Bringing a systematic element to complex focussed literature reviews

Article type
Authors
Swift SL1, Leadley RM1, Armstrong N1, Kleijnen J1
1Kleijnen Systematic Reviews Ltd, Escrick Business Park, York
Abstract
Background:
Focussed literature reviews represent a critical way to appraise a pool of published and grey literature to address questions related to incidence, prevalence, treatment patterns, mortality, progression or screening. They differ from systematic reviews of interventions in that they often aim to make a large review more manageable and they tend to be less resource-intensive. However, the decision-making process that informs the choice of which studies to focus on tends to be arbitrary and thus can be prone to review author bias.

Methods:
We have developed a new approach to improve the reliability of this decision-making process for large, complex research questions. In this approach, each study that is being considered for relevance at the full paper screening stage is subject to a mini data extraction focussed on five key elements: study size, country, study representativeness, study quality and the amount of data reported for each patient population of interest and outcome of interest. This approach was piloted on a complex set of research questions centred on the per-country incidence and prevalence of diabetic kidney disease worldwide. Two review authors performed the mini data extraction and quality assessment: one review author extracted the data and the other review author checked it. Quality assessment was performed using the Joanna Briggs Institute (JBI) critical appraisal checklist for studies reporting prevalence data. On a per-country basis, the two review authors then chose the three most relevant studies for each patient population and outcome, and made trade-off decisions based on the five key elements described above.

Results:
As an illustrative example, 10 studies were identified that reported the prevalence of diabetic kidney disease in the USA. Seven studies were discarded for the following reasons: two studies were of low quality, three studies had a small sample size and two studies were not representative of the USA. Three studies were finally included to inform the prevalence question. Despite the additional time required to perform mini data extractions on those studies that were ultimately excluded from the final focussed literature review, we found that this was not a particularly burdensome or time-consuming process.

Conclusions:
Ultimately, this approach can bring an important systematic element to a focussed literature review, and is designed to increase the transparency and reliability of the research data.