1

To the Editors and Reviewers:

Thank you for your thoughtful comments on our manuscript. We appreciate the opportunity to

revise the manuscript. Below, we provided point-by-point responses for each of the comments.

1. Please provide a more thorough explanation of the different insurance categories, as

these will not be as well understood by a non-U.S reader. (reviewers have also commented on

this). You might consider a box that contains this information.

Thank you for this suggestion. We have provided a box with a brief primer on the key types of

insurance available in the United States and noted which are included in the OLDW.

2. Several editors commented that it is difficult to make sense of this information without

knowing the types of opioids that were prescribed and whether these changed over time. For

example, did prescriptions for tramadol increase and those for oxycodone preparations

decrease over time, or has the distribution of prescriptions across opioid classes remained

roughly the same? Please provide more detail about the types of opioids prescribed, and

whether this has changed over time.

We agree that knowing the type of opioids used is of interest. We have added information to

the appendix with trends over time. We refer to this appendix in a new section of results

(Trends in opioid used), noting a few items of interest.

3. Please describe the most common diagnoses in the various groups of patients. How

many of the disabled elderly have cancer or chronic pain conditions, for example?

We have calculated and provided prevalence of the Elixhauser comorbidities in the Appendix

(Quan 2005). We calculated the comorbidities on a rolling 6-month basis and required the

diagnosis to appear on at least one inpatient date of service or two outpatient dates of service.

In the disabled Medicare population (which includes Medicare Advantage beneficiaries under

the age of 65), 1% were identified as having a metastatic cancer diagnosis, 4% solid tumor

without metastasis, and 1% had lymphoma. In the elderly Medicare Advantage group, the

prevalence of these conditions were 1%, 8%, and 1%, respectively.

On the prevalence of chronic pain: this is very difficult to identify using claims data. There is

currently no good algorithm to characterize chronic pain in claims data. Some analysts use ICD9

codes 338.2X (chronic pain) and 338.4 (chronic pain syndrome), however, these codes are not

consistently applied and represent a substantial underestimate of the total population of

chronic pain sufferers (see, e.g., Tian 2013 reporting sensitivity of 20.3% for those codes). The

Tian paper presents an algorithm to identify chronic pain, but it requires information not

available in claims and was designed for use with electronic health record data rather than

claims data.

Quan H, Sundararajan V, Halfon P, Fong A, Burnand B, Luthi JC, Saunders LD,

Beck CA, Feasby TE, Ghali WA. Coding algorithms for defining comorbidities in

2

ICD-9-CM and ICD-10 administrative data. Med Care. 2005 Nov;43(11):1130-9. PubMed

PMID: 16224307.

Tian TY, Zlateva I, Anderson DR. Using electronic health records data to

identify patients with chronic pain in a primary care setting. J Am Med Inform

Assoc. 2013 Dec;20(e2):e275-80. doi: 10.1136/amiajnl-2013-001856. Epub 2013 Jul

31. PubMed PMID: 23904323; PubMed Central PMCID: PMC3861913.

4. Please provide more information about whether the studied population is

representative of the U.S as a whole (The south Atantic region appears to be over-represented,

for instance,) Prescribing patterns vary substantially in different regions of the US, so how

generalisable is this information.

The OLDW includes 20% of the commercially insured population in the United States and 24%

of the Medicare Advantage population. The distributions of age, sex, and race/ethnicity are

similar to the United States commercial and Medicare Advantage populations (see below). The

geographic distributions are less similar, however, all parts of the country are represented and

we have adjusted the predictions for census division to mitigate the effect of the differences.

