Abstract
Introduction
Many aspects of sleep and circadian physiology appear to be sensitive to characteristics of the studied population, most notably sex. While recent research robustly highlights the importance of considering participant-level demographic information, it is not clear to what extent this information is available in the large body of already published literature. In this systematic review, we evaluated the study sample characteristics in the published sleep and chronobiology research over the past 40 years.
Methods
Articles published between 1979 and 2019 (odd years) in the top eight sleep and chronobiology journals, identified by their five-year Impact Factor, were found through MEDLINE. 6,777 articles were initially included for screening. Inclusion requirements included conducting original research, reporting human data, and recruiting volunteers. The reporting of sample size, age, sex, gender, ethnicity, level of education, socio-economic status, and profession of the study population was scored binarily (0 = not reported), and any reported aggregate summary statistics for these variables were recorded. Funding source, geographical location and clinical focus of the article were examined, as well as whether data were analyzed including any of the demographic variables as covariates.
Results
~75% of screened articles met inclusion criteria. While >90% of studies reported age or sex, all other variables were reported in <10% of cases. We found that sex balance greatly changed over the years, from a ~3:1 male to female ratio in the 1990s to a near-equal representation in the 2010s. Overall, ~75% of studies recruited both male and female participants. Of studies recruiting a single sex, ~50% all-female studies focused on a sex-dependent feature, compared to <5% in all-male studies.
Conclusion
In this comprehensive review, we found that the majority of studies report at least sex or age, while many other important variables are typically not reported. Reporting quality is highly variable, indicating an opportunity to standardize reporting guidelines for participant-level characteristics to facilitate disaggregated data analyses.
Support (if any)
Wellcome Trust (204686/Z/16/, 204686/Z/16/A); Linacre College, University of Oxford.