Skip to main content
. 2016 Mar 29;11(3):e0151724. doi: 10.1371/journal.pone.0151724

Table 5. Balancing achieved by different propensity score-based matching strategies for later generation fluoroquinolone use.

Covariate Later generation FQN No later generation FQN Before matching Matching within studyes, without replacement* Matching within studies, with replacement* Matching across studies without replacement* Matching across studies with replacement*
Mean (sd) Mean (sd) Std dif # (%) p-value+ Std dif (%) p-value+ Std dif (%) p-value Std dif (%) p-value+ Std dif (%) p-value+
Age 37.71 (14.75) 39.63 (13.53) 4.79 0.0017* 10.50 0.104 3.35 0.123 2.24 0.736 2.83 0.147
Male gender 0.59 (0.49) 0.69 (0.46) -20.7 0 3.20 0.6698 13.50 <.0001 -2.84 0.6949 3.46 0.0817
Extensive disease 0.8 (0.4) 0.72 (0.45) 18.2 0 1.76 0.7815 1.31 0.5529 0.00 1.0000 -3.79 0.0480
Prior TB 0.53 (0.5) 0.71 (0.45) -38.4 0 -1.52 0.7815 2.20 0.3508 -8.49 0.2008 13.55 <.0001
Prior MDR-TB 0.23 (0.42) 0.07 (0.26) 45.3 0 4.90 0.3173 -0.99 0.6858 -5.37 0.4054 -11.14 <.0001
Known HIV co-infection 0.02 (0.15) 0.17 (0.38) -51.5 0 6.18 0.1573 3.44 0.0578 22.64 0.0039 8.28 <.0001
No. drugs (intensive) 5.73 (1.58) 5.13 (1.36) -14.37 <0.0001 1.84 0.746 2.65 0.226 -10.53 <0.001 0.54 0.453
Total therapy (months) 26.07 (16.2) 16.84 (6.94) -26.2 <0.0001 -6.86 0.053 -6.52 <0.001 -14.70 0.016 -16.61 <0.001
Logit of predicted propensity score 3.39 (1.48) 2.86 (1.29) -13.35 <0.0001 -26.53 <0.001 -17.44 <0.001 -12.79 <0.001 0.13 0.14

*One to one matching.

# Std dif = standardized difference for unexposed relative to exposed. FQN = fluoroquinolone. The differences in characteristics between exposed and unexposed individuals receiving were tested for significance using a paired t-test for continuous variables and McNemar’s test for dichotomous variables [34]. MDR-TB = multi-drug resistant tuberculosis. TB = tuberculosis. HIV = human immunodeficiency virus.