Background: Systematic reviews are often used when attempting to understand large and evolving areas of investigation, such as dementia research. However, the publication rate of systematic reviews is increasing, raising the possibility that their number will become unmanageable. Our aim was therefore to forecast the future publication rate of dementia-related systematic reviews.
Methods: We based our forecasts on the number of prior relevant systematic reviews identified in our EMANATE database (updated 7 February 2017), based upon a systematic review of all dementia-related systematic reviews identified from five databases (MEDLINE, EMBASE, PsycINFO, CINAHL, and Cochrane Database of Systematic Reviews). Publications from 2016 onwards were excluded due to reduced numbers reflecting the time lag between publication and inclusion in the databases. Preliminary visual inspection revealed a nonlinear relationship. Therefore, we modelled the predicted number of publications and their 95% prediction intervals (PIs) using polynomial regression. Predicted number of publications were calculated for observed years (1989-2015), and forecast for the following 10 years (2016-2025).
Results: The observed number of publications ranged from 2 in 1989 to 356 in 2015. A quartic model gave the best fit to the observed data in comparison with linear, quadratic, cubic and quantic models (indicated by the lowest Root Mean Square Error and highest adjusted R2 value). The predicted number of publications fit the observed data very well (R2=0.99; Figure 1). Our model forecasts that the number of dementia-related systematic reviews will reach 892 by year 2020 (95% PI 765-1019), and 1896 by year 2025 (95% PI 1486-2306).
Conclusions: We forecast that the publication rate of dementia-related systematic reviews will be more than five times higher in 2025 in comparison with 2015. Methodological innovation may be needed to ensure that the value of systematic reviews is not diminished by their burgeoning number.