regress ion l ine
Level p=0.3094Trend p=0.0014
A
B
Oral total reimbursement
Time of intervention
0
1000
2000
3000
4000
0 10 20 30 40
Month
Cost/1000 Inhabitants
Cost Regression line
Level p=0.2738Trend p=0.0177
Insulin total reimbursement
Time of intervention
0
4000
8000
12000
0 10 20 30 40
Month
Cost/1
000 Inhabitants
Cost Regress ion l ine
Level p=0.3094Trend p=0.0014
A
B
Results
58
4.2.2. Quality aspects of glucose-lowering therapies
The study included a total of 430 Swedish adult patients with T2DM treated with a
combination of metformin and SUs. Their mean age was 69, and 60% of them were
men, Table 8. All patients were treated with metformin, with a mean daily dose of
1862 mg, and SUs (glibenclamid [64%], glimepirid [10%], and glipizid [26%]).
The goal of <6% HbA1c was attained by 40% of the total population, with 39% of
patients reporting no or only mild hypoglycemia, and 48% of patients reporting
experience of moderate or more severe hypoglycemia (p=0.14 between the latter two
groups). Mean HbA1c was lower in the group reporting moderate or worse
hypoglycemia compared to patients reporting no or mild hypoglycemia (6.1 mmol/l
and 6.4 mmol/l respectively, p=0.03), Table 8 .
Table 8. Patient characteristics in all patients and study groups with no/mild hypoglycemia and moderate/worse hypoglycemia, mean (SD), percentage.
In the total population, 34% reported that they had experienced hypoglycemia, with
19% experiencing moderate or worse hypoglycemia (17%, 1%, and 1% experienced
moderate, severe, and very severe hypoglycemia, respectively). Experience of
moderate or worse hypoglycemia was associated with a lower QoL measured by EQ‐
Results
59
5D summary scores (lower EQ‐5D scores indicate lower QoL). Both the weighted and
the unweighted EQ‐5D summary scores were lower in the group reporting moderate
or worse hypoglycemia than in the group with no/mild hypoglycemia (weighted:
0.81 vs. 0.88, p<0.001; unweighted: 0.75 vs. 0.83, p<0.001). The VAS score was lower
for the group reporting moderate or worse hypoglycemia than for the group with
no/mild hypoglycemia (0.71 vs. 0.76, p<0.01), Table 9.
Table 9. EQ5D scores, weighted and unweighted, visual analogue (VAS) scores and HFS-II worry score, mean (SD) in all patients and in group with no/mild and moderate or worse hypoglycemia.
The proportion of patients reporting problems in the EQ‐5D dimensions (mobility,
self care, usual activities, pain/discomfort, anxiety/depression) are shown in Figure
13. More patients in the group with moderate or worse hypoglycemia reported
problems in the dimensions ʺpain/discomfortʺ and ʺanxiety/depressionʺ compared to
the group with no/mild hypoglycemia symptoms (pain/discomfort: 56% vs. 43%,
p<0.01; anxiety/depression: 36 vs. 22%, p<0.05).
Results
60
Figure 13. Proportion (%) of patients with no/mild and moderate or worse hypoglycemia indicating problems per EQ-5D dimension.
QoL measured by the VAS score decreased, while fear of hypoglycemia increased,
with increasing severity of experienced hypoglycemia, Figure 14 and Figure 15.
Figure 14. EQ-5D VAS score by severity of hypoglycemia experienced.
Results
61
Figure 15. HFS-II worry score by severity of hypoglycemia experienced
In summary, only 40% of Swedish adult patients with T2DM treated with a
combination of metformin and SUs attained the goal of <6% HbA1c, and patients
reporting moderate or more severe hypoglycemia were found to have lower QoL
than patients reporting no or mild hypoglycemia.
Discussion
62
5. DISCUSSION
5.1. Main findings
5.1.1. Lipid-modifying therapies
About 836,000 patients in Sweden in 2010 were treated with lipid‐modifying
therapies, predominantly statins (98%). We found that changes following the new
reimbursement scheme had a large effect on the use of LMT. Patients initially treated
with low doses of statins that were eventually excluded from the reimbursement
scheme switched to higher doses or discontinued to a much greater extent in a period
following the new reimbursement scheme compared to a reference period. The new
reimbursement scheme had a moderate decreasing effect on costs, but the expected
savings were not realized.
The primary purpose of the reformed P&R system in Sweden is to improve the
rational and cost‐effective use of medicines. The TLV review of the lipid‐modifying
market concluded that regardless of the statin used, a decrease in LDL‐C is correlated
to the risk of CVD [103]. Consequently, the reimbursements for all lipid‐modifying
drugs were continued, restricted, or halted on the basis of their marginal cost‐
effectiveness on documented decrease in LDL‐C. According to our findings, this
approach accelerated the effect of reducing use of lower doses of atorvastatin and
rosuvastatin, and increased use of generic simvastatin; however, it was paired with
potentially less desirable effects related to switching behavior and discontinuation.
More patients initially treated with atorvastatin 10 mg and rosuvastatin 5 mg
switched to higher doses. Similar results were shown in Hungary and in Germany,
Discussion
63
where use of higher doses of statins was increased following the introduction of
therapeutic reference pricing [119, 121]. Increased doses could be beneficial if
therapeutically motivated, but high doses are also associated with more side
effects [144, 145]. However, neither study can exclude the possibility that patients
split the higher‐dose tablets to spread their use over more than one day, for example
using one 40 mg tablet split in half over two days instead of using two whole 20 mg
tablets, a strategy which is sometimes used to reduce costs [146]. Switching to higher
doses prior to the intervention could also have been influenced by the new treatment
guidelines issued in 2008 [58], where the use of patented statins to attain target levels
of LDL‐C was recommended for secondary prevention in high‐risk patients, as well
as by the increased awareness of treatment objectives induced by the coming
intervention. It has further been shown that guidelines accompanied by a change in
reimbursement rules had a significant influence on the prescribing of lipid‐lowering
drugs [11].
Total costs decreased following the intervention despite increased volumes, but the
expected savings were not fully realized. In addition, the calculated savings might be
overestimates, as they do not take into account any long‐term effects on costs of other
health care resources that might be caused by the policy, such as costs for GP visits
when patients switch, or costs for managing morbidity as a consequence of
discontinuation. This is a limitation of the studies in the present thesis as well as
other similar studies [117, 118]. On the other hand, evaluation of the health impact of
a policy is more important where drugs are not interchangeable [10], which is not the
case in LMT.
A full cost analysis should account for the total transition and other costs related to
the new reimbursement scheme. It should also be evaluated against the accumulated
savings generated over time, which may be substantial. The long‐term effectiveness
of lipid‐lowering drugs in clinical practice is therefore of major interest, and is the
subject of another ongoing project [147]. On the other hand, in this specific case,
because the Swedish patent for atorvastatin expires during 2011, the savings
Discussion
64
associated with low‐dose atorvastatin over time might be limited if the price for
generic atorvastatin was to follow that of generic simvastatin.
Economic evaluations, as employed in current practice in Sweden, use a static
approach in cost‐effectiveness analyses; they do not include assessment of products
using data on the total life cycle of a particular treatment. This approach might not be
an efficient resource allocation tool over time [148]. Lindgren et al. showed that if the
cost‐effectiveness data for statins were re‐estimated taking into consideration the
current availability of generics, the results would indicate savings for health care
systems; the cost savings from reductions in events are greater than the cost of the
drug [111], creating large social benefits over time. Producer surplus will only last for
the limited time until patent expiration, while the social net benefits will not only
increase with a reduction in price, but will also continue over a long time frame. The
producer appropriated 20–43% of the value during the on‐patent period, a figure
dropping to 1% following loss of exclusivity. The total producer surplus between
1987 and 2018 will be 2–5% of the total social surplus. The major part of the social
surplus generated comes from the value of improved quality‐adjusted survival. The
monetary value of the total surplus was estimated at between 2,368 and 1,135 million
euros per million inhabitants, based on value of statistically saved life and QALY
respectively. A review by Ward et al. of the cost‐effectiveness of statins
acknowledged the influence of prices on cost‐effectiveness, as the analyses were
pointed out as being sensitive to the cost of statins; the authors therefore advocated
for reviewing the cost‐effectiveness in the light of any significant changes in the price
of statins [24]. In contrast to static efficiency, a dynamic approach, where the future
value of a pharmaceutical is considered, is a more efficient criterion for resource
allocation [148]. Another method that could be considered to satisfy the criterion of
dynamic efficiency is to employ an anticipated average price of a pharmaceutical
over its total life cycle for CEA, based on both the price of a product during the
patented time and its price after patent expiration.
Discussion
65
Due to increasing discontinuations and switching to higher doses, and the long term
impact this might have on health outcomes and costs over time, we do not consider it
possible to clearly judge if use of lipid‐modifying therapies was improved following
the new reimbursement scheme. It has been suggested that soft demand‐side policies
might be a more effective tool to promote a cost‐effective use of drugs, since more
restrictive regulations might lead to unintended effects [54].
