Skip to main content
. Author manuscript; available in PMC: 2015 Dec 8.
Published in final edited form as: J Am Stat Assoc. 2012;107(499):901–915. doi: 10.1080/01621459.2012.682533

Table 3.

Counts of discordant pairs for two outcomes and their combination, with McNemar P-values, and Gart’s test for interaction with type. There are 2298 pairs, 1354 pairs that finely balance the 47 hospitals (type 1) and 944 pairs that contrast hospitals with different usual practices (type 2). Only discordant pairs are recorded here. For readmission, there were 13 deaths within 30 days, 6 with general anesthesia, 7 with regional anesthesia, and these are included in the unfavorable outcome category in each comparison.

Alive without deep vein thrombosis

All Finely Balanced Pairs (type 1) Usual Practice Pairs (type 2)
General 72 42 30
Regional 100 66 34

P-value 0.040 0.027 0.708

Odds Ratio (OR) 1.39 1.57 1.13
95% CI for OR [1.03, 1.88] [1.07, 2.31] [0.69, 1.85]

P-value for interaction 0.34

Alive without readmission at 30 days

All Finely Balanced Pairs (type 1) Usual Practice Pairs (type 2)

General 106 61 45
Regional 135 84 51

P-value 0.071 0.068 0.610

Odds Ratio (OR) 1.27 1.38 1.13
95% CI for OR [0.99, 1.64] [0.99, 1.91] [0.75, 1.69]

P-value for interaction 0.51

Alive without both deep vein thrombosis and readmission at 30 days

All Finely Balanced Pairs (type 1) Usual Practice Pairs (type 2)

General 152 84 68
Regional 200 128 72

P-value 0.012 0.003 0.800

Odds Ratio (OR) 1.32 1.52 1.06
95% CI for OR [1.07, 1.62] [1.16, 2.01] [0.76, 1.47]

P-value for interaction 0.10