[Skip to Content]
[Skip to Content Landing]
Figure 1.
Documentation in the Medical Office of the XXIst Century Electronic Health Record (MOXXI EHR)
Documentation in the Medical Office of the XXIst Century Electronic Health Record (MOXXI EHR)

If a medication (eg, gabapentin) is discontinued, the physician must choose a reason for discontinuing use of the drug. If the reason selected is an adverse drug event, the reaction is documented. TID indicates 3 times per day.

Figure 2.
Cumulative Hazard Ratios (HRs) of Adverse Drug Events (ADEs) According to On-label and Off-label Use
Cumulative Hazard Ratios (HRs) of Adverse Drug Events (ADEs) According to On-label and Off-label Use

The comparison has not been adjusted for drug and patient characteristics (HR, 1.48; 95% CI, 1.35-1.64). After adjustment for such variables on multivariate analysis and taking into account the correlation within the patient, the difference between the 2 groups was statistically significant (adjusted HR, 1.44; 95% CI, 1.30-1.60; P < .001).

Figure 3.
Cumulative Hazard Ratios (HRs) of Adverse Drug Events (ADEs) According to On-label and Off-label Use With and Without Strong Scientific Evidence
Cumulative Hazard Ratios (HRs) of Adverse Drug Events (ADEs) According to On-label and Off-label Use With and Without Strong Scientific Evidence

The comparison has not been adjusted for drug and patient characteristics. The unadjusted HRs (95% CI) for off-label use with and without strong scientific evidence were 1.04 (0.84-1.30) and 1.62 (1.45-1.80), respectively. After adjustment for such variables on multivariate analysis and taking into account the correlation within the patient, the adjusted HR (95% CI) for off-label use with and without strong scientific evidence were 1.10 (0.88-1.38; P = .40) and 1.54 (1.37-1.72; P < .001), respectively.