Quality aspects
Our results show that in terms of the guidelines, about 40% of the patients attained
goals in LDL‐C following treatment, but only 18% attained goals/normal levels in all
three lipid parameters. Improvement in TGs was moderate, and low HDL‐C
persisted, showing only modest improvement following therapy; this was most
notable in patients with T2DM, which could be explained by the limited use of LMT
targeting lipid parameters other than LDL‐C.
Studies have shown that LMT targeting multiple lipid abnormalities provides
additional benefits beyond statin monotherapy, where niacin was considered to be
the most effective HDL‐C modifying agent available [149, 150]. When used alone or
in combination with other LMT, niacin has been associated with a significant
reduction in cardiovascular events [151‐153]. On the other hand, a recent study
showed that there was no incremental clinical benefit from addition of niacin to
statin ± ezetimibe therapy in patients at target LDL‐C with established, non‐acute,
atherosclerotic cardiovascular disease, despite significant improvements in HDL‐C
and triglyceride levels. This creates doubt about the usefulness of niacin in reducing
residual risk [154].
In the logistic regression analysis, we found duration of statin therapy to be
negatively associated with attainment of goal/normal levels. For each additional year
on statin treatment, the odds of attaining LDL‐C goal decreased by about 10% (OR:
0.86; 95%CI: 0.84‐0.87). This could indicate lower efficacy with increasing time on
treatment; however, better medication possession ratio was found to be associated
Discussion
66
with a better goal attainment in TC and in LDL‐C [155]. Half of all patients on statin
treatment discontinue the medication by the end of the first year [156]. Our findings
might be influenced by factors related to patient compliance with treatment and
discontinuation. Patients in this study were assumed to be compliant with treatment,
as they fulfilled the criteria of refilling their prescriptions for at least one year;
however, this still might not accurately reflect real compliance with treatment.
Another possible explanation could be related to dosing. Almost 94% of all patients
in this study were treated with statins, of which 61% were treated with simvastatin.
The mean dose for those patients treated with simvastatin was 16.74 mg; 43% were
treated with 10 mg/day, 52% with 20 mg/day, 4.9% with 40 mg/day, and only 0.1%
with 80 mg/day. This dosing is in the lower range of what is recommended for
patients at high risk [73]. The doses of other statins were also in the lower ranges.
In a study assessing the impact of the new guidelines on lipid levels in a diabetes
patient population participating in the Västerbotten Intervention Programme, Fährm
et al. [78] showed that there was a marked decrease in mean plasma total cholesterol
levels among patients with diabetes after introduction of the guidelines in 1999, from
5.79±1.21 mmol/1 in 1995‐1999 to 5.07±1.00 mmol/1 in 2000‐2004 (p<0.001). They
found the trend in diabetes patients was influenced by increased use of lipid‐
lowering agents, even though only 25.3% of the diabetes patients received lipid‐
lowering treatment after the introduction of the new guidelines. We found the
discontinuation rates following the new reimbursement scheme were similar for the
total patient population and for a subpopulation of patients treated with glucose‐
lowering drugs; this indicates a non‐desirable effect, since patients with diabetes are
at higher risk for CVD than patients without diabetes.
Discussion
67
5.1.2. Glucose-lowering therapies
The total number of patients treated with glucose‐lowering drugs in 2010 was about
372,000, an increase of 17% since 2006. Of these, 183,000 were treated with insulins
and 265,000 with oral GLT, representing increases of 15% and 19% respectively since
2006 [115] .
We found the new reimbursement scheme had a minor impact on use and costs of
oral GLT. Despite restricted reimbursement for patients with T2DM, the use and
costs of insulin‐based GLT increased more rapidly following the intervention. The
use of insulin‐based treatment in patients ≥45 years of age has been reported to be
twice as high in Sweden as in its neighboring country Denmark [125], indicating
wider use of insulin‐based treatment for T2DM in Sweden compared to that in
Denmark. This difference might partly be explained by differences in reimbursement
system; in Sweden, insulins are reimbursed to 100% while the reimbursement of oral
GLT was reduced in 1997. In Denmark, the reimbursement is the same for oral and
for insulin‐based GLT.
The TLV review concluded that that the overall use of glucose‐lowering drugs was
already considered cost‐effective prior to the review, with a few exceptions, and so
the new reimbursement scheme was not designed to alter the overall use. However,
the proportion of use of insulin for T2DM might be excessive in Sweden [125], and
the costs for insulin amount to the largest part of costs for GLT. The use of insulins
could have been evaluated in the HTA review, with respect to cost‐effectiveness. If
the use was found to be excessive (not cost‐effective) or could be steered toward
cheaper oral treatment, this could have been a potential target to consider in the new
reimbursement scheme. This would have better reflected the stated aim of the
review: “The purpose of our reviews is to extract as much health as possible for each
tax crown expended on medicines” [105]5.
5 Page 4.
Discussion
68
Quality aspects
Failure to achieve the treatment goals could among other things be due to poor
adherence because of the side effects of certain antidiabetic treatments [89].
Hypoglycemia has been suggested to be the main limiting factor in achieving
adequate glycemic control [90], which might be explained by the impact on the
patient’s QoL when experiencing hypoglycemia, as shown in earlier studies in
Sweden and in France [95, 96].
About 50% of all patients with diabetes in Sweden reach the goal of HbA1c <6% [88]
according to guidelines, while in our findings only 40% of Swedish adult patients
with T2DM treated with a combination of metformin and SUs reached glycemic goal.
However, a review by Hemmingsen et al. found that the risks of both mild and
severe hypoglycemia were increased with intensive glycemic control, even if it was
found to reduce the risk of microvascular complications and suggested to reduce the
risk of non‐fatal myocardial infarction in trials exclusively dealing with glycemic
control in usual care settings [82].
Our findings confirm the results of earlier studies in which patients reporting
experience of moderate or more severe hypoglycemia were found to have lower QoL
summary scores, as measured using EQ5D and VAS. In health economic evaluations,
even seemingly minor numerical improvements in QoL measured by EQ‐5D might
translate into considerable gain in QALYs [157].
According to guidelines, QoL aspects in GLT should be taken into consideration (see
section 1.3.1.2.Glucose‐lowering therapies). Our findings indicate further that socio‐
economic factors could influence QoL; a larger proportion of patients reporting no or
only mild hypoglycemia had elementary school as highest education, compared to
the group of patients reporting experience of moderate or more severe
hypoglycemia. This contradicts the usual view that lower socioeconomic status leads
to lower QoL [158]. In this case, those patients with higher education seemed to
report more hypoglycemia and subsequently lower QoL; this could be explained by
Discussion
69
better disease control, which leads to hypoglycemia and lower QoL. The results are
confirmed by higher HbA1c in the group reporting experience of no or only mild
hypoglycemia. These findings could be interesting to consider from a health equity
standpoint, to ensure fair and just distribution of health care.
5.1.3. Use of preventive medicines
Prevention can be a cost‐effective and sometimes cost‐saving component of
managing established chronic conditions [159].
Use of preventive medicines depends on demand and supply. However, market
failures such as moral hazard and asymmetric information might drive the private
demand of use toward a non‐optimal level (lower level) from a societal perspective.
This might be even more evident in a publicly‐financed health care system, since
incentives for prevention might be partly shifted away from the insured to the
insurers.
Effective preventive drugs for coronary heart disease and stroke have been shown to
be underused globally, with striking variation between countries at different stages
of economic development [160]. Variation in use of new drugs could be explained by
macro‐level or system‐level determinants, service organization determinants, and
clinical practice determinants; the most important determinants are spending on
pharmaceuticals, the role and impact of health technology assessment, guidelines,
and clinical culture and attitudes [161]. It has been suggested that relaxation of
restriction of reimbursement policies for statins might increase their use [11, 117,
162].
The level of use of statins in Sweden compared to other similar countries was found
to be low during 2008‐2009 [113] and medium during 1997‐2003 [163]. In the PURE
study, even in high‐income countries (Sweden included), only 72.2% of patients with
coronary heart disease and 52.2% of patients who had a stroke were treated with
Discussion
70
statins [160], which might be an indication of underuse of LMT for secondary
prevention. On the other hand, overuse of statin therapy was found among 69% of
patients undergoing primary prevention in the US [164], but that might differ
substantially from the use in Sweden. Taylor et al. questioned the benefits of
treatment with statins for primary prevention; they found only limited evidence that
primary prevention with statins may be cost‐effective and improve patient
QoL [165], but acknowledged that cost‐effectiveness should be reviewed in the light
of changes in cost as suggested by Ward et al.
Underuse of preventive technologies might also arise from inadequate supply if
market forces discourage investment in R&D for prevention compared to R&D for
cure, for example if there are differences in the type of evidence required for market
authorization or for reimbursement. It has been suggested that many public and
private insurers use a double standard when evaluating new treatments; while the
adoption of a curative treatment seems to require evidence of safety and efficacy at
the least and cost‐effectiveness at the most, adoption of a preventive treatment seems
to require evidence of cost‐effectiveness at the least and often also evidence of cost
savings [166]. Furthermore, safety, efficacy, and cost‐effectiveness data on prevention
are likely to be more costly to gather compared to the process for cure, since the
benefits of prevention tend to be diffused and take a long time to develop, so clinical
trials of prevention require larger samples and longer durations than clinical trials of
cure.
