PURPOSE: To compare survival between patients enrolled in the Collaborative Ocular Melanoma Study (COMS) randomized trial ofpre-enucleation radiation therapy (PERT) for large choroidal melanoma and eligible patients who did not enroll. METHODS: COMS clinical center personnel prospectively reported to the COMS Coordinating Center all patients with choroidal melanoma examined between November 1986 and December 1994. Deaths of enrolled patients were reported prospectively by clinical center personnel. In a COMS ancillary study, we retrospectively searched medical records of participating clinical centers, the Social Security Death Index, and the National Death Index to determine vital status of eligible patients not enrolled. Cox proportional hazards analysis was used to compare survival within 10 years of baseline reporting and before July 31, 2000, of enrolled patients versus eligible patients not enrolled. RESULTS: Clinical centers that received local institutional review board approval to participate in this ancillary study prospectively reported on 129 of 299 eligible patients not enrolled in the COMS PERT trial. The baseline characteristics of the 129 patients included in this ancillary study were similar to those of the 170 patients not included; 73 patients were reported as deceased. Previously identified prognostic covariates, i.e., age and longest tumor diameter, were confirmed to predict survival in both enrolled patients and eligible patients not enrolled; trial enrollment was not predictive. After adjusting for prognostic covariates and stratifying by clinical center, the estimated hazard ratio (enrolled vs. not-enrolled) was 1.12 (95% confidence interval: 0.83 to 1.51). CONCLUSIONS: The results of the COMS PERT trial should be generalizable to all patients with choroidal melanoma meeting the eligibility criteria for that trial. While the methods we used may not be generalizable to all clinical trials because of unique features of the COMS, other researchers may be able to use similar methods to determine the generalizability of their trial results.
RCT Entities:
PURPOSE: To compare survival between patients enrolled in the Collaborative Ocular Melanoma Study (COMS) randomized trial of pre-enucleation radiation therapy (PERT) for large choroidal melanoma and eligible patients who did not enroll. METHODS: COMS clinical center personnel prospectively reported to the COMS Coordinating Center all patients with choroidal melanoma examined between November 1986 and December 1994. Deaths of enrolled patients were reported prospectively by clinical center personnel. In a COMS ancillary study, we retrospectively searched medical records of participating clinical centers, the Social Security Death Index, and the National Death Index to determine vital status of eligible patients not enrolled. Cox proportional hazards analysis was used to compare survival within 10 years of baseline reporting and before July 31, 2000, of enrolled patients versus eligible patients not enrolled. RESULTS: Clinical centers that received local institutional review board approval to participate in this ancillary study prospectively reported on 129 of 299 eligible patients not enrolled in the COMS PERT trial. The baseline characteristics of the 129 patients included in this ancillary study were similar to those of the 170 patients not included; 73 patients were reported as deceased. Previously identified prognostic covariates, i.e., age and longest tumor diameter, were confirmed to predict survival in both enrolled patients and eligible patients not enrolled; trial enrollment was not predictive. After adjusting for prognostic covariates and stratifying by clinical center, the estimated hazard ratio (enrolled vs. not-enrolled) was 1.12 (95% confidence interval: 0.83 to 1.51). CONCLUSIONS: The results of the COMS PERT trial should be generalizable to all patients with choroidal melanoma meeting the eligibility criteria for that trial. While the methods we used may not be generalizable to all clinical trials because of unique features of the COMS, other researchers may be able to use similar methods to determine the generalizability of their trial results.