Table 5.
Cost effectiveness analysis estimates for the missing data methods
RA | CCA (% bias) | MI-PMM (% bias) | Log MI-PMM (% bias) | Two-step MI (% bias) | |
---|---|---|---|---|---|
17 % Missing data | |||||
Correlation utility and costs | 0.0507 | 0.0591 | 0.0517 | 0.0509 | 0.0487 |
Covariance | 5.6 | 8.6 | 6.7 | 6.6 | 6.4 |
Mean (INB) | 14,422 | 16,026 (11) | 15,257 (6) | 15,654 (9) | 15,023 (4) |
SE INB | 6083 | 7270 (20) | 6438 (6) | 6457 (6) | 6504 (7) |
95 % CI lower limit | 4417 | 4069 | 4669 | 5034 | 4324 |
95 % CI upper limit | 24,427 | 27,983 | 25,846 | 26,274 | 25,721 |
Prob C-E | 0.99 | 0.99 (0) | 0.99 (0) | 0.99 (0) | 0.99 (0) |
ICER | −235,472 | −235,448 (0) | −220,988 (6) | −227,410 (3) | −217,656 (8) |
35 % Missing data | |||||
Correlation utility and costs | 0.0507 | 0.0251 | 0.0300 | 0.0292 | 0.028 |
Covariance | 5.6 | 4.9 | 4.4 | 4.4 | 4.4 |
Mean (INB) | 14,422 | 11,105 (23) | 13,029 (10) | 13,143 (9) | 12,841 (11) |
SE (INB) | 6083 | 8685 (43) | 6864 (13) | 7000 (15) | 6966 (15) |
95 % CI lower limit | 4417 | −3181 | 1738 | 1629 | 1383 |
95 % CI upper limit | 24,427 | 25,390 | 24,319 | 24,656 | 24,299 |
Prob C-E | 0.99 | 0.90 (9) | 0.97 (2) | 0.97 (2) | 0.97 (2) |
ICER | −235,472 | −134,488 (43) | −158,857 (33) | −156,289 (34) | −155,935 (34) |
50 % Missing data | |||||
Correlation utility and costs | 0.0507 | 0.0223 | 0.0433 | 0.0436 | 0.0406 |
Covariance | 5.6 | 5.5 | 7.0 | 7.0 | 6.7 |
Mean (INB) | 14,422 | 8786 (39) | 12,682 (12) | 12,867 (11) | 11,907 (17) |
SE (INB) | 6083 | 9584 (58) | 6858 (13) | 6770 (11) | 6962 (14) |
95 % CI lower limit | 4417 | −6978 | 1401 | 1731 | 456 |
95 % CI upper limit | 24,427 | 24,551 | 23,962 | 24,003 | 23,358 |
Prob C-E | 0.99 | 0.82 (17) | 0.97 (2) | 0.97 (2) | 0.96 (3) |
ICER | −235,472 | −155,561 (34) | −134,979 (43) | −144,317 (39) | −128,670 (45) |
SE standard error, INB incremental net benefit (euros), CI confidence interval, Prob C-E probability of cost-effectiveness, ICER incremental cost effectiveness ratio