No generalizable pattern has emerged when comparing cost‐effectiveness of
technologies for prevention and cure [17]. However, it is possible that general
methods in CEA, such as discounting of future health benefits, are less likely to
promote a preventive program than a curative one [38, 167].
In contrast to what is sometimes argued, Russell (1986) suggested that in many cases
prevention will not be cheaper than cure [168]. An analysis of 599 studies published
between 2000 and 2005 revealed that spending more on prevention increased
Discussion
71
medical spending over 80% of the time [169]. However, prevention is still attractive,
since it will be a better option than cure in cases where cure is incomplete, and in
many cases the uninsurable utility loss from health risks, for example pain and
suffering, far exceeds the insurable monetary loss.
We found that the impact on use and costs from the new reimbursement schemes
following the HTA reviews for LMT and GLT depended mostly on initial perception
of the efficiency in the market place, timing (when the review was undertaken and
implemented), and design (how precise it is), besides the characteristics of a specific
market (e.g. if generics are available in the market, new entries, competition). In the
HTA review of GLT, the TLV concluded that the overall use of glucose‐lowering
drugs was already cost‐effective prior to the review, and so the new reimbursement
scheme was not designed to alter the overall use. In contrast, the review of LMT was
aimed at reducing the use of lower doses of patented statins in favor of equivalent
doses of generic statins, which was considered to be more cost‐effective.
Consequently the new schemes differed in terms of design and precision. Timing of
the reviews is an important factor that influences the effectiveness of a policy,
because the potentials for savings might differ. This could be understood by the
results in Papers I and III, since the impact of the new reimbursement scheme for
LMT had a clearer effect on the market compared to the new reimbursement scheme
for GLT, where there was no potential for major gains from increased use of generics.
In a review, potential savings should be evaluated and weighted against the
potential unintended effects that may emerge following new restrictive
reimbursement schemes, as was shown in our studies. Soft demand‐side policies
might be more effective instruments under certain circumstances.
Discussion
72
5.2. Methodological considerations
5.2.1. Papers I & III
Time series design
The three most commonly used designs in guideline implementation studies are
uncontrolled before‐and‐after studies, time series designs, and controlled before‐and‐
after studies [132]. While the before‐and‐after design uses two time points, one before
and one after the intervention, the time series design makes use of multiple time
points to evaluate the effect of an intervention. Before‐and‐after design might
produce results confounded by other factors. One or both of the time points could be
atypical apart from the new program or regulation, or could be influenced by other
regulations or other factors such as introduction of new drugs, generic competition,
changes in treatment policies, marketing activities, introduction of new treatment,
and educational interventions by Drug and Therapeutic Committees. Even though
the time series design does not eliminate all problems of interpreting changes in
turnover rate, the extended time period strengthens the ability to attribute changes to
the intervention [131].
Internal validity could be threatened by historical events, that is, some other event
occurring at the same time as the intervention which could also explain the pattern of
change over time. Another source threatening internal validity is related to collection
of the data, if aspects of the record‐keeping procedures may change at the same time
as the intervention.
Interrupted segmented regression is best suited to testing immediate and sustained
changes associated with the intervention while controlling for trends, and to
capturing immediate effects. Capturing immediate effects could therefore be a
Discussion
73
problem using interrupted segmented regression if a program is implemented slowly
over time.
Drug utilization studies and the drug registry There are clear methodological issues related to drug utilization studies. In most of
them, units of drugs available (purchased) are used as a proxy for the volume
consumed. However, the quantity of prescribed drugs might differ widely from the
quantity consumed by those for whom they were intended. Compliance with
treatment differs between therapies, but in general is less than 100%, and the
compliance rate might also be lower in asymptomatic diseases, because patients do
not perceive an immediate need for their medication [170].
The indication for the prescription is not recorded in the database; it is therefore not
possible to evaluate average dosage and adherence to treatment guidelines, nor is it
possible to measure if the use of drugs is at an optimal level. It is further not possible
to capture use of drugs in patients who may have switched to over‐the‐counter
products.
5.2.2. Paper II
One possible limitation in the study on prevalence of lipid abnormalities was related
to the selection of patients. Only those 5,424 who met all predefined inclusion and
exclusion criteria were included in the analysis, comprising about 58% of the total
study population (n=5,424/9,384). The inclusion criterion of complete lipid profiles
caused an exclusion of about 21% (n= 1,933/9,384); this may have resulted in selection
bias, since patients with high CVD risk should have better documentation and thus a
greater probability of selection in the cohort. The problem of incomplete lipid profiles
has been reported by other researchers, and the analytical solution used in this study
was adapted from their prior work [155]. In short, they compared baseline lipid
values for those patients with complete lipid profiles to all included patients. If
Discussion
74
baseline lipid profiles were similar, they concluded that those patients with complete
profiles were representative of the entire cohort. Using the same methodology in this
study, no major differences in baseline lipid values were found between excluded
and included cases. Nonetheless, one should use caution when extrapolating these
data to the general population of patients using LMT. Retrospective data do not
permit controlling for a variety of confounding factors, thus limiting our ability to
make inferences about associations observed in the results. Furthermore, all included
patients should have had their LMT prescription refilled for at least one year as well
as having complete lipid profiles, which might represent a best‐case scenario of
goal/normal lipid level attainment. Another possible limitation is that it was not
possible to differentiate between fasting and non‐fasting TG measurements in this
study. To address this issue, LDL‐C measurements were considered invalid if the
triglyceride values were >4.5 mmol/L, since both fasting and non‐fasting TG act as
strong predictors of cardiovascular events [171].
5.2.3. Paper IV
Hypoglycemia was one of the main concepts in this study, but the literature shows a
lack of consensus on the definition of hypoglycemia [21]. Hypoglycemia definitions
may be heterogeneous between trials [82], which limits the possibility in general to
compare and integrate findings across studies on hypoglycemia. However, the
definition for hypoglycemia used in our study is consistent with the definitions used
in a recent Cochrane review [82], where hypoglycemia was defined as mild
(controlled by patient), moderate (daily activities interrupted but self‐managed), or
severe (requiring assistance).
Reasonable accuracy can be achieved when using a quantitative approach to study a
phenomenon using physiologic measurements such as blood pressure or body
temperature, but no comparable methods have yet been developed to measure
psychological phenomena such as hope or self esteem [131]. Quantitative
Discussion
75
methodology can thus be limited when it comes to explaining or giving in‐depth
insights into a phenomenon, which is important in research areas dealing with
patients’ experience and views related to different treatments. Patient‐reported
outcomes have become increasingly important, and many instruments and
questionnaires have been developed for use in different settings. However, using
questionnaires instead of interviews can result in missing information that was never
asked for in the questionnaire, while issues in general might more easily appear in
the more flexible form of an interview as compared to the highly structured form of a
questionnaire. This is also evident since the use of questionnaires is sometimes
criticized by patient organizations, because these instruments are developed for a
generic purpose and might not reflect the specific issues related to a disease area or
therapy as perceived by the patients. Compared with questionnaires, interviews are
superior in terms of response rates, audience, clarity, depth of questioning, providing
complete information, order of questions, sample control, and supplementary data.
The study utilized an observational, cross‐sectional, retrospective design. In a
retrospective study, the response is recorded at entry and an attempt is made to look
backwards in time for possible explanatory features [131]. Patients were asked about
their experience of hypoglycemia during the past six months; the answers and hence
the results were thus dependent on the patientsʹ ability to recall the episodes. While
the recall of severe hypoglycemic episodes may be reliable over a longer time period,
the reliable recall period for mild episodes is shorter than one week [25]. This is in
general a problem with retrospective studies, since they are subject to biases of recall;
but on the other hand they may often yield results much more quickly than
corresponding prospective studies.
The cross‐sectional design is a limitation, since observations taken at just one time
point are likely to be less enlightening than those taken over time [131].
Another potential weakness is that generic instruments like the EQ‐5D may not be
sensitive enough to detect features specific to a certain disease outcome; this was
Discussion
76
acknowledged in a recent report evaluating QoL in Swedish patients with
diabetes [97]. This may in fact be reflected by the fact that in our study the VAS score
showed a dose‐response relationship with hypoglycemia severity, which was not the
case for the EQ‐5D summary score.
Patients should have been prospectively and consecutively recruited to the study at
their usual visit to their GP. However, many of the investigators had difficulties
recruiting patients who were eligible for participation, and many potential study
patients were excluded because they had been already been switched to other
treatments. The QoL difference between the study groups might therefore be
underestimated due to a selection bias, in that patients with more severe
hypoglycemic problems had already been taken off their treatment as recommended.
This may indicate that patients included in this study tolerated the treatment better
than other patients, which would underestimate the problem of hypoglycemia.
Conclusions
77
6.CONCLUSIONS
The new pricing and reimbursement scheme for LMT had a substantial effect on the
use of LMT. Patients initially treated with low doses of statins that were eventually
excluded from the reimbursement scheme were switched to higher doses or
discontinued to a much greater extent following the new reimbursement scheme.
The new reimbursement scheme for GLT had a minor impact on the use and costs of
oral GLT, while there was an accelerated increasing trend in the number of patients
treated with restricted or retained insulins, as well as in costs for insulin‐based GLT.
