Skip to main content
. Author manuscript; available in PMC: 2020 Apr 1.
Published in final edited form as: Am J Epidemiol. 2019 Apr 1;188(4):709–723. doi: 10.1093/aje/kwy265

Table 6.

Empirical Example 2: Results for Time-to-Event Outcomes from Propensity Score-Adjusted Analyses using Different Combinations of Confounding Adjustment Method and Data-Sharing Approach, Non-TNFi vs TNFia

Confounding adjustment method & data-sharing approach Effectiveness outcome Safety outcome
Treatment switchingb Serious infectionsc
HR 95% CI HR 95% CI
Stratification
 Pooled individual-level 0.59 0.41, 0.86 0.94 0.50, 1.77
 Risk-set 0.59 0.41, 0.86 0.94 0.50, 1.77
 Summary-table 0.59 0.39, 0.85 0.94 0.45, 1.78
 Effect-estimate, fixed-effect 0.64 0.44, 0.93 1.04 0.55, 1.96
 Effect-estimate, random-effects 0.63 0.26, 1.50 1.04 0.55, 1.96
 Measure of heterogeneity, Qd 4.21 0.1215 0.83 0.6599
Matching
 Pooled individual-level 0.55 0.37, 0.83 1.07 0.51, 2.22
 Risk-set 0.55 0.37, 0.83 1.07 0.51, 2.22
 Summary-table 0.55 0.35, 0.82 1.07 0.47, 2.34
 Effect-estimate, fixed-effect 0.57 0.38, 0.84 1.23 0.58, 2.61
 Effect-estimate, random-effects 0.58 0.34, 1.02 1.23 0.58, 2.61
 Measure of heterogeneity, Qd 2.37 0.3053 0.03 0.9701
Inverse probability weighting
 Pooled individual-level 0.60 0.38, 0.93 2.67 1.86, 3.84
 Risk-set 0.60 0.35, 1.01 2.67 0.53, 13.52
 Effect-estimate, fixed-effect 0.69 0.44, 1.07 2.93 2.03, 4.22
 Effect-estimate, random-effects 0.79 0.26, 2.42 2.93 2.03, 4.22
 Measure of heterogeneity, Qd 7.61 0.0222 0.81 0.6656
Matching weighting
 Pooled individual-level 0.60 0.39, 0.94 0.85 0.38, 1.92
 Risk-set 0.60 0.41, 0.88 0.85 0.45, 1.63
 Effect-estimate, fixed-effect 0.62 0.40, 0.96 0.89 0.39, 2.05
 Effect-estimate, random-effects 0.60 0.30, 1.23 0.89 0.39, 2.05
 Measure of heterogeneity, Qd 2.62 0.2689 0.37 0.8304

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

a

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

b

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

c

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

d

Q 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.