Under 65 privately insured population, 2015

US privately insured OLDW

commercial

Race/ethnicity

White 66% 70%

Asian 7% 6%

Hispanic 15% 14%

Black 11% 11%

American Indian 1% N/A

Female 50% 49%

Age

0 to 17 24% 22%

18 to 24 11% 11%

25 to 34 16% 18%

35 to 44 15% 17%

45 to 54 17% 18%

55 to 59 9% 8%

60 to 64 7% 6%

Census Division

New England 5% 3%

3

Under 65 privately insured population, 2015

US privately insured OLDW

commercial

Mid Atlantic 13% 7%

East North Central 15% 16%

West North Central 6% 11%

South Atlantic 20% 22%

East South Central 6% 4%

West South Central 12% 18%

Mountain 7% 10%

Pacific 16% 9%

OLDW commercial includes people with both medical and prescription

coverage; excludes people with unknown race/ethnicity, year of birth, or

sex Source: CPS ASEC (Current Population Survey—Annual Social and

Economic Supplement), data for 2015, using

https://www.census.gov/cps/data/cpstablecreator.html; Includes only

people listed with a single race. Hispanic includes people of any single

race who indicated they were of Hispanic origin

Medicare Advantage, 2015

US Medicare

Advantage

OLDW

Medicare

Advantage

Race/ethnicity

White 76% 76%

Black 11% 12%

Hispanic 8% 9%

Other 6% 3%

Female 55% 57%

Census Division

New England 3% 6%

Mid Atlantic 15% 15%

East North Central 15% 18%

West North Central 6% 12%

South Atlantic 20% 31%

East South Central 6% 5%

4

Medicare Advantage, 2015

US Medicare

Advantage

OLDW

Medicare

Advantage

West South Central 10% 4%

Mountain 7% 5%

Pacific 18% 3%

OLDW Medicare Advantage includes people with both medical and

prescription coverage; excludes people with unknown race/ethnicity or

sex US Medicare Advantage source: Kaiser Family Foundation, 2015 data

https://www.kff.org/state-category/medicare/medicare-advantage/

5. Please think about the headline statistics in your paper and make sure they are in the

abstract where interested readers will be able to find them.

We have edited the results to report the key results. We are happy to modify them if the

editors or reviewers felt that reporting additional results might be helpful.

6. We thought use of the word "high" to characterize prescription rates was an

oversimplification. In some groups prescription rates appear to be quite low, for example. This

is quite subjective language, particularly when we do not know whether prescribing was

medically appropriate or not.

Thank you—this is an important point, as the quarterly commercial use prevalence could be

considered low. It is difficult to compare opioid use prevalence in different countries, because

the data are difficult to find. However, we do know that the United States uses roughly twice

the volume of prescription opioids as the next closest countries—Canada and Germany (see

chart below). Van Amsterdam (2015) gathers some opioid use data for the US, Canada, and

Europe, including a few estimates of opioid use prevalence. Opioid use prevalence in Germany

(Schubert 2013) is reported to be 4.5% (residents of Hesse with insurance, 2010, codeine

excluded) to 5.9% (2009, people insured by Barmer, including codeine). Weighting our 2009

numbers for adults by the approximate proportion of the US population with commercial and

Medicare insurance, 21% used prescription opioids.

In Ontario, a survey showed that 23% used any prescription opioid in 2015 (Ialomiteanu 2016).

This compares to a US survey reporting that 38% of adults used any prescription opioid in in

2015 (Han 2017). (Note that survey results are not directly comparable to claims data because

both surveys include both medical and non-medical use and that medical use can include

opioids that were filled months or years ago.)

5

Both of these data suggest that US figures reported in our study are high relative to the

countries nearest to the US in usage, and therefore would be considered very high relative to

the rest of the world.

We don’t know what proportion of US use is medically appropriate. We can’t exclude the

possibility that US usage rates are high relative to the rest of the world because no other

country is adequately managing pain. We have added a brief discussion of this difference to

the discussion. However, if the editors and reviewers feel strongly about this, we are happy to

make the change.

Han B, Compton WM, Blanco C, Crane E, Lee J, Jones CM. Prescription Opioid Use, Misuse, and

Use Disorders in U.S. Adults: 2015 National Survey on Drug Use and Health. Ann Intern Med.

2017;167:293–301. doi: 10.7326/M17-0865

Ialomiteanu, A. R., Hamilton, H. A., Adlaf, E. M., & Mann, R. E. (2016). CAMH Monitor e-Report:

Substance Use, Mental Health and Well-Being Among Ontario Adults, 1977–2015 (CAMH

Research Document Series No. 45). Toronto, ON: Centre for Addiction and Mental Health.

Available at: www.camh.ca/en/research/news_and_publications/

Pages/camh_monitor.aspx

Schubert I, Ihle P, Sabatowski R. Increase in opiate prescription in Germany

between 2000 and 2010: a study based on insurance data. Dtsch Arztebl Int. 2013

Jan;110(4):45-51. doi: 10.3238/arztebl.2013.0045. Epub 2013 Jan 25. PubMed PMID:

23413387; PubMed Central PMCID: PMC3570953.

6

Barmar study cited in Schubert 2013: Marschall U, L`hoest H. Opioidtherapie in der

Versorgungsrealität. >Ein Beitrag zur Diskussion um ein weitverbreitetes Arzneimittel. In:

Repschläger U, Schulte C, Osterkamp N, editors. Gesundheitswesen aktuell 2011. Düsseldorf:

Beiträge und Analysen; 2011. pp. 242–269.

van Amsterdam J, van den Brink W. The Misuse of Prescription Opioids: A Threat

for Europe? Curr Drug Abuse Rev. 2015;8(1):3-14. Review. PubMed PMID: 26084418.

6. In your discussion, please explore further the role of cash payments for opioids in your

findings. Do you perceive this to be an important factor in the pattern of results, or the

implications of your findings?

Data from pharmacy fills suggest that approximately 3% of opioid prescriptions for the

commercially insured were self paid in the 2014-2015 time period (Cepeda 2017). Given the

rates of opioid use in this paper, we would not expect the addition of self paid prescriptions to

affect the interpretation of the results.

The key concern with missing cash-paid fills among insured people is in cases where the

patient’s insurance company does not cover a specific drug or where the drug requires prior

approval from the insurance company. In the former case, we would not observe these fills. In

the latter, we might miss the first time the patient fills that drug, but once the insurance

company approves, we would observe the later fills.

