Table 3.

Cochrane Method Risk of Bias - Quantitative Studies

Study	Random sequence generation (selection bias)	Allocation concealment (selection bias)	Blinding of ppts and personnel (performan ce bias)	Blinding of outcome assessment (detection bias): self-report measures	Blinding of outcome assessment (detection bias): objective measures	Incomplete outcome data (attrition bias)	Selective reporting (reporting bias)	Other bias
Cumella, 2014	U	U	X	U	✓	✓	✓	✓
Lutter, 2008; Lutter, Smith-Osborne, 2011	N/a - No control	N/a - No control	U	✓	✓	✓	✓	✓
Schenk, 2009	X - Not random	N/a - Not random	X	✓	✓	U	✓	X
Stefanini, 2015	✓	✓	✓	✓	U	✓	✓	✓
Stefanini, 2016	✓	✓	✓	U	✓	✓	✓	✓

Note: ✓= low risk, X = high risk, U = unclear risk, N/a = not applicable (i.e. no randomization)

Cumella¹⁴ - Selection protocols were not described in detail. For performance bias, the treatment team decided how much equine therapy patients received. With detection bias, the BDI, BAI, EDI are all self-reported, robust measures not likely to be influenced by blinding. However, it did not say if administrators were blinded to EAP length. No one dropped out.

Lutter²⁷; Lutter & Smith-Osborne²⁸ - The method consisted of retrospective review without a control group. Blinding is likely since the study was retrospective, but there was no blinding to treatment during administration. For detection bias, self-report measures are well-validated. Attrition and selection bias are of low risk given the retrospective nature.

Schnek³² - Participants were “selected” not randomized and there was no mention of blinding (plus the AAT group was the only ones who traveled to Florida). Participant drop out was not reported.

Stefanini²⁹; Stefanini³⁰ - Random numbers were generated to assign participants to groups. Evaluators and clinicians were blinded and treatment participants received the same amount of AAT. Observation of AAT was scored by two blind, independent investigators.