Table 1.  
Drug and Patient Characteristics
Drug and Patient Characteristics
Table 2.  
Association of ADEs and Drug and Patient Characteristics
Association of ADEs and Drug and Patient Characteristics
1.
Radley  DC, Finkelstein  SN, Stafford  RS.  Off-label prescribing among office-based physicians.  Arch Intern Med. 2006;166(9):1021-1026.PubMedGoogle ScholarCrossref
2.
Eguale  T, Buckeridge  DL, Winslade  NE, Benedetti  A, Hanley  JA, Tamblyn  R.  Drug, patient, and physician characteristics associated with off-label prescribing in primary care.  Arch Intern Med. 2012;172(10):781-788.PubMedGoogle ScholarCrossref
3.
Connolly  HM, Crary  JL, McGoon  MD,  et al.  Valvular heart disease associated with fenfluramine-phentermine.  N Engl J Med. 1997;337(9):581-588.PubMedGoogle ScholarCrossref
4.
Flowers  CM, Racoosin  JA, Kortepeter  C.  Seizure activity and off-label use of tiagabine.  N Engl J Med. 2006;354(7):773-774.PubMedGoogle ScholarCrossref
5.
US Food and Drug Administration. FDA issues second warning against treating leg cramps with quinine. http://www.medscape.com/viewarticle/724798. Published July 8, 2010. Accessed September 28. 2015.
6.
Levi  M, Levy  JH, Andersen  HF, Truloff  D.  Safety of recombinant activated factor VII in randomized clinical trials.  N Engl J Med. 2010;363(19):1791-1800.PubMedGoogle ScholarCrossref
7.
Avorn  J, Kesselheim  A.  A hemorrhage of off-label use.  Ann Intern Med. 2011;154(8):566-567.PubMedGoogle ScholarCrossref
8.
Waller  PC, Evans  SJ.  A model for the future conduct of pharmacovigilance.  Pharmacoepidemiol Drug Saf. 2003;12(1):17-29.PubMedGoogle ScholarCrossref
9.
Schiff  GD, Rucker  TD.  Computerized prescribing: building the electronic infrastructure for better medication usage.  JAMA. 1998;279(13):1024-1029.PubMedGoogle ScholarCrossref
10.
Institute of Medicine, Committee on the Assessment of the US Drug Safety System. Baciu A, Stratton K, Burke SP, eds.  The Future of Drug Safety: Promoting and Protecting the Health of the Public. Washington, DC: National Academies Press; 2006.
11.
Dal Pan  GJ.  Monitoring the safety of medicines used off-label.  Clin Pharmacol Ther. 2012;91(5):787-795.PubMedGoogle ScholarCrossref
12.
Eguale  T, Winslade  N, Hanley  JA, Buckeridge  DL, Tamblyn  R.  Enhancing pharmacosurveillance with systematic collection of treatment indication in electronic prescribing: a validation study in Canada.  Drug Saf. 2010;33(7):559-567.PubMedGoogle ScholarCrossref
13.
Eguale  T, Tamblyn  R, Winslade  N, Buckeridge  D.  Detection of adverse drug events and other treatment outcomes using an electronic prescribing system.  Drug Saf. 2008;31(11):1005-1016.PubMedGoogle ScholarCrossref
14.
Stafford  RS.  Regulating off-label drug use: rethinking the role of the FDA.  N Engl J Med. 2008;358(14):1427-1429.PubMedGoogle ScholarCrossref
15.
Emmerich  J, Dumarcet  N, Lorence  A.  France’s new framework for regulating off-label drug use.  N Engl J Med. 2012;367(14):1279-1281.PubMedGoogle ScholarCrossref
16.
Tamblyn  R, Huang  A, Perreault  R,  et al.  The Medical Office of the 21st Century (MOXXI): effectiveness of computerized decision-making support in reducing inappropriate prescribing in primary care.  CMAJ. 2003;169(6):549-556.PubMedGoogle Scholar
17.
Tamblyn  R, Eguale  T, Huang  A, Winslade  N, Doran  P.  The incidence and determinants of primary nonadherence with prescribed medication in primary care: a cohort study.  Ann Intern Med. 2014;160(7):441-450.PubMedGoogle ScholarCrossref
18.
Health Canada.  Drug Product Database. 2009. http://www.hc-sc.gc.ca/dhp-mps/prodpharma/databasdon/index-eng.php. Modified June 18, 2015. Accessed September 28, 2015.
19.
Walton  SM, Schumock  GT, Lee  KV, Alexander  GC, Meltzer  D, Stafford  RS.  Prioritizing future research on off-label prescribing: results of a quantitative evaluation.  Pharmacotherapy. 2008;28(12):1443-1452.PubMedGoogle ScholarCrossref
20.
Hutchinson  TA, Flegel  KM, Kramer  MS, Leduc  DG, Kong  HH.  Frequency, severity and risk factors for adverse drug reactions in adult out-patients: a prospective study.  J Chronic Dis. 1986;39(7):533-542.PubMedGoogle ScholarCrossref
21.
Budnitz  DS, Pollock  DA, Weidenbach  KN, Mendelsohn  AB, Schroeder  TJ, Annest  JL.  National surveillance of emergency department visits for outpatient adverse drug events.  JAMA. 2006;296(15):1858-1866.PubMedGoogle ScholarCrossref
22.
Weber  JCP. Epidemiology of adverse reactions to nonsteroidal anti-inflammatory drugs. In: Rainsford  KD, Velo  GP, eds.  Advances in Inflammation Research. New York, NY: Raven Press; 1984:1-6.
23.
Charlson  ME, Pompei  P, Ales  KL, MacKenzie  CR.  A new method of classifying prognostic comorbidity in longitudinal studies: development and validation.  J Chronic Dis. 1987;40(5):373-383.PubMedGoogle ScholarCrossref
24.
Zhang  M, Holman  CD, Price  SD, Sanfilippo  FM, Preen  DB, Bulsara  MK.  Comorbidity and repeat admission to hospital for adverse drug reactions in older adults: retrospective cohort study.  BMJ. 2009;338:a2752.PubMedGoogle ScholarCrossref
25.
Martin  RM, Biswas  PN, Freemantle  SN, Pearce  GL, Mann  RD.  Age and sex distribution of suspected adverse drug reactions to newly marketed drugs in general practice in England: analysis of 48 cohort studies.  Br J Clin Pharmacol. 1998;46(5):505-511.PubMedGoogle ScholarCrossref
26.
Pouyanne  P, Haramburu  F, Imbs  JL, Bégaud  B; French Pharmacovigilance Centres.  Admissions to hospital caused by adverse drug reactions: cross sectional incidence study.  BMJ. 2000;320(7241):1036.PubMedGoogle ScholarCrossref
27.
Jacubeit  T, Drisch  D, Weber  E.  Risk factors as reflected by an intensive drug monitoring system.  Agents Actions Suppl. 1990;29:117-125.PubMedGoogle ScholarCrossref
28.
Sarkar  U, López  A, Maselli  JH, Gonzales  R.  Adverse drug events in US adult ambulatory medical care.  Health Serv Res. 2011;46(5):1517-1533.PubMedGoogle ScholarCrossref
29.
Gill  JM, Mainous  AG  III, Nsereko  M.  The effect of continuity of care on emergency department use.  Arch Fam Med. 2000;9(4):333-338.PubMedGoogle ScholarCrossref
30.
Tamblyn  R, Abrahamowicz  M, Dauphinee  D,  et al.  Physician scores on a national clinical skills examination as predictors of complaints to medical regulatory authorities.  JAMA. 2007;298(9):993-1001.PubMedGoogle ScholarCrossref
31.
Chuang  SK, Wei  LJ, Douglass  CW, Dodson  TB.  Risk factors for dental implant failure: a strategy for the analysis of clustered failure-time observations.  J Dent Res. 2002;81(8):572-577.PubMedGoogle ScholarCrossref
32.
Bellera  CA, MacGrogan  G, Debled  M, de Lara  CT, Brouste  V, Mathoulin-Pélissier  S.  Variables with time-varying effects and the Cox model: some statistical concepts illustrated with a prognostic factor study in breast cancer.  BMC Med Res Methodol. 2010;10:20.PubMedGoogle ScholarCrossref
33.
Chen  DT, Wynia  MK, Moloney  RM, Alexander  GC.  US physician knowledge of the FDA-approved indications and evidence base for commonly prescribed drugs: results of a national survey.  Pharmacoepidemiol Drug Saf. 2009;18(11):1094-1100.PubMedGoogle ScholarCrossref
34.
Schiff  GD.  Computerized prescriber order entry: models and hurdles.  Am J Health Syst Pharm. 2002;59(15):1456-1460.PubMedGoogle Scholar
35.
Zopf  Y, Rabe  C, Neubert  A, Hahn  EG, Dormann  H.  Risk factors associated with adverse drug reactions following hospital admission: a prospective analysis of 907 patients in two German university hospitals.  Drug Saf. 2008;31(9):789-798.PubMedGoogle ScholarCrossref
36.
Steinman  MA, Landefeld  CS, Rosenthal  GE, Berthenthal  D, Sen  S, Kaboli  PJ.  Polypharmacy and prescribing quality in older people.  J Am Geriatr Soc. 2006;54(10):1516-1523.PubMedGoogle ScholarCrossref
37.
Johnell  K, Klarin  I.  The relationship between number of drugs and potential drug-drug interactions in the elderly: a study of over 600 000 elderly patients from the Swedish Prescribed Drug Register.  Drug Saf. 2007;30(10):911-918.PubMedGoogle ScholarCrossref
38.
Corsonello  A, Pedone  C, Corica  F, Mussi  C, Carbonin  P, Antonelli Incalzi  R; Gruppo Italiano di Farmacovigilanza nell’Anziano (GIFA) Investigators.  Concealed renal insufficiency and adverse drug reactions in elderly hospitalized patients.  Arch Intern Med. 2005;165(7):790-795.PubMedGoogle ScholarCrossref
39.
Bates  DW, Miller  EB, Cullen  DJ,  et al; ADE Prevention Study Group.  Patient risk factors for adverse drug events in hospitalized patients.  Arch Intern Med. 1999;159(21):2553-2560.PubMedGoogle ScholarCrossref
40.
Bégaud  B, Martin  K, Fourrier  A, Haramburu  F.  Does age increase the risk of adverse drug reactions?  Br J Clin Pharmacol. 2002;54(5):550-552.PubMedGoogle ScholarCrossref
41.
Gurwitz  JH, Avorn  J.  The ambiguous relation between aging and adverse drug reactions.  Ann Intern Med. 1991;114(11):956-966.PubMedGoogle ScholarCrossref
42.
Galanter  W, Falck  S, Burns  M, Laragh  M, Lambert  BL.  Indication-based prescribing prevents wrong-patient medication errors in computerized provider order entry (CPOE).  J Am Med Inform Assoc. 2013;20(3):477-481.PubMedGoogle ScholarCrossref
43.
Forster  AJ, Auger  C; ISTOP ADE Investigators.  Using information technology to improve the monitoring of outpatient prescribing.  JAMA Intern Med. 2013;173(5):382-384.PubMedGoogle ScholarCrossref
44.
Gandhi  TK, Weingart  SN, Borus  J,  et al.  Adverse drug events in ambulatory care.  N Engl J Med. 2003;348(16):1556-1564.PubMedGoogle ScholarCrossref
45.
Classen  DC, Bates  DW.  Finding the meaning in meaningful use.  N Engl J Med. 2011;365(9):855-858.PubMedGoogle ScholarCrossref
46.
Rodríguez-Monguió  R, Otero  MJ, Rovira  J.  Assessing the economic impact of adverse drug effects.  Pharmacoeconomics. 2003;21(9):623-650.PubMedGoogle ScholarCrossref
47.
Al Hamid  A, Ghaleb  M, Aljadhey  H, Aslanpour  Z.  A systematic review of hospitalization resulting from medicine-related problems in adult patients.  Br J Clin Pharmacol. 2014;78(2):202-217.PubMedGoogle ScholarCrossref
Original Investigation
Health Care Reform
January 2016

