Systematic reviews (SRs) publications focusing on physical therapy rehabilitation have significantly increased. SRs are known to present the highest level of scientific evidence, thus constituting the most reliable type of research to be used in clinical decision-making in healthcare. In these studies, subgroup analysis is usually used as a statistical control technique to investigate sources of heterogeneity and explore treatment effects in individualized subgroups. However, the analyses recorded in the protocol are not always reported in published SRs, with complete absence, partial reduction in the number of analyses, and even the inclusion of new subgroups not protocolled.

To evaluate the frequency with which physical therapy intervention SRs, published in high-impact journals, perform subgroup analyses that are previously reported in protocols or add post-publication unplanned analyses.

The Rayyan software was used by two independent authors to select all SRs published between March 2020 and August 2022 in the 10 highest impact rehabilitation journals according to the Journal Citation Reports (JCR). Disagreements were resolved by an experienced third reviewer. Subgroup analysis described in the protocol and reported in final publications were compared using descriptive statistics.

3,032 records were identified, of which 2,927 were excluded for not meeting the inclusion criteria. 105 SRs published in journals with impact factors ranging from 4.76 to 10.71 (JCR, 2021) were included. Of these, 60 (57.1%) reported subgroup analyses that were consistent with what was recorded in the protocol; 29 (27.6%) did not report any of the previously registered analyses, and 16 SRs (15.3%) added unplanned analyses in the protocol, with an average of 1.6 new subgroup analyses included in the final publication.

The findings indicate that 43% of SRs present significant discrepancy between the subgroup analyses planned in registered protocols and those reported in published SRs, even in high-impact scientific journals. Thus, it is essential that SRs conducted in the physical therapy preserve as much as possible in the final text, the subgroup analyses planned in their respective protocols, making their results more reliable and accurate for researchers and clinicians in the field.

This study has the potential to highlight shortcomings in the methodological strategies used in SRs in the physical therapy field and, consequently, raise awareness for greater care in the planning and execution of studies that are more transparent and faithful to previously registered protocols, as well as greater caution in interpreting SR results, even if they come from sources considered to be reliable.