Characteristics of studies using PS matching (N=210)
| Yes (n (%)) | Not reported (n (%)) | |
| Matches: | 8 (3.8) | |
| 1:1 | 178 (84.7) | |
| 1:2 | 8 (3.8) | |
| 1:3 | 5 (2.4) | |
| 1:4 | 5 (2.4) | |
| Other | 4 (1.9) | |
| Multiple matching ratios | 2 (1) | |
| Calliper scale (nearest neighbour matching with callipers): | 2 (1.7) | |
| PS | 52 (43.3) | |
| SD of Logit PS | 39 (32.4) | |
| SD of PS | 21 (17.5) | |
| Logit PS | 3 (3.3) | |
| Mahalanobis distance | 2 (1.7) | |
| Matches found without replacement | 79 (37.6) | 120 (57.1) |
| Matching algorithm: | 120 (57.1) | |
| Greedy | 80 (38.1) | |
| Optimal | 10 (4.8) | |
| Assessment of balance | ||
| Balance of covariates assessed by: | 17 (8.1) | |
| Hypothesis testing | 68 (32.4) | |
| SMDs | 57 (27.1) | |
| Both | 60 (28.6) | |
| Stated balance was assessed | 8 (3.8) | |
| Threshold for successful balance explicitly stated (eg, 10% SMDs, p>0.05) | 142 (67.6) | --- |
| SMDs threshold: | 39 (33.3) | |
| <10% | 71 (60.7) | |
| <20% | 3 (2.6) | |
| Other | 4 (3.4) |
PS, propensity score; SMD, standardised mean difference.