The effects of new reimbursement schemes depend mostly on the timing (when the
review is undertaken and implemented) and design (how precise it is), besides the
characteristics of a specific market.
Our findings show that about 40% of the patients attained goal levels for LDL‐C
following treatment, but only 18% attained goals/normal levels in all three lipid
parameters. Improvement in TGs was moderate, and low HDL‐C persisted, showing
only modest improvement following therapy; this was most notable in patients with
T2DM, which could be explained by the limited use of LMT targeting lipid
parameters other than LDL‐C.
These findings show that treatment patterns for LMT were more in line with
treatment guidelines considering LDL‐C compared to treatment of other lipid
parameters.
Hypoglycemia was found to be associated with lower QoL in Swedish adult T2DM
patients treated with GLT as a combination of metformin and SU. This should be
considered in clinical practice.
Implications for the future
78
7. IMPLICATIONS FOR THE FUTURE
Changes in P&R policies to steer use of pharmaceuticals should be carefully
evaluated with respect to timing and design, in order to avoid unintended and
unwanted effects; this should be weighted against potential savings. Softer demand‐
side policies might be a better option to steer more precisely towards a cost‐effective
use of medicines.
Independent follow‐up is needed to assess the impact of new reimbursement
decisions. At the same time, the TLV should have a follow‐up plan to make their own
assessments of the impact from their decisions, since our findings show that
unintended effects might emerge.
Focusing dyslipidemia therapy on LDL‐C reduction allows 40% of all patients to
successfully achieve LDL‐C goal and also helps reduce triglyceride levels, whereas
HDL‐C and/or triglyceride abnormalities mainly persist. About 60% of all patients
starting statin therapy could be considered for addition of treatments that target
multiple lipid disorders. This option is most urgent for patients with T2DM.
Experience of hypoglycemia was found to be associated with lower QoL in patients
with T2DM on dual treatment with metformin and SU. This should be taken into
consideration when selecting treatment for these patients in clinical practice, perhaps
through introduction of a specific questionnaire at the regular visits with health care
providers.
Future research
79
8. FUTURE RESEARCH
The effects of a policy on costs of health care resources other than drugs are often not
accounted for; this constitutes a limitation of our studies as well as other similar
studies [117, 118]. Employing a wider societal perspective might result in different
conclusions when evaluating the cost‐effectiveness of a treatment, since it would take
into account all costs, whereas if only costs for drugs are included, the results
generated might be inconclusive. A full cost analysis should account for the total
transition and other costs related to the new reimbursement scheme, and should be
evaluated against the accumulated savings generated over time, which may be
substantial. Research on long‐term effectiveness and cost‐effectiveness of LMT is
therefore of major interest.
The economist’s view that the optimal level of prevention is where the marginal
benefits equal the marginal costs remains somewhat foreign and controversial to
health professionals who encourage greater use of prevention [17], even if the
opinion varies between countries. Primary prevention is heavily debated from both
economic and ethical perspectives, since it may drive costs and be of only minor
benefit for individuals that are actually healthy. However, recent research shows a
substantial underuse of effective secondary prevention drugs despite the low costs of
these drugs, which are generic in most parts of the world [160]. Further research on
the causes of this underuse could give further information useful in determining the
optimal level of prevention.
The outcomes of cost‐effectiveness analysis where initial prices are used would differ
widely compared to analysis using a price averaged over a product’s total life cycle.
In the case of statins, where prices have dropped considerably since the Swedish
patent for simvastatin expired in 2003, results on cost‐effectiveness using a dynamic
analytical approach showed that simvastatin generates considerable societal
Furure research
80
surplus [111] if total life cycle value is captured, which is different to the static
approach generally employed in economic evaluations [43]. Further research should
consider using a dynamic approach to better inform on the cost‐effectiveness of a
therapy over time.
In health economic evaluations, even seemingly minor numerical improvements in
QoL measured by EQ‐5D might translate into considerable gain in QALYs [157]. Due
to methodological challenges related to QoL studies in general, further research on
methodological issues could help to better reflect the value and cost‐effectiveness of
medicines with benefits predominately related to QoL improvements.
Acknowledgements
81
9. ACKNOWLEDGEMENTS
I have many people to thank for achieving this goal. First, I am grateful for the opportunity given to me by my main supervisor, Lars‐Åke Levin, and for all the support provided by my co‐supervisors, Per Wändell and Mikael Hoffmann. Thanks to my colleagues and Per Carlsson at CMT, for feedback and comments on earlier versions of the dissertation. Thanks to my children, Emil and Joséphine, and to my partner, Torsten Ericsson, for your patience and encouragement. Iʹll be able to cook more now, Emil! Thanks to my employer, MSD, for providing me with the financial and other support essential to this achievement. Thanks also to the study groups and co‐authors on Primula and Exhype. I enjoyed working with all of you! Thanks to my mentor, Gunilla Journath, for invaluable support in many dimensions. I really miss you at work! A special thanks to statisticians Daniel Åberg and David Andersson, for your help and assistance. To my sisters and brothers and rest of the family and to my dear friend Camilla Eriksson, for all your support throughout the years, whatever challenges I was facing. Thanks also to F&S Täby; exercising was sometimes essential to clear the mind. Last but not least, thanks to all my friends, for your patience with my considerable absence. I hope to see more of you now!
References
82
10. REFERENCES
1. Allender, S., Scarborough, P., Peto, V., Rayner, M., Leal, J., Luengo‐Fernandez, R., Gray, A., European cardiovascular disease statistics 2008. 2008, Department of public health, University of Oxford: Oxford.
2. Bjorck, L., A. Rosengren, K. Bennett, G. Lappas, and S. Capewell, Modelling the decreasing coronary heart disease mortality in Sweden between 1986 and 2002. Eur Heart J, 2009. 30(9): p. 1046‐56.
3. Ford, E.S., U.A. Ajani, J.B. Croft, J.A. Critchley, D.R. Labarthe, et al., Explaining the decrease in U.S. deaths from coronary disease, 1980‐2000. N Engl J Med, 2007. 356(23): p. 2388‐98.
4. Newhouse, J.P., Medical care costs: how much welfare loss? J Econ Perspect, 1992. 6(3): p. 3‐21.
5. Docteur, E., Oxley, H., Health‐Care Systems: Lessons from the Reform Experience. OECD Health working papers. 2003, OECD: Paris.
6. OECD, Pharmaceutical pricing policies in a global market, in OECD Health Policy Studies, OECD, Editor. 2008, OECD: Paris.
7. Wettermark, B., B. Godman, K. Andersson, L.L. Gustafsson, A. Haycox, et al., Recent national and regional drug reforms in Sweden: implications for pharmaceutical companies in Europe. Pharmacoeconomics, 2008. 26(7): p. 537‐50.
8. Drummond, M., B. Jonsson, and F. Rutten, The role of economic evaluation in the pricing and reimbursement of medicines. Health Policy, 1997. 40(3): p. 199‐215.
9. Sorenson, C., Drummond M., Kanavos, P., Ensuring value for money in health care‐ The role of health technology assessment in the European Union. 2008, Cornwall: World Health Organization on behalf of the European Observatory on health systems and Policies.
10. Green, C.J., M. Maclure, P.M. Fortin, C.R. Ramsay, M. Aaserud, et al., Pharmaceutical policies: effects of restrictions on reimbursement. Cochrane Database Syst Rev, 2010(8): p. CD008654.
11. Bjerrum, L., J. Larsen, and J. Kragstrup, Guidelines accompanied by changes in reimbursement rules. Effects on lipid‐lowering drug prescribing. Scand J Prim Health Care, 2001. 19(3): p. 158‐62.
References
83
12. Jansson, S.P., D.K. Andersson, and K. Svardsudd, Prevalence and incidence rate of diabetes mellitus in a Swedish community during 30 years of follow‐up. Diabetologia, 2007. 50(4): p. 703‐10.
13. Wirehn, A.B., H.M. Karlsson, and J.M. Carstensen, Estimating disease prevalence using a population‐based administrative healthcare database. Scand J Public Health, 2007. 35(4): p. 424‐31.
14. Causes of death in Sweden. 2010, Swedish board of health and welfare: [cited; Available from: http://www.socialstyrelsen.se/publikationer2011/2011‐7‐6.
15. Gaining health. The European Strategy for the Prevention and Control of Noncommunicable Diseases. 2006, World Health Organization: [cited 2011 20 September]; Available from: http://www.euro.who.int/__data/assets/pdf_file/0008/76526/E89306.pdf.
16. The National public health strategy for Sweden in brief. 2003, Swedish National Institute of Public health [cited 2011 July]; Available from: http://www.fhi.se/PageFiles/4411/public%20health%20strat(1).pdf.
17. Kenkel, D.S., Prevention, in Handbook of Health Economics, A.J. Culyer, Newhouse, J.P., Editor. 2000, Elsevier Science B.V: Amsterdam.
18. Gravelle, H., Rees, R., Microeconomics. 2nd ed. 1992, London and New York: Longman group Uk Limited.
19. Pauly, M.V., Taxation, health insurance, and market failure in the medical economy. J Econ Lit, 1986. 24(2): p. 629‐75.