Cepeda MS, Fife D, Denarié M, Bradford D, Roy S, Yuan Y. Quantification of

missing prescriptions in commercial claims databases: results of a cohort study.

Pharmacoepidemiol Drug Saf. 2017 Apr;26(4):386-392. doi: 10.1002/pds.4165. Epub

2017 Jan 25. PubMed PMID: 28120552; PubMed Central PMCID: PMC5396298.

7. Please expand your discussion to comment on your interpretation and perceived

significance of your results (as mentioned by reviewers).

We have added further discussion to place our results in the context of opioid use outside the

US; specifically, we have added a paragraph comparing US opioid use to Germany (the next

highest user of opioids per capita).

8. Data sharing:: We strongly prefer that you commit to making the data underlying the

analyses in the paper available upon reasonable request so that other researchers can replicate

the main findings. There are ways to share anonymized data.

OptumLabs data are deidentified and available for research through a virtual data warehouse.

OptumLabs has agreed to make the dataset for this study and the accompanying code available

7

to interested researchers interested in replicating the findings through a virtual data

warehouse. Interested researchers can contact the corresponding author for the manuscript

and the author will facilitate access to the data and code for the researchers.

When you revise your paper, please respond to all of the editorial comments above, as well as

those of the reviewers, which are available at the end of this letter, below.

Please provide a "track changes" and "clean" version of your revised paper, and in your

response to reviewers and editors please indicate what changes, if any, you have made in

response to suggestions, and where in the manuscript these changes are located. In your

response please provide, point by point, your replies to the comments made by the reviewers

and the editors, explaining how you have dealt with them in the paper.

** Comments from the external peer reviewers**

Reviewer: 1

Recommendation:

Comments:

BMJ Referee report

Trends in opioid use in commercially insured and Medicare Advantage populations: a

retrospective cohort study

Thanks for the opportunity to review this study, which used data from Optum Labs on

commercially insured individuals, those in Medicare Advantage (MA), and disabled MA

beneficiaries, to study trends in opioid prescribing from 2007 to 2016.

This is an important and interesting study that adds to prior studies by comparing trends in

opioid utilization rates across there important patient populations, bringing to bear probably

the largest set of data to date, and providing recent estimates of prescription opioid use which

are made possible by use of Optum data.

The overall writing is excellent, the study is well motivated, the description of the data and

methods are clear, and the results make sense and are nicely discussed. From a methodological

perspective, I don't have anything to add.

Thank you very much for your review and helpful comments.

A few comments/suggestions.

9. Because the BMJ has an international audience, it’s not uncommon in studies like this in

the BMJ to describe the institutional background a bit more, perhaps directly in the text or in a

sidebox. Non-US readers may not know what commercially insured populations are, Medicare

Advantage, etc.

Thank you for the suggestion. We have added a text box.

8

10. I like the point that using individual level data is important for studying the opioid

epidemic. Reporting rates of MME in a population obscures the fact that some patients

disproportionately use more opioids than others, which is why analyses like these are revealing,

because they tell us what proportion of individuals in a given category are using prescription

opioids over time.

We agree—using individual level data provide important insights including trends and patterns

in opioid use.

11. On a related point, the authors refer to the analysis by Sun and Jena on the

concentration of opioid use over time (i.e., is use becoming more or less concentrated within a

select group of individuals over time). The authors report this concentration information in the

Discussion in the cross-section, but perhaps an appendix exhibit that plots the change in

concentration over time (e.g., X percent of individuals in each group – commercial, MA, MA

disabled – account for 10%, 20%, etc. of opioid prescriptions or MME in a given year). Not a

critical thing to do but since there are a lot of citable descriptive statistics here, this would be a

good one to add.

Thank you for the suggestion; we have added it to the Appendix and referred to it in the text in

the “Trends by age group” section. We agree that it would be of interest and hope that this

article will serve as a good source for a broad range of statistics on some groups that have not

been extensively evaluated.

12. A brief discussion on what MA plans are doing, if anything, to curb the opioid epidemic

may be useful in the discussion. Similarly, may be worth speculating as to why there is huge

growth in MA disabled prescription opioid use.

We have added to the discussion a reference to the very recently announced CMS-proposed

hard and soft edits.

The growth in the disabled MA beneficiaries’ use is likely multifactorial, but it’s difficult to

speculate without more study. We hope to address this issue in future work.

13. Is it worthwhile to do a sensitivity analysis on how long-term use is defined by allowing

more than 30 days between opioid prescriptions to close out a long-term episode (e.g., 40

days?).

First, we would like to clarify that closing out an episode of opioid use requires 30 days to lapse

between the end of one prescription and the start of the next. In other words, to close out an

episode, we count 30 days from (fill date of the last prescription + count days supply of the last

prescription). In other words, our definition is robust to someone making one month’s worth of

fills that don’t appear in our claims.

