Changes in industry marketing payments to physicians during the covid-19 pandemic: quasi experimental, difference-in-difference study

Data sources

We matched the Centres for Medicare and Medicaid Services' Open Payments databases in 2018-202019 to its National Plan and Provider Enumeration System (NPPES) database20 and collated physicians’ full name and the zip code for the primary practice location, as previous studies have done.16 18

Physician characteristics

We extracted information on physicians’ gender(male and female), specialty (taxonomy code), and practice location (US state) from the NPPES database. Physician specialty was classified into 32 categories; four in primary care physicians (internal medicine, family medicine, pediatrics, and hospitalist), nine in surgeons (obstetrics and gynecology, general, orthopedic, otolaryngology, urology, neurosurgery, plastic, thoracic and vascular, and colorectal), 18 in specialists (emergency medicine, psychiatry, anesthesiology, radiology, cardiology, ophthalmology, pediatric specialty, neurology, rehabilitation, dermatology, gastroenterology, hematology and oncology, nephrology, pulmonology, infectious diseases, endocrinology, rheumatology, and allergy and immunology), and one called others (eg, surgical oncology, oral surgery, radio oncology, pathology, and nuclear medicine). As a result of the variation of stay-at-home orders across states during the pandemic,21 information about practice location by state was also extracted from the NPPES database.

Payment data

We identified all non-research payments to physicians, which included non-research forms of payment such as meals, travel, consulting fees, speaker compensation, honorariums, and others (education, grant, gifts, entertainment, and space rental or facility fees). Research payments, royalty or license payments, and ownership interests, such as stocks and partnership shares, were not included in this analysis because these payments were not likely to be targeted to specific drugs at a specific time point. Payments are presented in nominal US dollars.

Statistical analysis

First, we examined the trends in the value and the number of industry payments between 2018 and 2020, according to the type of payments (ie, meals, travel, consulting, speaking fee, honorariums, and others). We then used a quasi experimental, difference-in-difference method22 23 to investigate changes in monthly industry payments due to the covid-19 pandemic. We compared before the pandemic (January-February 2020) to after (April-December 2020; March 2020 was excluded as the transition period) in 2020, using the same period in 2019 as the control, adjusting for physician characteristics (ie, gender, specialty, and US state). The changes in outcomes attributable to the covid-19 pandemic were represented by regression coefficients of the interaction terms between the year indicator (ie, 2019 v 2020) and the month indicator (ie, January-February v April-December) in the multivariable negative binomial models (to account for the distribution skewed to the right of the value and the number of industry payments). In all models, we adjusted for physicians’ gender, specialty, and the state in which the practice is located. We report difference-in-difference estimates in relative percentage changes as well as absolute changes in outcomes using average marginal effects. We found the average marginal effects by calculating the differences in predicted outcomes at each level of the interaction terms for each observation and averaging over the entire sample.24 The parallel trend assumption of the difference-in-difference model was formally tested.

Each type of industry marketing payment requires different resources and might have different influences on clinical practice,25 26 therefore, we stratified analyses by the type of payments (meals, travel, consulting fees, speaker compensation, honorariums, and others). Because the receipt of industry marketing payments also varies depending on physicians’ characteristics,16–18 we also stratified our analyses by physicians’ gender and specialties. Additionally, to assess whether the trends in industry payments during the pandemic vary by geographical regions, we stratified analyses by the four main areas (northeastern, midwestern, southern, and western regions) in the US according to the Census Bureau.27 P values were adjusted for multiple comparisons using the Benjamini-Hochberg method.28 Statistical analyses were done with Stata software version 16.0 (StataCorp, TX).

Sensitivity analysis

To evaluate the robustness of our findings, we carried out three main sensitivity analyses. We used the mean value and number of payments in 2018 and 2019 (instead of 2019 only) as the control group. We reanalyzed the data using ordinary least squares regression models with Huber-White robust standard errors adjusting for physicians’ characteristics. We also used the event study design23 29 to compare the change in the value and the number of industry marketing payments in the event year with the previous year as the control year. In the event study design, we used the multivariable negative binomial models, including indicator variables for time (reference category was defined as February—ie, a month before the event), a binary variable for the intervention year group (2018-19 v 2019-20), the interaction term between these two regressors, and physicians’ characteristics.

