A National Survey of Physician–Industry Relationships

Survey Design

The IMAP survey of U.S. physicians was designed on the basis of information from a single focus group of eight physicians, four interviews with individual physicians, and a review of the literature; it was pretested with the use of eight cognitive interviews. The survey included 47 questions about professionalism, 3 of which focused on physician–industry relationships; 1 of these 3 questions comprised 10 subitems (see the Supplementary Appendix, available with the full text of this article at www.nejm.org). The survey was approved by the institutional review board at Massachusetts General Hospital.

Dependent Measures and Variables

We asked, “Which of the following have you received in the last year from drug, device, or other medically related companies?” The possible answers were food or beverages in the workplace; free drug samples; honoraria for speaking; payment for consulting services; payment for service on a scientific advisory board or board of directors; payment in excess of costs for enrolling patients in industry-sponsored trials; costs of travel, time, meals, lodging, or other personal expenses for attending meetings; gifts received as a result of prescribing practices; free tickets to cultural or sporting events; and free or subsidized admission to meetings or conferences for which continuing medical education (CME) credits were awarded.

Four post hoc categories of physician–industry relationships were created. The first category was free drug samples. The second category was gifts, which included food or beverages in the workplace, free tickets to cultural or sporting events, and gifts given because of prescribing practices. The third category was reimbursements for expenses, including the costs of travel, time, meals, lodging, or other personal expenses for attending meetings and free or subsidized admission to meetings for which CME credits were awarded. The fourth category was payments for consulting, serving on a scientific advisory board or board of directors, speaking at a professional meeting, or enrolling patients in industry-sponsored clinical trials.

In addition to obtaining data regarding the categories of physician–industry relationships, the survey measured the frequency of meetings between physicians and industry representatives. The survey asked, “In an average month, how many times do you meet with representatives from drug, device, or other medically related companies?”

Survey Sample

From the 2003 Physician Masterfile of the American Medical Association, we identified all U.S. physicians in primary care (internal medicine, family practice, and pediatrics) and those in three non–primary care specialties chosen to represent a medical specialty, a surgical specialty, and an inpatient specialty (cardiology, general surgery, and anesthesiology, respectively). Then, we excluded all doctors of osteopathy, residents, physicians working in federally owned hospitals, those with no listed address, those who requested not to be contacted, and those who were retired. From the resulting list of 271,148 physicians, we randomly selected 584 physicians in each specialty, for a total sample of 3504.

Of the 3504 physicians in the sample, 337 were ineligible to participate in the survey because they were deceased, out of the country, practicing in a specialty that was not included in the survey, on leave, or not providing patient care. This yielded a raw eligibility estimate of 90%. Of the 3167 eligible physicians, 1662 completed a questionnaire, for an overall raw response rate of 52% (the response rate among cardiologists was 42%; anesthesiologists, 55%; physicians in family practice, 54%; surgeons, 52%; internists, 50%; and pediatricians, 63%). The weighted overall response rate was 58% (cardiologists, 43%; anesthesiologists, 57%; physicians in family practice, 55%; surgeons, 54%; internists, 52%; and pediatricians, 64%).²² To calculate the weighted overall response rate, we divided the number of completed interviews by the number of completed and partial interviews, plus the number of physicians who declined to participate or who could not be contacted, plus the eligibility estimate multiplied by the number of physicians with unknown eligibility. Physicians were classified as having unknown eligibility if no information was obtained about their eligibility either directly from the physician or from a gatekeeper. Also, in this study no physicians were classified as “other.” In calculating the specialty-specific weighted response rates, we accounted for the differences in eligibility rates among the specialties.

Survey Administration

Between November 2003 and June 2004, physicians received a mailed survey, cover letter, postcard with the subject's name on it, postage-paid return envelope, and check for $20. The subjects were asked to return the completed survey separately from the postcard. This system permitted the tracking of nonrespondents while preserving the anonymity of respondents, since the questionnaire had no identifying information. Nonrespondents were contacted by mail and telephone and were encouraged to participate.

Statistical Analysis

Logistic-regression models were used to assess the multivariate associations between types of physician–industry relationships and physicians' sex, race or ethnic group, number of years in practice, specialty, and primary practice type, as well as the financial status of their patients (whether 25% or more were receiving Medicaid or were uninsured and unable to pay) and their other professional roles (a preceptor of physicians in training, reviewer for a professional journal, or developer of clinical guidelines). We used the 25% cutoff because it divided the respondents into two groups of approximately equal size. Odds ratios with 95% confidence intervals were calculated, and all P values were two-tailed.

We performed pairwise comparisons of each specialty with every other specialty and then compared each primary practice type with every other type. Bonferroni corrections were used to adjust for multiple comparisons (a total of 37). A P value of less than 0.0014 (0.05 divided by 37) was considered to indicate statistically significant differences between specialties and between primary practice types.

To examine the association between the physicians' characteristics and the frequency of their meetings with industry representatives, we used the same independent variables in a negative binomial regression model, with adjusted odds ratios for the frequency of meetings. The negative binomial regression was used because the number of meetings with industry representatives most closely approximated a negative binomial distribution rather than a Poisson or normal distribution. Similar results were obtained when logistic regression was used. All of the analyses, except for the analysis of characteristics of the respondents, were weighted to adjust for differences in the rates of sampling and nonresponse according to specialty.

Table 1. Table 1. Characteristics of 1662 Survey Respondents.

Table 2. Table 2. Frequency of Physician–Industry Relationships According to Benefit Received.

