Chemotherapy for the primary treatment of osteosarcoma: population effectiveness over 20 years.

Examination of the survival rate of 5-year cohorts with primary osteosarcoma registered by the Thames Cancer Registry between 1963 and 1982 and followed up to the end of 1984 showed a borderline significant improvement in survival (test for trend p = 0.05) for patients aged under 25 years at registration, but not for patients 25-64 years. A Cox's proportional hazards model was used to identify effects of sex, period of treatment, whether the patient received chemotherapy, and experience of the hospital. For patients aged under 25 years, a significant (p = 0.02) trend of improved survival was seen among the most recent cohort, and the greater the treatment experience of the hospital, possibly the better the results (p = 0.11), although selective referral cannot be excluded as a reason for this finding. Chemotherapy was not independently associated with survival. Period of treatment, chemotherapy, and experience of hospital were not associated with improved survival for patients aged 25-64. Recent clinical trials for primary osteosarcoma have contributed to an apparent improved survival through selection of patients with good prognosis. Claims for the efficacy of new regimens should be assessed in clinical practice by population monitoring through the analysis of cancer registers.