34 Chapter 2 aforementioned inclusion criteria, but were regarded as potentially discussing APRs or sleep problems as part of toxicity registration during trials. Therefore, the full texts of these articles were reviewed as well. Data extraction Data from each study were extracted independently by two authors (Annelienke M. van Hulst and Shosha H.M. Peersmann) based on predefined items: study design, number of participants, mean age, type and dose of steroids, type of APR/sleep outcome, method of measuring APR/sleep outcome, risk factors, method of measuring risk factors and results (often descriptive/percentages). Disagreements were resolved by consensus (Annelienke M. van Hulst and Shosha H.M. Peersmann). If necessary, a third reviewer was consulted (Raphaële R. L. van Litsenburg). Assessment of risk of bias of individual studies To assess risk of bias, the Quality In Prognosis Studies (QUIPS) tool was used. The QUIPS systematically appraises risk of bias in individual studies of prognostic (risk) factors.34 The Cochrane Prognosis Methods Group recommends the use of this instrument.35 The QUIPS ascertains high, moderate or low risk of bias on six domains: (1) study participation, (2) study attrition, (3) prognostic factor measurement, (4) outcome measurement, (5) study confounding, and (6) statistical analysis and reporting. Each study was independently rated using the QUIPS tool by Annelienke M. van Hulst and Shosha H.M. Peersmann after which the scores were discussed to resolve any disagreements. A third reviewer was available when necessary (Raphaële R. L. van Litsenburg). In line with the recommendations of Hayden and colleagues (2013), we assessed each domain and did not rate a summated risk of bias score for individual studies based on the six domains.34 See Supplemental Table 2 for definitions and application of the QUIPS domains. To summarize the quality of individual study results, we took into account: the number of QUIPS domains scoring high on risk of bias, the sample size of APRs/sleep outcomes and whether a study was a priori designed to study risk factors of steroid-induced APRs or sleep problems. We considered a study of lower quality when it entailed more high risk of bias domains, was not a priori designed and had a small sample size. A color-coding was used to indicate our considerations: red (lower quality), orange (medium quality), and green (higher quality). Assessment of grading evidence across studies and synthesis of results To systematically evaluate the quality of summated evidence for each study question and to identify the level of evidence for each risk factor of either APR or sleep problems, we used the Grading of Recommendations Assessment, Development and Evaluation
RkJQdWJsaXNoZXIy MTk4NDMw