1. Searching and selecting relevant studies
Our search resulted in 20091 records. After duplicate removal, title and abstract screening, 4412 were excluded and 5549 were assessed for eligibility based on the inclusion criteria described above. We extracted data on 4384 eligible studies and mapped their content onto our pre-established coding scheme.
(Insert Figure 1)
2. Categorizing relevant studies according to characteristics and content
i. Study characteristics
The following study characteristics were retained and made available on the database for every article: year of publication, language, study location and study design. There has been an increase in the mean annual number of peer-reviewed publications in Cameroon during our study period. Most studies in the database were published in English (n=3494, 79.7%), conducted in the Centre region (n=1972, 45.0%); with an observational study design (n=3144, 71.7%) ( See Table 2 and Figure 2)
Table 2
Study Characteristics
Characteristics
|
N (%)
|
Language, N (%)
English
French
|
3494
890
|
(79.7)
(20.3)
|
Study Location
|
|
|
Adamawa
|
195
|
(4.4)
|
Centre
|
1972
|
(45.0)
|
East
|
211
|
(4.8)
|
Far North
|
226
|
(5.2)
|
Littoral
|
747
|
(17.0)
|
North
|
216
|
(4.9)
|
North West
|
480
|
(10.9)
|
South
|
239
|
(5.5)
|
South West
|
561
|
(12.8)
|
West
|
331
|
(7.5)
|
Nationwide
|
90
|
(2.1)
|
Not reported
|
812
|
(18.6)
|
Study Designs
|
|
|
Experimental
|
250
|
(5.7)
|
Observational
|
3144
|
(71.7)
|
Secondary Analysis
|
283
|
(6.5)
|
Qualitative
|
181
|
(4.1)
|
Mixed Methods Studies
|
88
|
(2.0)
|
Other
|
437
|
(10.0)
|
(Insert Figure 2)
ii. Content
The full CAMHRED coding scheme or taxonomy consists of 10 main domains divided into subdomains. Each domain represents a content category used to describe the focus of research output from Cameroon. These domains include Disability, Diseases and Health Conditions, Health Systems, Medical Specialties, Pharmaceutics, Public Health, Providers, Population, Social Determinants of Health, Sexual and Reproductive Health.
The top four most coded domains in the CAMHRED were Diseases and conditions (n = 3524, 80.4%); Medical Specialties (n=3903,89.0%); Population (n = 2267,51.7%) and Public Health (n =2253,51.4%). Within these domains, the most common subdomains were Infectious and parasitic diseases (n = 1194, 27.2%); Infectiology (n = 2005, 45.7%); Children (n = 654,14.9%); Disease surveillance (n = 1661,37.9%) respectively. Subdomains within the same domain were not mutually exclusive.
Case study: using the database for local evidence mapping and knowledge gap analysis
We adapted the global evidence mapping core tasks and steps (see figure 3) to conduct an evidence gap analysis on sexual and reproductive health in Cameroon. We used the database to identify relevant studies to produce gap maps and recommendations for future research. A summarised account of the methods we used is available below and can inform future protocols for local evidence mapping.
(Insert Figure 3)
i. Developing questions
We used a list of research priorities drawn up by researchers and stakeholders during a deliberative dialogue on sexual and reproductive health in Cameroon in April 2018.(34) Other sources for broad themes which can inform question development include:
- National health priorities available through Ministries of Health, National health policy and strategy documents,
- Consultations, online surveys, evidence mapping workshops with expert and stakeholders
- Preliminary literature searches
It might be necessary to convert broad themes from the sources above into specific research questions depending on the intended use. Research questions might differ depending on whether you are creating an evidence gap map or conducting a narrative synthesis.
For our case study, we selected the following research question from the stakeholder dialogue: “what is the contraceptive prevalence among adolescents in Cameroon?” To conduct an evidence, gap analysis, we transformed this question into “what are the gaps in research on contraceptive prevalence among adolescents in Cameroon?”
ii. Prioritisation
Prioritisation of research questions can be a formal or informal process. Specific questions might be prioritized based on salient issues from stakeholders’ perspective or the availability of funding to address a specific population or problem.
Specific populations identified based on stakeholder concerns (particularly decision makers) included adolescents, persons living in rural areas and regions with poor sexual and reproductive health indicators. (Reference stakeholder dialogue report)
iii. Searching and selecting studies
Our search strategy was based primarily on a keyword search of the main concepts from our research question. We conducted the search exclusively in CAMHRED 1.0 (https://camhred.org/) as the database provided a comprehensive one-stop-shop for health research and evidence from Cameroon. (35)
There are two main options for searching CAMHRED. The first is a free text search which involves searching combinations of keywords relevant to your research question and selecting filtering parameters (language, year, province, study design) based on your inclusion criteria. For example, searching “contraception” OR “family planning” will yield all articles in which the terms family planning or contraception are mentioned in the title or abstract, in French or English. (See Appendix 2 for screen captures from the database)
The second option is a guided search which takes advantage of the content taxonomy used to categorize articles in the database.
For example:
Select content domains: Sexual and Reproductive Health > Subdomain: Contraception, Family Planning
Check content domains : Population > subdomain : adolescents (optional)
This search will yield all the articles tagged as “Contraception” or “family planning” within the topic of “Sexual and Reproductive Health” and in which the population of interested was “Adolescent and Youth”.
For our case study, we used the guided search and retrieved 174 studies coded as contraception OR family planning. Applying the population (adolescents and youth) code resulted in 68 studies including 10 studies which had been coded both as contraception and family planning. The results were downloaded as CSV files from the database for full text screening and data extraction. After excluding studies not available as full text (n= 6) and those in which contraceptive prevalence was not measured (n = 8); we included 44 studies for our evidence maps and gap analysis. A detailed description of the methods and results for our case study is reported elsewhere. (36)
iv. Data extraction and creating evidence maps
After downloading search results as CSV format, we created a mapping protocol to determine further inclusion criteria, full text procurement, data extraction modalities and analysis relevant to the prioritized research question as detailed in Appendix 3.
The complete results of our evidence mapping, and gap analysis are reported elsewhere.(36) However, we have included illustrations of the findings here to provide a glimpse into the applications of local evidence mapping. For instance, we identified knowledge gaps on the contraceptive prevalence of adolescents within specific age categories, regions and places of residence (urban versus rural). We found that there is paucity of disaggregated data on contraceptive use among young adolescents (10-14 years old) and there are geographical and residential disparities (rural areas) in the availability of data.
(Insert Figure 4)
In addition to identifying gaps in the literature, we were able to make recommendations for future research including suggestions on the type of study design, content (research question) and reporting.
(Insert Figure 5)