Abstract
Purpose
The purpose of the study was to determine whether access to drug samples influences resident prescribing decisions.
Subjects and methods
The authors observed 390 decisions to initiate drug therapy by 29 internal medicine residents over a 6-month period in an inner-city primary care clinic. By random selection, half of the residents agreed not to use available free drug samples. Five drug class pairs were chosen for study prospectively. Highly advertised drugs were matched with drugs commonly used for the same indication that were less expensive, available over-the-counter, or available in generic formulation.
Results
Resident physicians with access to drug samples were less likely to choose unadvertised drugs (131/202 decisions) than residents who did not have access to samples (138/188 decisions; P = .04) and less likely to choose over-the-counter drugs (51/202, 73/188; P = .003). There was a trend toward less use of inexpensive drugs.
Conclusion
Access to drug samples in clinic influences resident prescribing decisions. This could affect resident education and increase drug costs for patients.
Keywords
There are few objective data describing how drug samples affect resident physicians. When samples were removed from 1 clinic, usage of recommended antihypertensive drugs increased.
1
In another study, residents in 2 programs with restrictions on samples prescribed more recommended nonsteroidal anti-inflammatory drugs (NSAIDs) than residents in a comparable program without restrictions.2
Neither study was randomized.This is an important information gap, because drug samples are widely available in residents’ clinics,
3
and making treatment decisions based on which samples are available could affect both resident education and patient care.4
Although samples may provide short-term economic benefit for some patients, their primary purpose is to market new and expensive drugs.5
, We report a prospective randomized trial investigating whether access to drug samples influences prescribing decisions by internal medicine residents.
Methods
We asked all residents working in a primary care clinic affiliated with a teaching hospital in inner-city Minneapolis for written permission to study their prescribing habits. Residents were told only that the study was related to the observation that many patients could not afford their drugs. By random selection, permission documents for half of the residents in each postgraduate year included a second paragraph asking them not to use free drug samples available in clinic and reminding them that social workers were available to find other ways to help patients. All residents signed these agreements, 14 with the second paragraph (no sample access) and 15 without the second paragraph (sample access allowed). There was no further intervention. Thus, study endpoints and randomization were unknown to the 30 residents and the 8 staff doctors seeing patients in this clinic, other than the authors. The hospital’s institutional review board approved the study. No funding was received for the study.
Drug-class pairs were prospectively selected for study if they met 3 criteria: at least 10 pages of advertising in 4 journals during the month of May 2003 (Annals of Internal Medicine, Archives of Internal Medicine, JAMA, and The New England Journal of Medicine), availability of samples in the clinic, and the existence of an alternative class of drugs commonly used for the same indication, but lower priced or available over-the-counter (OTC) or in generic formulation. A pharmacy located 2 blocks from the clinic, patronized by many clinic patients, prepared lists of drugs available generically and OTC. The 5 drug-class pairs meeting these criteria were cyclooxygenase-2 (COX2) inhibitors/nonselective NSAIDs, newer antidepressants/tricyclic antidepressants, proton-pump inhibitors (PPIs)/H2 receptor blockers (H2 blockers), angiotensin receptor blockers/angiotensin-converting enzyme inhibitors, and thiazolidinediones/metformin. Omeprazole became available OTC during the study and was counted as an OTC drug for the duration of the study.
We evaluated resident prescribing behavior from July to December 2003 by chart inspection during the billing process, looking for instances in which drugs in these classes were prescribed for the first time for each patient. “Assessment and plan” and medication lists were searched for each patient encounter. Decisions to refill previously prescribed drugs or restart drugs that patients had stopped on their own were disregarded. Prescribing decisions made by the authors were excluded. All decisions about which drugs to study, how to classify drugs, and when to count them as newly prescribed were made before the study was started. The prospectively chosen primary hypothesis was that residents with sample access would be more likely to choose the heavily advertised drugs and less likely to choose the inexpensive, OTC, and generic alternatives.
Pharmaceutical company representatives stock a 2 × 2 m sample cabinet in clinic and are not allowed contact with resident physicians. There was no attempt to influence which drugs were stocked during the study. The cabinet was monitored periodically to ensure that advertised study drug classes were available; in all cases, they were. To avoid altering behavior by the process of studying it, residents were not reminded to use the sample sign-out sheets. Consequently, we collected data on which drugs were started but not on whether drugs were provided by prescription or as samples.
We obtained data about the residents by questionnaire at the conclusion of the study (Table 1).
