Boxplots include the interquartile range (IQR), including the 25th percentile (bottom of the box) and the 75th percentile (top of the box), the mean (dot), and the upper fence/whisker (largest observation within 1.5 × IQR). Owing to the skewness of this distribution, outliers (observations outside of 1.5 × IQR) have been suppressed. All medians are 0.
Customize your JAMA Network experience by selecting one or more topics from the list below.
Horstman AA, Niziol LM, Chimonas S, Lichter PR. Association of Mandatory Disclosure Policies and Laws With Physician-Industry Financial Relationships. JAMA Ophthalmol. 2019;137(5):523–530. doi:10.1001/jamaophthalmol.2019.0085
Since the implementation of laws and policies to enhance physician financial disclosures of industry relationships in 2008, how have disclosures changed?
This 8-year cross-sectional study (through 2015) did not identify a decreasing number of financial disclosures of potentially beneficial physician-industry ties. Over this period, an increasing number of physician-industry relationships were reported.
Since this study shows that current disclosure policies and laws have not been associated with a decrease in ethically questionable physician-industry relationships and may instead have created a permissive atmosphere for physician-industry relationships, perhaps disclosure policies and laws should be reconsidered.
Beside the goal of increasing transparency to the public, disclosure policies and laws have been established with a goal to also reduce ethically questionable financial relationships between physicians and the medical industry. Data on these relationships should be reviewed to understand the association between these policies and laws and the attainment of reduced relationships.
To assess whether disclosure policies and laws have been associated with a decrease in financial disclosure reporting by physicians.
Design, Setting, and Participants
This cross-sectional study uses yearly data from 2008 through 2015 from the participants in theAmerican Academy of Ophthalmology’s Annual Meeting. Trends in financial disclosures over time were investigated for the association of disclosure policies and laws with potentially beneficial, as well as ethically questionable physician-industry ties. Linear regression models were used to estimate the annual change in financial disclosures and are reported with 95% CIs.
Disclosure policies and laws.
Main Outcomes and Measures
The annual aggregate financial disclosures by type (ie, consultant, lecturer, employee, grant support, equity owner, patent holder).
Financial disclosures increased from 3966 in 2008 to 5266 in 2015 (P < .001). The number of disclosures reported in the categories consultant, equity owner, patents, and grant support all increased from 2008 to 2015 (consultant disclosures, 121 [95% CI, 88-155] per year; P < .001; equity owner disclosures, 32 [95% CI, 22-42] per year; P < .001; patent disclosures, 19 [95% CI, 13-26] per year; P < .001; grant support disclosures, 78 [95% CI, 48-107] per year; P < .001), while the employee and lecturer categories did not change significantly. The percentage of financial disclosures in the lecturer category decreased relative to the total (estimate, −1.1% [95% CI, −1.3% to −0.8%] per year; P < .001), owing to the number of financial disclosures for this category remaining stable while most other types increased.
Conclusions and Relevance
Disclosure was not associated with a chilling effect (decrease in financial disclosures associated with potentially beneficial physician-industry ties). Disclosure was associated with a possible disinfecting effect, whereby the percentage of ethically questionable disclosures (ie, lecturers) decreased, although the frequency remained stable. A permissive effect (physicians becoming more inclined to having industry relationships) was also observed. Thus, disclosure rules should be enhanced or alternative approaches to disclosure reconsidered to promote a decrease in ethically questionable relationships.
Since the turn of this century, there has been increasing interest regarding the medical practice outcomes associated with financial ties between physicians and pharmaceutical and medical device companies (or industry). A growing literature documents the importance of these relationships on physician prescribing: specifically, more total prescribing and more prescribing of expensive medications in particular.1-9 To be sure, grants and contractually defined consulting relationships can have beneficial effects on patient care by fostering innovations in drug discovery and technology. However, other common relationships, including speakers’ bureaus and token consulting arrangements, may promote activities of questionable medical benefit by favoring industry’s interests over those of patients, such as by promoting brand-name drug prescriptions and use of new devices and tests with uncertain value to patients.10-18
Disclosure policies and laws were established to address this issue, with the goal of increasing transparency for the public and discouraging ethically questionable ties to industry. Many physician-led professional societies and academic medical centers have clarified and strengthened their disclosure policies19,20 whereby physician-led specialty societies21-28 follow the Accreditation Council for Continuing Medical Education’s (ACCME) disclosure guidelines.29
Recent state and federal laws, too,30,31 have increasingly mandated disclosure. The Physician Payment Sunshine Act (PPSA), enacted in 2010 as part of the Affordable Care Act, has made physician-industry financial ties since August 2013 a matter of public record. Payment data are accessible through the Centers for Medicare and Medicaid Services Open Payments Data website (http://openpaymentsdata.cms.gov).
