Change in Intended Management by Indication and Cancer Type with Comparison to Prostate Cancer
| Indication | Breast | NSCLC† | Others | Prostate‡ | P‡ |
| IS | |||||
| Participants (n) | 181 | 166 | 223 | 2,301 | |
| Change in intended management (%) | 42.5 (35.3–49.7) | 54.2 (46.6–61.8) | 52.0 (45.5–58.6) | 46.4 (44.4–48.5) | 0.059 |
| Imaging-adjusted frequency of change (%) | 11.0 (6.5–15.6) | 13.9 (8.6–19.1) | 11.2 (7.1–15.4) | 10.3 (9.0–11.5) | 0.52 |
| Suspected FOM | |||||
| Participants (n) | 781 | 380 | 653 | 4686 | |
| Change in intended management (%) | 24.3*** (21.3–27.3) | 36.0** (31.2–40.9) | 31.1*** (27.5–34.6) | 43.6 (42.2–45.0) | <0.0001 |
| Imaging-adjusted frequency of change (%) | 7.7*** (5.8–9.5) | 8.7** (5.8–11.5) | 8.0*** (5.9–10.0) | 15.0 (13.9–16.0) | <0.0001 |
| Suspected POM | |||||
| Participants (n) | 199 | – | 236 | 1297 | |
| Change in intended management (%) | 60.3 (53.5–67.1) | – | 52.1 (45.7–58.5) | 53.0 (50.2–55.7) | 0.14 |
| Imaging-adjusted frequency of change (%) | 11.6 (7.1–16.0) | – | 9.3 (5.6–13.0) | 10.9 (9.2–12.6) | 0.72 |
↵† For suspected POM stratum, NSCLC participants were grouped into other cancer type.
↵‡ For each comparison, logistic regression was performed to test difference of rates across specified cancer types on change (or imaging adjusted) in intended management, respectively. Prostate cancer group was used as reference level in regression. P value was calculated using global Wald test.
If global Wald test from logistic regression was significant (P < 0.05), individual tests were performed to find out which cancer types were different from prostate cancer (reference) in terms of change rates. Multiple comparisons were corrected for within this analysis, such that cutoff value for significance level was 0.0167 (0.05/3).
*P value of individual test was smaller than 0.0167.
↵** P value was smaller than 0.01.
↵*** P value was smaller than 0.001.
Data in parentheses are 95% confidence intervals, computed using normal approximation for binomial proportion.