The full content of Annals is available to subscribers

Subscribe/Learn More  >
Original Research |

Opioid Prescribing After Nonfatal Overdose and Association With Repeated Overdose: A Cohort StudyOpioid Prescribing After Nonfatal Overdose

Marc R. Larochelle, MD, MPH; Jane M. Liebschutz, MD, MPH; Fang Zhang, PhD; Dennis Ross-Degnan, ScD; and J. Frank Wharam, MB, BCh, BAO, MPH
[+] Article, Author, and Disclosure Information

This article was published online first at www.annals.org on 29 December 2015.

From Harvard Medical School, Harvard Pilgrim Health Care Institute, Boston University School of Medicine, and Boston Medical Center, Boston, Massachusetts.

Grant Support: Dr. Larochelle was supported by the Health Resources and Services Administration (grants T32 HP10251 and T32 HP12706), Ryoichi Sasakawa Fellowship Fund, and Harvard Pilgrim Health Care Institute.

Disclosures: Dr. Larochelle reports grants from the Health Resources and Services Administration and Ryoichi Sasakawa Fellowship Fund and nonfinancial support from the Harvard Pilgrim Health Care Institute during the conduct of the study. Authors not named here have disclosed no conflicts of interest. Forms can be viewed at www.acponline.org/authors/icmje/ConflictOfInterestForms.do?msNum=M15-0038.

Editors' Disclosures: Christine Laine, MD, MPH, Editor in Chief, reports that she has no financial relationships or interests to disclose. Darren B. Taichman, MD, PhD, Executive Deputy Editor, reports that he has no financial relationships or interests to disclose. Cynthia D. Mulrow, MD, MSc, Senior Deputy Editor, reports that she has no relationships or interests to disclose. Deborah Cotton, MD, MPH, Deputy Editor, reports that she has no financial relationships or interest to disclose. Jaya K. Rao, MD, MHS, Deputy Editor, reports that she has stock holdings/options in Eli Lilly and Pfizer. Sankey V. Williams, MD, Deputy Editor, reports that he has no financial relationships or interests to disclose. Catharine B. Stack, PhD, MS, Deputy Editor for Statistics, reports that she has stock holdings in Pfizer.

Reproducible Research Statement:Statistical code: Available from Dr. Larochelle (e-mail, marc.larochelle@bmc.org). Study protocol and data set: Not available.

Requests for Single Reprints: Marc R. Larochelle, MD, MPH, Boston Medical Center, 801 Massachusetts Avenue, 2nd Floor, Boston, MA 02118; e-mail, marc.larochelle@bmc.org.

Current Author Addresses: Drs. Larochelle and Liebschutz: Boston Medical Center, 801 Massachusetts Avenue, 2nd Floor, Boston, MA 02118.

Dr. Zhang, Mr. Ross-Degnan, and Mr. Wharam: Department of Population Medicine, Harvard Medical School, 133 Brookline Avenue, 6th Floor, Boston, MA 02215.

Author Contributions: Conception and design: M.R. Larochelle, J.M. Liebschutz, D. Ross-Degnan, J.F. Wharam.

Analysis and interpretation of the data: M.R. Larochelle, J.M. Liebschutz, D. Ross-Degnan.

Drafting of the article: M.R. Larochelle, J.F. Wharam.

Critical revision of the article for important intellectual content: M.R. Larochelle, J.M. Liebschutz, F. Zhang, D. Ross-Degnan, J.F. Wharam.

Final approval of the article: M.R. Larochelle, J.M. Liebschutz, F. Zhang, D. Ross-Degnan, J.F. Wharam.

Statistical expertise: F. Zhang.

Obtaining of funding: J.F. Wharam.

Administrative, technical, or logistic support: M.R. Larochelle, J.F. Wharam.

Collection and assembly of data: M.R. Larochelle, J.F. Wharam.

Ann Intern Med. 2016;164(1):1-9. doi:10.7326/M15-0038
Text Size: A A A

Background: Nonfatal opioid overdose is an opportunity to identify and treat substance use disorders, but treatment patterns after the overdose are unknown.

Objective: To determine prescribed opioid dosage after an opioid overdose and its association with repeated overdose.

Design: Retrospective cohort study.

Setting: A large U.S. health insurer.

Participants: 2848 commercially insured patients aged 18 to 64 years who had a nonfatal opioid overdose during long-term opioid therapy for noncancer pain between May 2000 and December 2012.

Measurements: Nonfatal opioid overdose was identified using International Classification of Diseases, Ninth Revision, Clinical Modification, codes from emergency department or inpatient claims. The primary outcome was daily morphine-equivalent dosage (MED) of opioids dispensed from 60 days before to up to 730 days after the index overdose. We categorized dosages as large (≥100 mg MED), moderate (50 to <100 mg MED), low (<50 mg MED), or none (0 mg MED). Secondary outcomes included time to repeated overdose stratified by daily dosage as a time-varying covariate.