Nonetheless, the value of disclosure remains contested. While proponents believe, to paraphrase Louis Brandeis, “sunlight is the best disinfectant,”32(p92) critics have warned of divergent negative consequences. Some, seeing disclosure as a necessary but insufficient remedy for industry bias, cautioned that it could have an unintended permissive effect33 whereby physicians would feel free to pursue ethically questionable ties because they know that there is government-mandated full disclosure of their industry financial relationships.34,35 Furthermore, opponents of disclosure also warned of a chilling effect whereby physicians might decline to participate in potentially valuable industry interactions (eg, research, drug development, technological innovation36,37) because the lack of context for these industry ties could suggest impropriety and harm reputations.38-41 In fact, while the PPSA was pending, some physicians indicated that they might avoid financial relationships with industry, fearing public scrutiny would sully them.42-44
Using data from annual meeting programs of the American Academy of Ophthalmology (AAO) between 2008 and 2015, we investigated whether increasing pressure for transparency is associated with changes in frequency or type of physician-industry financial ties. Among medical specialty society meetings, the AAO is one of the largest, with attendance ranging from 22 169 to 28 355 over the 2008 to 2015 meeting years (AAO staff, written communication, June 30, 2017). In addition, ophthalmology widely uses drugs, devices, and equipment in medical practice, for which the AAO has a financial disclosure (FD) policy for physicians to report ties with industry. Through 2015, the AAO disclosure guidelines required FDs from all program participants (including first authors and coauthors).45 In short, the AAO Annual Meeting can serve as a prototype for analyzing whether greater disclosure has been associated with a disinfecting, permissive, or chilling effect on physician-industry financial relationships. Was there a decline, stability, or increase in the aggregate number and percentages of reported FDs over time by program participants? Furthermore, were there associations of disclosure rules and policies with potentially beneficial ties (eg, grants and patents36,37) compared with ties of questionable benefit to patients (eg, speakers’ bureaus34,35)?
In an attempt to answer these questions, files of the AAO’s Annual Meeting Program for the years 2008 to 2015 were used to extract study data. Each year’s Participant Index (PI) and participant Financial Disclosure Index (FDI) were converted into Excel databases using formulas, macros, and text-to-column functions, and then cleaned for consistency. The PI is an alphabetical listing of program participants (first authors and coauthors) and it includes full last names, first initials, middle initials, suffix, and all page numbers where a participant is listed in the program. The FDI is an alphabetical listing of program participants with 1 or more reported FDs and includes full first and last names, middle initial(s), suffix, degree(s), and the type of disclosure and associated company. Persons identified in the PI but not in the FDI were assumed to have no FDs, unless indicated in the body of the program book by an asterisk denoting they did not report their FDs by the deadline set by the AAO to be included in the printed program book.
To link program participants to their FDs, each name listed in the FDI had to have an exact match in the PI, so that indices from the same year could be merged. This required manually adding full name information in the PI and rigorously cleaning names for consistency and accuracy. Merging files identified several issues, including misspelled names between indices, the addition or omission of information in 1 index but not the other (eg, middle initial or suffix), and program participants listed in the FDI who were not found in the PI. To resolve these discrepancies, we hand-searched details within the body of the program book to ensure that the individual listed in the FDI was the same as the comparable program participant found in the PI. Name listings were corrected as necessary, so that entries would merge between indices within a given year.
Throughout our study period, the AAO’s disclosure requirements remained consistent, stating that all contributors to AAO educational activities must self-report all financial relationships with industry whether or not they were relevant to the topic of presentation. Participants were to report companies with which they had FDs and the corresponding FD type(s) (consultant, lecture fees, equity owner, grant support, or patent). For the AAO’s purposes, the term company included not only commercial entities, but also granting agencies, such as the National Institutes of Health.
