Academic research productivity of post-graduate students at Makerere University College of Health Sciences, Uganda, from 1996 to 2010: A retrospective review

Background: Research is a core business of universities globally, and is crucial in the scientific process as a precursor for knowledge uptake and use. We aimed to assess the academic productivity of post-graduate students in a university located in a low-income country. Methods: This is an observational retrospective documentary analysis using hand searching archives, Google Scholar and PubMed electronic databases. The setting is Makerere University College of Health Sciences, Uganda. Records of post-graduate students (Masters) enrolled from 1996 to 2010, and followed to 2016 for outcomes were analysed. The outcome measures were publications (primary), citations, electronic dissertations found online or conference abstracts (secondary). Descriptive and multivariable logistic regression analyses were performed using Stata 14.1. Results: We found dissertations of 1172 Masters students over the 20-year period of study. While half (590, 50%) had completed clinical graduate disciplines (surgery, internal medicine, paediatrics, obstetrics and gynaecology), Master of Public Health was the single most popular course, with 393 students (31%). Manuscripts from 209 dissertations (18%; 95% CI, 16-20%) were published and approximately the same proportion was cited (196, 17%; 95% CI, 15-19%). Very few (4%) policy-related documents (technical reports and guidelines) cited these dissertations. Variables that remained statistically significant in the multivariable model were students’ age at enrolment into the Masters programme (adjusted coefficient -0.12; 95% CI, -0.18 to -0.06; P < 0.001) and type of research design (adjusted coefficient 0.22; 0.03 to 0.40; P = 0.024). Cohort studies were more likely to be published compared to cross-sectional designs (adjusted coefficient 0.78; 95% CI, 0.2 to 1.36; P = 0.008). Conclusions: The productivity and use of post-graduate students' research conducted at the College of Health Sciences Makerere University is considerably low in terms of peer-reviewed publications and citations in policy-related documents. The need for effective strategies to reverse this 'waste' is urgent if the College, decision-makers, funders and the Ugandan public are to enjoy the 'return on investment' from post-graduate students research. Our study was driven by the hypothesis that there is an association between the type of post-graduate degree and the main outcome, which we defined as the proportion of dissertations from which at least one manuscript was published. We conducted a retrospective cohort documentary analysis and reviewed hard copy records (admissions, research proposals and dissertations) from 1996 to 2010. Further, we performed an electronic search for publications, citations, abstracts and conference presentations in Google Scholar, PubMed and the Makerere University online repository of electronic copies of dissertations from 1996 until June 2016. A systematic review estimated that experimental studies with positive results were published approximately 4 to 5 years after completion, and those with null or negative results in 6 to 8 years [17]. Thus, we considered a range of 6 years (from 2010) to 20 years (from 1996) of study period sufficient to identify publication outputs. We take cognisance of the many meanings and definitions of research productivity [18]. Our choice of outcome variables is informed after review of bibliometric literature, as the more suitable measures of academic research productivity. We conducted this study at MakCHS, formerly Makerere Medical School. MakCHS is the oldest health sciences education and research institution in East Africa, having been established in 1924 [19]. Further, Makerere University is one of the highest ranking post-graduate education and research institutions in Africa [20], providing courses in basic sciences and in clinical and public health disciplines. As such, MakCHS has substantially contributed to the research pool by post-graduate students in Africa. Our unit of analysis was an academic research dissertation. We included all Masters level dissertations whose proposals were reviewed by the higher degrees research and ethics committee between 1996 and 2010. These included projects about health-related research topics in the field of veterinary medicine, education, agriculture, or the social sciences or humanities in the spirit of the ‘one-health’ concept to integrate animal, environmental and human health [21], as well as social determinants of health. We excluded research projects conducted by doctorate students or established faculty, except where registered as a Masters student. We included all the available 1172 dissertations for Masters students admitted between 1996 and 2010. We opted to study the whole sample since this population was sufficiently small and available. We used a data extraction form (Additional file 1) and extracted data from hard copies of research proposals and dissertation reports as well as electronic manuscript publications. We used Google scholar to obtain electronic data about citations and types of documents citing the post-graduate student research. We obtained dissertation records from the directorate of research and graduate training and the office of the academic registrar at Makerere University. We corroborated data on the type and year of post-graduate qualifications with publicly available registers