Association of Off-label Drug Use and Adverse Drug Events in an Adult Population

Author Affiliations
  • 1Department of Epidemiology, Biostatistics, and Occupational Health, McGill University, Montreal, Quebec, Canada
  • 2Department of Medicine, McGill University, Montreal, Quebec, Canada
  • 3Division of General Medicine and Primary Care, Brigham and Women’s Hospital, Harvard Medical School, Boston, Massachusetts
  • 4Department of Social and Administrative Pharmacy, Schoool of Pharmacy, MCPHS University, Boston, Massachusetts
  • 5Department of Mathematics and Statistics, McGill University, Montreal, Quebec, Canada
JAMA Intern Med. 2016;176(1):55-63. doi:10.1001/jamainternmed.2015.6058
Abstract

Importance  Off-label use of prescription drugs has been identified as an important contributor to preventable adverse drug events (ADEs) in children. Despite concerns regarding adverse outcomes, to date, no systematic investigation of the effects of off-label drug use in adult populations has been performed.

Objective  To monitor and evaluate off-label use of prescription drugs and its effect on ADEs in an adult population.

Design, Setting, and Participants  A cohort of 46 021 patients who received 151 305 incident prescribed drugs was assembled from primary care clinics in Quebec, Canada, using the Medical Office of the XXIst Century electronic health record, which supports documentation of treatment indications and treatment outcomes. Prescriptions dispensed from January 1, 2005, through December 30, 2009, were followed up from the date of the prescription to the date the drug use was discontinued, the end of treatment, or the end of follow-up (December 30, 2010). Data were analyzed from January 5, 2012, to March 15, 2015.

Exposures  Off-label prescription drug use with and without strong scientific evidence.

Main Outcomes and Measures  Adverse drug events in off-label use with and without strong scientific evidence. Analysis used multivariate marginal Cox proportional hazards regression for clustered data with the drug as the unit of analysis.

Results  A total of 3484 ADEs were found in the 46 021 study patients, with an incidence rate of 13.2 per 10 000 person-months. The rate of ADEs for off-label use (19.7 per 10 000 person-months) was higher than that for on-label use (12.5 per 10 000 person-months) (adjusted hazard ratio [AHR], 1.44; 95% CI, 1.30-1.60). Off-label use lacking strong scientific evidence had a higher ADE rate (21.7 per 10 000 person-months) compared with on-label use (AHR, 1.54; 95% CI, 1.37-1.72). However, off-label use with strong scientific evidence had the same risk for ADEs as on-label use (AHR, 1.10; 95% CI, 0.88-1.38). The risks for ADEs were higher for drugs approved from 1981 to 1995 (14.4 per 10 000 person-months; AHR, 1.62; 95% CI, 1.45-1.80) and for those used by women (14.3 per 10 000 person-months; AHR, 1.17; 95% CI, 1.06-1.28), patients receiving 5 to 7 drugs (12.1 per 10 000 person-months; AHR, 3.23; 95% CI, 2.66-3.92), and patients receiving cardiovascular drugs (15.9 per 10 000 person-months; AHR, 3.30; 95% CI, 2.67-4.08) and anti-infectives (66.2 per 10 000 person-months; AHR, 6.33; 95% CI, 4.58-8.76). Patients with a 1-unit increase in the continuity of care index had a 19% increase in ADEs (AHR, 1.19; 95% CI, 1.12-1.26).

Conclusions and Relevance  Off-label use of prescription drugs is associated with ADEs. Caution should be exercised in prescribing drugs for off-label uses that lack strong scientific evidence. Future electronic health records should be designed to enable postmarket surveillance of treatment indications and treatment outcomes to monitor the safety of on- and off-label uses of drugs.

Introduction

Off-label prescribing of drugs is common1,2 and has been identified as a potentially important contributor to preventable adverse drug events (ADEs). Significant deleterious effects can occur with off-label use of some drugs, such as cardiac valve damage with fenfluramine and phentermine (fen-phen),3 status epilepticus with tiagabine hydrochloride,4 thrombocytopenia with quinine sulfate,5 and thromboembolic events with recombinant factor VIIa.6,7 Despite concerns about adverse outcomes, no systematic investigation of the effect of real-life off-label use on adverse events has been performed.