9

That said, we were interested in pursuing this suggestion to see how sensitive the definition

was. Overall, increasing from 30 to 40 days to close out an episode decreased the total number

of observed opioid episodes by 6.8%, including a 7.2% reduction in the number of non-long-

term episodes and a 1.5% reduction in the number of long-term use episodes. However, the

proportion of episodes that represented long-term use did not change substantively—it went

from 4.5% to 4.7%. Because this difference was so small, we did not conduct additional

sensitivity analysis. We would be happy to add this to the manuscript or appendix if the

reviewers or editors felt it was important.

Additional Questions:

Please enter your name: Anupam Jena

Job Title: Ruth L Newhouse Associate Professor

Institution: Harvard Medical School

Reimbursement for attending a symposium?: No

A fee for speaking?: No

A fee for organising education?:

Funds for research?: No

Funds for a member of staff?: No

Fees for consulting?: Yes

Have you in the past five years been employed by an organisation that may

in any way gain or lose financially from the publication of this paper?: No

Do you hold any stocks or shares in an organisation that may in any way

gain or lose financially from the publication of this paper?: No

If you have any competing interests <A HREF='http://www.bmj.com/about-bmj/resources-

authors/forms-policies-and-checklists/declaration-competing-interests'target='_new'> (please

see BMJ policy) </a>please declare them here: I have received consulting fees unrelated to this

work from Pfizer, Novartis, Hill Rom, Vertex, Bristol Myers Squibb, and Precision Health

Economics.

Reviewer: 2

10

Recommendation:

Comments:

14. This is a well-written manuscript describing an important, well-performed epidemiological

study using an administrative data set. It provides clinically useful evidence that informs the

generalizations made about opioid prescribing in the “opioid epidemic” era.

Thank you very much for your very helpful review and comments.

Introduction

15. General comment: I think non-US readers need some brief clarification about the clinically

relevant differences between insurance types, and the break down , eg commercial implies

employed people (young and healthy) and covers their dependents.

Thank you for this suggestion. We have added a text box to explain the types of insurance

available in the US. This will provide non-US readers a better overview of the types of

populations included in each of the insurance types.

16. And the US population breakdown percentage wise by insurance types.

We have added a breakdown to the box on US insurance types.

P5, line 34. Any opioid data available on original Medicare beneficiaries (I think reference 26,

but please confirm; any others?)

We are aware of no similar study published that describes opioid use of fee-for-service (FFS)

beneficiaries. We cite Morden 2014 which looks only at the disabled group of FFS beneficiaries

and compare our findings, which are similar. Jena 2014 reports that in 2010, 35% of Medicare

beneficiaries with prescription drug coverage filled any opioid prescription. In our sample, 30%

of Medicare beneficiaries filled any opioid prescription in 2010. These rates are somewhat

different, however, all Medicare Advantage beneficiaries have prescription drug coverage

unlike FFS Medicare, where only about 46% had part D in 2010;{Chronic Conditions Data

Warehouse, 2017 #673} we might expect the FFS population who select part D coverage to be a

higher than average risk of prescription drug use, so the somewhat higher rate of opioid use in

that population compared to our sample is unsurprising.

Chronic Conditions Data Warehouse. Medicare Enrollees in Part D, 2006-2015 2017 [Available from: https://www.ccwdata.org/web/guest/medicare-charts/medicare-part-d-charts accessed 3/6/2018].

Jena AB, Goldman D, Weaver L, Karaca-Mandic P. Opioid prescribing by multiple

providers in Medicare: retrospective observational study of insurance claims.

BMJ. 2014 Feb 19;348:g1393. doi: 10.1136/bmj.g1393. PubMed PMID: 24553363; PubMed

Central PMCID: PMC3928962.

Morden NE, Munson JC, Colla CH, Skinner JS, Bynum JP, Zhou W, Meara E.

Prescription opioid use among disabled Medicare beneficiaries: intensity, trends,

and regional variation. Med Care. 2014 Sep;52(9):852-9. doi:

11

10.1097/MLR.0000000000000183. PubMed PMID: 25119955; PubMed Central PMCID:

PMC4151179.

Methods

17. Variables – general: what about other variable, e.g. ICD and CPT codes?

Thank you for the suggestion. We’ve provided additional information on the prevalence of

some common comorbidities among opioid users using ICD9 and ICD10 diagnosis codes in

Appendix 6.

CPT codes, which describe procedures received by beneficiaries, may be more difficult to

characterize, since this group includes a mixed cohort of long-term and short-term opioid users.

18. Prescriber variables? I think it would be very informative to see if prescribing patterns are

changing in different clinician types (specialty e.g. pain specialist vs internist, age, gender).

