Suzanne E. Proctor, MSPH Richard J. Klein, MPH NCHS Data User’s Conference July 12, 2004
description
Transcript of Suzanne E. Proctor, MSPH Richard J. Klein, MPH NCHS Data User’s Conference July 12, 2004
Methodological Issues in Producing CDC Data for the
National Healthcare Quality and Disparities Reports
Suzanne E. Proctor, MSPHRichard J. Klein, MPH
NCHS Data User’s ConferenceJuly 12, 2004
Background
Preliminary set of measures for the 1st National Healthcare
Quality Report (NHQR) and National Healthcare Disparities
Report (NHDR) presented to NCHS in summer 2002
High degree of overlap between measures in two reports (same Quality of Care measures)
Majority of requested measures were Healthy People 2010 measures
• Measures already gone through consensus process • Established national data source• Measures widely recognized & used at Federal & State levels
Background
Inclusion of HP2010 measures simplified the project
• Operational definitions developed
• Technical issues (e.g., age adjustment) resolved
• SAS/SUDAAN programs written
• Some baseline & update data available for most measures
However…
• Definitions for some measures differed slightly from HP2010
• Population subgroup data and cross-tabulations requested beyond what is
regularly programmed for HP2010
• Additional analytic issues arose during the production of data
Scope of NCHS Contribution
Summary of Measures Provided for the 2003-2004 Reports
NHQR only
NHDR only
Both reports
HP2010 measure (55) 18 7 30
Modified HP2010 measure (13) 4 0 9
Not a HP2010 measure (39) 6 31 2
TOTAL MEASURES: 107 28 38 41
Summary of Measures Provided for the 2003-2004 Reports
National & State data
National data only
State data only
HP2010 measure (55) 23 28 4
Modified HP2010 measure (13) 2 10 1
Not a HP2010 measure (39) 0 39 0
TOTAL MEASURES: 107 25 77 5
Distribution of Measures by Healthy People 2010 Focus Area
6
33
18
243
44
108
Number of measures (both reports)
3. Cancer5. Diabetes
16. Maternal, Infant & Child Health14. Immunization & Infectious Diseases
13. HIV12. Heart Disease & Stoke
18. Mental Health
1. Access to Quality Health Services
21. Oral Health24. Respiratory Diseases
Not a HP2010 measure
Distribution of Measures by National Data Source
11
4555
66
710
1519
23
Behavioral Risk Factor Surveillance System
National Ambulatory Medical Care Surveys
National Immunization Survey
National Health Interview Survey
National Hospital Discharge Survey
National Vital Statistics System
HIV/AIDS Reporting System
National Nosocomial Infections Surveillance System
Number of measures (both reports)
Adult Spectrum of Disease Surveillance Project
Children with Special Healthcare Needs
National Health and Nutrition Examination Survey
National Home and Hospice Care Survey
National Nursing Home Survey
Types of Measures Provided
Selected NHQR measures
Effectiveness
• Percent of women (40 and over) who report they had a mammogram in the past 2 years
• Hospital admissions for pediatric asthma per 10,000 population under 18 years
• HIV-infection deaths per 100,000 population
Safety
• Central line-associated bloodstream infection in Intensive Care Unit patients
Selected NHQR measures
Timeliness
• Percent of persons who have a usual source of medical care, by place of care
• Percent of patients who wait 1 or more hours to be seen by a physician, by urgency of needed care
Patient Centeredness
(No measures provided)
Selected NHDR measures
Quality of Care
• Effectiveness: Hospital admissions for lower extremityamputations in patients with diabetes per 1,000 population
• Safety: (No measures provided)
• Timeliness: Percent of persons in fair or poor health whohave a specific source of on-going care
• Patient centeredness: (No measures provided)
Selected NHDR measures
Access to Care
• Entry barriers: Percent of persons < 65 years with public health insurance only
• Structural barriers: (No measures provided)
• Patients’ perceptions: (No measures provided)
• Use: Emergency department visits per population
Depth of Data Provided
Data Templates
Specific tabulations requested by AHRQ differed forthe NHQR and NHDR
Typical NHQR Data Template for Population-Based Objectives
Total Total
AgeAge RaceRace
American Indian or Alaska Native Asian or Pacific Islander
Asian Native Hawaiian or Other Pacific Islander Black or African American White
Hispanic origin and raceHispanic origin and raceHispanic or Latino Not Hispanic or Latino Black or African American White
GenderGender Female Male
Socioeconomic StatusSocioeconomic Status
Family Income Level Poor Near poor Middle/high income
oror
Education Level Less than high school High school graduate At least some college
NHDR Data Templates for Population-Based Objectives
Race Race x NHQR variables
EthnicityEthnicity x NHQR variables
Socioeconomic statusSocioeconomic status x NHQR variables
Data Templates
Specific tabulations requested by AHRQ differed forthe NHQR and NHDR
Data quality/small number issues limited ability toprovide data for some cross-tabulations in the
NHDR
Availability of data to complete templates varied bydata source
• SES and health insurance information not available from all
data sources
• Collection & tabulation of race/ethnic data differed by data source
Tracking Trends
Tracking Trends – Key Issues
Change in Federal standards for collecting and tabulating race/ethnicity data
Change in Census population denominators
National/State data comparability
Periodicity of data collection
Sample size and data quality issues
Population Denominators
For the 2003 reports, NCHS provided 1999 and 2000 data from NVSS
Denominators for these data were post-censal estimates (based on the 1990 Census)
For the 2004 reports, NCHS reran the 1999 and 2000 NVSS data using 2000 census data and
revised 1999 inter-censal estimates
National/State Data Comparability
National and State data provided for 25 measures
In general, the national and State data werecomparable
• NVSS national and State data based on same source reports • NIS national and State data based on same source reports• NHIS (national) and BRFSS (state) surveys – compare cautiously
However, not all of the BRFSS questions were asked by all States each year
• Some BRFSS questions contained in optional modules
Periodicity of Data
For most measures, NCHS was able to provide a new data point for each data year shown
However …
• Some surveys are fielded on a periodic basis (e.g. NNHS)
• Questions used to track some measures were not included inthe survey each year (e.g. NHIS, BRFSS)
• For some measures, needed to combine 2 or more years ofdata to obtain reliable estimates (e.g. NHANES)
Periodicity of Data - Example
NHIS measure:
12-12. Percent of adults (18 years and over) who have had a blood pressure measurement within the past 2 years and can state whether it was normal or high
Data available in 1998, 2003
BRFSS measure:
12-15. Percent of adults (18 years and over) who have had their blood cholesterol checked within the past 5 years
2001: 50 States + D.C.2002: 12 States + D.C.
Sample Size/Data Quality Issues
Data for smaller subpopulation groups may be statistically unreliable due to small sample size
Even when the subpopulation groups size is sufficient for univariate analysis, cross-tabulation with other variables in
the template can spread the data too thin
For some data sources, the sample design does not produce representative estimates for some sub-population
groups
Large number of missing responses
DSU – data are suppressed
Summary
Reports provided an opportunity for intra-agency and cross-agency collaboration on two major Federal
public health reports
Reports provided a rich source of additional data for HP2010 (e.g. Progress Reviews)
Excellent example of how the existing Federal data infrastructure can be readily used for in-depth
examination of critical public health issues