20. Neil, N., T. Malmfors, and P. Slovic, Intuitive toxicology: expert and lay judgments of chemical risks. Toxicol Pathol, 1994. 22(2): p. 198‐201.
21. Tversky, A. and D. Kahneman, Judgment under Uncertainty: Heuristics and Biases. Science, 1974. 185(4157): p. 1124‐31.
22. Johannesson, M., Theory and methods of economic evaluation of health care. Developments in health economics and public policy, ed. P. Sweifel, Frech, H.E. Vol. 4. 1996, Dordrecht: Kluwer Academic publisher.
23. Weisbrod, B.A., The Health‐Care Quadrilemma ‐ an Essay on Technological‐Change, Insurance, Quality of Care, and Cost Containment. Journal of Economic Literature, 1991. 29(2): p. 523‐552.
24. Ward, S., M. Lloyd Jones, A. Pandor, M. Holmes, R. Ara, et al., A systematic review and economic evaluation of statins for the prevention of coronary events. Health Technol Assess, 2007. 11(14): p. 1‐160, iii‐iv.
25. Prevention of cardiovascular disease. 2007, WHO: [cited 2011 6 December]; Available from: http://www.who.int/cardiovascular_diseases/guidelines/PocketGL.ENGLISH.AFR‐D‐E.rev1.pdf.
References
84
26. Black, H., R., Cardiovascular risk factors, in Yale university school of medicine heart book, M.D. Barry L. Zaret, Marvin Moser, M.D., Lawrence S. Cohen, M.D., Editor. 1992, Hearst Books: New York.
27. Khot, U.N., M.B. Khot, C.T. Bajzer, S.K. Sapp, E.M. Ohman, et al., Prevalence of conventional risk factors in patients with coronary heart disease. JAMA, 2003. 290(7): p. 898‐904.
28. Pearson, T.A., S.N. Blair, S.R. Daniels, R.H. Eckel, J.M. Fair, et al., AHA Guidelines for Primary Prevention of Cardiovascular Disease and Stroke: 2002 Update: Consensus Panel Guide to Comprehensive Risk Reduction for Adult Patients Without Coronary or Other Atherosclerotic Vascular Diseases. American Heart Association Science Advisory and Coordinating Committee. Circulation, 2002. 106(3): p. 388‐91.
29. De Backer, G., D. De Bacquer, C. Brohet, S. De Ceukelier, A. Franck, et al., [Recommendations for the prevention of cardiovascular diseases in clinical practice]. J Pharm Belg, 2005. 60(2): p. 41‐50.
30. Definitions. 2011, International Network of Agencies for Health Technology Assessment (INAHTA): [cited 2011 28 December]; Available from: http://www.inahta.org/HTA/.
31. Battista, R., Towards a paradigm for technology assessment, in The scientific basis of health services´, S.R. Peckham M, Editor. 1996, BMJ Publishing Group.: London.
32. Banta, D. and E. Jonsson, History of HTA: Introduction. Int J Technol Assess Health Care, 2009. 25 Suppl 1: p. 1‐6.
33. Banta, D., E. Jonsson, and P. Childs, History of the international societies in health technology assessment: International Society for Technology Assessment in Health Care and Health Technology Assessment International. Int J Technol Assess Health Care, 2009. 25 Suppl 1: p. 19‐23.
34. Banta, D., F.B. Kristensen, and E. Jonsson, A history of health technology assessment at the European level. Int J Technol Assess Health Care, 2009. 25 Suppl 1: p. 68‐73.
35. Jonsson, E., H.D. Banta, C. Henshall, and L. Sampietro‐Colom, Executive summary of the ECHTA/ECAHI project. European Collaboration for Health Technology Assessment/Assessment of Health Interventions. Int J Technol Assess Health Care, 2002. 18(2): p. 213‐7.
36. Velasco Garrido M., B.K.F., Palmhøj Nielsen C., Busse R., Health Technology Assessment and Health Policy‐Making in Europe‐ Current status, challenges and potential, in Observatory Studies Series No 14. 2008, World Health organization (WHO): Copenhagen.
References
85
37. Zentner, A., M. Velasco‐Garrido, and R. Busse, Methods for the comparative evaluation of pharmaceuticals. GMS Health Technol Assess, 2005. 1: p. Doc09.
38. Drummond, Methods for the Economic Evaluation of Health Care Programmes. Third ed. 2005, Norfolk: Oxford Medical Publications.
39. Berger ML, B.K., Hedblom EC, Pashos CL, Torrance GW. , Health Care Cost, Quality, and Outcomes: ISPOR Book of Terms. 2003, Lawrenceville, NJ: International Society of Pharmacoeconomics and Outcomes Research (ISPOR).
40. Acquadro, C., R. Berzon, D. Dubois, N.K. Leidy, P. Marquis, et al., Incorporating the patientʹs perspective into drug development and communication: an ad hoc task force report of the Patient‐Reported Outcomes (PRO) Harmonization Group meeting at the Food and Drug Administration, February 16, 2001. Value Health, 2003. 6(5): p. 522‐31.
41. Drummond, Methods for the Economic Evaluation of Health Care Programmes. Second ed. 2004, Norfolk: Oxford Medical Publications.
42. Pharmacoeconomic guidelines around the world. 2004, International Society for Pharmacoeconomics and Outcomes Research ( ISPOR): [cited 2011 29 August]; Available from: http://www.ispor.org/PEguidelines/index.asp.
43. General guidelines for economic evaluations from the Pharmaceutical Benefits Board (LFNAR 2003:2). 2003, International Society for Pharmacoeconomics and Outcomes Research: [cited 2011 August]; Available from: http://www.ispor.org/PEguidelines/source/Guidelines_in_Sweden.pdf.
44. Brouwer, W.B., A.J. Culyer, N.J. van Exel, and F.F. Rutten, Welfarism vs. extra‐welfarism. J Health Econ, 2008. 27(2): p. 325‐38.
45. Brouwer, W.B. and M.A. Koopmanschap, On the economic foundations of CEA. Ladies and gentlemen, take your positions! J Health Econ, 2000. 19(4): p. 439‐59.
46. Johannesson, M. and G. Karlsson, The friction cost method: a comment. J Health Econ, 1997. 16(2): p. 249‐55; discussion 257‐9.
47. Hill, S., Henry D, ed. National Medicinal Drug Policies: Their Relationship to Pharmacoepidemiology. Fourth ed. Pharmacoepidemiology, ed. S. BL. 2005, Wiley&sons: Chichester.
48. Anell, A. and U. Persson, Reimbursement and clinical guidance for pharmaceuticals in Sweden: do health‐economic evaluations support decision making? Eur J Health Econ, 2005. 6(3): p. 274‐9.
References
86
49. Andersson, Swedish Pharmaceutical Benefit Reforms ‐Analyses of implementation, pharmaceutical sales patterns and expenditures, in Department of Public health and Community Medicine. 2006, University of Gothenburg: Gothenburg.
50. Anell, A., Swedish healthcare under pressure. Health Econ, 2005. 14(Suppl 1): p. S237‐54.
51. Gerdtham, U.G. and D. Lundin, Why did drug spending increase during the 1990s? A decomposition based on Swedish data. Pharmacoeconomics, 2004. 22(1): p. 29‐42.
52. Andersson, K., M.G. Petzold, C. Sonesson, K. Lonnroth, and A. Carlsten, Do policy changes in the pharmaceutical reimbursement schedule affect drug expenditures? Interrupted time series analysis of cost, volume and cost per volume trends in Sweden 1986‐2002. Health Policy, 2006. 79(2‐3): p. 231‐43.
53. Godman, B., B. Wettermark, M. Hoffmann, K. Andersson, A. Haycox, et al., Multifaceted national and regional drug reforms and initiatives in ambulatory care in Sweden: global relevance. Expert Rev Pharmacoecon Outcomes Res, 2009. 9(1): p. 65‐83.
54. Wettermark, B., B. Godman, B. Jacobsson, and F.M. Haaijer‐Ruskamp, Soft regulations in pharmaceutical policy making: an overview of current approaches and their consequences. Appl Health Econ Health Policy, 2009. 7(3): p. 137‐47.
55. Godman, B., S. Sakshaug, C. Berg, B. Wettermark, and A. Haycox, Combination of prescribing restrictions and policies to engineer low prices to reduce reimbursement costs. Expert Rev Pharmacoecon Outcomes Res, 2011. 11(1): p. 121‐9.
56. Cheah, T.S., The impact of clinical guidelines and clinical pathways on medical practice: effectiveness and medico‐legal aspects. Ann Acad Med Singapore, 1998. 27(4): p. 533‐9.
57. Worrall, G., P. Chaulk, and D. Freake, The effects of clinical practice guidelines on patient outcomes in primary care: a systematic review. Cmaj, 1997. 156(12): p. 1705‐12.
58. Nationella riktlinjer för hjärtsjukvård. Tabellbilaga till beslutsstödsdokument – tillstånds‐ och åtgärdslistor för hjärtsjukdomar. 2008, National Board of Health and Welfare: [cited; June 2009. Available from: http://www.socialstyrelsen.se/Lists/Artikelkatalog/Attachments/8592/2008‐102‐7_20081027_bilaga_1.pdf.
