Our study was based on a protocol (Supplementary File 1) using methodology similar to previous studies on COI policies at medical schools in US and Canada [8, 12]. The protocol specifies the methods for the identification, assessment, and analysis of COI policies at Scandinavian medical schools.
Identification of conflict of interest policies
Two pairs of coders (one pair for Sweden and one pair for both Denmark and Norway) developed a list of the medical schools in the three included countries using Google searches. The names and number of medical schools per country was further verified by searching official national educational websites. (see Supplementary File 2) We included Scandinavian medical schools with full Bachelor and Master’s program allowing graduates to work as physicians after completing the program. Schools where the majority of the program took place in another country were excluded. Two pairs of coders independently searched the website of each included school and its parent University in July-August 2020 to identify COI policies. Disagreements were resolved by discussion. We use the term medical school to describe the institution hosting the medical education, typically a faculty of health sciences.
The websites were searched using a list of keywords in appropriate language (e.g., policy, conflict of interest, industry) and was complemented with Google searches (see Supplementary File 3 for the full details of the search strategy). The name of each policy and the date of adoption or most recent amendment were recorded. Both policies of the medical school and University-wide policies were included in line with the previous study by Shnier et al. . In some cases, the relevant information was only stated on a website and not contained in a separate document. In those cases, we included the website information as a form of informal policy.
A letter in the appropriate language was sent to the Dean of each medical school to inform them of the study. The letter explained the aim of the study, listed the COI policies identified via the website searches and asked for confirmation that we had not missed any relevant documents. We asked for both publicly available and non-publicly available policies. The first email was sent in September 2020 and was followed by up to two e-mail reminders in case the Dean did not reply. In only one case we did not receive a reply nor a delivery notification that confirmed the receipt of the email after two e-mail reminders. We therefore undertook additional contact by multiple telephone calls and one additional email, but were unable to determine whether the Dean had received our email.
We did not include specific COI policies of the various teaching hospitals affiliated with a particular medical school as these institutions are typically not under the authority of the medical school (i.e., faculty of health sciences). Similarly, we did not include regional or national policies or regulation (even when Deans sent us copies of those) unless they were explicitly mentioned in the included COI policies or on the websites of the University or medical school. When a national or external policy was mentioned in an included COI policy or on the medical school or University websites, it was used for the assessment, but it did not contribute to the final count of institutional policies in order to avoid double-counting.
Assessment of content and strength of conflict of interest policies
We based our assessment of the content and strength of COI policies on the 12-item adaptation of the AMSA scorecard developed by Shnier et al. for medical schools in Canada . We modified the system slightly to adapt it to the Scandinavian context. For example, we excluded the item on “Industry support for scholarships and funds for trainees” because industry does not provide scholarships to medical students in Scandinavian countries as education is free for EU/EEA students.
Our revised 11-item assessment system included the following items:
gifts (including meals)
consulting relationships (excluding funding for scientific research and speaking fees)
industry-funded speaking relationships and speakers’ bureaus
honoraria (beyond consulting or speaker fees)
disclosure of financial relationships with industry
industry sales representatives
on-site education activities
compensation for travel or attendance at off-site lectures and meetings
medical school curriculum (or other documentation of educational objectives aimed at training students in industry-interactions. In order to assess this item, we looked not only at the COI policies but also at the medical curriculum and learning objectives of the school of medicine. However, those documents were not included in our count of the number of institutional COI policies for each school)
Each item was graded using a rating scale of 0 to 2 (0 = no policy or permissive policy, 1 = moderate, and 2 = restrictive policy). The system also included two final questions on oversight and sanction that were graded as “Yes” or “No”.
We developed a standardised guidance on how to assess the different items. Supplementary File 4 shows the detailed assessment criteria for each item. We pilot tested the system using three randomly selected Canadian medical schools included in the study by Shnier et al. . This allowed training of the data collectors and to address ambiguities identified through coding disagreements.
Two pairs of coders (one pair for Sweden and one pair for both Denmark and Norway) assessed the included COI policies of each medical school independently. Disagreements were resolved by discussion. If consensus could not be reached, a third assessor adjudicated.
For each included country, we reported scores for the 11 policy items and for oversight and sanctions. We also reported the mean item score per country.
Since the study focused on institutional policies, rather than personal conflicts of interest information, no ethics approval was required according to Scandinavian law.