Patient and public involvement

No patients were involved in setting the research question or the outcome measures, nor were they involved in developing plans for design or implementation of the study. No patients were asked to advise on interpretation or writing up of results. We have no plans to disseminate the results of the research to study participants or the relevant patient community.

Physician characteristics

Across 1 029 149 physicians in the merged database, we excluded 148 541 physicians without information about practice location in the US. Of 785 237 physicians were unmatched in the Open Payments Profile database. We also excluded 19 physicians who received industry payments of more than $1 million (£850 000; €1 005 000) within a month (as outliers), resulting in the final analytical sample of 880 589 physicians. The flow of sample selection for the study is provided in online supplemental figure A. Among 880 589 physicians included in our study, 63.7% were men, 38.2% were primary care physicians, 16.8% were surgeons, and 39.5% were specialists (table 1). Surgeons and most of the specialists (except emergency medicine, psychiatry, anesthesiology, radiology, pediatric specialty, and rehabilitation) were likely to receive industry payments across 2018-20 (ie, they showed a larger proportion among physicians who received industry payments than the proportion among physicians in each specialty in the NPPES database). The distribution of physicians’ gender and specialty among physicians who received industry payments was similar across the study period.

Industry payments between 2018 and 2020

In 2020, 267 463 physicians in the US received 4 117 482 non-research payments (except royalty, license, and ownership interests) to the value of $626m, which was a 40-44% reduction from $1115m in 2019 and $1047m in 2018 (table 2). From 2018 to 2020, meals constituted 90% of the total number of industry payments, and speaking fees constituted the largest value of payments (35-38%). The proportion of travel fees decreased in 2020 compared with previous years for both the number (5.1% in 2018, 5.6% in 2019, and 2.7% in 2020) and the value (12.1% in 2018, 12.4% in 2019, and 5.0% in 2020) of payments.

Change in the value of industry payments

The unadjusted monthly trends in the value of industry payments between 2018 and 2020 are shown in figure 1 (top panel). The results of the tests for the parallel trend assumption can be found in online supplemental table A. We found a significant decline in the monthly value of payments due to the covid-19 pandemic (adjusted percentage change of −48.4%; 95% confidence interval −50.6% to −46.2%; P<0.001; and adjusted absolute change of −$75.7, 95% confidence interval −81.2 to −70.3; P<0.001; table 3). We also found a significant reduction in the value of payments during the pandemic across any types of payments except others (table 3, online supplemental figure B). We found no evidence that changes in the value of payments due to the pandemic varied by gender (online supplemental table B). When we stratified by physicians’ specialty, the change was noted across all specialties, with the largest reduction among neurologists and rheumatologists on both the relative and the absolute scales compared with internal medicine physicians. The decline in the value of industry payments was consistently found across all regions in the US (online supplemental table C).

Request permission

Changes in the number of industry payments

The unadjusted monthly trends in the number of industry payments between 2018 and 2020 are shown in figure 1 (bottom panel). We found a significant decline in the monthly number of payments per 100 physicians due to the covid-19 pandemic (adjusted percentage change of −47.4%; 95% confidence interval −47.7 to −47.1; P<0.001; and adjusted absolute change of −43.6, 95% confidence interval −44.2 to −43.0; P<0.001; table 3). Although we found a significant reduction in the number of payments, particularly for meals, travel fees, and speaking fees, the reduction was not observed for consulting and honorariums (table 3, online supplemental figure C). We found no evidence that changes in the number of payments due to the pandemic varied by gender (online supplemental table D). When we stratified by physicians’ specialty, most surgeons and specialists, except rehabilitation physicians and pulmonologists, showed a larger reduction on the relative scale during the pandemic compared with physicians in internal medicine (a specialty with the highest proportion among US physicians). The reduction in the number of industry payments was consistently noted across all regions in the US (online supplemental table C).

Sensitivity analyses

The results did not qualitatively change when we used the average value and number of payments in 2018 and 2019 instead of 2019 only (online supplemental table E) and when we reanalyzed the data using ordinary least squares instead of the negative binomial model (online supplemental table F). Our findings were also supported by the event study design, which provided visual evidence to show that the industry marketing payments dropped at the event month (ie, pandemic in March 2020; figure 2).

