This is a small study and dependent on the respondents answering accurately. We were careful to try and establish the denominator at each hospital, and had a reasonable response rate. The study sites are among the largest government funded training institutions in East and West Africa; Internet access is more likely to be available at these institutions. The results of the survey may be generalisable to other large teaching hospitals in countries in Africa with similar facilities.
We were surprised that postgraduate doctors in training were so dependent on text books as their main source of information. However, this probably reflects the structure of most postgraduate medical courses around clinical competence and knowledge, with limited expectations on understanding of current research and policy debates. Postgraduate training for doctors in the study countries involves a research based dissertation, but accessing up to date online articles might not be a strict requirement. Nevertheless, Internet access was high, and particularly in Muhimbili and Mulago hospitals, which have a more extensive research portfolio.
The high use of commercial outlets was astonishing, with Internet cafés being the main place of Internet use for postgraduate doctors in Lagos teaching hospital and Yaoundé University. This may well reflect the limited investment in cabling and other infrastructure in these postgraduate teaching institutions; other institutions may have invested more. The question specifically stated this was last place of use "for medical and scientific literature" so we are fairly sure this is how they responded, but were unable to validate this against actual access. The majority of all respondents stated that the last time they retrieved a full text article was through the Internet, illustrating the importance of this source of medical knowledge for postgraduates in these countries.
The results show low use of CD-ROM resources across all sites and this could be because this type of resource is complicated to distribute, can become out of date rapidly, and there are fewer organisations supplying material in this format now.
We found awareness of initiatives varied widely by country and initiative. Low awareness of the BMJ website in Yaoundé University, Cameroon could be to do with provision of material in English, although we do not have specific data on this. Awareness of the Cochrane Library appeared higher in Mulago hospital, Uganda compared to other sites; this could be attributed to awareness raising workshops on evidence informed practice by visiting researchers.
The log in data from HINARI (table ) correlate with higher reported awareness of HINARI and use of a computer within an institution. Fewer people in Lagos and Yaoundé had heard of HINARI, and used Internet cafés for access-presumably as institutional access was poor. The most use made of HINARI was through the externally funded research institution in The Gambia and Mulago hospital, Uganda where HINARI has been extensively promoted via institutional training.
The qualitative findings indicate that provision of passwords could be better managed and teaching hospitals could do more to publicise the various electronic resources available to students and staff. What was surprising was the degree to which librarians controlled access: it seemed that some librarians saw it as their job to provide the articles for the doctors, rather than allow the doctors to access websites themselves and in these institutions passwords were not freely available. The qualitative data also suggest users have difficulty with passwords and access, and that HINARI may wish to review the ease of use of its interface. Respondents described problems accessing free full text articles via Medline which may be because they are not logged into HINARI. Once logged in, users can search for articles through Medline in the usual way and have direct access to the full text. Staff at the WHO are currently improving the interface and publicising HINARI to help improve access to passwords in institutions.
Our findings suggest power interruptions and inadequate computing facilities continue to be major constraints; combined with apparent problems logging in to Internet based initiatives and difficulties acquiring passwords from institutional gate keepers, these factors represent considerable barriers to the effective use of initiatives providing access to online journals at the teaching hospitals in our study. Relatively small investments in connectivity which are carefully managed by institutions will increase access to large amounts of information. In addition, HINARI would be more effective with strong institutional endorsement and management to promote and ensure access.