59. Greenland, P., M.D. Knoll, J. Stamler, J.D. Neaton, A.R. Dyer, et al., Major risk factors as antecedents of fatal and nonfatal coronary heart disease events. JAMA, 2003. 290(7): p. 891‐7.
References
87
60. Randomised trial of cholesterol lowering in 4444 patients with coronary heart disease: the Scandinavian Simvastatin Survival Study (4S). Lancet, 1994. 344(8934): p. 1383‐9.
61. Baigent, C., A. Keech, P.M. Kearney, L. Blackwell, G. Buck, et al., Efficacy and safety of cholesterol‐lowering treatment: prospective meta‐analysis of data from 90,056 participants in 14 randomised trials of statins. Lancet, 2005. 366(9493): p. 1267‐78.
62. Castelli, W.P., Cholesterol and lipids in the risk of coronary artery disease‐‐the Framingham Heart Study. Can J Cardiol, 1988. 4 Suppl A: p. 5A‐10A.
63. Lewington, S., G. Whitlock, R. Clarke, P. Sherliker, J. Emberson, et al., Blood cholesterol and vascular mortality by age, sex, and blood pressure: a meta‐analysis of individual data from 61 prospective studies with 55,000 vascular deaths. Lancet, 2007. 370(9602): p. 1829‐39.
64. Hokanson, J.E. and M.A. Austin, Plasma triglyceride level is a risk factor for cardiovascular disease independent of high‐density lipoprotein cholesterol level: a meta‐analysis of population‐based prospective studies. J Cardiovasc Risk, 1996. 3(2): p. 213‐9.
65. Patel, A., F. Barzi, K. Jamrozik, T.H. Lam, H. Ueshima, et al., Serum triglycerides as a risk factor for cardiovascular diseases in the Asia‐Pacific region. Circulation, 2004. 110(17): p. 2678‐86.
66. Chirovsky, D.R., V. Fedirko, Y. Cui, V. Sazonov, and P. Barter, Prospective studies on the relationship between high‐density lipoprotein cholesterol and cardiovascular risk: a systematic review. Eur J Cardiovasc Prev Rehabil, 2009. 16(4): p. 404‐23.
67. Isomaa, B., P. Almgren, T. Tuomi, B. Forsen, K. Lahti, et al., Cardiovascular morbidity and mortality associated with the metabolic syndrome. Diabetes Care, 2001. 24(4): p. 683‐9.
68. Rizzo, M. and K. Berneis, Small, dense low‐density‐lipoproteins and the metabolic syndrome. Diabetes Metab Res Rev, 2007. 23(1): p. 14‐20.
69. Tan, K.C., Management of dyslipidemia in the metabolic syndrome. Cardiovasc Hematol Disord Drug Targets, 2007. 7(2): p. 99‐108.
70. Gordon, D.J., J.L. Probstfield, R.J. Garrison, J.D. Neaton, W.P. Castelli, et al., High‐density lipoprotein cholesterol and cardiovascular disease. Four prospective American studies. Circulation, 1989. 79(1): p. 8‐15.
71. Anderson, K.M., P.M. Odell, P.W. Wilson, and W.B. Kannel, Cardiovascular disease risk profiles. Am Heart J, 1991. 121(1 Pt 2): p. 293‐8.
References
88
72. Assmann, G., P. Cullen, and H. Schulte, Simple scoring scheme for calculating the risk of acute coronary events based on the 10‐year follow‐up of the prospective cardiovascular Munster (PROCAM) study. Circulation, 2002. 105(3): p. 310‐5.
73. Prevention av aterosklerotisk hjärtkärlsjukdom med lipidreglerande läkemedel. 2005, Medical Products Agency: [cited 2011; Available from: http://www.lakemedelsverket.se/malgrupp/Halso‐‐‐sjukvard/Behandlings‐‐rekommendationer/Behandlingsrekommendation‐‐‐listan/Hyperlipidemi/.
74. Jansson, J.H., K. Boman, and T. Messner, Trends in blood pressure, lipids, lipoproteins and glucose metabolism in the Northern Sweden MONICA project 1986‐99. Scand J Public Health Suppl, 2003. 61: p. 43‐50.
75. Eliasson, M., U. Janlert, J.H. Jansson, and B. Stegmayr, Time trends in population cholesterol levels 1986‐2004: influence of lipid‐lowering drugs, obesity, smoking and educational level. The northern Sweden MONICA study. J Intern Med, 2006. 260(6): p. 551‐9.
76. Public Health Report 2009. 2009, The National Board of Health and Welfare Stockholm.
77. Norberg, M., S. Wall, K. Boman, and L. Weinehall, The Vasterbotten Intervention Programme: background, design and implications. Glob Health Action, 2010. 3.
78. Fharm, E., O. Rolandsson, and L. Weinehall, Guidelines improve general trend of lowered cholesterol levels in type 2 diabetes patients in spite of low adherence. Scand J Public Health, 2008. 36(1): p. 69‐75.
79. Public Health Report 2009. 2009, The National Board of Health and Welfare [cited; June 2011 Available from: http://www.socialstyrelsen.se/publikationer2009/2009‐126‐71/Documents/7_Overvikt.pdf.
80. Conroy, R.M., K. Pyorala, A.P. Fitzgerald, S. Sans, A. Menotti, et al., Estimation of ten‐year risk of fatal cardiovascular disease in Europe: the SCORE project. Eur Heart J, 2003. 24(11): p. 987‐1003.
81. Intensive blood‐glucose control with sulphonylureas or insulin compared with conventional treatment and risk of complications in patients with type 2 diabetes (UKPDS 33). UK Prospective Diabetes Study (UKPDS) Group. Lancet, 1998. 352(9131): p. 837‐53.
82. Hemmingsen, B., S.S. Lund, C. Gluud, A. Vaag, T. Almdal, et al., Targeting intensive glycaemic control versus targeting conventional glycaemic control for type 2 diabetes mellitus. Cochrane Database Syst Rev, (6): p. CD008143.
References
89
83. Nathan, D.M., J.B. Buse, M.B. Davidson, E. Ferrannini, R.R. Holman, et al., Medical management of hyperglycemia in type 2 diabetes: a consensus algorithm for the initiation and adjustment of therapy: a consensus statement of the American Diabetes Association and the European Association for the Study of Diabetes. Diabetes Care, 2009. 32(1): p. 193‐203.
84. Nationella riktlinjer för diabetesvården 2010 – Stöd för styrning och ledning ( English translation: National guidelines for diabetes care 2010). 2010, National board for health and welfare: [cited 2011 20 July]; Available from: http://www.socialstyrelsen.se/publikationer2010/2010‐2‐2/Sidor/default.aspx.
85. Läkemedelsbehandling vid typ 2‐diabetes– ny rekommendation;( English translation: Pharmaceutical treatment for type 2 diabetes‐ new recommendation). 2010, Medical Products Agency: [cited 29 October 2011]; Available from: http://www.lakemedelsverket.se/upload/halso‐och‐sjukvard/behandlingsrekommendationer/L%c3%a4kemedelsbehandling%20vid%20typ%202‐diabetes_rek.pdf.
86. Landin‐Olsson, M., J.O. Jeppsson, and G. Nordin, [HbA1c‐‐new standardization introduced in Sweden. The new unit is mmol/mol]. Lakartidningen. 107(51‐52): p. 3282‐5.
87. Årsrapport‐ 2009 års resultat (English translation: Annual report‐ results for 2009) 2009, The National Diabetes Registry in Sweden (NDR) [cited 2011 28 December]; Available from: https://www.ndr.nu/pdf/Arsrapport_NDR_2010.pdf.
88. Öppna jämförelser av hälso‐ och sjukvårdens kvalitet och effektivitet. Jämförelser mellan landsting 2009. 2009, The Swedish Association of Local Authorities and Regions (Sverges kommuner och landsting) and The National Board of Health and Welfare (Socialstyrelsen) [cited 2011 27 December]; Available from: http://www.socialstyrelsen.se/Lists/Artikelkatalog/Attachments/17787/2009‐11‐2.pdf.
89. Zarowitz, B.J. and C. Conner, The intersection of safety and adherence: new incretin‐based therapies in patients with type 2 diabetes mellitus. Pharmacotherapy, 2009. 29(12 Pt 2): p. 55S‐67S.
90. Cryer, P.E., Hypoglycemia is the limiting factor in the management of diabetes. Diabetes Metab Res Rev, 1999. 15(1): p. 42‐6.
91. Jonsson, L., B. Bolinder, and J. Lundkvist, Cost of hypoglycemia in patients with Type 2 diabetes in Sweden. Value Health, 2006. 9(3): p. 193‐8.
92. Belsey, J. and G. Krishnarajah, Glycaemic control and adverse events in patients with type 2 diabetes treated with metformin + sulphonylurea: a meta‐analysis. Diabetes Obes Metab, 2008. 10 Suppl 1: p. 1‐7.
References
90
93. Burge, M.R., V. Sood, T.A. Sobhy, A.G. Rassam, and D.S. Schade, Sulphonylurea‐induced hypoglycaemia in type 2 diabetes mellitus: a review. Diabetes Obes Metab, 1999. 1(4): p. 199‐206.