OptumLabs Data Warehouse has limited information on prescribers, which unfortunately does

not include prescriber age or gender. We would be able to look at specialty; however it does

not include enough granularity to include the pain specialties. Many pain specialists are

certified in other areas such as family medicine or anesthesiology or psychiatry. We would not

be able to separate these specialists out. Due to this limitation, we did not add the results by

specialty. Other researchers have done excellent work describing trends in opioid prescriber

specialties. See in particular,

Levy B, Paulozzi L, Mack KA, Jones CM. Trends in Opioid Analgesic-Prescribing

Rates by Specialty, U.S., 2007-2012. Am J Prev Med. 2015 Sep;49(3):409-13. doi:

10.1016/j.amepre.2015.02.020. Epub 2015 Apr 18. PubMed PMID: 25896191.

19. Re accuracy of fill dates. One of the limitations of administrative data base research is the

accuracy of dates and coding. How accurate are the fill dates? If it relies on the pharmacist

entering it, it may not be accurate. Pease clarify

Fill dates are based on the date the pharmacist submitted the claim for payment. They are

generally the day the product was filled and placed on the shelf for patient pick up. Thus,

theremay be several days between the fill date and the date the patient picked up the

prescription. However, we do not believe this will substantially affect the analysis because the

fill dates should be within 2 weeks of patient receipt of the fill.

20. Statistical methods p 9, line 52. Re Outcomes. I don’t think filling opioid prescriptions should

be called an “outcome”. Another limitations of administrative data base research is the lack of

clinical outcomes (e.g. benefits, harms) and so it’s confusing calling prescribing an outcome. I

12

think the sentence should be “Opioid-related variables were defined at the person-quarter

level”.

Thank you for this suggestion. We have made the language more precise by referring to these

measures as “endpoints.”

Discussion

21. Comparison to literature, p.14, L. 43 you have made the point that your data does not

include cash-paid fills (up to 20% as per p. 16 L. 34). But there is no comment on how this

missing data might affect your results. Seems to me this could lead to significant under-

estimation, especially in people with aberrant drug taking on chronic opioid therapy who are

seeking early refills. Prescription Monitoring Programs which have come in during the study

period could also drive this up as people seek to avoid detection.

This is an interesting point that may represent a difference in the US healthcare system. People

seeking early refills would not receive any benefit from paying cash vs. submitting their fill to

their insurance company. Regardless of the payment method, they need a valid prescription

from a provider, and their fill will be entered in the local prescription monitoring program. The

key concern with missing cash-paid fills is in cases where the patient’s insurance company does

not cover a specific drug or where the drug requires prior approval from the insurance

company. In the former case, we would not observe these fills. In the latter, we might miss the

first time the patient fills that drug, but once the insurance company approves, we would

observe the later fills.

We expect little impact on our results from these cash paid fills. Drugs excluded from coverage

are all branded drugs that have a generic available.

In addition, as noted above, data from pharmacy fills suggest that approximately 3% of opioid

prescriptions filled by commercially insured people were self paid in the 2014-2015 time period

(Cepeda 2017). Given the rates of opioid use in this paper, we would not expect the addition of

self paid prescriptions to affect the interpretation of the results.

Cepeda MS, Fife D, Denarié M, Bradford D, Roy S, Yuan Y. Quantification of

missing prescriptions in commercial claims databases: results of a cohort study.

Pharmacoepidemiol Drug Saf. 2017 Apr;26(4):386-392. doi: 10.1002/pds.4165. Epub

2017 Jan 25. PubMed PMID: 28120552; PubMed Central PMCID: PMC5396298.

Additional Questions:

Please enter your name: Paul Glare

Job Title: Professor of pain medicine

Institution: university of sydney

13

Reimbursement for attending a symposium?: No

A fee for speaking?: No

A fee for organising education?: No

Funds for research?: No

Funds for a member of staff?: No

Fees for consulting?: No

Have you in the past five years been employed by an organisation that may

in any way gain or lose financially from the publication of this paper?: No

Do you hold any stocks or shares in an organisation that may in any way

gain or lose financially from the publication of this paper?: No

If you have any competing interests <A HREF='http://www.bmj.com/about-bmj/resources-

authors/forms-policies-and-checklists/declaration-competing-interests'target='_new'> (please

see BMJ policy) </a>please declare them here:

Reviewer: 3

Recommendation:

Comments:

22. This is a carefully done assessment of trends in opioid prescribing in 3 large insured

populations in the United States. There are two findings of interest: the absence of

large reductions in opioid use or dose in any of the populations, and the substantially higher

rates of use and higher doses in the Medicare disabled patients. A third result confirms prior

reports: that a small percentage of opioid use episodes accounts for a large share of opioids

dispensed on a population basis. The paper is generally well written, although the methods

employed, and their description, is inherently complex.

The results are useful because, as the authors note, most trend data on opioid use is either

aggregate amounts of opioids prescribed without person-level analyses, or trends for single

health plans. The results seem generally consistent with available trend data which showed

large increases in opioid prescribing from 1999 to 2009 (largely before the study period

employed in this report), with a plateau in opioid prescribing attained around 2010-12. This

study indicates that there had not been a sudden drop in opioid use or dose among long-term

14

users by 2013. One would expect trend changes to be gradual, particularly reversal from

increasing use to decreasing, so these results seem plausible.

