. Author manuscript; available in PMC: 2013 Jul 1.

Published in final edited form as: Pharmacoepidemiol Drug Saf. 2011 Dec 8;21(7):697–709. doi: 10.1002/pds.2256

Table 2.

a. Empirical results for the NSAID study. Key within-PS interactions were included in PS models.

Model		Entire Cohort (n=49,711)	Males (n=7,854)	Females (n=41,857)	Age <75 (n=14,103)	Age ≥75 (n=35,608)	High Risk (n=12,627)^a
Exposed outcomes	N	503	71	432	83	420	348
Unexposed outcomes	N	243	57	186	49	194	157
Unadjusted	OR [CI]	1.14 [0.98, 1.33]	0.92 [0.65, 1.30]	1.21 [1.02, 1.44]	1.30 [0.91, 1.86]	1.04 [0.87, 1.23]	0.91 [0.75, 1.10]
Age/sex adjusted	OR [CI]	1.08 [0.92, 1.26]	0.87 [0.61, 1.24]	1.14 [0.95, 1.35]	1.32 [0.93, 1.89]	1.02 [0.86, 1.21]	0.88 [0.73, 1.07]
Adjusted by all covariates	OR [CI]	0.87 [0.71, 1.08]	0.68 [0.42, 1.12]	0.93 [0.73, 1.17]	1.07 [0.67, 1.72]	0.83 [0.65, 1.05]	0.74 [0.53, 1.04]
PS estimated within subgroup (PS_SS) "Referent Standard"
Adj. by deciles of PS_SS	OR [CI]	0.93 [0.79, 1.09]	0.80 [0.55, 1.15]	0.96 [0.80, 1.15]	1.04 [0.72, 1.50]	0.91 [0.76, 1.09]	0.90 [0.74, 1.10]
	Distance^b	0.026	0.116	0.030	0.062	0.036	0.082
Matched on PS_SS	N Matched	33,188	5,812	27,368	10,758	22,370	7,204
	OR [CI]	0.97 [0.81, 1.17]	0.83 [0.55, 1.27]	0.95 [0.77, 1.16]	0.98 [0.64, 1.48]	0.89 [0.73, 1.10]	0.91 [0.72, 1.14]
	Distance	0.001	0.003	0.001	0.002	0.001	0.003
*PS estimated overall (PS_COH) and applied within subgroup*
Adj. by deciles of PS_COH	OR [CI]	0.93 [0.79, 1.09]	0.79 [0.55, 1.13]	0.96 [0.81, 1.15]	1.01 [0.70, 1.46]	0.92 [0.77, 1.10]	0.90 [0.74, 1.09]
	Difference (%) ^c	0.00 (0.0%)	−0.01 (−1.4%)	0.00 (0.0%)	−0.03 (−2.7%)	0.01 (1.3%)	−0.01 (−0.7%)
	Distance	0.026	0.127	0.032	0.069	0.036	0.089
Matched on PS_COH	N Matched	33,188	5,826	27,368	10,810	22,398	7,232
	OR [CI]	0.97 [0.81, 1.17]	0.74 [0.49, 1.13]	1.03 [0.84, 1.26]	0.98 [0.64, 1.49]	0.96 [0.79, 1.18]	0.79 [0.62, 1.01]
	Difference (%)	0.00 (0.0%)	−0.09 (−11.0%)	0.08 (8.8%)	−0.00 (−0.1%)	0.07 (7.7%)	−0.11 (−12.5%)
	Distance	0.001	0.006	0.001	0.006	0.001	0.015
Matched pairs from PS_COH matches
Matched on PS_COH	N Matched	33,188	1,278	22,886	5,000	16,482	2,112
	OR [CI]	0.97 [0.81, 1.17]	1.00 [0.45, 2.24]	0.99 [0.79, 1.23]	0.52 [0.26, 1.05]	1.05 [0.83, 1.33]	0.87 [0.54, 1.39]
	Difference (%)	0.00 (0.0%)	0.17 (19.9%)	0.04 (4.4%)	−0.46 (−46.9%)	0.16 (17.7%)	−0.04 (−4.2%)
	Distance	0.001	0.024	0.001	0.014	0.002	0.036

b. Empirical results for the Pennsylvania APM study. Key within-PS interactions were included in PS models.