Results: Over a median follow-up of 299 days, opioids were dispensed to 91% of patients after an overdose. Seven percent of patients (n = 212) had a repeated opioid overdose. At 2 years, the cumulative incidence of repeated overdose was 17% (95% CI, 14% to 20%) for patients receiving high dosages of opioids after the index overdose, 15% (CI, 10% to 21%) for those receiving moderate dosages, 9% (CI, 6% to 14%) for those receiving low dosages, and 8% (CI, 6% to 11%) for those receiving no opioids.

Limitation: The cohort was limited to commercially insured adults.

Conclusion: Almost all patients continue to receive prescription opioids after an overdose. Opioid discontinuation after overdose is associated with lower risk for repeated overdose.

Primary Funding Source: Health Resources and Services Administration.


Grahic Jump Location
Figure 1.

Definition of long-term opioid episodes and daily opioid dosage attribution.

MED = morphine-equivalent dose.

Grahic Jump Location
Grahic Jump Location
Appendix Figure 2.

Daily opioid dosage before and after index overdose.

MED = morphine-equivalent dose.

Grahic Jump Location
Grahic Jump Location
Figure 2.

Movement among categories of opioid dosage over time.

The arrows and values depict the percentage of movement from 1 dosage category to another, with the solid arrow representing the path with the most patients. Percentages may not sum to 100 due to rounding. MED = morphine-equivalent dose.

Grahic Jump Location
Grahic Jump Location
Appendix Figure 3.

Proportion of patients with an active benzodiazepine or buprenorphine dispensing on each day after overdose, stratified by whether the patient also had an active opioid dispensing on that respective day.

Grahic Jump Location
Grahic Jump Location
Figure 3.

Extended Kaplan–Meier cumulative incidence curves for repeated opioid overdose and claim for drug withdrawal after overdose stratified by category of opioid dosage (large, moderate, low, or none).

Grahic Jump Location




Citing articles are presented as examples only. In non-demo SCM6 implementation, integration with CrossRef’s "Cited By" API will populate this tab (http://www.crossref.org/citedby.html).


Submit a Comment/Letter
Concern Regarding Extended Kaplan-Meier Analysis
Posted on February 17, 2016
Charles T. Sullivan
NF/SG Veterans Health System
Conflict of Interest: None Declared
An extended Kaplan-Meier approach was used to evaluate the influence of time-varying dosage (by category) on the number of days until another overdose, to establish cumulative incidence curves.

This approach has been shown to be flawed(1). Kaplan-Meier estimates are suitable for time-invariant covariates unless strict criteria are met (conditions with regard to state occupation). As noted, transition probabilities affect the hazards and are vital to this type of analysis(1).

It is worth noting that the Kaplan-Meier method used in this report has been used in many others (ref 2 for example). To my knowledge, its critique is not recognized in any of them.

There was an event rate of 7% and 72% of the cohort seems to have been lost to follow-up, or censored for one reason or another. Neither the report nor the literature suggests that this would be uninformative. The event and risk sets for the Kaplan-Meier calculations are due to state transitions where individuals vary according to dosage and presence or absence in the response function. Changes in these sets are governed by censoring, subject traits, prior treatment, health status, addiction, and so forth. Because these transitions are not dealt with, as they might be in a marginal structural model for example, we cannot derive much information from the Kaplan-Meier analysis.

As pointed out, it is more accurate to interpret the Kaplan-Meier plot as a probability of being in the response function(1). How an individual got there is unresolved. The results from the Cox model support this ambiguity is present. Note that the authors rather smartly included dosage prior to index, sex, age, mental health diagnosis, and substance abuse diagnosis into the model because they are tied overdose events. But here, none of these factors emerged significant. However, the hazard ratios for daily dose after index mirrored the Kaplan-Meier plot and reaffirmed that large intake, or “too much”, is an overdose.


Charles T. Sullivan

The opinions and content above are those of the author and they do not represent the views of the United States Department of Veterans Affairs or the United States Government.

1. Beyersmann J, Gerds T, Schumacher M. Letter to the editor: comment on ‘Illustrating the impact of a time-varying covariate with an extended Kaplan-Meier estimator’ by Steven Snapinn, Qi Jiang, and Boris Iglewicz in the November 2005 issue of The American Statistician. Am Stat. 2006;60(3):295-296.

