Paradigm shift in clubfoot care strategies in Pakistan: A meta-analysis over three decades

Table 2 Risk of bias summary, n (%)

	Low risk	Moderate risk	Serious risk	Critical risk	No information
Confounding	12 (9.6)	48 (38.4)	55 (44)	6 (4.8)	4 (3.2)
	No adjustment for age/severity in older studies; better in recent Ponseti
Selection	25 (20)	60 (48)	35 (28)	3 (2.4)	2 (1.6)
	Retrospective designs were common pre-2010; prospective post-2015 improved
Classification	95 (76)	20 (16)	6 (4.8)	0	4 (3.2)
	Clear definitions in most (e.g., Ponseti protocol)
Deviations	18 (14.4)	55 (44)	45 (36)	5 (4)	2 (1.6)
	Bracing non-compliance is high in 40; co-interventions are unbalanced
Missing data	10 (8)	40 (32)	65 (52)	8 (6.4)	2 (1.6)
	Short follow-up (< 1 year in 70); loss to follow-up in 25
Measurement	30 (24)	70 (56)	20 (16)	1 (0.8)	4 (3.2)
	Pirani scoring standard, but no blinding in 60
Reported result	85 (68)	30 (24)	6 (4.8)	0	4 (3.2)
	Most pre-specified, selective in older surgical
Overall	8 (6.4)	52 (41.6)	53 (42.4)	8 (6.4)	4 (3.2)
	Moderate/Serious dominant; sensitivity excluded serious/critical (estimates stable)
Based on ROBINS-I, n = 125 native studies; grade downgraded for serious bias in confounding/missing data

ROBINS-I: Risk of bias in non-randomized studies - of interventions.