This blog was written by Wadhwani Nimarta, Senior Research Assistant at Aga Khan University Institute for Educational Development (AKU-IED); Waqas Faraz and Anila Hassan, both MPhil scholar at AKU-IED; and Sadia Bhutta, Associate Professor at AKU-IED, Pakistan.
Among the different types of literature reviews, systematic reviews are one of the distinct and useful ways to identify the trends and map out the key evidence available for a particular research area. Despite the emergent use of systematic reviews across academic fields, there is uneven capacity for conducting reviews between high-income and low-and middle-income countries due to technical and human resource differences. Despite several challenges, including restricted access to some of the databases and unsubscribed accounts due to high activation charges, there has been an increase in the number of systematic reviews from developing countries, largely from Asia.
Recently, our team from Pakistan initiated a systematic review project using existing open-access databases and free tools available. We selected four interdisciplinary, peer-reviewed, and education-relevant publisher platforms and databases—PubMed, Taylor & Francis, Scopus and EBSCOhost to guide our search. Little did we know that the first step of systematic review – ‘identification’ – which appeared mechanical to just click and download the citation, would turn out to involve a number of technicalities that require careful handling. Therefore, this blog aims to share the technical challenges encountered by researchers from Pakistan during database search and citation download, and share strategies we opted for to move forward.
- Uneven filters
‘Every database works differently’, was the initial disclaimer by the librarian who was consulted in designing the search query and selecting relevant databases. It ultimately became our mantra while navigating different databases for citation extraction. Some databases, such as EBSCOhost, offer an extensive list of filtering options (e.g., date, geography, language, type of publication), whereas other platforms, such as Taylor & Francis, provide very few filters (date, language). This variation in filtering options underscored an important lesson: each database requires a tailored approach. This reinforced the need for flexibility and careful documentation to maintain consistency across databases.
- Variability in export formats
The story of database differences does not end with just uneven filter numbers. Different databases also have a preferred citation export format. For example, PubMed only allows downloading in a particular format (.nbib), whereas some other databases provide an option to export in different formats (.ris; .nbib). The dilemma is not the differential citation export formats, but the requirement of citation management software, which might, in some cases, have a preferred style for citation formats to be imported. We were using the Rayyan citation manager, which accepted citations in different formats, hence we did not face difficulties in citation imports. Otherwise, it would have been an additional struggle to convert the citation format of different databases and make them uniform for being accepted by a particular citation management software. The key lesson here is to confirm the specifications for acceptable citation import formats of the citation management software you are using and then export citations in those compatible formats for smooth importing.
- Export limitation
Downloading the citation itself is a tedious task. It gets more hectic when some databases, such as PubMed, are limited to one citation download at a time. When the search yields a large number of records, managing large search results can become a time-intensive process. At the same time, some platforms make life a little easier by giving an option to download 10-20-50 citations in one click. For example, Scopus contains an option to export all citations at once, allowing for the saving of time and energy. The key lesson is to anticipate platform-specific download limitations and allocate additional time for managing large search outputs that do not allow bulk citation exports.
- Conflict in numbers
Some databases indicated a discrepancy between the number of searches they displayed and the number of studies they allowed you to download. These databases limit the export numbers, particularly when one does not have a subscription-based account. In such a case, no matter how many searches the particular query entails, it will show you a limited number. For example, Taylor & Francis did not allow exporting citations beyond two thousand, despite the fact that the number of searches was higher than two thousand. In this case, we had to play around with the database by applying different filters. Because our scoping review aimed to capture studies published between 2022 and 2025, we conducted separate searches for each year. This led to a reduction in search numbers as the system treated each yearly search as a separate query rather than aggregating the results across years. The lesson is clear: familiarise yourself with database limitations.
- Handling database slowdowns
Along with displaying a limited number of studies, the exporting process of some platforms, such as EBSCOhost, slows down after 300 citation exports. Surprisingly, this database contained three thousand searchers, which were half of the total citations found on the research topic. Again, in this case, we had to be tactful and boil down our searches not only by years but also by month. We had to pool all of our reviewers on to this database to make the exports quick and efficient. The key lesson is to understand system performance thresholds and structure your team accordingly.
Conclusion
To sum up, the identification stage of a scoping review is rarely a straightforward process, particularly when someone is a novice in the field of systematic reviews, belongs to the global south, and does not have a subscribed database account. This blog does not aim to discourage, but rather presents guidelines and solutions to work under these constraints. Overall, our experience illustrates that each database comes with its own technical constraints. Understanding and resolving these challenges is essential for maintaining the accuracy, efficiency and transparency of the review.