of the national health professional councils (doctors, pharmacists and nurses) who constituted by far the majority of students. We did not corroborate this information for students who qualified with non-medical degrees at undergraduate training, all of whom pursued non-clinical post-graduate degrees, as there is no public registry for them. We measured the primary outcome (dependent variable) for productivity as the proportion of dissertations of research projects by postgraduate students from which at least one manuscript was published in a peer-reviewed journal. Our main secondary outcome was citation of research, measured as the proportion of dissertations by postgraduate students from which the dissertations or manuscripts from dissertations were cited in policy-related documents (technical reports or guidelines) or peer-reviewed journals. As we expected some dissertations to be cited more than once, we considered the first citation by time. Additional outcomes were presentation of abstracts in scientific meetings and uploading into the electronic repository of dissertations at Makerere University Library. We considered publication a key step in the dissemination of research, and thus a precursor for use in decision-making or other applications as appropriate. In addition, we reported the time to any of these outcome events to better understand the average shelf life of student research projects before being translated into tangible outputs or shared for broader usage (publications, conference presentations, citations). The main independent variable was the type of university degree (biomedical, clinical or public health). We grouped these degrees according to the similarities in the duration of the post-graduate training, nature of the training, time resource available to conduct research as well as size of the programme. As such, Masters of Public Health, Masters of Health Services Research and Masters of Science in Clinical Epidemiology and Biostatistics emphasised research and statistical methods, and were allotted between 2 (full time) or 3 (part time) years to completion. The clinical disciplines (Masters of medicine in surgery, or medicine, or paediatrics, or obstetrics and gynaecology) prioritised practical training in hospital attending to patients, with relatively less time for developing research methodology over a 3-year period of training. Finally, the basic sciences graduate degrees had a substantial component of laboratory-based work. Additional independent variables were demographics of post-graduate students (age, sex, marital status), research environment (funding, period), research design (quantitative or qualitative), the level of research (sub-individual, individual and population) and priorities (Millennium Development Goals (MDGs) or the health system strengthening building blocks of WHO) [22, 23]. This description of levels of research pertains to sub-individual (laboratory), individual (clinical) and population (public health) levels [24]. In terms of health priorities, we categorised studies according to the health-related MDGs [22], with MDG 1 on ending hunger; MDG 4 on child health; MDG 5 on maternal health; MDG 6 on HIV/AIDS, malaria, TB and other diseases; and MDG 7 on environmental health (water and sanitation). We used the framework of the WHO pillars for health system strengthening to assess the post-graduate research projects (governance, financing, human resources, service delivery, information systems and access to medicines, vaccines health technologies) [23]. Secondly, we organised the student projects into the three periods that may have influenced the research and health policy environment. The first period was from 1996 to 2000, after a call for research investment considering the 10/90 gap [25]. In brief, this highlights inequity in research investment (10%), where the bulk (90%) of health problems are commonly in low-income countries. The second period was from 2001 to 2005, at the start of MDGs [22], while the third was from 2006 to 2010, which would depict more recent events such as the 2004 and 2008 inter-ministerial summits in Mexico and Mali, respectively, about the use of research in health decision-making and policy [26]. Noteworthy, the Makerere Medical School transitioned into the collegiate system in 2007, the third period that could have attracted or altered the distribution of human and other resources, as well as prioritisation of research at the school and department levels. Our final classification addressed the types of research design (quantitative or qualitative) and research level as described under the variables section [24]. We analysed the data using Stata version 14.1 software and relevant health research priorities frameworks for WHO and the MDGs. We used frequencies, proportions and measures of central tendency to conduct initial analyses and tabulated these. We used multivariable logistic regression to explore the determinants of productivity (publication) and use (citations) of student’s research. In the multivariable logistic regression model we included covariates that altered the relationship between the primary outcome (publication) and main exposure (degree type) by more than 10%. We chose these covariates for testing a priori guided by literature but also our expert knowledge in the field of health systems and health policy. In the time-to-event analysis, we analysed for events after completion of dissertation project. We used two sided tests with P < 0.05 as the significance level.