As you note, we do not observe substantial decreases in the 3 to 4 years after rates plateaued

in 2012 to 2013. This is in contrast to recent reports of market-level data that saw decreases in

fills on a per capita basis (Guy 2017). We agree that we would expect trend changes to be

gradual, so it is important that by the end of 2016 we have not seen use rates dropping.

Guy GP Jr, Zhang K, Bohm MK, Losby J, Lewis B, Young R, Murphy LB, Dowell D.

Vital Signs: Changes in Opioid Prescribing in the United States, 2006-2015. MMWR

Morb Mortal Wkly Rep. 2017 Jul 7;66(26):697-704. doi: 10.15585/mmwr.mm6626a4.

PubMed PMID: 28683056; PubMed Central PMCID: PMC5726238.

23. Improved surveillance of prescription opioid use along with illicit opioid use is needed to

understand trends in opioid overdose. Fatal opioid overdoses have continued to increase, with

recent increases in deaths involving heroin and illicit fentanyl. It will be important to

understand whether changes in medical prescribing of opioids contribute to changes (either

favorable or unfavorable) in opioid overdose rates, but the overdose trend data, but the

determinants of recent trends in opioid overdose remain unclear.

We agree that the trends in prescription opioid use do not tell the story of the epidemic of

opioid overdoses, as they cannot directly address overdoses due to intentional or accidental

use of illicit fentanyl.

However, we believe that large scale surveillance of individual prescription opioid use can help

us identify opportunities to improve prescribing practices. For example, work by Mary Willy

(2014) in the Medicare population and Marc Larochelle (2017) in a commercial population has

found large numbers of non-opioid-tolerant people receiving long-acting/extended release

formulations of opioids, and our own work in progress showing large proportions of people

receiving benzodiazepines while they are taking opioids.

Larochelle MR, Cocoros NM, Popovic J, Dee EC, Kornegay C, Ju J, Racoosin JA.

Opioid tolerance and urine drug testing among initiates of extended-release or

long-acting opioids in Food and Drug Administration's Sentinel System. J Opioid

Manag. 2017 Sep/Oct;13(5):315-327. doi: 10.5055/jom.2017.0400. PubMed PMID:

29199397.

Willy ME, Graham DJ, Racoosin JA, Gill R, Kropp GF, Young J, Yang J, Choi J,

MaCurdy TE, Worrall C, Kelman JA. Candidate metrics for evaluating the impact of

prescriber education on the safe use of extended-release/long-acting (ER/LA)

opioid analgesics. Pain Med. 2014 Sep;15(9):1558-68. doi: 10.1111/pme.12459. Epub

2014 May 15. PubMed PMID: 24828968.

15

24. This paper helps us understand recent trends in opioid prescribing at what is likely the

plateau in opioid prescribing in the United States, but does not signal whether a large drop is

beginning or whether the currently high rates of opioid use in the United States will be

sustained at these levels indefinitely. While there is considerable professional and public

interest in the opioid epidemic in the United States, the trend data reported in this paper may

be of greatest interest to scientists tracking and trying to understand these trends.

We agree that these data will be of interest to scientists and public health officials, which is one

reason we have provided extensive appendix tables (and have provided more on the

recommendation of reviewers). We believe it is also important to establish detailed information

about the current shape of the opioid epidemic for clinicians and members of the public. With

some market-level data suggesting that opioid use may be falling, we think it is important to

demonstrate that at the individual level this trend is not apparent. We have instead seen

several years of stable rates of opioid fills. The US uses double the volume of opioids per capita

as the next nearest countries (Germany and Canada). We have a substantial opportunity to

continue to improve opioid prescribing. We hope this study will highlight that.

Additional Questions:

Please enter your name: Michael VonKorff

Job Title: Senior Investigator

Institution: Kaiser Permanente Washington Health Research Institute

Reimbursement for attending a symposium?: No

A fee for speaking?: No

A fee for organising education?: No

Funds for research?: Yes

Funds for a member of staff?: No

Fees for consulting?: No

Have you in the past five years been employed by an organisation that may

in any way gain or lose financially from the publication of this paper?: No

Do you hold any stocks or shares in an organisation that may in any way

gain or lose financially from the publication of this paper?: No

If you have any competing interests <A HREF='http://www.bmj.com/about-bmj/resources-

authors/forms-policies-and-checklists/declaration-competing-interests'target='_new'> (please

16

see BMJ policy) </a>please declare them here: In the past 3 years, I was the PI of grants to

Kaiser Permanente Washington Health Research Institute from Pfizer Inc concerning

identification of problem opioid use among chronic pain patients using opioids long-term, and

evaluation of opioid risk reduction initiatives. I am a co-investigator on FDA-mandated

surveillance studies assessing risks among chronic pain patients using extended release opioids

funded by the Campbell Alliance, a consortium of opioid manufacturers supporting this

research as part of the opioid REMS program.

