Reporting practices for genomic epidemiology of tuberculosis: a systematic review of the literature using STROME-ID guidelines as a benchmark

Abstract

Summary Background Pathogen genomics have become increasingly important in infectious disease epidemiology and public health. The Strengthening the Reporting of Molecular Epidemiology for Infectious Diseases (STROME-ID) guidelines were developed to outline a minimum set of criteria that should be reported in genomic epidemiology studies to facilitate assessment of study quality. We evaluate such reporting practices, using tuberculosis as an example. Methods For this systematic review, we initially searched MEDLINE, Embase Classic, and Embase on May 3, 2017, using the search terms “tuberculosis” and “genom* sequencing”. We updated this initial search on April 23, 2019, and also included a search of bioRxiv at this time. We included studies in English, French, or Spanish that recruited patients with microbiologically confirmed tuberculosis and used whole genome sequencing for typing of strains. Non-human studies, conference abstracts, and literature reviews were excluded. For each included study, the number and proportion of fulfilled STROME-ID criteria were recorded by two reviewers. A comparison of the mean proportion of fulfilled STROME-ID criteria before and after publication of the STROME-ID guidelines (in 2014) was done using a two-tailed t test. Quasi-Poisson regression and tobit regression were used to examine associations between study characteristics and the number and proportion of fulfilled STROME-ID criteria. This study was registered with PROSPERO, CRD42017064395. Findings 976 titles and abstracts were identified by our primary search, with an additional 16 studies identified in bioRxiv . 114 full texts (published between 2009 and 2019) were eligible for inclusion. The mean proportion of STROME-ID criteria fulfilled was 50% (SD 12; range 16–75). The proportion of criteria fulfilled was similar before and after STROME-ID publication (51% [SD 11] vs 46% [14], p=0·26). The number of criteria reported (among those applicable to all studies) was not associated with impact factor, h-index, country of affiliation of senior author, or sample size of isolates. Similarly, the proportion of criteria fulfilled was not associated with these characteristics, with the exception of a sample size of isolates of 277 or more (the highest quartile). In terms of reproducibility, 100 (88%) studies reported which bioinformatic tools were used, but only 33 (33%) reported corresponding version numbers. Sequencing data were available for 86 (75%) studies. Interpretation The reporting of STROME-ID criteria in genomic epidemiology studies of tuberculosis between 2009 and 2019 was low, with implications for assessment of study quality. The considerable proportion of studies without bioinformatics version numbers or sequencing data available highlights a key concern for reproducibility.

Adherence of SARS-CoV-2 seroepidemiologic studies to the ROSES-S reporting guideline during the COVID-19 pandemic

“…The median adherence to ROSES-S items was 48.1% per study, which is comparable to estimates from studies evaluating adherence to STROBE and STROME-ID. [7][8][9][10][11] There was variation in the completeness of reporting for different reporting items. Domains of the ROSES-S guideline with the least frequently reported items were laboratory methods and results, while the least reported individual items related to serological assays and statistical methods.…”

“…The sample size was determined based on a precision level of ± 7% with 95% confidence, and a 60% estimated adherence to checklist items. 11,17 To ensure a comparable sample size before and after the ROSES-S guideline's publication date, we created strata of six month discretized time periods and randomly sampled 34 studies from each strata. If a given month had only one included study then we conducted additional random sampling of studies from that month to select an additional study -bringing the total to two.…”

“…For instance, estimates of adherence to STROBE's 22 reporting criteria have ranged from 51.4% to 76.5%, [7][8][9][10] with similar estimates of adherence (50%) to STROME-ID. 11 In June 2021, the World Health Organization (WHO) published the "Reporting of Sero-Epidemiologic Studies -SARS-CoV-2" (ROSES-S) guideline as a content-specific extension of the STROBE checklist. ROSES-S is a checklist of 22 criteria for reporting SARS-CoV-2 seroepidemiologic studies regardless of study design.…”

Adherence of SARS‐CoV‐2 Seroepidemiologic Studies to the ROSES‐S Reporting Guideline During the COVID‐19 Pandemic

“…We included a random time‐stratified sample ( n = 199) of peer‐reviewed, pre‐print studies, and gray literature (government and institutional reports) from March 2020 to December 2022 to capture studies before and after the publication of the ROSES‐S guideline (June 2021). The sample size was determined based on a precision level of ±7% with 95% confidence and a 60% estimated adherence to checklist items [ 11 , 17 ]. To ensure a comparable sample size before and after the ROSES‐S guideline's publication date, we created strata of six month discretized time periods and randomly sampled 34 studies from each strata.…”

“…Previous studies suggest that the use of these types of guidelines improves reporting [ 5 , 6 ], yet adherence to such guidelines still appears suboptimal. For instance, estimates of adherence to STROBE's 22 reporting criteria have ranged from 51.4% to 76.5% [ 7 , 8 , 9 , 10 ], with similar estimates of adherence (50%) to STROME‐ID [ 11 ].…”