Table 2. Subgroups examined in the sensitivity analyses.

Group	Rationale for Evaluation	Studies, n	I2, %	Overall HR (95% CI)
DCGL reported as an outcome	To investigate the effect of TG on loss of graft, where patients who died were excluded from the graft survival analysis	12 [34, 36–41, 43, 46, 48, 49, 51]	70.0	3.30 (2.43–4.50)
DCGF reported (graft loss and other laboratory indicators of graft failure)	To investigate the effect of TG on graft loss/functional deterioration	5 [26, 42, 45, 47, 52]	80.5	3.31 (1.55–7.11)
Graft loss and patient death as outcomes	To investigate the effect of TG in studies that did not censor death	4 [35, 44, 50, 53]	11.4	2.58 (1.80–3.71)
Studies that included published HRs and CIs	To investigate the effect of TG in studies that presented HR and CIs, avoiding uncertainties around the estimation of the HR	13 [26, 34, 36–38, 40, 42–44, 46, 51–53]	62.7	4.11 (2.94–5.73)
Follow-up reported as commencing at time of biopsy	To investigate the effect of TG in studies that defined follow-up initiation at the time of biopsy	17 [26, 35, 36, 38–40, 42, 44–53]	65.5	3.24 (2.45–4.28)
Follow-up reported as commencing at time of biopsy and published HR	To investigate the effect of TG in studies that met the strict criteria where follow-up was reported as commencing at the time of biopsy and HR was published	10 [26, 36, 38, 40, 42, 44, 46, 51–53]	50.3	4.53 (3.19–6.43)
For-cause biopsy and time to biopsy data reported	To investigate the effect of TG in studies that were used in the meta-regression analysis	16 [34–37, 40–42, 44, 45, 47–53]	61.6	2.89 (2.20–3.80)
For-cause biopsy reported, DC graft outcome and time to biopsy data	To investigate the effect of TG in a subgroup of studies used in the meta-regression analysis that censored death	12 [34, 36, 37, 40–42, 45, 47–49, 51, 52]	69.1	3.09 (2.18–4.37)

CI, confidence interval; DC, death-censored; DCGF, death-censored graft failure; DCGL, death-censored graft loss; HR, hazard ratio; TG, transplant glomerulopathy.