TABLE 3.
Statistical models predicting the appearance of pathologic changes in the eye after long-duration spaceflight
| Outcome | Test | df | P | AIC |
|---|---|---|---|---|
| Choroidal folds, total n = 48 | Wald χ2 | |||
| Block 1 | 3.97 | 2 | 0.140 | 58.91 |
| Block 2 | 12.90 | 5 | 0.020 | 44.52 |
| Block 3 | 6.18 | 3 | 0.100 | 45.28 |
| Globe flattening, total n = 34 | Wald χ2 | |||
| Block 1 | 5.70 | 2 | 0.060 | 77.03 |
| Block 2 | 7.04 | 5 | 0.220 | 72.42 |
| Block 3 | 8.04 | 3 | 0.050 | 46.49 |
| Disc edema, total n = 48 | Wald χ2 | |||
| Block 1 | 3.79 | 2 | 0.150 | 90.77 |
| Block 2 | 9.51 | 5 | 0.090 | 79.63 |
| Block 3 | 1.93 | 3 | 0.590 | 82.54 |
| Change in diopters, total n = 31 | F | |||
| Block 1 | 3.70 | 2 | 0.040 | 99.06 |
| Block 2 | 2.04 | 5 | 0.100 | 86.88 |
| Block 3 | 13.61 | 3 | 0.001 | 70.38 |
Block 1, base model (days in space, age at launch); block 2, genetics + base vs. base; block 3, nutrition + genetics + base vs. genetics + base. Regression-based statistical methods were used to determine whether, as a block, additional predictors entered in blocks contributed significantly to the prediction of each of the outcomes shown above. See “Statistical methods” for details. The base model (block 1) was compared to a model (block 2) that, in addition to age and flight duration, included a block of key genetic factors, and block 2 was then compared to a final model (block 3) that also included several nutrition variables. The Wald test statistic allows for comparing nested models, and the AIC is provided as a descriptive measure of relative model fit. Lower AIC values indicate a better fitting model.