The Institute of Medicine and drug regulatory agencies have envisioned a postmarket surveillance system in which patterns of drug use, indications for the use (on- and off-label), and associated ADEs could be tracked.810 However, an explicit link between prescribed drugs and their indication is rarely documented, making it challenging to measure off-label use and its effects.11 Recently, methods to support the documentation of treatment indication, reasons for discontinuation of treatment orders, and ADEs have been developed, and evidence supports requiring these elements in electronic prescription.12,13 These new features provide a unique opportunity to monitor and evaluate off-label use and its effect on ADEs systematically. The results from such monitoring can provide important guidance to initiatives aimed at regulating off-label use and strengthening the role of drug regulatory bodies (eg, the US Food and Drug Administration and Health Canada).14,15

Methods
Context

The study was conducted in Quebec, Canada, a province that provides health insurance for all 8.5 million residents. In 2003, the Medical Office of the XXIst Century (MOXXI), a community-based clinical information system, was created to link beneficiary, medical billing, and pharmacy claim data to create a longitudinal electronic health record (EHR) for patients.16 Evidence of validity for the data has been shown, and these data have been used frequently for health services and epidemiologic research.17 One hundred thirteen primary care physicians participate in the MOXXI research program, and these physicians work in office-based practices in Montreal or Quebec City. The MOXXI data overrepresent older patients and patients with frequent physician visits.16

In the MOXXI EHR, physicians must document the treatment indication for each new electronic prescription, the reasons for dose changes and drug discontinuation orders, and the nature of any ADEs (Figure 1 and the eFigure in the Supplement). After a new prescription is entered through the MOXXI system, the patient may fill the prescription at a pharmacy and may experience an ADE. At the next visit to a MOXXI physician, if the prescription is not renewed, the reason for discontinuation must be documented within the MOXXI system. In this way, the reasons for all discontinuations of drug use are recorded. The reasons for discontinuation are grouped into the following 5 domains: safety (eg, adverse drug reaction, allergic response, or drug interaction), effectiveness (eg, ineffective treatment, no longer necessary), prescribing or dispensing errors, factors associated with adherence to the medication regimen (eg, simplifying treatment, substitution of a less expensive drug), and factors such as patient request and discontinuations by other physicians. The ADEs are documented using a prepopulated, drug-specific ADE list that is mapped to the Medical Dictionary for Regulatory Activities (MedDRA) classification. Physicians can also search from the global list of ADEs or can use free text to enter a specific ADE description. To support clinical decision making, treatment indications automatically populate the patient’s list of medical problems. Prior failed treatment and reasons for discontinuations of drug treatment (eg, ADE, hypotension) are listed under each indication.

Design and Study Population

To evaluate the association between off-label use and ADEs, a prospective cohort of 46 021 adult patients (aged ≥18 years) with an incident prescription for a drug was assembled from January 1, 2005, through December 30, 2009. A drug prescription was considered incident if the same drug had not been prescribed or dispensed in the past 12 months. Prescriptions were followed up from the date of the prescription to the date the drug use was discontinued, the end of treatment, or the end of follow-up (December 30, 2010). The MOXXI research program was approved by the institutional review board of McGill University. Patients and physicians provided written informed consent to be part of the research program.

Adverse Drug Events

Adverse drug events were defined as discontinuations of drug use made by physicians owing to an adverse drug reaction or an allergic reaction. Prior evaluation of the validity of these data found an 85.7% concordance between the electronic documentation of ADEs with the medical record.13 The ADEs were categorized using MedDRA, which includes system organ classes and preferred terms. The system organ classes represent the highest hierarchy that provides the broadest concept for data retrieval and analysis, and preferred terms provide a distinct descriptor of sign, symptom, diagnosis, or surgical or medical procedure.

Risk for ADEs and Off-label Use

The indication recorded for each prescription was classified as on-label or off-label use according to the Health Canada drug approval database.18 Compared with the medical record, the sensitivity (completeness) and the positive predictive value (correctness) of the documentation of electronic treatment indication in the MOXXI system were 98.5% and 97.0%, respectively.12 Indications were considered approved by Health Canada (or on-label) if they could be matched to the therapeutic indication reported in the drug’s package insert as of December 2010. Any indication that could not be matched to the labeled indication was considered off-label. For each off-label drug indication, the level of evidence supporting the drug’s overall efficacy was categorized using the DrugPoints System (Thomson Reuters). Strong evidence exists when (1) the drug is effective or favors efficacy for the off-label treatment indication, (2) the drug is recommended for at least most patients with the off-label treatment indication, and (3) the studies used to evaluate efficacy and the strength of evidence include at least 1 randomized controlled trial.2,19 Accordingly, we created the following 2 variables: (1) on- or off-label use and (2) off-label use with and without strong scientific evidence. In eTable 1 in the Supplement, we tabulated selected drugs and off-label indications with and without strong scientific evidence.

We measured drug class as a potential risk factor for ADEs because the central nervous system (CNS), cardiovascular, and anti-infective classes were more often implicated in ADEs.20,21Drug age, defined as the year the drug was approved for marketing, was included because ADEs were more likely to be reported with recently approved drugs than older drugs.22 Drug age was categorized into the following 3 groups: before 1981; 1981 through 1995; and after 1995 (recent approval).

For patient characteristics, we included age and measures of comorbidity (Charlson comorbidity index; range, 0-25; 1 or greater indicates increased risk for death)23 because older patients6,21 and those with more than 1 comorbidity24 may have a higher risk for ADEs. Patient sex was included because higher rates of ADEs were reported for women than men.25,26 The number of drugs the patient was taking was included because it was shown to be the most important risk factor for ADEs.20,27,28 The continuity of care (COC) index was included to correct for possible surveillance bias in the opportunity to detect ADEs because patients with better continuity have fewer emergent visits,29 possibly because ADEs and disease exacerbations are more likely to be detected and averted. The COC index was defined as the ratio of the number of a patient’s visits to the primary care physician to the square root of the patient’s total outpatient visits and calculated using the medical services claims.30

Statistical Analysis

Data analysis was performed from January 5, 2012, to March 15, 2015. Incidence rates of discontinuations of drug use owing to ADEs were calculated by dividing the number of incident cases by the number of person-months of follow-up, both overall and by exposure classification. Person-months were calculated per drug-patient pair starting from the first day of prescription to the discontinuation of drug use, the end of treatment, or the end of the study (December 30, 2010). The hazard ratio (HR) was calculated with the drug as the unit of analysis, and drugs were nested within patients. In univariate and multivariate analyses, we fit a marginal Cox proportional hazards regression model using a robust sandwich covariance estimator to construct 95% CIs that accounted for intracluster dependence.31 The 2 off-label variable indicators were included in the Cox proportional hazards regression model separately. Proportionality assumptions were tested by examining covariate-time interaction terms and Schoenfeld and Martingale residuals.32

