Association between industry payments and prescribing costly medications: an observational study using open payments and medicare part D data

Prescription costs in the U.S. are the highest in the world [1]. These high costs have been attributed not only to the complex drug development process but also to the restriction of price negotiation and government-protected monopolies granted to drug manufacturers [2]. Pharmaceutical manufacturers practice many methods to innovate and maintain market shares, including reformulating existing molecules and combining drugs [3]. While many new brand name drugs have clear medical benefits over alternatives already on the market, some do not, especially when considering the lower costs of existing alternatives. For example, Vimovo, a branded combination of naproxen and esomeprazole, has the potential benefit of improving tolerance of naproxen by co-administration with esomeprazole. However, with tolerance being the only clear benefit of the combined formulation, the listed wholesale acquisition cost (WAC) of $2259 for a bottle for 60 tablets⁴ may be hard to justify compared to the WAC of 60 tablets of naproxen ($6.36) and 60 tablets of esomeprazole ($225) [4].

The marketing of drugs to physicians includes various practices such as free drug samples, gifts, meals, speaker and consulting fees, and travel sponsorships. Because of the inherent risks for conflicts of interest and bias in physician-industry relations [5‐9], policy makers have long called for increased transparency in these relations. The Physician Payments Sunshine Act requires that drug or device manufacturers report all payments made to physicians to the Centers for Medicare and Medicaid Services (CMS). CMS releases these reports in the Open Payments dataset (OPD) each year [10].

Most studies of physician-industry relationships have reported survey findings and qualitative data, until recent years. The release by CMS of the Medicare Part D Prescriber Public Use Files (MPDPUF) [11] has allowed studies on prescribing patterns of individual physicians to beneficiaries in the Medicare Part D system. Linkage of the OPD and the MPDPUF enables studies of the association between physician-industry relations and prescribing patterns using real-world data. Recently, studies using these linked datasets have found associations between industry payments and brand name prescribing [12, 13], regional patterns in prescribing of marketed drugs [14], and prescribing by ophthalmologists and the use of newer, more expensive medications [15].

While prior work has shown the presence of physician-industry relationships associated with prescribing, we aimed to expand this body of literature by focusing specifically on the prescribing of newer, more expensive drugs that offer little to no incremental clinical benefit to patients and adjusting for factors affecting the association between payments and prescribing. Thus, our objective was to determine the association between pharmaceutical industry payments to physicians and the physicians’ likelihood of prescribing costly brand name drugs for which there are similarly effective, less costly alternatives – medications that we defined as drugs of uncertain medical benefit. Our hypothesis was that payment from the pharmaceutical industry would be associated with greater likelihood of prescriptions of drugs of uncertain medical benefit.

From 837,679 individual prescribers who prescribed drugs to Medicare Part D beneficiaries in 2014, 667,278 met inclusion and exclusion criteria, 38.6% (n = 257,719) of whom received at least one payment from the pharmaceutical industry. Figure 2 shows the selection of physicians included in the study, the number of physicians evaluated for each drug of uncertain medical benefit, alternative brand, and alternative generic, and the percent of physicians who received any payment for each trio. Among prescribers of the drugs of uncertain medical benefit, 58% to 63% received at least one payment. Among prescribers of brand control and generic alternative drugs, 48% to 59% and 39% to 51%, respectively, received at least one payment. The total drug cost and total days’ supply for the selected drugs of uncertain benefit for Part D beneficiaries from the MPDPUF data for 2014 is shown in Table 1. The range of cost for the selected drugs of uncertain medical benefit were from $5 to $26 per day [4]. The summary of number and percent of prescribers that received payments for each drug are presented in Table 2.

Table 3 shows the multivariable analysis for the odds of prescribing each drug by physicians who received any payment vs. those who did not. Any industry payment was associated with increased odds of prescribing of all the drugs of uncertain medical benefit, with non-significant findings for lovastatin ER, almotriptan and ibuprofen+famotidine. Payment was associated with increased odds of prescribing of all brand name controls except for rizatriptan. For all generic controls, payment was associated with a decrease in the odds of prescribing of the generic drug, with non-significant findings for sumatriptan.

Table 4 presents the odds ratios for prescribing the selected drugs by physicians with payments from the individual drug manufacturer. Payment by the drug manufacturer was associated with increased odds of prescribing of the drug manufactured by that company for all drugs except lovastatin ER and almotriptan.

