Nelson Systematic Reviews to Answer Healthcare, Questions, 2e

68

Chapter 6 • Selecting Studies for Inclusion

must decide whether to include the trial and, if so, refine their eligibility criteria. One solution is adopting a best evidence approach where studies with less clearly defined populations are used only if there is insufficient evidence from studies with more precisely defined populations. Intervention Some studies combine participants with different treatments into a single intervention group. Combining treatments is acceptable for systematic reviews examining a class effect , but treat ments must be similar enough that they can be considered a related group for major outcomes, such as bisphosphonates for treatment of osteoporotic fractures. However, studies with com bined treatment groups may be less relevant when interventions are dissimilar or comparisons are intended. Whether these studies are included depends on the purpose of the systematic review. For example, a review focusing on the effects of individual drugs could include studies examining a class effect when there are gaps in evidence. Comparator Comparators are often described broadly or imprecisely in eligibility criteria. Difficulties occur when comparators include nonspecific treatments, such as usual care or no treatment. Systematic reviewers may select studies based on their interpretations of imprecise criteria, introducing bias when there is variation across studies. Usual care may differ by study setting or when the study was done, for example. Understanding the key elements of usual care that differ is essential to interpreting their effects on outcomes. For example, in a systematic review of hyperbaric oxy gen to treat traumatic brain injury, furosemide and mannitol were given to all participants as standard therapy to reduce presumed elevated intracranial pressure in older studies. 8,9 Newer studies used invasive intracranial pressure monitoring to guide treatment more selectively and provided phenytoin to all participants for prophylaxis of seizures. Outcome When outcomes are precisely defined in the eligibility criteria and consistently reported in study publications, variability among systematic reviewers is less likely. However, important details are often missing or inadequate. For example, a systematic review of screening for developmental delay in young children would include an overwhelming number of studies with a broad range of outcomes requiring further definition, whereas a review of the effectiveness of screening for speech and language delay in preschool age children would focus on very specific outcomes. Study Design Systematic reviewers’ decisions about the inclusion of observational studies commonly intro duce variability. 2–5 Although there is broad agreement about the hierarchy of study designs, 10–12 thoughts vary about the role of observational studies 1,13,14 and classification of observational study designs. 15 The specific study designs eligible for inclusion for each research question need to be clearly defined and refined as necessary, particularly if higher-level evidence is not available. Other Selection Criteria Additional selection criteria include how multiple publications stemming from an individual, unique study will be handled. Care should be taken to identify the primary and related publications to prevent double counting and possibly inclusion of duplicate data in analyses. Systematic review ers need to establish rules and processes for handling multiple publications that may vary by topic. The selection of studies based on the availability of data for meta-analysis also contributes to variability. Decisions include whether data derived from graphs should be included, statistical imputation methods will be used when precise variance data are not reported, and missing data will be sought by contacting study authors. These decisions will affect study selection and must

Copyright © 2024 Wolters Kluwer, Inc. Unauthorized reproduction of the content is prohibited.