Abstract
Purpose::
There is the perception that commercial support may bias research results. In our pilot study at the 2006 ARVO meeting we analyzed a subset of the abstracts from the previous year to assess whether commercial support led to a greater incidence of statistically significant outcomes. Aside from one small subset, we found no significant difference overall. Herein we review all the abstracts from the 2006 ARVO meeting to ascertain the effect of commercial support on research results.
Methods::
A review of all the abstracts from the ARVO 2006 international meeting was performed. We documented the type of study, commercial support, whether a pharmaceutical agent was studied, whether statistical significance was achieved, and the corresponding p-value. The study was considered to have commercial support if a commercial entity was listed under the support section. We excluded studies that did not quantitatively state significance with a p-value.
Results::
1576 of the abstracts stated significance with a p-value. 418 of these documented commercial support. A total of 1284 outcomes were studied in these commercially-supported abstracts. 990 of these outcomes were statistically significant. 2816 of the non-commercially supported outcomes were significant. Chi-square analysis was performed to compare these two groups. There was no statistical difference in the two groups (p=0.1175). We then subdivided the data to analyze clinical studies of pharmaceutical agents. 347 abstracts met the criteria, and of those 123 reported commercial support. 291 of the 401 commercially supported outcomes were statistically significant. The remaining 224 non-commercially supported clinical studies of pharmaceutical agents studied 728 outcomes. 586 of these were statistically significant. This difference was found to be statistically significant by chi-square analysis (p=0.0022).
Conclusions::
Our analysis of the abstracts from the 2006 ARVO meeting reveal no overall statistically significant difference between the commercially and non-commercially funded studies. Similar to last year’s results, we noted a higher rate of statistically significant outcomes in non-commercially funded studies of pharmaceutical agents. On the whole, it appears that commercial funding did not exert an undue influence on the reporting of results at the 2006 ARVO meeting.
Keywords: clinical (human) or epidemiologic studies: health care delivery/economics/manpower • clinical (human) or epidemiologic studies: biostatistics/epidemiology methodology • clinical (human) or epidemiologic studies: systems/equipment/techniques