Model		Entire Cohort (n=46,659)	Males (n=7,841)	Females (n=38,818)	Age <75 (n=10,035)	Age ≥75 (n=36,624)	High Risk (n=29,994)^a
Exposed outcomes	N	6214	1546	4668	1213	5001	4454
Unexposed outcomes	N	2061	510	1551	127	1934	1667
Unadjusted	OR [CI]	1.36 [1.29, 1.44]	1.44 [1.28, 1.61]	1.35 [1.27, 1.43]	2.26 [1.87, 2.74]	1.37 [1.29, 1.45]	1.37 [1.29, 1.46]
Age/sex adjusted	OR [CI]	1.58 [1.50, 1.67]	1.62 [1.44, 1.83]	1.57 [1.47, 1.67]	2.36 [1.95, 2.86]	1.51 [1.43, 1.61]	1.56 [1.46, 1.66]
Adjusted by all covariates	OR [CI]	1.25 [1.16, 1.33]	1.21 [1.05, 1.40]	1.25 [1.16, 1.35]	1.29 [1.02, 1.63]	1.26 [1.17, 1.36]	1.27 [1.18, 1.38]
PS estimated within subgroup (PS_SS) "Referent Standard"
Adj. by deciles of PS_SS	OR [CI]	1.24 [1.16, 1.33]	1.19 [1.03, 1.37]	1.27 [1.17, 1.37]	1.19 [0.94, 1.49]	1.26 [1.17, 1.35]	1.27 [1.18, 1.38]
	Distance^b	0.365	0.828	0.359	2.037	0.296	0.420
Matched on PS_SS	N Matched	18,548	3,166	15,372	2,668	15,766	12,660
	OR [CI]	1.27 [1.17, 1.37]	1.18 [1.00, 1.39]	1.28 [1.17, 1.40]	1.11 [0.84, 1.47]	1.26 [1.16, 1.36]	1.26 [1.15, 1.38]
	Distance	0.004	0.017	0.005	0.024	0.004	0.005
*PS estimated overall (PS_COH) and applied within subgroup*
Adj. by deciles of PS_COH	OR [CI]	1.24 [1.16, 1.33]	1.18 [1.03, 1.36]	1.27 [1.17, 1.37]	1.44 [1.15, 1.81]	1.24 [1.16, 1.33]	1.26 [1.16, 1.36]
	Difference (%)^c	0.00 (0.0%)	−0.01 (−0.9%)	0.00 (0.0%)	0.25 (21.3%)	−0.02 (−1.4%)	−0.02 (−1.3%)
	Distance	0.365	1.091	0.365	1.973	0.306	0.397
Matched on PS_COH	N Matched	18,548	3,164	15,402	2,700	15,848	12,674
	OR [CI]	1.27 [1.17, 1.37]	1.10 [0.93, 1.30]	1.26 [1.15, 1.38]	1.27 [0.97, 1.67]	1.24 [1.14, 1.34]	1.22 [1.12, 1.34]
	Difference (%)	0.00 (0.0%)	−0.08 (−6.5%)	−0.02 (−1.2%)	0.16 (14.7%)	−0.02 (−1.6%)	−0.03 (−2.7%)
	Distance	0.004	0.041	0.006	0.093	0.011	0.014
Matched pairs from PS_COH matches
Matched on PS_COH	N Matched	18,548	578	12,868	462	13,454	8,402
	OR [CI]	1.27 [1.17, 1.37]	1.27 [0.86, 1.89]	1.28 [1.16, 1.41]	1.28 [0.69, 2.38]	1.23 [1.13, 1.35]	1.31 [1.18, 1.47]
	Difference (%)	0.00 (0.0%)	0.10 (8.1%)	0.01 (0.5%)	0.17 (15.5%)	−0.03 (−2.1%)	0.06 (4.5%)
	Distance	0.004	0.151	0.007	0.257	0.015	0.019

c. Empirical results for the British Columbia APM study. Key within-PS interactions were included in PS models.