Results

A total of 46 021 patients received 151 305 incident prescriptions from January 1, 2005, through December 30, 2009. Off-label use was reported in 17 847 prescriptions (11.8%); in 14 431 of these cases (80.9%), the off-label uses lacked strong scientific evidence (Table 1). Mean (SD) patient age was 58.2 (16.8) years. Patients were predominantly women (27 982 [60.8%]); 11 383 patients (24.7%) had a Charlson comorbidity index of 1 or greater, and 19 342 (42.0%) used more than 4 unique drugs during the study period. Of the 630 prescribed drugs, 292 (46.3%) had been approved for market before 1981, whereas 180 (28.6%) were approved after 1995. The mean (SD) COC index was 1.09 (0.64). The follow-up time for a prescription ranged from 1 day to almost 6 years (median, 386 days; mean, 530 days).

The distribution of off-label use (with and without strong scientific evidence) by American Hospital Formulary Service drug class is shown in eTable 2 in the Supplement. For the 38 370 CNS drugs found, off-label use and off-label use without scientific evidence were reported in 9814 of 38 370 prescribed drugs (25.6%) and 8157 prescribed drugs (21.3%), respectively. Among the CNS drugs, off-label use was highest for anticonvulsants (2158 [65.6%]), antipsychotics (669 [51.4%]), and antidepressants (3657 [33.5%]); off-label use without strong scientific evidence for these drugs was reported in 2116 (64.3%), 668 (51.3%), and 3063 (28.1%), respectively.

Physicians discontinued 3484 drug treatments owing to ADEs. Most ADEs (2456 [70.5%]) occurred in the first year of treatment and, among these, 1228 (50%) occurred in the first 3 months. eTable 3 in the Supplement shows selected drugs with high levels of off-label use and the proportion of drugs with discontinuation of use owing to ADEs. The overall incidence rate of ADEs for all drugs was 13.2 per 10 000 person-months. The ADE rate for on-label use was lower (12.5 per 10 000 person-months) than for off-label use (19.7 per 10 000 person-months), a 44% increase in the risk for ADEs with off-label use (adjusted HR [AHR], 1.44; 95% CI, 1.30-1.60) (Table 2 and Figure 2). When off-label use was stratified to off-label use with and without strong scientific evidence, the ADE rates were 13.2 and 21.7 per 10 000 person-months, respectively. Compared with on-label use, the AHRs (95% CI) for off-label use with and without strong scientific evidence were 1.10 (0.88-1.38) and 1.54 (1.37-1.72), respectively (Table 2 and Figure 3). Proportionality assumptions for the HRs were not violated. The median time to discontinue drug use owing to an ADE was shorter for drugs used off-label (146 days) compared with those used on-label (179 days).

Anti-infectives had the highest ADE rate (66.2 per 10 000 person-months), more than a 6-fold increase in risk for ADEs compared with gastrointestinal tract drugs (AHR, 6.33; 95% CI, 4.58, 8.76). Central nervous system and cardiovascular drugs had ADE rates of 18.1 and 15.9 per 10 000 person-months, respectively. Drugs approved after 1995 had ADE rates that were 55% higher than drugs approved before 1981 (AHR, 1.55; 95% CI, 1.39-1.73), and the same is true for drugs approved from 1981 to 1995 (AHR, 1.62; 95% CI, 1.45-1.80) (Table 2).

Dose-response relationships were identified between the number of drugs the patient used and the risk for ADEs, in which patients using 8 or more drugs had more than a 5-fold increased risk for ADEs compared with patients using 1 to 2 drugs (AHR, 5.29; 95% CI, 4.39-6.38). Patients 65 years and older exhibited an increased risk for ADEs in an unadjusted analysis compared with younger cohorts; however, the increased risk did not persist after adjusting for the number of drugs and patients’ comorbidities. Women also had a higher risk for ADEs than men (14.3 vs 11.7 per 10 000 person-months, respectively; AHR, 1.17; 95% CI, 1.06-1.28). After adjusting for other patient characteristics, the patients with a Charlson comorbidity index of 1 or greater had a lower risk for ADEs compared with patients with an index of 0 (AHR, 0.89; 95% CI, 0.81-0.98). The COC index was an important determinant of ADE detection, with a 19% increase in ADE rate per 1-unit increase in the COC index (AHR, 1.19; 95% CI, 1.12-1.26) (Table 2). In a subanalysis, we fit a model to drugs predominantly used for chronic conditions (by removing the anti-infective group) and found a slight change in the AHR for off-label use without strong scientific evidence (1.58; 95% CI, 1.41-1.77).

The ADEs related to the gastrointestinal tract, nervous, respiratory, and musculoskeletal system organ classes were documented the most frequently (eTable 4 in the Supplement). Selected examples of ADEs associated with the most frequently used off-label drugs include akathisia resulting from the use of gabapentin for neurogenic pain; agitation associated with the use of amitriptyline hydrochloride for migraine; hallucinations with the use of trazodone hydrochloride for insomnia; QT interval prolongation with the use of quetiapine fumarate for depression; and weight gain with the use of olanzapine for depression.

Discussion

This study is, to our knowledge, the first to systematically evaluate the association between off-label use of drugs and the risk for ADEs in an adult population. The study is also the first to use electronically documented treatment indications and treatment outcomes to measure off-label drug use and ADE occurrence.2,12,13 We found that off-label use of drugs was associated with ADEs after adjusting for important patient and drug characteristics. Moreover, we noted a risk gradient with higher rates of ADEs for off-label uses lacking strong scientific evidence.

