P-318 Application of the Standards for Reporting of Diagnostic Accuracy studies (STARD) 2015 to research on blood biomarkers of endometriosis

Abstract

Abstract Study question How completely reported are studies on blood biomarkers of endometriosis, according to their adherence to the STARD 2015 checklist? Summary answer Studies on blood biomarkers of endometriosis are often incompletely reported, with only an overall adherence rate of 28.5% to the STARD 2015 checklist. What is known already Laparoscopy, the gold standard diagnostic method for determining endometriosis, is invasive, expensive, requires anesthesia and is associated with risk of vascular/ visceral damage. Therefore, less invasive diagnostics are needed. Many studies have been published on blood biomarkers of endometriosis with variable methodological quality. To assess applicability and risk of bias of diagnostic accuracy studies, completeness of study reporting must first be ensured. STARD 2015 checklist was introduced to guide reporting of diagnostic accuracy studies. The degree of adherence in studies on blood biomarkers of endometriosis to this 30-item STARD 2015 checklist is not known. Study design, size, duration Our systematic review included 126 studies from the 2016 Cochrane review on blood biomarkers of endometriosis. We added 84 more studies after updating the search in Pubmed, CINHAL and Web of Science, for the period May 2015- July 2021, with the keywords: endometriosis and biomarkers/ non-invasive diagnosis/ laboratory tests/ blood test, using the same eligibility criteria as in the Cochrane review. Totally, 210 studies were assessed for adherence to STARD 2015 checklist. Participants/materials, setting, methods STARD 2015 items were scored 1 if reported, or 0 if unreported. Each study received an adherence rate (percent of reported STARD items). Every item received individual item’s reporting rate (percent of its reporting across studies). Effect of publication year on study adherence and item reporting rates was examined by linear and logistic regressions, respectively. Journal impact factors were classified to tertiles and their relation to study adherence rate examined with one-way ANOVA. Main results and the role of chance Adherence rate was 28.5%, with 8.8± 3.1 reported items per study (range 2-18). Study adherence to STARD 2015 significantly increased with advancing publication years [r = 0.344 (95% CI 0.219 – 0.458), p < 0.001]. Journal impact factor had no effect on adherence to STARD 2015 (P = 0.274). Items with reporting rates >50% included #1 (identification as diagnostic accuracy study), #5 (prospective or retrospective), #7 (participants identification), #10a (index test), #10b (reference standard), #14 (diagnostic accuracy measures), #21a (disease severity), #21b (alternate diagnosis), and #22 (time between index test and reference standard). Items with reporting rates 20 to 50% included #6 (eligibility criteria), #8 (when/ where participants identifies), #12a (test positivity cutoff), #20 (participant characteristics) and #26 (limitations). Reporting rate < 20% seen in items #3 (background), #4 (hypothesis), #9 (consecutive /random recruitment), #13a/b (blinding of index test/ reference standard performers), #15 (intermediate results), #16 (missing data), #17 (analysis of variability), #18 (sample size), #19 (flow diagram), #23 (cross tabulation), #24 (diagnostic accuracy precision), #25 (adverse events), #27 (implications), #28 (registration), #29 (protocol), #30 (funding). Reporting improved for items: #1, #8, #10a, #10b, #12a, #14, #17, #18, #19, #20 and #25 and declined for items #9, #21a and #27 over publication years. Limitations, reasons for caution As the consequence of our choice to adhere to the methodology of the original Cochrane review, a limitation of our study is that we excluded non-English language studies and limited the search to the previously chosen 3 databases only. Wider implications of the findings With inadequately reported diagnostic accuracy studies, the derived evidence can be misleading. Our findings represent a baseline evaluation of the reporting status of studies on blood biomarkers of endometriosis. Our results can guide researchers and editors about poorly reported STARD 2015 items, aiming to improve their future reporting quality. Trial registration number PROSPERO # CRD42021259990