94. Berne, C., Fritz, T. Diabetes Mellitus, in Läkemedelsboken 2011/2012, H. Ramström, Editor. 2011, Läkemedelsverket (Swedish Medical Products Agency): Uppsala.
95. Vexiau, P., P. Mavros, G. Krishnarajah, R. Lyu, and D. Yin, Hypoglycaemia in patients with type 2 diabetes treated with a combination of metformin and sulphonylurea therapy in France. Diabetes Obes Metab, 2008. 10 Suppl 1: p. 16‐24.
96. Lundkvist, J., C. Berne, B. Bolinder, and L. Jonsson, The economic and quality of life impact of hypoglycemia. Eur J Health Econ, 2005. 6(3): p. 197‐202.
97. Nationell utvärdering 2011. DIABETESVÅRD Rekommendationer, bedömningar och sammanfattning. ( In English : National evaluation 2011. Diabetes care, recommendations, assessments and summary. 2012, Socialstyrelsen: Stockholm.
98. Genomgången av läkemedel vid diabetes ( English translation: The review of diabetes treatments). 2010, Tandvårds‐ och läkemedelsförmånsverket (TLV): [cited 2011 27 December]; Available from: http://www.tlv.se/Upload/Genomgangen/rapport‐diabetes.pdf.
99. Sjögren, E., Reasonable Drugs‐ Making Decisions with Ambiguous Knowledge, in The Economic Research Institute. 2006, Stockholm School of Economics: Stockholm.
100. Polain, M., Franken, M., Koopmanshap, M., Cleemput, I., Drug reimbursement systems: international comparison and policy recommendations. , in KCE reports 147C. 2010, Belgian Health Care Knowlege Centre: Brussels.
101. The Pharmaceutical reviews. 2011, The Dental and Pharmaceutical Benefits Agency (TLV): [cited June 17 2011]; Available from: http://www.tlv.se/in‐english‐old/in‐english/.
102. Wettermark, B., B. Godman, M. Neovius, N. Hedberg, T.O. Mellgren, et al., Initial effects of a reimbursement restriction to improve the cost‐effectiveness of antihypertensive treatment. Health Policy, 2010. 94(3): p. 221‐9.
103. The review of medicines for treating lipid disorders. 2009, The Dental and Pharmaceutical Benefits Agency (TLV): [cited 2011 17 June]; Available from: http://www.tlv.se/Upload/Genomgangen/review‐lipid‐disorder.pdf.
References
91
104. Hoffmann, M. Blodfettsänkande läkemedel ‐effekter av Tandvårds‐ och läkemedelsförmånsverkets genomgång februari 2009. 2010, The NEPI Foundation ‐ The Swedish Network for Pharmaco‐epidemiology: [cited 2011 17 Juni]; Available from: http://www.nepi.net/res/dokument/NEPI‐rapport‐blodfettsaenkande‐behandling‐och‐TLV‐20100601.pdf
105. The review of medicines used for treating diabetes‐ A summary. 2010, The dental and Pharmaceutical Benefits Agency (TLV): [cited 2011 11 November]; Available from: http://www.tlv.se/Upload/Genomgangen/summary‐diabetes.pdf.
106. Hoffmann, M. Diabetesmedel ‐ effekter av Tandvårds‐ och läkemedelsförmånsverkets genomgång december 2009 2010, The NEPI Foundation – The Swedish Network for Pharmaco‐epidemiology: [cited 2011 5 June]; Available from: http://www.nepi.net/res/dokument/NEPI‐rapport‐diabetesmedel‐100831.pdf.
107. Third Report of the National Cholesterol Education Program (NCEP) Expert Panel on Detection, Evaluation, and Treatment of High Blood Cholesterol in Adults (Adult Treatment Panel III) final report. Circulation, 2002. 106(25): p. 3143‐421.
108. Statins for the prevention of cardiovascular events in patients at increased risk of developing cardiovascular disease or those with established cardiovascular disease. 2006, National Instiute for Health and Clinical Excellence: [cited 2011 25 July]; Available from: http://www.nice.org.uk/TA094.
109. Mills, E.J., P. Wu, G. Chong, I. Ghement, S. Singh, et al., Efficacy and safety of statin treatment for cardiovascular disease: a network meta‐analysis of 170,255 patients from 76 randomized trials. Qjm, 2011. 104(2): p. 109‐24.
110. Jonsson, B., M. Johannesson, J. Kjekshus, A.G. Olsson, T.R. Pedersen, et al., Cost‐effectiveness of cholesterol lowering. Results from the Scandinavian Simvastatin Survival Study (4S). Eur Heart J, 1996. 17(7): p. 1001‐7.
111. Lindgren, P. and B. Jonsson, From 4S to IDEAL: the health economics of the statin trials. Eur J Cardiovasc Prev Rehabil, 2009. 16(2): p. 138‐43.
112. Lindgren, P., M. Buxton, T. Kahan, N.R. Poulter, B. Dahlof, et al., Cost‐effectiveness of atorvastatin for the prevention of coronary and stroke events: an economic analysis of the Anglo‐Scandinavian Cardiac Outcomes Trial‐‐lipid‐lowering arm (ASCOT‐LLA). Eur J Cardiovasc Prev Rehabil, 2005. 12(1): p. 29‐36.
113. Richards, M., Extent and casues of international variations in drug usage. 2010, CBE: London.
114. Walley, T., P. Folino‐Gallo, U. Schwabe, and E. van Ganse, Variations and increase in use of statins across Europe: data from administrative databases. Bmj, 2004. 328(7436): p. 385‐6.
References
92
115. The Swedish Drug Registry. 2011, The National Board of Health and Welfare: [cited 2011 17 June]; Available from: http://www.socialstyrelsen.se/register/halsodataregister/lakemedelsregistret
116. Godman, B., W. Shrank, B. Wettermark, M. Andersen, I. Bishop, et al., Use of Generics‐A Critical Cost Containment Measure for All Healthcare Professionals in Europe? Pharmaceuticals, 2010. 3(8): p. 2470‐2494.
117. Sakshaug, S., K. Furu, O. Karlstad, M. Ronning, and S. Skurtveit, Switching statins in Norway after new reimbursement policy: a nationwide prescription study. Br J Clin Pharmacol, 2007. 64(4): p. 476‐81.
118. Martikainen, J.E., L.K. Saastamoinen, M.J. Korhonen, H. Enlund, and A. Helin‐Salmivaara, Impact of restricted reimbursement on the use of statins in Finland: a register‐based study. Med Care, 2010. 48(9): p. 761‐6.
119. Kalo, Z., N. Muszbek, J. Bodrogi, and J. Bidlo, Does therapeutic reference pricing always result in cost‐containment? The Hungarian evidence. Health Policy, 2007. 80(3): p. 402‐12.
120. Winkelmayer, W.C., M. Asslaber, A. Bucsics, T. Burkhardt, A. Schautzer, et al., Impact of reimbursement changes on statin use among patients with diabetes in Austria. Wien Klin Wochenschr, 2010. 122(3‐4): p. 89‐94.
121. Stargardt, T., The impact of reference pricing on switching behaviour and healthcare utilisation: the case of statins in Germany. European Journal of Health Economics, 2010. 11(3): p. 267‐277.
122. Ostini, R., D. Hegney, C. Jackson, M. Williamson, J.M. Mackson, et al., Systematic review of interventions to improve prescribing. Ann Pharmacother, 2009. 43(3): p. 502‐13.
123. Godman, B., W. Shrank, M. Andersen, C. Berg, I. Bishop, et al., Comparing policies to enhance prescribing efficiency in Europe through increasing generic utilization: changes seen and global implications. Expert Rev Pharmacoecon Outcomes Res, 2010. 10(6): p. 707‐22.
124. Giuliani, G., G. Selke, and L. Garattini, The German experience in reference pricing. Health Policy, 1998. 44(1): p. 73‐85.
125. Melander, A., P. Folino‐Gallo, T. Walley, U. Schwabe, P.H. Groop, et al., Utilisation of antihyperglycaemic drugs in ten European countries: different developments and different levels. Diabetologia, 2006. 49(9): p. 2024‐9.
126. Ringborg, A., P. Lindgren, M. Martinell, D.D. Yin, S. Schon, et al., Prevalence and incidence of Type 2 diabetes and its complications 1996‐2003‐‐estimates from a Swedish population‐based study. Diabet Med, 2008. 25(10): p. 1178‐86.
127. Jonsson, B., Revealing the cost of Type II diabetes in Europe. Diabetologia, 2002. 45(7): p. S5‐12.
References
93
128. Fleming, D.M., F.G. Schellevis, and V. Van Casteren, The prevalence of known diabetes in eight European countries. Eur J Public Health, 2004. 14(1): p. 10‐4.
129. Landerholm, L., Stenberg, A‐M., Läkemedelsförmånerna, in Läkemedelsboken. 2011‐2012, Läkemedelsverket (MPA): Uppsala. p. 1416‐1422.