Reviewer: 4

Recommendation:

Comments:

25. This was well-conducted and well-written study that adds important information to the

existing literature. I just have a few minor comments aimed at improving the manuscript.

Thank you for your review and helpful comments and suggestions

As worded the “What this study adds” sections makes it sound like data was only analyzed

through 2013.

Thank you for pointing this out. We have revised that wording.

26. Very little information is included in the methods section on the regression analyses. The

authors should explain why regression analysis was used. For example, why was it important to

adjust the result and not just provide unadjusted use? Also, why were certain variables and

interactions included in the regression models.

We provide results adjusted for geography and basic demographics to reduce noise from small

changes in OLDW enrollment over time. Unadjusted use is not substantially different, but is a

bit noisier. We would be happy to add unadjusted graphs to the Appendix if reviewers and

editors believe it would be helpful.

27. The authors should consider testing the significance of trends across the study period. For

example, the use of Joinpoint regression, which has been used in previous studies, seems like it

might be appropriate.

We have not conducted extensive statistical or trend testing because even very small changes

are significant because of the size of the population. If the reviewers are interested in a specific

statistical test, we would be happy to provide that.

28. Figure 3 has labels “Chronic” and Non-chronic” should this be long-term and not long term?

Yes—thank you for pointing this out. We have made this change.

29. More information on the impact of the results on practice and policy would be helpful. The

authors mention a strength of this study is that is looks at opioid use across different

17

populations. I agree with the authors and believe they have some interesting findings across

different groups. However, I feel that more information is needed on the implications of these

findings specific to these different groups.

Thank you for this suggestion. We have added further discussion of the implications for the

disabled Medicare group, who will likely be most affected by policies limiting opioid dispensing

(like those proposed by CMS).

Additional Questions:

Please enter your name: Gery Guy

Job Title: Health Economist

Institution: Centers for Disease Control and Prevention

Reimbursement for attending a symposium?: No

A fee for speaking?: No

A fee for organising education?: No

Funds for research?: No

Funds for a member of staff?: No

Fees for consulting?: No

Have you in the past five years been employed by an organisation that may

in any way gain or lose financially from the publication of this paper?: No

Do you hold any stocks or shares in an organisation that may in any way

gain or lose financially from the publication of this paper?: No

If you have any competing interests <A HREF='http://www.bmj.com/about-bmj/resources-

authors/forms-policies-and-checklists/declaration-competing-interests'target='_new'> (please

see BMJ policy) </a>please declare them here: None

**Information for submitting a revision**

"Please remember that the author list and order was finalised upon initial submission. The BMJ

has an open peer review process. Reviewers and editors know the names of a paper's authors

18

and are able to judge the work in light of the authors' conflicts of interest. If authors are added

or removed after the evaluation is done, that process is subverted. We reserve the right to

require the formation of an authorship group when there are a large number of authors"

Deadline: Your revised manuscript should be returned within one month.

How to submit your revised article: Log into http://mc.manuscriptcentral.com/bmj and enter

your Author Center, where you will find your manuscript title listed under "Manuscripts with

Decisions." Under "Actions," click on "Create a Revision." Your manuscript number has been

appended to denote a revision.

You will be unable to make your revisions on the originally submitted version of the manuscript.

Instead, revise your manuscript using a word processing program and save it on your computer.

Once the revised manuscript is prepared, you can upload it and submit it through your Author

Center. When submitting your revised manuscript, you will be able to respond to the comments

made by the reviewer(s) and Committee in the space provided. You can use this space to

document any changes you make to the original manuscript and to explain your responses. In

order to expedite the processing of the revised manuscript, please be as specific as possible in

your response to the reviewer(s). As well as submitting your revised manuscript, we also

require a copy of the manuscript with changes highlighted. Please upload this as a

supplemental file with file designation ‘Revised Manuscript Marked copy’. Your original files are

available to you when you upload your revised manuscript. Please delete any redundant files

before completing the submission.

When you revise and return your manuscript, please take note of all the following points about

revising your article. Even if an item, such as a competing interests statement, was present and

correct in the original draft of your paper, please check that it has not slipped out during

revision. Please include these items in the revised manuscript to comply with BMJ style

(see: http://www.bmj.com/about-bmj/resources-authors/article-submission/article-

requirements and

http://www.bmj.com/about-bmj/resources-authors/forms-policies-and-checklists).

Items to include with your revision (see http://www.bmj.com/about-bmj/resources-

authors/article-types/research):

1. What this paper adds/what is already known box (as described

at http://resources.bmj.com/bmj/authors/types-of-article/research)

2. Name of the ethics committee or IRB, ID# of the approval, and a statement that

participants gave informed consent before taking part. If ethics committee approval was not

required, please state so clearly and explain the reasons why

(see http://resources.bmj.com/bmj/authors/editorial-policies/guidelines.)