Companies listed in the FDI were self-reported by program participants, and therefore company names lacked uniformity. Accordingly, company names were rigorously cleaned to create consistency before analysis at the company level. This included correcting spelling, capitalization, and punctuation errors, as well as condensing alternative names of the same company.
Simple tabulations were performed regarding the number of AAO program participants each year overall and stratified by FD status (none, ≥1, or missing), as well as the number of companies with at least 1 FD attributed to them. The distribution of number of FDs per program participant over time is displayed using boxplots. Because this distribution was skewed, outliers more than 1.5 times the interquartile range (IQR) are not presented in the figure (but descriptive statistics of the distribution are provided in Table 1). This allows for better visualization of mean, median, and IQR of the distributions. The distribution of total number of FDs over time, stratified by FD type, is displayed using stacked bar charts. Linear regression models were used to assess trends over 8 years for the number of program participants, percentage of participants with and without FDs, mean number of FDs per participant, mean number of companies with FDs per participant, number of FDs overall and by type, percentage of FD types, and number of companies with reported FDs. Results are reported as β estimates (change per year) and 95% CIs. We used SAS version 9.4 (SAS Institute) for all analyses. Data analysis took place from June 2016 through May 2018.
From 2008 through 2015, the number of AAO Annual Meeting program participants ranged from 3594 to 4304 (Table 1) and showed no clear trend over time, but modest fluctuations from year to year. However, the number of program participants reporting at least 1 FD showed an increase (estimate, 33 [95% CI, 5-62] participants per year; P = .03). Also, there was a nonsignificant increase in the percentage of program participants reporting at least 1 FD (from 26.0% in 2008 to 26.7% in 21015; P = .08). Of the yearly program participants, a range of 826 participants (19.2%) in 2009 and 1156 participants (30.7%) in 2013 reported at least 1 FD; a range of 1786 participants (41.5%) in 2009 to 2366 participants in 2013 (63.1%) reported no FDs, and 293 participants in 2013 (7.8%) to 1692 in 2009 (39.3%) did not submit their FD information by the deadline set by the AAO to be included in the printed program.
The mean (SD) number of FDs per program participant was 1.2 (3.2) in 2008 (median, 0) and increased to a mean (SD) of 1.7 (4.3) in 2015 (median, 0; estimated β, 0.05 [95% CI, 0.01-0.09] per year; P = .02; Figure 1 and Table 1). The number of companies with which program participants reported having FDs also increased from 460 in 2008 to 767 in 2015 (estimated β, 48 [95% CI, 39-56] per year; P < .001). Program participants reported FDs with a mean (SD) of 0.90 (2.15) companies in 2008, increasing to 1.27 (3.02) by 2015 (estimated β, 0.04 [95% CI, 0.01-0.07] per year; P = .01).
The total number of FDs showed an increasing trend from 3966 in 2008 to 5266 in 2015 (estimate, 259 [95% CI, 155-363] per year; P < .001; Table 2 and Figure 2). Most FDs were reported as consultants (1642 [41.4%] in 2008 and 2352 [44.7%] in 2015), followed by lecturers (1069 [27.0%] in 2008 and 939 [17.8%] in 2015) and recipients of grant support (794 [20.0%] in 2008 and 1233 [23.4%] in 2015). Increases over time were found for the number of program participants reporting 1 FD or more as a consultant (estimated β, 26 [95% CI, 11-41] per year; P = .005; 0.7% [95% CI, 0.1%-1.4%] per year; P = .03), equity owner (estimated β, 12 [95% CI, 4-19] per year; P = .009; 0.3% [95% CI, 0.2%-0.6%] per year; P = .02), patent holder (estimated β, 10 [95% CI, 7-14] per year; P < .001; 0.3% [95% CI, 0.2%-0.4%] per year; P = .002), and grant recipient (estimated β, 18 [95% CI, 10-26] per year; P = .002; 0.5% [95% CI, 0.2%-0.9%] per year; P = .01). The mean number of each of these respective disclosure types per participant also showed increases over time (consultant: estimate, 0.03 [95% CI, 0.01-0.04] per year; P = .01; equity owner: estimate, 0.008 [95% CI, 0.004-0.011] per year; P = .002; patents: estimate, 0.005 [95% CI, 0.003-0.007] per year; P < .001; grant support: estimate, 0.0218 [95% CI, 0.004-0.032] per year; P = .02). Although the number and percent of program participants reporting 1 FD or more as a lecturer remained relatively stable over time, the percentage of FDs reported in the lecturer category showed a decrease over time (from 27.0% in 2008 to 17.8% in 2015; estimate, −1.1% [95% CI, −1.3% to −0.8%]; P < .001).