This study of 667,278 physicians prescribing to Medicare Part D beneficiaries found that receiving any payment from the pharmaceutical industry was associated with increased odds of prescribing drugs of uncertain medical benefit. Brand name alternatives were also more likely to be prescribed when physicians received any industry payment. There was a pattern of lower likelihood of prescribing the generic alternative with receipt of any payment. While our study findings are consistent with prior studies showing the relationship between pharmaceutical industry payments and the prescription of brand name drugs [14], this study evaluates the prescription of selected expensive drugs that add no apparent additional benefit to a market already full of alternatives. This study adds to the literature by finding that this association is consistent in direction and magnitude for new drugs of uncertain medical benefit while adjusting for multiple factors including physician specialty and prescribing volume. The magnitude of the association between prescription and payments was even higher when the payment from the specific manufacturer was evaluated with prescription for that drug.

This study evaluated those drugs that were of the least discernible benefit in light of significant increases in costs. The drugs chosen in this study were combination drugs, in which convenient dosing and adherence might be the only benefits, namely, Kombiglyze XR and Azor. Also included were long-acting formulations of existing drugs (Altoprev) and newer drugs structurally similar to existing drugs (Axert). Finally, as is the case with Duexis and Vimovo, this study included drugs available without prescription at negligible cost that are marketed as combination branded drugs, costing more than $2000 per month in 2016 [24].

Our findings are consistent with other recent studies using the OPD and MPDPUF data [12, 13, 25, 26]. Yeh et al. found that pharmaceutical payments were associated with higher rates of prescribing brand name statins [12]. Perlis et al. found an association between payments and brand name prescribing as well as total prescription costs in the Medicare Part D program, an association that was found across multiple physician specialties [13]. Moreover, Fleischman et al. found an association between pharmaceutical payments and regional prescribing of “marketed” drugs – drugs for which there were 100 or more payments made by industry between August 2013 and December 2014 – in the Part D program [14]. DeJong et al. found a significant association between brand name prescribing and a corresponding industry-sponsored meal for the drug [27]. Qian et al. reported a decreased rate of generic drug prescribing with receipt of payments [28]. Finally, Propublica reported findings of their analysis of the OPD and MPDPUF data, showing that payment was associated with brand name prescribing [26]. It is clear from the published literature that payment from the pharmaceutical industry impacts prescribing practices.

The major strength of this study was the focus on expensive, brand name drugs for which there are reasonable alternatives and uncertain medical benefit. Our interest was in evaluating drugs whose purpose based on our inclusion/exclusion criteria was more related to gaining market share without meaningful innovation. These drugs were priced higher than alternatives available in the market at the same time and may have been promoted to physicians with financial incentives. An additional strength of our study was the adjustment for multiple potential confounders. In our analysis, we found that all bivariate associations for available covariates in the datasets were consistently significantly associated with prescribing of drugs of interest. We adjusted for all covariates in multivariable analysis, and notably found that controlling for physician specialty and therapeutic category volume markedly decreased the apparent effect size of the association between payment and prescribing. Another advantage of our study was the use of publicly available datasets that included 99.9% of Medicare Part D beneficiaries, enhancing generalizability. Lastly, a significant strength of our study is the analysis of the association between payments from specific manufacturer of the selected brand name drugs and prescribing of the selected brand name drugs.

Our study adds to the accumulating evidence of the influence of pharmaceutical industry payments on the use of newer, expensive brand name medications that have similarly effective, less costly alternatives. Given the possible conflict of interest inherent in physicians receiving pharmaceutical payments, the transparency of the data and the availability of tools such as Dollars for Docs [29] will make it easier for patients and consumers to understand the possible influence of the pharmaceutical industry on a physician’s decision to use a newer, more expensive drug, including drugs of uncertain medical benefit [30]. There may be a perception among the public that a new drug, and its implied novelty, works better than existing drugs in the same class. However, using evidence-guided criteria, our study identified drugs where this may not be the case and found a significant association between pharmaceutical industry payment and physician prescribing of these drugs. These findings contribute to growing evidence of the role of pharmaceutical industry payment in prescribing practice. Patients could use these findings to be more informed as healthcare consumers in order to advocate for shared decision-making at the time of prescribing new drugs. Providers could use these findings to be more aware of the explicit or implicit effects of receiving payments on their prescribing practices. Policymakers could use these findings to develop policies to improve interactions between the pharmaceutical industry and providers in order to improve the quality of care and reduce drug costs.