Although the intrinsic nature of the drug to cause ADEs is the same for on-label and off-label uses, it may be modified by a number of factors, including the off-label disease condition. In addition, the lack of approval from a regulatory body implies a lack of safe dose ranges and inadequate information on contraindications, which in aggregate make ADEs more likely. We found that 4 in 5 off-label prescriptions lacked strong scientific evidence, and this group had higher rates of ADEs. Off-label use may be clinically appropriate given the complexity of the patient’s condition, the lack of alternative effective drugs, or after exhausting approved drugs. However, a lack of physician knowledge of approved treatment indications33 was shown to be 1 factor for off-label prescribing. Physicians are finding it difficult to keep up with rapidly changing medication information, and this lack of knowledge is affecting treatment of patients.34 Studies20,27,28,35 have shown that the number of drugs used by a patient strongly influences the risk for ADEs owing to an increased risk for inappropriate prescribing,36 drug-drug interaction,37 and drug-disease interaction.38 Similarly to other studies,20,21,39 we found the rates of ADEs were significantly higher among patients using drugs from the anti-infective, cardiovascular, and CNS drug classes. We also found the same age effect that was reported from the French pharmacovigilance study,40 with the oldest group having a higher crude incidence rate of ADEs; however, after adjusting for the number of drug treatments, the effect disappears. Overall, this finding shows that the number of prescribed drugs is the strongest risk factor for ADEs.20,41 In our study, a higher risk for ADEs was observed for women, even after accounting for the COC index to mitigate the effect of high consultation rates among women. We also identified patients with a higher COC index as having a high risk for ADEs, possibly owing to surveillance detection biases with more opportunity to identify ADEs in their earlier stages.

Using appropriately configured computerized decision support systems, we could fill the knowledge gap in off-label prescribing by supplying clinicians with information on drug approval status and the degree of scientific evidence at the point of care. In addition, the inclusion of treatment indication in the electronic prescription would facilitate communication among physicians, pharmacists, and patients and may decrease medication errors42 (eg, wrong-patient errors and look-alike and sound-alike drugs). At the same time, these technologies could allow physicians to participate actively in the systematic collection of data to evaluate the benefits and risks associated with the off-label use of drugs. Moreover, monitoring patients in the community with a nurse or an interactive voice response system to detect drug-related problems43 and relaying this information to a community pharmacist or the treating physician would be one option of detecting and ameliorating these risks.

The study has several limitations. First, ADE identification depends on physicians and patients. Physicians are known to miss medication-related symptoms, and patients may not inform their physicians about all of their symptoms.44 Second, ADEs among patients with comorbidities may not be as easily attributed to the drug (vs the concurrent disease) and may not be identified. Third, patients with severe ADEs may visit hospitals and other physicians, and the ADEs might not be recorded in their EHR. All these limitations might have resulted in an underestimation of ADE incidence. However, the underestimation of ADE rates will likely affect on-label and off-label uses equivalently and thus not bias the association between drug approval status and ADEs.33 Although the ADEs were severe enough to warrant discontinuation of drug use, the ADEs were not documented with a severity scale, which limited subgroup analysis.

When classifying off-label use, we only considered the drug and the treatment indication. Our off-label definition did not consider the dosage, frequency, route of administration, duration of treatment, and patients’ age range. A wider definition of off-label use could have increased the prevalence of off-label use.

In recent years, drug regulatory bodies have created mechanisms to expedite the drug review process and to give the public rapid access to drugs. In addition, they have mandated the life-cycle approach to drug evaluation with ongoing pharmacovigilance and risk evaluation and mitigation strategies. For example, France has passed a law to monitor off-label drug use and study the benefit-risk ratios of these uses in collaboration with the pharmaceutical industry.15 In this study, we demonstrated that an EHR system tailored to document treatment indications and the reason for treatment discontinuation can be used to monitor off-label drug use and its outcomes. Such a system can help to tackle fundamental problems in postmarketing surveillance of drugs, namely, the lack of an explicit link between prescribed drugs and their indication for use and the underreporting of ADEs. Large-scale documentation of drug indications, ADEs, and clinical variables within the EHR can facilitate powerful observational studies, comparative safety and effectiveness trials, and decision support systems. Moreover, linking a prescribed drug with an indication and outcome (eg, ADE) could be a meaningful-use objective, thereby advancing meaningful use to meaningful benefit.45

In our study, the cost of ADEs was not measured. However, the cost can be estimated using previously published figures. Adverse effects that occur in nonhospital settings may lead to emergency department visits and hospital admissions. The mean cost of an emergency department visit without hospital admission has been estimated to range from $549 to $704 per ADE, whereas the estimated mean cost of treating an ADE resulting in hospital admission has been estimated to range from $5118 to $11 789 (in 2014 US currency).46 The median prevalence rate of hospitalizations resulting from ADEs has been estimated at 4.6%.47 The present study found 3484 ADEs in 46 021 patients; thus, the mean cost per ADE would range from $759 to $1214. Likewise, the mean cost per study patient would be $51 to $77.

Conclusions

Off-label drug use, and particularly off-label use without strong scientific evidence, is a risk factor for ADEs. Hence, physicians and physician organizations should recognize the enormity of the problem and be active participants in the promotion of cautious prescribing of drugs for off-label uses lacking strong scientific evidence. Future EHRs should be designed to enable postmarketing surveillance of treatment indications and treatment outcomes to monitor the safety of on- and off-label uses of drugs.

Back to top
Article Information

Corresponding Author: Tewodros Eguale, MD, PhD, Department of Epidemiology, Biostatistics, and Occupational Health, McGill University, 1140 Pine Ave W, Morrice House, Montreal, QC H3A 1A3, Canada (tewodros.eguale@mail.mcgill.ca and tewodros.eguale@mcphs.edu).

Accepted for Publication: September 10, 2015.

Published Online: November 2, 2015. doi:10.1001/jamainternmed.2015.6058.

Author Contributions: Drs Eguale and Tamblyn had full access to all the data in the study and take responsibility for the integrity of the data and the accuracy of the data analysis.

Study concept and design: Eguale, Buckeridge, Verma, Winslade, Hanley, Tamblyn.

Acquisition, analysis, or interpretation of data: Eguale, Buckeridge, Verma, Benedetti, Hanley, Tamblyn.

Drafting of the manuscript: Eguale, Verma, Hanley, Tamblyn.

Critical revision of the manuscript for important intellectual content: Eguale, Buckeridge, Verma, Winslade, Benedetti, Tamblyn.

Statistical analysis: Eguale, Buckeridge, Verma, Benedetti, Hanley.