130. DiPietro, N.A., Methods in epidemiology: observational study designs. Pharmacotherapy. 30(10): p. 973‐84.
131. Polit, D., Tatano Beck, C., Nursing Research: Generating and Assessing Evidence for Nursing Practice. 8th ed. 2008, Philadelphia: Lippincott Williams & Wilkins.
132. Grimshaw, J., M. Campbell, M. Eccles, and N. Steen, Experimental and quasi‐experimental designs for evaluating guideline implementation strategies. Fam Pract, 2000. 17 Suppl 1: p. S11‐6.
133. Wagner, A.K., S.B. Soumerai, F. Zhang, and D. Ross‐Degnan, Segmented regression analysis of interrupted time series studies in medication use research. J Clin Pharm Ther, 2002. 27(4): p. 299‐309.
134. Decision 1699/2007. 2009, Tandvårds ‐och läkemedelsförmånsverket (TLV): [cited 2011 20 September]; Available from: http://www.tlv.se/Upload/Beslut_2009/bes090211‐zocord.pdf.
135. Furu, K., B. Wettermark, M. Andersen, J.E. Martikainen, A.B. Almarsdottir, et al., The Nordic countries as a cohort for pharmacoepidemiological research. Basic Clin Pharmacol Toxicol, 2009. 106(2): p. 86‐94.
136. Guidelines for ATC classification 2009, WHO Collaborating Centre for Drug Statistics Methodology: [cited 2012 20 January]; Available from: http://www.whocc.no/atc_ddd_publications/guidelines/.
137. DʹAgostino, R.B., Sr., S. Grundy, L.M. Sullivan, and P. Wilson, Validation of the Framingham coronary heart disease prediction scores: results of a multiple ethnic groups investigation. JAMA, 2001. 286(2): p. 180‐7.
138. Guidelines for ATC classification and DDD assignment 2010. 2009, WHO Collaborating Centre for Drug Statistics Methodology: [cited 2011 17 June]; Available from: http://www.whocc.no/atc_ddd_publications/guidelines/.
139. Reimbursement decision, Levemir. 2007, The Dental and Pharmaceutical benefits Agency (TLV): [cited 2011 20 September]; Available from: http://www.tlv.se/Upload/Beslut_2007/BES_070618_levemir.pdf.
140. EuroQol‐‐a new facility for the measurement of health‐related quality of life. The EuroQol Group. Health Policy, 1990. 16(3): p. 199‐208.
141. Dolan, P., C. Gudex, P. Kind, and A. Williams, The time trade‐off method: results from a general population study. Health Econ, 1996. 5(2): p. 141‐54.
References
94
142. Cox, D.J., A. Irvine, L. Gonder‐Frederick, G. Nowacek, and J. Butterfield, Fear of hypoglycemia: quantification, validation, and utilization. Diabetes Care, 1987. 10(5): p. 617‐21.
143. Polonsky, W.H., C.L. Davis, A.M. Jacobson, and B.J. Anderson, Correlates of hypoglycemic fear in type I and type II diabetes mellitus. Health Psychol, 1992. 11(3): p. 199‐202.
144. Escobar, C., R. Echarri, and V. Barrios, Relative safety profiles of high dose statin regimens. Vasc Health Risk Manag, 2008. 4(3): p. 525‐33.
145. Atar, D., R. Carmena, P. Clemmensen, K.L. A, S. Wassmann, et al., Clinical review: impact of statin substitution policies on patient outcomes. Ann Med, 2009. 41(4): p. 242‐56.
146. Stafford, R.S. and D.C. Radley, The potential of pill splitting to achieve cost savings ‐ In reply. American Journal of Managed Care, 2002. 8(11): p. 925‐926.
147. Ongoing projects at Centre for Pharmacoepidemiology, Karolinska Institutet, Sweden, November 2010. 2010, Karolinska Institutet: [cited 2011 20 September]; Available from: http://www.nhv.se/upload/dokument/forskning/N%C3%A4tverk/NorPEN/Ongoing%20projekt%20NorPEN.pdf.
148. Lundin, D., Ramsberg, J., Dynamic Cost‐Effectiveness: A More Efficient Reimbursement Criterion. Forum for Health Economics & Policy, 2008. 11 (2).
149. Abdel‐Maksoud, M., V. Sazonov, S.W. Gutkin, and J.E. Hokanson, Effects of modifying triglycerides and triglyceride‐rich lipoproteins on cardiovascular outcomes. J Cardiovasc Pharmacol, 2008. 51(4): p. 331‐51.
150. Singh, I.M., M.H. Shishehbor, and B.J. Ansell, High‐density lipoprotein as a therapeutic target: a systematic review. JAMA, 2007. 298(7): p. 786‐98.
151. Canner, P.L., K.G. Berge, N.K. Wenger, J. Stamler, L. Friedman, et al., Fifteen year mortality in Coronary Drug Project patients: long‐term benefit with niacin. J Am Coll Cardiol, 1986. 8(6): p. 1245‐55.
152. Carlson, L.A. and G. Rosenhamer, Reduction of mortality in the Stockholm Ischaemic Heart Disease Secondary Prevention Study by combined treatment with clofibrate and nicotinic acid. Acta Med Scand, 1988. 223(5): p. 405‐18.
153. Brown, B.G., X.Q. Zhao, A. Chait, L.D. Fisher, M.C. Cheung, et al., Simvastatin and niacin, antioxidant vitamins, or the combination for the prevention of coronary disease. N Engl J Med, 2001. 345(22): p. 1583‐92.
154. Boden, W.E., J.L. Probstfield, T. Anderson, B.R. Chaitman, P. Desvignes‐Nickens, et al., Niacin in patients with low HDL cholesterol levels receiving intensive statin therapy. N Engl J Med. 365(24): p. 2255‐67.
References
95
155. Phatak, H., C. Wentworth, V. Sazonov, and T. Burke, Prevalence and predictors of lipid abnormalities in patients treated with statins in the UK general practice. Atherosclerosis, 2009. 202(1): p. 225‐33.
156. Goldenberg, N. and C. Glueck, Efficacy, effectiveness and real life goal attainment of statins in managing cardiovascular risk. Vasc Health Risk Manag, 2009. 5(1): p. 369‐76.
157. McEwan, P., M. Evans, and K. Bergenheim, A population model evaluating the costs and benefits associated with different oral treatment strategies in people with type 2 diabetes. Diabetes Obes Metab. 12(7): p. 623‐30.
158. Ross, C.E. and C.L. Wu, Education, age, and the cumulative advantage in health. J Health Soc Behav, 1996. 37(1): p. 104‐20.
159. Russell, L.B., Preventing chronic disease: an important investment, but donʹt count on cost savings. Health Aff (Millwood), 2009. 28(1): p. 42‐5.
160. Yusuf, S., S. Islam, C.K. Chow, S. Rangarajan, G. Dagenais, et al., Use of secondary prevention drugs for cardiovascular disease in the community in high‐income, middle‐income, and low‐income countries (the PURE Study): a prospective epidemiological survey. Lancet. 378(9798): p. 1231‐43.
161. Nolte, E., Newbould J., Conklin A., International variation in the usage of medicines‐ A review of the litterature, in Rand corporation technical report series. 2010, Rand Corporation Europe: Cambridge.
162. Hartung, D.M., K.L. Ketchum, and D.G. Haxby, An evaluation of Oregonʹs evidence‐based Practitioner‐Managed Prescription Drug Plan. Health Aff (Millwood), 2006. 25(5): p. 1423‐32.
163. Walley, T., P. Folino‐Gallo, P. Stephens, and E. Van Ganse, Trends in prescribing and utilization of statins and other lipid lowering drugs across Europe 1997‐2003. Br J Clin Pharmacol, 2005. 60(5): p. 543‐51.
164. Abookire, S.A., A.S. Karson, J. Fiskio, and D.W. Bates, Use and monitoring of ʺstatinʺ lipid‐lowering drugs compared with guidelines. Arch Intern Med, 2001. 161(1): p. 53‐8.
165. Taylor, F., K. Ward, T.H. Moore, M. Burke, G. Davey Smith, et al., Statins for the primary prevention of cardiovascular disease. Cochrane Database Syst Rev, (1): p. CD004816.
166. Warner, K.E. and P.A. Warner, Is an ounce of prevention worth a pound of cure? Disease prevention in health care reform. J Ambul Care Manage, 1993. 16(4): p. 38‐49.
167. Nord, E. Discounting future health benefits: the poverty of consistency arguments. Health Econ. 20(1): p. 16‐26.
168. Russel, L. Is prevention Better than Cure? 1986, Washington D.C.: The Brookings institution.
References
96
169. Cohen, J.T., P.J. Neumann, and M.C. Weinstein, Does preventive care save money? Health economics and the presidential candidates. N Engl J Med, 2008. 358(7): p. 661‐3.
170. Vermeire, E., H. Hearnshaw, P. Van Royen, and J. Denekens, Patient adherence to treatment: three decades of research. A comprehensive review. J Clin Pharm Ther, 2001. 26(5): p. 331‐42.
171. Ridker, P.M. Fasting versus nonfasting triglycerides and the prediction of cardiovascular risk: do we need to revisit the oral triglyceride tolerance test? Clin Chem, 2008. 54(1): p. 11‐3.
