Definition of eligibility for non-randomised studies in 113 Cochrane Reviews of interventions to prevent healthcare-associated infections

Article type
Authors
López-Alcalde J1, Stallings E2, Cabir Nunes S3, Daheron M3, Fernández-Chávez A4, Bonfill Cosp X5, SEXCOMPLEX W6, Zamora J7
1Cochrane Associate Center of Madrid (Universidad Francisco de Vitoria-Madrid, Hospital Universitario Ramón y Cajal-IRYCIS); Department of Paediatrics, Obstetrics and Gynaecology and Preventative Medicine, UAB
2Clinical Biostatistics Unit, Hospital Universitario Ramón y Cajal (IRYCIS)
3Independent consultant
4Preventive Medicine Unit, Hospital Universitario Ramón y Cajal
5Department of Paediatrics, Obstetrics and Gynaecology and Preventative Medicine, Universitat Autònoma de Barcelona
6Hospital Ramón y Cajal
7Clinical Biostatistics Unit, Hospital Universitario Ramón y Cajal (IRYCIS)/CIBER Epidemiology and Public Health (CIBERESP)
Abstract
Background:
Randomized clinical trials (RCT) are difficult - or even unethical - in infection-control research. Thus, systematic reviews (SR) of interventions to prevent healthcare-associated infections (HAIs) can define non-randomized studies (NRS) as eligible. NRS cover different study designs designated with different labels that are not used consistently among researchers. Review authors make relevant decisions based on study design classification and labels.

Objectives:
To determine how many Cochrane Reviews of interventions to prevent HAIs defined NRS as eligible. To describe the justification provided to include NRS, the NRS defined as eligible, and the design labels used.

Methods:
We conducted a descriptive study. We retrieved all active Cochrane Reviews of interventions to prevent HAIs published in the Cochrane Library before 1 January 2017. Two authors independently used EPPI-Reviewer 4 to select the reviews and to extract data. We resolved disagreements through consensus. We extracted the description of the NRS and the reasons for their inclusion, which were appraised against the criteria presented in Chapter 13 of the Cochrane Handbook (CH).

Results:
We included 113 Cochrane Reviews out of the 7156 screened. A total of 55/113 (49%) reviews considered NRS to be eligible. Only 7/55 (13%) justified this decision with at least one reason. The justifications that were provided varied, and, according to the CH were inadequate in 4/7 (57%) reviews. The eligible NRS were quasi-RCTs, controlled clinical trials, cluster quasi-RCTs, controlled cohort before-after studies, interrupted-time-series studies, controlled interrupted-time-series studies, repeated measures studies, crossover studies, observational studies (cohort and case-control designs), and ‘any other comparative design’. The study design labels used were heterogeneous: indeed 38/55 (69%) reviews used the design label with no further explanation about the study design; 17/55 (31%) reviews described at least one of the key study design features suggested in the CH, but no review described all the features.

Conclusions:
NRS are usually defined as eligible in Cochrane Reviews of interventions to prevent HAIs. However, this decision is rarely justified, the design labels used are heterogeneous, and there is often no description of the key study design features.

Patient or healthcare consumer involvement:
We assessed reviews relevant for decision making.