Obtained funding: Tamblyn.

Administrative, technical, or material support: Verma, Winslade.

Study supervision: Buckeridge, Tamblyn.

Conflict of Interest Disclosures: None reported.

Additional Contributions: Rosa Rodriguez-Monguio, PhD, and Enrique Seoane-Vasquez, PhD, provided input in estimation of the cost of adverse drug events. Neither was financially compensated.

References
1.
Radley  DC, Finkelstein  SN, Stafford  RS.  Off-label prescribing among office-based physicians.  Arch Intern Med. 2006;166(9):1021-1026.PubMedGoogle ScholarCrossref
2.
Eguale  T, Buckeridge  DL, Winslade  NE, Benedetti  A, Hanley  JA, Tamblyn  R.  Drug, patient, and physician characteristics associated with off-label prescribing in primary care.  Arch Intern Med. 2012;172(10):781-788.PubMedGoogle ScholarCrossref
3.
Connolly  HM, Crary  JL, McGoon  MD,  et al.  Valvular heart disease associated with fenfluramine-phentermine.  N Engl J Med. 1997;337(9):581-588.PubMedGoogle ScholarCrossref
4.
Flowers  CM, Racoosin  JA, Kortepeter  C.  Seizure activity and off-label use of tiagabine.  N Engl J Med. 2006;354(7):773-774.PubMedGoogle ScholarCrossref
5.
US Food and Drug Administration. FDA issues second warning against treating leg cramps with quinine. http://www.medscape.com/viewarticle/724798. Published July 8, 2010. Accessed September 28. 2015.
6.
Levi  M, Levy  JH, Andersen  HF, Truloff  D.  Safety of recombinant activated factor VII in randomized clinical trials.  N Engl J Med. 2010;363(19):1791-1800.PubMedGoogle ScholarCrossref
7.
Avorn  J, Kesselheim  A.  A hemorrhage of off-label use.  Ann Intern Med. 2011;154(8):566-567.PubMedGoogle ScholarCrossref
8.
Waller  PC, Evans  SJ.  A model for the future conduct of pharmacovigilance.  Pharmacoepidemiol Drug Saf. 2003;12(1):17-29.PubMedGoogle ScholarCrossref
9.
Schiff  GD, Rucker  TD.  Computerized prescribing: building the electronic infrastructure for better medication usage.  JAMA. 1998;279(13):1024-1029.PubMedGoogle ScholarCrossref
10.
Institute of Medicine, Committee on the Assessment of the US Drug Safety System. Baciu A, Stratton K, Burke SP, eds.  The Future of Drug Safety: Promoting and Protecting the Health of the Public. Washington, DC: National Academies Press; 2006.
11.
Dal Pan  GJ.  Monitoring the safety of medicines used off-label.  Clin Pharmacol Ther. 2012;91(5):787-795.PubMedGoogle ScholarCrossref
12.
Eguale  T, Winslade  N, Hanley  JA, Buckeridge  DL, Tamblyn  R.  Enhancing pharmacosurveillance with systematic collection of treatment indication in electronic prescribing: a validation study in Canada.  Drug Saf. 2010;33(7):559-567.PubMedGoogle ScholarCrossref
13.
Eguale  T, Tamblyn  R, Winslade  N, Buckeridge  D.  Detection of adverse drug events and other treatment outcomes using an electronic prescribing system.  Drug Saf. 2008;31(11):1005-1016.PubMedGoogle ScholarCrossref
14.
Stafford  RS.  Regulating off-label drug use: rethinking the role of the FDA.  N Engl J Med. 2008;358(14):1427-1429.PubMedGoogle ScholarCrossref
15.
Emmerich  J, Dumarcet  N, Lorence  A.  France’s new framework for regulating off-label drug use.  N Engl J Med. 2012;367(14):1279-1281.PubMedGoogle ScholarCrossref
16.
Tamblyn  R, Huang  A, Perreault  R,  et al.  The Medical Office of the 21st Century (MOXXI): effectiveness of computerized decision-making support in reducing inappropriate prescribing in primary care.  CMAJ. 2003;169(6):549-556.PubMedGoogle Scholar
17.
Tamblyn  R, Eguale  T, Huang  A, Winslade  N, Doran  P.  The incidence and determinants of primary nonadherence with prescribed medication in primary care: a cohort study.  Ann Intern Med. 2014;160(7):441-450.PubMedGoogle ScholarCrossref
18.
Health Canada.  Drug Product Database. 2009. http://www.hc-sc.gc.ca/dhp-mps/prodpharma/databasdon/index-eng.php. Modified June 18, 2015. Accessed September 28, 2015.
19.
Walton  SM, Schumock  GT, Lee  KV, Alexander  GC, Meltzer  D, Stafford  RS.  Prioritizing future research on off-label prescribing: results of a quantitative evaluation.  Pharmacotherapy. 2008;28(12):1443-1452.PubMedGoogle ScholarCrossref
20.
Hutchinson  TA, Flegel  KM, Kramer  MS, Leduc  DG, Kong  HH.  Frequency, severity and risk factors for adverse drug reactions in adult out-patients: a prospective study.  J Chronic Dis. 1986;39(7):533-542.PubMedGoogle ScholarCrossref
21.
Budnitz  DS, Pollock  DA, Weidenbach  KN, Mendelsohn  AB, Schroeder  TJ, Annest  JL.  National surveillance of emergency department visits for outpatient adverse drug events.  JAMA. 2006;296(15):1858-1866.PubMedGoogle ScholarCrossref
22.
Weber  JCP. Epidemiology of adverse reactions to nonsteroidal anti-inflammatory drugs. In: Rainsford  KD, Velo  GP, eds.  Advances in Inflammation Research. New York, NY: Raven Press; 1984:1-6.
23.
Charlson  ME, Pompei  P, Ales  KL, MacKenzie  CR.  A new method of classifying prognostic comorbidity in longitudinal studies: development and validation.  J Chronic Dis. 1987;40(5):373-383.PubMedGoogle ScholarCrossref
24.
