Table 7.

Empirical Example 2: Results from Disease Risk Score^a-Adjusted Analyses using Different Combinations of Confounding Adjustment Method and Data-Sharing Approach, Non-TNFi vs TNFi^b

Confounding adjustment method & data-sharing approach	Effectiveness outcome		Safety outcome
	Treatment switchingc		Serious infectionsd
	OR	95% CI	OR	95% CI
Stratification
Pooled individual-level	0.53	0.35, 0.78	0.88	0.42, 1.64
Risk-set	0.53	0.36, 0.78	0.88	0.42, 1.64
Summary-table	0.53	0.35, 0.78	0.88	0.42, 1.64
Effect-estimate, fixed-effect	0.56	0.38, 0.82	0.97	0.52, 1.82
Effect-estimate, random-effects	0.51	0.28, 0.96	0.97	0.52, 1.82
Measure of heterogeneity, Qe	2.67	0.2624	0.24	0.8832
Matching
Pooled individual-level	0.41	0.26, 0.63	0.86	0.37, 1.93
Risk-set	0.41	0.26, 0.63	0.86	0.37, 1.93
Summary-table	0.41	0.26, 0.63	0.86	0.37, 1.93
Effect-estimate, fixed-effect	0.41	0.27, 0.63	0.89	0.40, 1.98
Effect-estimate, random-effects	0.41	0.27, 0.63	0.89	0.40, 1.98
Measure of heterogeneity, Qe	1.77	0.4115	0.00	0.9961
	HR	95% CI	HR	95% CI
Stratification
Pooled individual-level	0.59	0.41, 0.84	0.86	0.47, 1.57
Risk-set	0.59	0.41, 0.84	0.86	0.47, 1.57
Summary-table	0.58	0.39, 0.83	0.86	0.42, 1.57
Effect-estimate, fixed-effect	0.63	0.44, 0.91	0.95	0.52, 1.75
Effect-estimate, random-effects	0.59	0.26, 1.32	0.95	0.52, 1.75
Measure of heterogeneity, Qe	3.83	0.1466	0.21	0.8989
Matching
Pooled individual-level	0.46	0.31, 0.68	0.89	0.42, 1.86
Risk-set	0.46	0.31, 0.68	0.89	0.42, 1.86
Summary-table	0.45	0.30, 0.68	0.88	0.38, 1.95
Effect-estimate, fixed-effect	0.47	0.32, 0.70	0.91	0.42, 2.00
Effect-estimate, random-effects	0.52	0.22, 1.25	0.91	0.42, 2.00
Measure of heterogeneity, Qe	4.01	0.1345	0.00	1.0000

Note: TNFi= tumor necrosis factor-alpha inhibitor; OR= odds ratio; CI= confidence interval; HR= hazard ratio

^aThe disease risk score was estimated using Cox proportional hazards regression on patients receiving tumor necrosis factor-alpha inhibitor biologics.

^bThere were 407 (5.2%) patients who initiated non-TNFi, and 7,419 (94.8%) patients who initiated TNFi.

^cThe incidences for treatment switching were 7.6% in new users of non-TNFi and 11.2% in new users of TNFi.

^dThe incidences for serious infections were 2.9% in new users of non-TNFi and 3.1% in new users of TNFi.

^eQ is a measure of heterogeneity among the three data-contributing sites. The summary statistic and p-value from Cochran’s Q test are shown here.