Table 3.
Demographics, smoking history, and overall survival (OS) statistics across subphenotypes in the development sub-cohort
| Variables | Subphenotype 1 | Subphenotype 2 | Subphenotype 3 | P-value | Post-hoc pair-wise analysis with p-value < 0.05 |
|---|---|---|---|---|---|
| Patients, n (%) | 1355 (42.0%) | 450 (14.0%) | 1420 (44.0%) | ||
| Age at index date, years, mean ± SD | 68.71 ± 9.39 | 68.77 ± 9.54 | 68.64 ± 9.28 | 0.966 | - |
| Gender, n (%) | 1.61E-29 | 1 vs. 2, 1 vs. 3 | |||
| Female | 752 (55.50%) | 152 (33.78%) | 500 (35.21%) | ||
| Male | 603 (44.50%) | 298 (66.22%) | 920 (64.79%) | ||
| Race, n (%) | 0.321 | - | |||
| White | 1051 (77.56%) | 342 (76.00%) | 1125 (79.23%) | ||
| Non-White | 304 (22.44%) | 108 (24.00%) | 295 (20.77%) | ||
| Practice type, n (%) | 3.47E-06 | 1 vs.2, 2 vs. 3, 1 vs. 3 | |||
| Academic | 209 (15.42%) | 100 (22.22%) | 175 (12.32%) | ||
| Community | 1146 (84.58%) | 350 (77.78%) | 1245 (87.68%) | ||
| Smoking status, n (%) | 0.181 | - | |||
| History of smoking | 1201 (88.63%) | 407 (90.44%) | 1282 (90.28%) | ||
| No history of smoking | 139 (10.26%) | 34 (7.56%) | 123 (8.66%) | ||
| Overall Survival | 1.11E-40 | 1 vs. 2, 2 vs. 3, 1 vs. 3 | |||
| Survival days, mean ± SD | 676 ± 543 | 454 ± 446 | 321 ± 387 | ||
| Survival days, median (Q1-Q3) | 516 (248, 982) | 305 (137, 611) | 180 (60, 442) | ||
| Observed events, n (%) | 795 (58.67%) | 341 (75.78%) | 1137 (80.07%) | ||
P-value: The p value was computed by testing differences of each variable across the three subphenotypes. Continuous normally and non-normally distributed variables were tested using one-way analysis of variance or Kruskal–Wallis tests, respectively, and categorical variables were tested using the one-sided Fisher’s exact test. Post-hoc pairwise analysis: If the overall p-value across groups was statistically significant (p < 0.05) for a variable, post-hoc pairwise analysis was performed to identify pairwise significance. Continuous normal and non-normal variables were tested using Tukeyʼs honestly significant difference or pairwise Wilcoxon rank sum test, respectively, and categorical variables were tested by pairwise Fisherʼs test. For instance, 1 vs. 2 indicates statistically significant differences between subphenotypes 1 and 2 on the variable with p < 0.05. SD standard deviation.