Zhang  M, Holman  CD, Price  SD, Sanfilippo  FM, Preen  DB, Bulsara  MK.  Comorbidity and repeat admission to hospital for adverse drug reactions in older adults: retrospective cohort study.  BMJ. 2009;338:a2752.PubMedGoogle ScholarCrossref
25.
Martin  RM, Biswas  PN, Freemantle  SN, Pearce  GL, Mann  RD.  Age and sex distribution of suspected adverse drug reactions to newly marketed drugs in general practice in England: analysis of 48 cohort studies.  Br J Clin Pharmacol. 1998;46(5):505-511.PubMedGoogle ScholarCrossref
26.
Pouyanne  P, Haramburu  F, Imbs  JL, Bégaud  B; French Pharmacovigilance Centres.  Admissions to hospital caused by adverse drug reactions: cross sectional incidence study.  BMJ. 2000;320(7241):1036.PubMedGoogle ScholarCrossref
27.
Jacubeit  T, Drisch  D, Weber  E.  Risk factors as reflected by an intensive drug monitoring system.  Agents Actions Suppl. 1990;29:117-125.PubMedGoogle ScholarCrossref
28.
Sarkar  U, López  A, Maselli  JH, Gonzales  R.  Adverse drug events in US adult ambulatory medical care.  Health Serv Res. 2011;46(5):1517-1533.PubMedGoogle ScholarCrossref
29.
Gill  JM, Mainous  AG  III, Nsereko  M.  The effect of continuity of care on emergency department use.  Arch Fam Med. 2000;9(4):333-338.PubMedGoogle ScholarCrossref
30.
Tamblyn  R, Abrahamowicz  M, Dauphinee  D,  et al.  Physician scores on a national clinical skills examination as predictors of complaints to medical regulatory authorities.  JAMA. 2007;298(9):993-1001.PubMedGoogle ScholarCrossref
31.
Chuang  SK, Wei  LJ, Douglass  CW, Dodson  TB.  Risk factors for dental implant failure: a strategy for the analysis of clustered failure-time observations.  J Dent Res. 2002;81(8):572-577.PubMedGoogle ScholarCrossref
32.
Bellera  CA, MacGrogan  G, Debled  M, de Lara  CT, Brouste  V, Mathoulin-Pélissier  S.  Variables with time-varying effects and the Cox model: some statistical concepts illustrated with a prognostic factor study in breast cancer.  BMC Med Res Methodol. 2010;10:20.PubMedGoogle ScholarCrossref
33.
Chen  DT, Wynia  MK, Moloney  RM, Alexander  GC.  US physician knowledge of the FDA-approved indications and evidence base for commonly prescribed drugs: results of a national survey.  Pharmacoepidemiol Drug Saf. 2009;18(11):1094-1100.PubMedGoogle ScholarCrossref
34.
Schiff  GD.  Computerized prescriber order entry: models and hurdles.  Am J Health Syst Pharm. 2002;59(15):1456-1460.PubMedGoogle Scholar
35.
Zopf  Y, Rabe  C, Neubert  A, Hahn  EG, Dormann  H.  Risk factors associated with adverse drug reactions following hospital admission: a prospective analysis of 907 patients in two German university hospitals.  Drug Saf. 2008;31(9):789-798.PubMedGoogle ScholarCrossref
36.
Steinman  MA, Landefeld  CS, Rosenthal  GE, Berthenthal  D, Sen  S, Kaboli  PJ.  Polypharmacy and prescribing quality in older people.  J Am Geriatr Soc. 2006;54(10):1516-1523.PubMedGoogle ScholarCrossref
37.
Johnell  K, Klarin  I.  The relationship between number of drugs and potential drug-drug interactions in the elderly: a study of over 600 000 elderly patients from the Swedish Prescribed Drug Register.  Drug Saf. 2007;30(10):911-918.PubMedGoogle ScholarCrossref
38.
Corsonello  A, Pedone  C, Corica  F, Mussi  C, Carbonin  P, Antonelli Incalzi  R; Gruppo Italiano di Farmacovigilanza nell’Anziano (GIFA) Investigators.  Concealed renal insufficiency and adverse drug reactions in elderly hospitalized patients.  Arch Intern Med. 2005;165(7):790-795.PubMedGoogle ScholarCrossref
39.
Bates  DW, Miller  EB, Cullen  DJ,  et al; ADE Prevention Study Group.  Patient risk factors for adverse drug events in hospitalized patients.  Arch Intern Med. 1999;159(21):2553-2560.PubMedGoogle ScholarCrossref
40.
Bégaud  B, Martin  K, Fourrier  A, Haramburu  F.  Does age increase the risk of adverse drug reactions?  Br J Clin Pharmacol. 2002;54(5):550-552.PubMedGoogle ScholarCrossref
41.
Gurwitz  JH, Avorn  J.  The ambiguous relation between aging and adverse drug reactions.  Ann Intern Med. 1991;114(11):956-966.PubMedGoogle ScholarCrossref
42.
Galanter  W, Falck  S, Burns  M, Laragh  M, Lambert  BL.  Indication-based prescribing prevents wrong-patient medication errors in computerized provider order entry (CPOE).  J Am Med Inform Assoc. 2013;20(3):477-481.PubMedGoogle ScholarCrossref
43.
Forster  AJ, Auger  C; ISTOP ADE Investigators.  Using information technology to improve the monitoring of outpatient prescribing.  JAMA Intern Med. 2013;173(5):382-384.PubMedGoogle ScholarCrossref
44.
Gandhi  TK, Weingart  SN, Borus  J,  et al.  Adverse drug events in ambulatory care.  N Engl J Med. 2003;348(16):1556-1564.PubMedGoogle ScholarCrossref
45.
Classen  DC, Bates  DW.  Finding the meaning in meaningful use.  N Engl J Med. 2011;365(9):855-858.PubMedGoogle ScholarCrossref
46.
Rodríguez-Monguió  R, Otero  MJ, Rovira  J.  Assessing the economic impact of adverse drug effects.  Pharmacoeconomics. 2003;21(9):623-650.PubMedGoogle ScholarCrossref
47.
Al Hamid  A, Ghaleb  M, Aljadhey  H, Aslanpour  Z.  A systematic review of hospitalization resulting from medicine-related problems in adult patients.  Br J Clin Pharmacol. 2014;78(2):202-217.PubMedGoogle ScholarCrossref
×