This analysis of 8 years of FD data from a major specialty society’s annual meeting did not show trends in FD reporting over time that could be interpreted as a chilling effect on potentially beneficial physician-industry interactions (eg, grants and patents), in the way that some worried might happen as a result of disclosure mandates.36,38-40 Instead, we found that the frequency and percentage of program participants with at least 1 FD increased over the years, overall and for those reporting FDs as consultants, equity owners, patent holders, and grant support recipients. Additionally, the mean number of these FDs per participant showed an increasing trend over time. Thus, rather than a chilling effect, disclosure policies and laws may be associated with a permissive effect, creating an atmosphere in which physicians, feeling liberated by mandatory disclosure, became less constrained about having greater financial relationships with industry. The increasing trend in the number of participants reporting at least 1 FD as a consultant while the number of participants reporting at least 1 FD as a lecturer remained about the same indicates that companies are signing up new consultants but not new lecturers. This may be a specific example of a permissive effect associated with consultant positions.
Had there been an overall disinfecting effect because of disclosure mandates, it might have been expected that the mean number of lecturer FDs per participant would have declined rather than remaining stable. Additionally, a disinfecting effect might have been borne out had we seen an increase over the years of program participants with no FDs. Rather, we found that the number and percentage of program participants with no FDs stayed about the same. While the number of program participants over the years stayed roughly the same, the mean number of companies with which a program participant had FDs increased over the years.
The lecturer (speakers’ bureau) category deserves a closer look, because it is known that these FDs are unlikely to benefit patients.9,46 The data show that over the years, the proportion of lecturer FDs decreased, while both the proportion and total number of equity owners, patent holders, and grant support recipients (which are ethically acceptable ties, broadly speaking) all increased. While the decrease in proportion of lecturer FDs may suggest the intended disinfecting effect of disclosure policies and laws (most notably the PPSA), it is difficult to ascribe a specific association with the PPSA because some physicians were expressing reservations about having industry financial relationships in anticipation of but prior to passage of the Act. Thus, comparing pre-2013 (pre-PPSA) data with post-2013 (post-PPSA) data and ascribing a cause-and-effect relationship would be oversimplified and inaccurate. Rather, the overall atmosphere of disclosure policies and laws and their general association with changes in behavior seems to be a more fitting description. While the proportion of FDs reported as lecturers decreased, there could be mitigating circumstances surrounding this finding. For example, while an association is uncertain, a recent decline in industry funding of continuing medical education47 could help explain the decrease in the percentage of FDs reported as lecturers.
We do not have FD data for 7.8% to 39.3% of meeting program participants across the years of data collection, with 39.3% being an outlier year (2009); otherwise, the missing data did not exceed 26.0%. Despite the missing data, our study showed increasing numbers and percentages of FDs over time and mean number of FDs per participant over time.
The AAO provided the following statement with regard to their collecting FDs: “In accordance with Academy policy and ACCME requirements in effect during the years covered in this study, all presenters at the annual meeting were required to disclose financial relationships they had with commercial interests. The Academy actively pursued missing disclosures and collected and disclosed them prior to the event being conducted. In the event a presenter refused to disclose, the presenter was not permitted to participate in the CME [continuing medical education] activity” (AAO staff, written communication, July 25, 2018). The missing FD data were considered by the AAO to be owing to the deadline for the printed program being much earlier than the date of a participant’s presentation. The data in this study came only from the printed program.