Request permission

Principal findings

Using national data for industry marketing payments to US physicians, we found that the value and the number of the payments significantly declined by almost half during the covid-19 pandemic. A similar pattern was observed across physicians’ gender, all specialties, and regions in the US. To our knowledge, this study is the first to elucidate how the covid-19 pandemic has changed the relations between industry and physicians. Given that the reduction in these payments was the largest among those related to meals and travel fees, this effect was observed partially due to restricted in-person interactions during the pandemic. Our findings should be informative for policy makers interested in understanding the financial association between industry and physicians, and how these changes might eventually affect individual physicians’ prescription practice and patient outcomes, which should be the subject of future research.

The possible influence of the financial association between the pharmaceutical industry and physicians has received great concern,30 mainly because inappropriate prescriptions could have been introduced through misguided information and potential financial competing interests. For example, a previous study showed that physicians who received meals sponsored by the pharmaceutical industry were more likely to prescribe brand name drug than generic alternatives.31 Additionally, the receipt of the industry payments related to opioids was associated with increased prescription of opioids and opioid overdose deaths, indicating the possible over prescription of opioids related to the industry marketing.32–34 However, industry marketing to physicians might also have positive aspects related to continuing medical education activities. Across the US, the pharmaceutical industry often engages and supports the provision of educational opportunities for physicians to obtain the latest evidence on old and new drugs. For example, a previous work focusing on a new cardiovascular drug that was under strict prior-authorization requirements indicated that industry marketing to physicians might have contributed to facilitating the judicious and early adoption of the high-cost but effective drugs as a key driver of their prescribing behavior.26

Of note, we found the largest reduction in the value and the number of industry payments in the form of meals and travel fees during the pandemic. Previous studies have consistently shown that meals constituted more than 90% of the total number of industry payments.11 26 These studies suggested that encounters with industry marketing, even in the inexpensive form of meals, drive prescriptions, possibly through educational opportunities, drive sharing updated information on the drug, and improve physicians’ familiarity with the drug.11 26 31 Our findings indicate that these educational opportunities have been reduced due to the physical restrictions enforced during the pandemic and cancellations of local and national medical conferences. Although a decrease in revenues could also be a reason for decreased marketing, the average profit margin of pharmaceutical industries did not decrease during the pandemic.35 Because this trend might be affected by profits related to covid-19 for some specific pharmaceutical organizations, future research is needed to assess whether and to what extent the change in revenues of each organization influences their financial relations with physicians.

We did not find a reduction in the number of payments for consulting and honorariums. These payments do not necessarily require physical interactions between the pharmaceutical industry and physicians. Because consulting and honorariums are likely to be targeted to key opinion leaders (whereas meals are generally targeted to a broad type of physicians), the possible influence of the pandemic on prescription patterns through reduced industry marketing payments needs to be carefully assessed and monitored according to physicians’ role in academia or clinical practice.

The reduction in industry payments during the pandemic was consistently observed, even when categorized by physician gender and across all physician specialties and geographical regions in the US. Although the absolute reductions in the value and the number of payments were larger in male physicians than in female physicians, this amount is probably because male physicians are generally more likely to receive industry payments than female physicians at baseline (ie, before the pandemic),16–18 and the relative reduction was similar between physicians’ gender. Our consistent findings of the stratified analyses by physician specialties indicate that the pandemic has influenced the interaction between the pharmaceutical industry and individual physicians regardless of their specialties.

Limitations of the study

Our study has limitations. First, although we used the nationwide database of licensed US physicians (NPPES), the matching rate of physicians in the Open Payments database to the NPPES database was 88%; thus, we might not be able to capture all physicians who received industry marketing payments in the US. Second, measurement error is a possibility due to misreporting or under-reporting of payments. However, this bias would be small given the effort by the Centres for Medicare and Medicaid Services to improve the accuracy of the Open Payments data (eg, encouraging all physicians to review the information, facilitating education activities to increase awareness of the program), making the database comprehensive and reliable.7 Third, because the present study focused on non-research payments, our findings might not be generalizable to other forms of payments such as ownership interests, royalty, and license payments. These types of payments are targeted to a small number of physicians18 and not specific to a drug at a specific time,19 thus beyond the scope of this study. Finally, because data for physicians’ prescription and patients’ outcomes during the covid-19 pandemic are not publicly available yet, we could not evaluate these outcomes via the decreased industry marketing payments, which should be the subject of future research.