19

3. Patient confidentiality forms when appropriate

(see http://resources.bmj.com/bmj/authors/editorial-policies/copy_of_patient-confidentiality).

4. Competing interests statement (see http://resources.bmj.com/bmj/authors/editorial-

policies/competing-interests)

5. Contributorship statement+ guarantor

(see http://resources.bmj.com/bmj/authors/article-submission/authorship-contributorship)

6. Transparency statement: (see http://www.bmj.com/about-bmj/resources-

authors/forms-policies-and-checklists/transparency-policy)

7. Copyright statement/licence for publication (see http://www.bmj.com/about-

bmj/resources-authors/forms-policies-and-checklists/copyright-open-access-and-permission-

reuse)

8. Data sharing statement (see http://www.bmj.com/about-bmj/resources-

authors/article-types/research)

9. Funding statement and statement of the independence of researchers from funders

(see http://resources.bmj.com/bmj/authors/article-submission/article-requirements).

10. Patient involvement statement (see http://www.bmj.com/about-bmj/resources-

authors/article-types/research).

11. Please ensure the paper complies with The BMJ’s style, as detailed below:

a. Title: this should include the study design eg "systematic review and meta-analysis.”

b. Abstract: Please include a structured abstract with key summary statistics, as explained

below (also see http://resources.bmj.com/bmj/authors/types-of-article/research). For every

clinical trial - and for any other registered study- the last line of the abstract must list the study

registration number and the name of the register.

c. Introduction: This should cover no more than three paragraphs, focusing on the

research question and your reasons for asking it now.

d. Methods: For an intervention study the manuscript should include enough information

about the intervention(s) and comparator(s) (even if this was usual care) for reviewers and

readers to understand fully what happened in the study. To enable readers to replicate your

work or implement the interventions in their own practice please also provide (uploaded as one

or more supplemental files, including video and audio files where appropriate) any relevant

20

detailed descriptions and materials. Alternatively, please provide in the manuscript urls to

openly accessible websites where these materials can be found.

e. Results: Please report statistical aspects of the study in line with the Statistical Analyses

and Methods in the Published Literature (SAMPL) guidelines http://www.equator-

network.org/reporting-guidelines/sampl/. Please include in the results section of your

structured abstract (and, of course, in the article's results section) the following terms, as

appropriate:

i. For a clinical trial: Absolute event rates among experimental and control groups; RRR

(relative risk reduction); NNT or NNH (number needed to treat or harm) and its 95% confidence

interval (or, if the trial is of a public health intervention, number helped per 1000 or 100,000.)

ii. For a cohort study: Absolute event rates over time (eg 10 years) among exposed and

non-exposed groups; RRR (relative risk reduction.)

iii. For a case control study:OR (odds ratio) for strength of association between exposure

and outcome.

iv. For a study of a diagnostic test: Sensitivity and specificity; PPV and NPV (positive and

negative predictive values.)

v. For a systematic review and/or meta-analysis: Point estimates and confidence intervals

for the main results; one or more references for the statistical package(s) used to analyse the

data, eg RevMan for a systematic review. There is no need to provide a formal reference for a

very widely used package that will be very familiar to general readers eg STATA, but please say

in the text which version you used. For articles that include explicit statements of the quality of

evidence and strength of recommendations, we prefer reporting using the GRADE system.

f. Discussion: To minimise the risk of careful explanation giving way to polemic, please

write the discussion section of your paper in a structured way. Please follow this structure: i)

statement of principal findings of the study; ii) strengths and weaknesses of the study; iii)

strengths and weaknesses in relation to other studies, discussing important differences in

results; iv) what your study adds (whenever possible please discuss your study in the light of

relevant systematic reviews and meta-analyses); v) meaning of the study, including possible

explanations and implications for clinicians and policymakers and other researchers; vi) how

your study could promote better decisions; vi) unanswered questions and future research

g. Footnotes and statements

Online and print publication: All original research in The BMJ is published with open access. Our

open access policy is detailed here: http://www.bmj.com/about-bmj/resources-authors/forms-

policies-and-checklists/copyright-open-access-and-permission-reuse. The full text online

version of your article, if accepted after revision, will be the indexed citable version (full details

are at http://resources.bmj.com/bmj/about-bmj/the-bmjs-publishing-model). The print and

iPad BMJ will carry an abridged version of your article. This abridged version of the article is

essentially an evidence abstract called BMJ pico, which we would like you to write using the

template downloadable at http://resources.bmj.com/bmj/authors/bmj-pico. Publication of

21

research on bmj.com is definitive and is not simply interim "epublication ahead of print", so if

you do not wish to abridge your article using BMJ pico, you will be able to opt for online only

publication. Please let us know if you would prefer this option. If your article is accepted we will

invite you to submit a video abstract, lasting no longer than 4 minutes, and based on the

information in your paper’s BMJ pico evidence abstract. The content and focus of the video

must relate directly to the study that has been accepted for publication by The BMJ, and should

not stray beyond the data.