This study’s greatest strength is the 8-year database we created that spans many of the predisclosure and postdisclosure policies and laws, including the PPSA. We have unique longitudinal data on physician-industry financial interactions from a medical specialty society annual meeting enabling the study of the association of disclosure policies and laws on those interactions. Additionally, throughout the 8 years of this study, the AAO went beyond the ACCME’s disclosure guidelines, providing us with financial disclosures on all program participants. Studies on FDs such as this one rely on access to complete disclosure information, such that that ACCME’s current disclosure policies appear to fall short in addressing full disclosure, in that these policies “require everyone who is in a position to control the content of an education activity…has disclosed all relevant financial relationships with any commercial interest…”29 This apparently allows authors to decide whether or not they are in a position to control educational content and thus to potentially decide not to disclose.
It is not clear why the ACCME29 does not require disclosure from all program participants (all authors and coauthors) so that there is not confusion. To remedy this uncertainty, perhaps the ACCME could revise its disclosure guidelines to require all authors to submit their complete FDs whether or not the individual believes he or she is in a position to control CME content. It would seem that more disclosure is preferable to less, in that more disclosure promotes increased transparency.
Study limitations include analyses on aggregated yearly data rather than following individual presenters across all years. Missing data were an uncertain limitation; however, it may be reasonable to surmise that many presenters who did not submit their FD information in time for the printed program did not have any FDs to report. Additionally, our data do not provide us with a way to determine whether a consultancy was contractually defined, or 1 of the token variety that might result in a lecturer being called a consultant to perhaps obfuscate a lecturer role. Also, the FD consultant category includes money given to an individual to attend a meeting, further obscuring what a consultant does. Therefore, conclusions about associations between disclosure policies and laws with trends in the consultant category are not on firm ground. Future work should assess individual participant change in FDs over time, including whether a lecturer in a given year became a consultant in the next year. Importantly, our findings may have occurred irrespective of mandated disclosure policies and laws. Also, we did not assess payment data associated with the self-reported financial disclosures; thus, it is possible that payment amounts and/or the type of payment could in some way have a bearing on the overall increase in physician-industry financial disclosures. Finally, it would be important to know whether there are associations between FDs, the companies involved in them, and the prominence and prevalence of presenters in the program.
This analysis of 8 years of self-reported physician-industry relationships showed that an increase in disclosure policies and laws was not clearly associated with a disinfecting effect, albeit that ethically questionable lecturing showed a decreasing percentage of overall FDs while not showing a decrease in the number of lecturers. There was, however, an associated permissive effect of disclosure policies and laws shown by increasing numbers of FDs overall. Perhaps most importantly, we observed no associated chilling effect on potentially beneficial physician-industry relationships, as evidenced by increases in reporting of putatively beneficial activities, including grants and patents, that can lead to new drug and device development.
Accepted for Publication: December 22, 2018.
Corresponding Author: Paul R. Lichter, MD, Kellogg Eye Center, 1000 Wall St, Ann Arbor, MI 48105 (email@example.com).
Published Online: February 28, 2019. doi:10.1001/jamaophthalmol.2019.0085
Author Contributions: Mss Horstman and Niziol had full access to all of the data in the study and take responsibility for the integrity of the data and the accuracy of the data analysis.
Concept and design: Niziol, Horstman, Chimonas.
Study concept and design: Lichter.
Acquisition, analysis, or interpretation of data: All authors.
Drafting of the manuscript: Lichter, Niziol, Horstman.
Critical revision of the manuscript for important intellectual content: All authors.
Statistical analysis: Niziol, Horstman.
Obtained funding: Lichter.
Administrative, technical, or material support: Lichter, Horstman.
Study supervision: Lichter.
Conflict of Interest Disclosures: None reported.
Funding/Support: This work was supported in part by a Research to Prevent Blindness Unrestricted Grant to the University of Michigan Department of Ophthalmology and Visual Sciences.
Role of the Funder/Sponsor: The funder had no role in the design and conduct of the study; collection, management, analysis, and interpretation of the data; preparation, review, or approval of the manuscript; and decision to submit the manuscript for publication.
Additional Contributions: We are grateful to the American Academy of Ophthalmology for providing us with copies of Annual Meeting Programs and in being responsive to our questions concerning Annual Meeting financial disclosure details. In addition, we thank Sameer Tammana, MSE, University of Michigan, for creating the critical process by which we transferred the Annual Meeting program PDFs to Excel files to allow for analysis. He was compensated for his contributions.