Table.4.
Mean hazard ratio (2.5th and 97.5th percentiles) for the composite outcome during a 3-year follow-up of patients on in-center hemodialysis
| Baseline characteristics adjusted in the analysis | Mean HR | Width of CI a | Type 1 errorc |
|---|---|---|---|
| (2.5th and 97.5th percentile) | |||
| Unadjusted analyses | |||
| Simple (i.e., unconstrained) randomization | 1.01 (0.87, 1.16) | 0.280 | 5.9%e |
| Constrained on a minimal set of baseline characteristicsb | 1.00 (0.89, 1.12) | 0.233 | 3.2% |
| Constrained on a minimal set of baseline characteristicsb and historical rate of the primary outcome | 1.00 (0.88, 1.13) | 0.250 | 4.4%e |
| Constrained on a minimal set of baseline characteristicsb and cluster size at time of randomization | 1.00 (0.88, 1.14) | 0.260 | 5.2%e |
| Constrained on a minimal set of baseline characteristicsb, historical rate of the primary outcome, and cluster size at time of randomization | 1.00 (0.88, 1.13) | 0.247 | 4.5%e |
| Constrained on 10 principal components | 1.01 (0.89, 1.12) | 0.234 | 3.3% |
| Constrained on 10 principal components and historic rate of primary outcome | 1.00 (0.88, 1.14) | 0.261 | 5.2%e |
| Constrained on 10 principal components and cluster size at time of randomization | 1.00 (0.87, 1.14) | 0.264 | 4.1%e |
| Constrained on ten principal components, the historical rate of the primary outcome, and cluster size at time of randomization | 1.00 (0.89, 1.13) | 0.239 | 3.1% |
| Adjusted for constrained baseline characteristicsd | |||
| Constrained on a minimal set of baseline characteristicsb | 1.00 (0.89, 1.12) | 0.232 | 8.6% |
| Constrained on a minimal set of baseline characteristicsb and historical rate of the primary outcome | 1.00 (0.89, 1.12) | 0.223 | 8.3% |
| Constrained on a minimal set of baseline characteristicsb and cluster size at time of randomization | 1.00 (0.89, 1.11) | 0.221 | 9.8% |
| Constrained on a minimal set of baseline characteristicsb, historical rate of the primary outcome, and cluster size at time of randomization | 1.00 (0.90, 1.11) | 0.216 | 9.6% |
| Constrained on 10 principal components | 1.00 (0.90, 1.11) | 0.203 | 5.2%e |
| Constrained on 10 principal components and historic rate of primary outcome | 1.00 (0.90, 1.11) | 0.201 | 6.0%e |
| Constrained on 10 principal components and cluster size at time of randomization | 1.00 (0.90, 1.11) | 0.203 | 6.3%e |
| Constrained on ten principal components, the historical rate of the primary outcome, and cluster size at time of randomization | 1.00 (0.91, 1.11) | 0.201 | 6.4%e |
All randomization methods had 1000 randomization schemes. The cohort included patients on dialysis as of April 1, 2014, and any patient who started in-center hemodialysis at one of the 72 participating centers during the 3-year follow-up
HR hazard ratio, width of CI width of confidence interval (i.e., upper minus lower confidence limit)
aThe confidence interval’s width may not equal the difference between the lower and upper confidence limits because of rounding
bIncluded patient-level information: age, living in a rural area, Black race, Modified Charlson comorbidity index, number of hospital visits in the previous 12 months, number of unique drugs the patient was prescribed in the 6 months before the index date, as well as history in the last 5 years of diagnosis for peripheral vascular disease, congestive heart failure, coronary artery disease, myocardial infarction, and number of nephrology consults in the previous 12 months before the index date
cType 1 error in the various constrained scenarios. Note: The nominal type 1 error is 5%. The observed type 1 error was within an “acceptable range” if it fell within the 95% confidence interval of the nominal value (i.e., between 3.6% and 6.4%)
dAdjusted analyses included baseline characteristics used in the constraining process
eAn acceptable type 1 error was observed for this method (i.e., between 3.6% and 6.4%)