Table 1Characteristics of residents with and without sample access
| No sample access | Sample access | |
|---|---|---|
| Number | 14 | 15 |
| Male/female | 9/5 | 9/6 |
| Likely career choice | ||
| Primary care | 4 | 4 |
| Other/undecided | 10 | 11 |
| Reported no face-to-face interaction with drug representatives during study in clinic | 13 | 13 |
| Reported no face-to-face interaction with drug representatives during study in any location | 7 | 7 |
| Reported no gift >$5 value received during study | 14 | 14 |
| Think “samples do not influence my choice of drug” | 9 | 11 |
| Median number of 4 index journals read regularly | 2 | 2 |
Analysis of categoric data by Fischer’s exact test was performed on a Macintosh computer using Statview 4.5 (SAS Institute, Cary, NC).
Results
Resident physicians randomized to sample access and no sample access were similar (Table 1).
The 29 residents initiated new drug therapy 390 times during the 6-month study period, whether by writing prescriptions or giving samples from the drug cabinet.
Residents assigned to sample access were less likely to initiate treatment with unadvertised drugs (131/202 decisions) than residents assigned to no sample access (138/188 decisions; P = .04) and OTC drugs (51/202, 73/188; P = .003). There was a trend toward less use of inexpensive drugs (115/202, 121/188; P = .08) (Table 2).
Table 2Prescribing behavior in residents with and without sample access
| No sample access | Sample access | P (difference) | |
|---|---|---|---|
| Unadvertised | 138 (73%) | 131 (65%) | .04 |
| Advertised | 50 (27%) | 71 (35%) | |
| Inexpensive | 121 (64%) | 115 (57%) | .08 |
| Expensive | 67 (36%) | 87 (43%) | |
| OTC | 73 (39%) | 51 (25%) | .003 |
| Prescription | 115 (61%) | 151 (75%) | |
| Generically available | 113 (60%) | 109 (54%) | .13 |
| Trade name only | 75 (40%) | 93 (46%) | |
| NSAIDs | 72 (90%) | 67 (82%) | .10 |
| COX2 inhibitors | 8 (10%) | 15 (18%) | |
| Tricyclic antidepressants | 25 (45%) | 16 (38%) | .33 |
| Newer antidepressants | 31 (55%) | 26 (62%) | |
| H2 blockers | 9 (27%) | 7 (15%) | .15 |
| PPIs | 24 (73%) | 39 (85%) | |
| ACEIs | 11 (73%) | 13 (72%) | .63 |
| ARBs | 4 (27%) | 5 (28%) | |
| Metformin | 4 (100%) | 12 (86%) | .59 |
| TZDs | 0 (0%) | 2 (14%) |
OTC = over-the-counter; NSAID = nonsteroidal anti-inflammatory drug; COX2 = cyclooxygenase-2; PPI = proton-pump inhibitor; ACEI = angiotensin-converting enzyme inhibitor; ARB = angiotensin receptor blocker; TZD = thiazolidinedione.
A post hoc analysis considering only drugs commonly used for short-term therapy (NSAIDs including COX2 inhibitors, PPIs, and H2 blockers) showed that access to samples of these drugs seemed to be especially influential, associated with less use of inexpensive drugs (63/126, 74/115; P = .02), OTC drugs (53/126, 71/115; P = .002), and generic drugs (70/126, 80/115; P = .02). There was little difference in use of unadvertised drugs (89/126, 87/115).
Discussion
We found that resident physicians with access to drug samples in clinic were more likely to write new prescriptions for heavily advertised drugs and less likely to recommend OTC drugs than their peers. There was also a trend toward less use of inexpensive drugs.
These results, although not surprising, add to the existing body of information because they come from direct observations of behavior in a randomized controlled trial. We performed an English-language Medline search from 1992 to 2004, using the keywords “internship and residency,” “drug samples,” “pharmaceutical industry,” “medical ethics,” and “gift giving.” Most articles consisted of resident opinion
7
, 8
, 9
, 10
, 11
, 12
or surveys of self-reported behavior.4
, 9
, 13
Residents see themselves as less influenced by marketing than their peers.8
, 10
They consider samples more acceptable than gifts for personal use.9
, 11
However, when samples run out they report writing prescriptions for sampled drugs rather than switching to drugs they would otherwise prefer.4
, 14
Attitudes of physicians in practice are similar.
4
, 15
They acknowledge the potential influence of samples but see them as helpful to patients.9
They tend to underestimate their personal response to marketing.5
Because all 4 prospectively chosen endpoints were affected in a direction that would increase drug costs, it is reasonable to predict that the short-term economic benefit of free samples would be replaced by higher costs later on, considering the likely development of brand loyalty.
