Table 2.

Sensitivity analyses specific to sample size ≥100 or intervention of TM with SF.

Methods	Main endpoints	No. of studies	No. of patients	MD/RR	95% CI	P	I 2
Sample size ≥100	ER visits	5 (11, 12, 27–29)	1,057	MD = −0.76	[−1.47 to −0.05]	0.04	96%
	AE-related readmissions	5 (10–13, 31)	1,195	RR = 0.82	[0.69, 0.97]	0.02	66%
TM with SF intervention	ER visits	3 (27–29)	670	MD = −0.98	[−1.26 to −0.69]	0.00	97%
	AE-related readmissions	4 (10, 13, 17, 31)	591	RR = 0.87	[0.79, 0.94]	0.01	26%

The effect of the pooled estimates for the studies in which the sample size was over 100 or the intervention was TM plus SF did not significantly alter the effect on ER visits and AE-related readmissions. The power of studies with small sample size (<100) (5, 17, 30) did not meet the minimum requirements of the research (0.34, 0.37, and 0.66, respectively). So, after the exclusion of the above studies with small sample size, the conclusion was in accordance to the overall conclusion. Similarly, the alike pattern was found in TM-plus-SF populations. In other words, after the sensitivity analysis, it was proved that our final conclusion was consistent with the conclusion from the subgroup analysis of specific populations (patient number ≥100 or TM-plus-SF intervention group). MD, mean difference; RR, risk ratio; CI, confidence interval.