The characteristics of the respondents are shown in Table 1. Table 2 shows the frequency of various types of physician–industry relationships. Overall, 94% of the respondents reported some kind of relationship with industry during the previous year.

Multivariate Analyses

Table 3. Table 3. Multivariate Predictors of Physician–Industry Relationships.

The results of multivariate analyses are shown in Table 3. The frequency of physician–industry relationships differed significantly according to the specialty and the primary practice organization. For example, after adjustment for other factors, pediatricians were less likely than internists to have received reimbursements or payments. Anesthesiologists were less likely to have received samples, reimbursements, or payments than were family practitioners, internists, or cardiologists. Cardiologists were more than twice as likely as family practitioners to receive payments for professional services and were also significantly more likely to receive payments than were pediatricians, anesthesiologists, or surgeons (P<0.002 for all comparisons).

As compared with physicians in hospitals or clinics and those in staff-model health maintenance organizations (HMOs), physicians in group practices were six times as likely to receive samples, three times as likely to receive gifts, and nearly four times as likely to receive payments for professional services (P<0.002 for all comparisons). Physicians in solo or group practices and those in university or medical-school practices also had much higher odds of receiving payments than did physicians in hospitals or clinics and those in staff-model HMOs. Other factors significantly associated with the receipt of payments were male sex, a patient population in which less than 25% of patients were receiving Medicaid or were uninsured, and a role as a preceptor of physicians in training or a developer of clinical guidelines.

Meetings with Industry Representatives

Family practitioners reported the highest average number of meetings with industry representatives (16 meetings per month), followed by internists (10 per month), cardiologists (9 per month), pediatricians (8 per month), surgeons (4 per month), and anesthesiologists (2 per month). Multivariate analyses showed that, as compared with family practitioners, physicians in all other specialties met significantly less frequently with industry representatives. For example, for every 10 meetings that family practitioners had with industry representatives, internists met 7.3 times, cardiologists 5.8 times, pediatricians 4.9 times, surgeons 2.6 times, and anesthesiologists 1.6 times (P<0.05 for all comparisons).

Our data show that physician–industry relationships are common in medicine, as are relationships between professionals and industrial organizations in the health sciences and many other sectors of the U.S. economy.^15,23 Furthermore, our data suggest that physicians' relationships with industry vary according to physicians' personal and professional characteristics and according to their practice setting. For example, pediatricians and anesthesiologists were significantly less likely than family practitioners to receive samples, reimbursements, and payments for professional services. We can only speculate about the reasons for these variations by specialty. Further research should consider factors such as the number and costs of drugs prescribed by physicians in the specialties in question, the accessibility of physicians in each specialty to company representatives, and the influence of physicians on the prescribing practices of their peers.

With regard to peer influence in particular, our findings suggest that industry may focus marketing efforts on physicians who are perceived as influencing the prescribing behaviors of other physicians. Cardiologists, whose prescribing patterns as specialists and opinion leaders are thought to influence the prescribing patterns of nonspecialists, are significantly more likely to receive direct payments from companies than are physicians in other specialties.¹ A Dutch study showed that two thirds of family practitioners' prescriptions for cardiovascular drugs were for regimens initiated by specialists.²⁴ The possibility that companies may target opinion leaders for marketing is further suggested by the higher frequency of industry payments to physicians who have developed clinical practice guidelines and to those who have served as preceptors for doctors in training.

Our survey showed that physicians in solo, two-person, or group practices were significantly more likely to have all types of relationships with industry than were physicians in hospitals or clinics. There are at least three possible explanations for this finding related to the practice setting. First, physicians in solo, two-person, or group practices may have more freedom in their prescribing choices than physicians in hospitals and clinics, which frequently use drug formularies that limit the prescribing autonomy of physicians. Second, hospitals and clinics may be more likely to have policies designed to restrict physician–industry relationships. Third, hospitals and large clinics are more likely to provide medical information through educational programs such as grand rounds and CME events, which may make the physicians at these facilities feel less dependent on industry representatives as the source of medical information. Further research is needed to explore these and other possible explanations for the influence of the practice setting on physician–industry relationships.

We found that in all specialties except anesthesiology, physicians met more frequently with industry representatives than the average of 4.4 meetings per month reported by Wazana in 2000.¹³ The reason for this apparent increase in meeting rates is unknown, but it may reflect an intensification of industry marketing since the 1990s or result from differences in study design.⁴

Our study has several limitations. First, the respondents may have underreported their associations with industry, a phenomenon known in the survey literature as social desirability bias.²⁵ Second, our results may not apply to specialties other than those we studied. Third, the results may be influenced by unmeasured factors, such as working hours or patient characteristics. Fourth, our overall response rate (52% when unweighted and 58% when weighted) reflects the increasing difficulty of obtaining physicians' responses to surveys, even when using the most professional techniques. This limitation may be especially salient with regard to cardiology, which had the lowest response rate (43%) among the specialties we surveyed. Finally, this study did not assess the risks, benefits, or overall appropriateness of various types of physician–industry relationships. Such judgments are the logical next step in discussions regarding physician–industry relationships, but they cannot be made solely on the basis of the data from this study, although they have been described extensively elsewhere.¹

Despite these limitations, the high prevalence of physician–industry relationships underscores the need to consider their implications carefully. The variations in the nature and frequency of physician–industry relationships among specialties and practice settings suggest that specialties, organizations, and practice leaders with an interest in reporting and managing physician–industry relationships may need to develop guidelines and recommendations that are specific to the context of each specialty and setting.