4
, 5
, This could be especially burdensome for low-income patients.From an educational perspective, residents preparing for practice in a world where many patients cannot afford their medications should become familiar with a variety of drugs, including less expensive ones that are available generically or OTC. Prescribing habits, once learned, may be difficult to change.
4
, Some caution is appropriate in interpreting these results. The study was small in size, and significant differences were found only for groups of drugs. However, these group endpoints were prospectively chosen and the primary analysis did not deviate from what had been planned.
The intentionally nonintrusive study methods limited our ability to analyze how samples affected decisions to start individual drugs. In the group of COX2 inhibitors, in which only 3 drugs were available, Vioxx (Merck and Company, Inc., White House Station, NJ) was consistently available in the sample cabinet, whereas Celebrex (Pfizer Inc., New York, NY) and Bextra (Pfizer Inc.) (each more heavily advertised than Vioxx) were present intermittently. Vioxx was chosen for 78% of new COX2 therapy and 86% of samples as recorded on sign-out sheets.
We did not evaluate appropriateness of prescribing. Some patients probably received samples inappropriately (gastric complaints were absent in 43% of patients recorded as receiving COX2 samples). For others, samples permitted access to appropriate therapy that patients would not have been able to afford.
Local factors exerted some influence, including low socioeconomic status of the patients, availability of social workers, and policies restricting interaction between residents and pharmaceutical company representatives. However, the patients and residents studied are not different in any obvious way from those in many other residency clinics. Knowing that prescribing behavior was being observed could have altered resident drug choices. However, the physicians working in this clinic (other than the authors) did not know what data were being collected, and resident randomization was concealed. Different language in the consent forms could have affected the results by raising consciousness of drug costs. Studying drugs in pairs that are not exactly equivalent is imperfect but probably approximates real-world decision making.
None of these factors seem important enough to invalidate the main result: Access to drug samples influenced prescribing decisions of resident physicians, something that would seem to violate published national guidelines on physician interactions with the pharmaceutical industry.
5
, 16
, 17
This finding contradicts 2 widespread beliefs: Drug samples are inherently different from other forms of marketing, and samples help patients manage drug costs in the long term. They raise questions about whether drug samples belong in clinics where residents are learning or where low-income patients are receiving care.
Acknowledgments
We thank Willie Richards for help with the literature search, Pat Oberembt for lists of generic and OTC drugs, and Terry Rosborough, MD, for statistical analyses.
References
- Effect of antihypertensive samples on physician prescribing habits.Fam Med. 2002; 34: 729-731
- The effect of drug sampling policies on residents’ prescribing.Fam Med. 1998; 30: 482-486
- Impact of pharmaceutical company representatives on internal medicine residency programs.Arch Intern Med. 1992; 152: 1009-1013
- A physician survey of the effect of drug sample availability on physicians’ behavior.J Gen Intern Med. 2000; 15: 478-483
- Physician-industry relations. Part 1: individual physicians.Ann Intern Med. 2002; 136: 396-402
- Closing the door on sample closets.Minn Med. 2001; 84: 17-24
- What residents don’t know about physician-pharmaceutical industry interactions.Acad Med. 2004; 79: 432-437
- Of principles and pens.Am J Med. 2001; 110: 551-557
- A comparison of physicians’ and patients’ attitudes toward pharmaceutical industry gifts.J Gen Intern Med. 1998; 13: 151-154
- Effects of an educational intervention on residents’ knowledge and attitudes toward interactions with pharmaceutical representatives.J Gen Intern Med. 1997; 12: 639-642
- Interactions with the pharmaceutical industry.CMAJ. 1995; 153: 553-559
- Pharmaceutical representatives and emergency medicine residents.Ann Emerg Med. 1993; 22: 1593-1596
- Drug samples and family practice residents.Ann Pharmacother. 1997; 31: 1296-1300
- Sample medication dispensing in a residency practice.J Fam Pract. 1992; 34: 42-48
- Physicians and the pharmaceutical industry.JAMA. 2000; 283: 373-380
- Physicians and the pharmaceutical industry (update 2001).Can Med Assoc J. 2001; 1164: 1339-1344
- Gifts to physicians from industry.JAMA. 1991; 265: 501
Article info
Publication history
Accepted:
February 11,
2005
Received in revised form:
February 11,
2005
Received:
October 22,
2004
Identification
Copyright
© 2005 Elsevier Inc. Published by Elsevier Inc. All rights reserved.