Model		Entire Cohort (n=42,565)	Males (n=16,790)	Females (n=25,775)	Age <75 (n=11,899)	Age ≥75 (n=30,666)	High Risk (n=6,519)^a
Exposed outcomes	N	3883	2052	1831	1190	2693	619
Unexposed outcomes	N	3015	1445	1570	430	2585	568
Unadjusted	OR [CI]	2.41 [2.29, 2.54]	2.31 [2.14, 2.50]	2.37 [2.21, 2.55]	4.37 [3.88, 4.91]	2.09 [1.97, 2.22]	1.79 [1.58, 2.03]
Age/sex adjusted	OR [CI]	2.44 [2.32, 2.58]	2.41 [2.23, 2.60]	2.47 [2.29, 2.66]	4.44 [3.95, 5.00]	2.06 [1.94, 2.19]	1.84 [1.62, 2.09]
Adjusted by all covariates	OR [CI]	2.10 [1.97, 2.23]	1.96 [1.79, 2.16]	2.22 [2.04, 2.43]	2.83 [2.44, 3.27]	1.90 [1.77, 2.04]	1.60 [1.38, 1.86]
PS estimated within subgroup (PS_SS) "Referent Standard"
Adj. by deciles of PS_SS	OR [CI]	1.92 [1.81, 2.04]	1.80 [1.65, 1.97]	2.04 [1.88, 2.22]	2.51 [2.20, 2.86]	1.79 [1.67, 1.92]	1.53 [1.33, 1.77]
	Distance^b	0.112	0.148	0.142	0.246	0.104	0.287
Matched on PS_SS	N Matched	22,358	9,074	13,234	6,558	15,612	3,670
	OR [CI]	1.89 [1.76, 2.03]	1.74 [1.58, 1.92]	2.15 [1.95, 2.38]	2.48 [2.13, 2.89]	1.72 [1.59, 1.87]	1.45 [1.23, 1.71]
	Distance	0.003	0.004	0.004	0.005	0.002	0.009
*PS estimated overall (PS_COH) and applied within subgroup*
Adj. by deciles of PS_COH	OR [CI]	1.92 [1.81, 2.04]	1.82 [1.67, 1.99]	2.03 [1.87, 2.21]	2.70 [2.37, 3.07]	1.76 [1.64, 1.89]	1.40 [1.21, 1.61]
	Difference (%)^c	0.00 (0.0%)	0.02 (1.1%)	−0.01 (−0.5%)	0.19 (7.7%)	−0.03 (−1.7%)	−0.13 (−8.7%)
	Distance	0.112	0.151	0.136	0.304	0.127	0.382
Matched on PS_COH	N Matched	22,358	9,090	13,266	6,640	15,664	3,708
	OR [CI]	1.89 [1.76, 2.03]	1.76 [1.60, 1.95]	2.09 [1.89, 2.31]	2.55 [2.20, 2.96]	1.71 [1.58, 1.86]	1.25 [1.06, 1.47]
	Difference (%)	0.00 (0.0%)	0.02 (1.3%)	−0.06 (−2.8%)	0.08 (3.1%)	−0.01 (−0.5%)	−0.20 (−14.0%)
	Distance	0.003	0.013	0.005	0.061	0.015	0.078
Matched pairs from PS_COH matches
Matched on PS_COH	N Matched	22,358	4,140	8,182	2,492	11,382	620
	OR [CI]	1.89 [1.77, 2.03]	1.75 [1.51, 2.03]	2.08 [1.83, 2.36]	2.51 [1.95, 3.23]	1.68 [1.52, 1.84]	1.10 [0.75, 1.60]
	Difference (%)	0.01 (0.3%)	0.01 (0.6%)	−0.08 (−3.5%)	0.04 (1.5%)	−0.05 (−2.6%)	−0.35 (−24.3%)
	Distance	0.003	0.032	0.009	0.090	0.020	0.163

Patients with prior gastrointestinal-related diagnoses or hospitalizations.

Mahalanobis distance between patients in the exposure and referent categories. In the decile analyses, the distance is measured within each decile and then averaged.

Difference between the observed odds ratio after adjusting by or matching on PS_COH versus PS_SS.

Patients with a history of cerebrovascular disease, myocardial infarction, or arrhythmias.

Mahalanobis distance between patients in the exposure and referent categories. In the decile analyses, the distance is measured within each decile and then averaged.

Difference between the observed odds ratio after adjusting by or matching on PS_COH versus PS_SS.

Patients with a history of cerebrovascular disease, myocardial infarction, or arrhythmias.

Mahalanobis distance between patients in the exposure and referent categories. In the decile analyses, the distance is measured within each decile and then averaged.

Difference between the observed odds ratio after adjusting by or matching on PS_COH versus PS_SS.