2. Vigen R, O'Donnell CI, Baron AE, Grunwald GK, Maddox TM, Bradley SM, et al. Association of testosterone therapy with mortality, myocardial infarction, and stroke in men with low testosterone levels. JAMA. 2013;310(17):1829-36.
Author's Response
Posted on March 15, 2016
Marc R. Larochelle, MD, MPH, Jane M. Liebschutz, MD, MPH, Fang Zhang, PhD, Dennis Ross-Degnan, ScD, J. Frank Wharam, MB, BCh, BAO, MPH
Boston Medical Center
Conflict of Interest: None Declared
We appreciate the comment from Dr. Sullivan posted February 17, 2016 regarding the use of an extended Kaplan-Meier estimator to generate cumulative incidence curves for repeated overdose by time varying opioid dosage category. We chose to employ the extended Kaplan-Meier approach to provide a descriptive summary, both graphical and numerical, of the association between daily opioid dosage category after an opioid overdose and risk of repeated overdose. Traditional Kaplan-Meier approaches require a time-invariant covariate for each subject. Snapinn et al. summarize well the flaws of past approaches that have chosen a single time point (i.e. baseline, intermediate or end) for time-varying covariates and compare that to their suggested extended Kaplan-Meier approach we employed in this study (1). Opioid prescriptions are typically dispensed monthly and may be adjusted frequently, thus there is potential for great variation in dosage over a long follow-up period such as the two years used in this study. The extended Kaplan-Meier approach allowed us to depict the cumulative incidence of repeated overdose based on the “dosage on hand” at each day in the follow-up period.

Despite the advantages stated, as with any analysis, the results need to be interpreted with caution based on the limitations inherent to the approach. In the case of the extended Kaplan-Meier approach Dr. Sullivan’s comment and related citation discuss several concerns. First, the extended Kaplan-Meier results do not reflect fixed cohorts of subjects as they are allowed to move between categories at each time point. Large movements in or out of categories over time along with shifts in risk level over time may lead to inappropriate interpretations of the cumulative incidence curves as presented. Snapinn et al. recognize this limitation and suggest a more appropriate term may be “pseudo percent of patients with an event” (2). Secondly, the extended Kaplan-Meier approach – as with standard Kaplan-Meier approach – is an “unadjusted” analysis that does not take into account potential confounders of the relationship. As pointed out by Dr. Sullivan, there are many factors that may influence movement from one opioid dosage category to the next. We did build a complementary Cox-regression model taking into account several baseline member characteristics that reassuringly presented very similar results to the extended Kaplan-Meier approach. We agree with Dr. Sullivan that the dosage transition periods (e.g. from moderate to high dosage) are of interest and should be pursued in future work. Finally, there is the potential for informative censoring whereby there are differential reasons and rates of member dropout by opioid dosage category. As a further sensitivity analysis, we reran the Cox-regression model, limiting the sample to 1406 members without censoring in the first 365 days of follow-up. The effect estimates of the association between opioid dosage category and risk of repeated overdose are similar to the original analysis, suggesting this type of informative censoring is less likely to have influenced our results (Table).

In summary, we recognize there are limitations to the extended Kaplan-Meier approach employed in our manuscript; however, when interpreted in the context of these limitations, we feel it offers readers a useful graphical and numerical summary of the association between daily opioid dosage and risk of a repeated overdose over the follow-up period. Given the limitations herein, we emphasize, as we do in the manuscript, that the relationship presented is associative and not causal. Further work addressing these limitations is needed to move toward causality.

Marc R. Larochelle, MD, MPH
Jane M. Liebschutz, MD, MPH
Fang Zhang, PhD
Dennis Ross-Degnan, ScD
J. Frank Wharam, MB, BCh, BAO, MPH
Submit a Comment/Letter

Summary for Patients

Clinical Slide Sets

Terms of Use

The In the Clinic® slide sets are owned and copyrighted by the American College of Physicians (ACP). All text, graphics, trademarks, and other intellectual property incorporated into the slide sets remain the sole and exclusive property of the ACP. The slide sets may be used only by the person who downloads or purchases them and only for the purpose of presenting them during not-for-profit educational activities. Users may incorporate the entire slide set or selected individual slides into their own teaching presentations but may not alter the content of the slides in any way or remove the ACP copyright notice. Users may make print copies for use as hand-outs for the audience the user is personally addressing but may not otherwise reproduce or distribute the slides by any means or media, including but not limited to sending them as e-mail attachments, posting them on Internet or Intranet sites, publishing them in meeting proceedings, or making them available for sale or distribution in any unauthorized form, without the express written permission of the ACP. Unauthorized use of the In the Clinic slide sets will constitute copyright infringement.


Buy Now for $32.00

to gain full access to the content and tools.

Want to Subscribe?

Learn more about subscription options

Related Articles
Journal Club
Related Point of Care
Topic Collections
PubMed Articles
Forgot your password?
Enter your username and email address. We'll send you a reminder to the email address on record.