Skip to main content

Table 4 Risk of bias in randomized control trials

From: Systematic review of structural interventions for intimate partner violence in low- and middle-income countries: organizing evidence for prevention

  Abramsky (2014) [6] Green (2015) [27] Gupta (2013) [28] Hidrobo (2013) [29] Hossain (2014) [30] Jewkes (2008) [31] Kim (2007) [33] Kim (2009) [34] Kyegombe (2014) [35] Pronyk (2006) [37]
Study design  
 Prospective identification of intervention and comparison groups Yes Yes Yes Yes Yes Yes Yes Partial1 Yes Yes
 Baseline and follow-up measurement of intervention and comparison groups Yes Yes Yes Yes Yes Yes Yes Partial Yes Yes
Selection bias  
 Sample size calculation Yes Yes Yes NR NR Yes2 Yes3 NR NR Yes3
 Random sequence generation4 Yes Yes + 5 Yes Yes Yes Yes+ Yes Yes Yes Yes
 Allocation concealment Yes6 No No NR NR No NR NR NR NR
 Blinding of outcome assessment NR NR No NR No NR NR NR NR NR
Detection bias  
 Consistent outcome measurement across intervention and comparison groups Yes Yes Yes Yes Yes Yes Yes Yes Yes Yes
Analysis  
 Statistical control for confounding Yes Yes No7 Yes Yes Yes Yes Yes Yes Yes
 Intention to treat analysis Yes Yes Yes Yes Yes Yes NR NR Yes NR
Reporting bias  
 Complete reporting of outcomes described in methods in results Yes Yes Yes Yes Yes Yes Yes Yes No Yes
Reporting bias: conflicts of interest  
 Implementation and analysis independent from funders NR NR Yes NR NR NR Yes Yes NR Yes
Reporting bias: adherence to recommendations for IPV research  
 Age ≥15 for IPV questions Yes No Yes NR Yes Yes Yes Yes NR NR
 IPV-specific training for interviewers Yes Yes Yes NR Yes NR Yes Yes NR Yes
 IPV referral information or protocols Yes NR Yes NR Yes NR Yes NR Yes NR
  1. NR not reported
  2. 1The authors did not report baseline measurement for the microfinance only intervention
  3. 2A sample size calculation was performed for HIV incidence, not IPV
  4. 3A sample size calculation was performed, but investigators were not able to enroll a sufficient number of clusters to adhere to minimum sample size calculations
  5. 4NR not reported, Yes = randomization reported, and Yes + = randomization and randomly generated allocation sequence reported
  6. 5A public lottery was used for WINGS v. control, and a randomization algorithm was used for W+ v. WINGS
  7. 6Interviewers were blinded to allocation at baseline, not follow-up
  8. 7The authors report that no covariates were included in intention to treat analyses because randomization was successful
Back to article page

BMC Public Health

ISSN: 1471-2458

Contact us