Trends in prostate cancer survival in Sweden, 1960 through 1988: evidence of increasing diagnosis of nonlethal tumors.

BACKGROUND: The incidence of prostate cancer has increased during the past 30 years but has been paralleled by increases in survival rates from this disease, despite the absence of documented major improvement in curative treatment. Since a high prevalence of microscopic prostate cancer has been observed in autopsied men and because many prostate cancers may never surface clinically, increased diagnostic activities might have led to increased detection of less aggressive tumors.
PURPOSE: This study was conducted to elucidate whether the trends in prostate cancer incidence and patient survival may be due to increasing diagnoses of nonlethal tumors.
METHODS: We analyzed a population-based cohort comprising all cases of prostate cancer (n = 80,901) detected in Sweden during the period of 1960 through 1988. Five hundred eighteen patients (0.64% of the total number) who could not be followed because of emigration or an incomplete national registration number were excluded. Observed and relative survival rates were calculated for the entire cohort of 80,383 assessable patients per 5-year age group in 5-year periods of diagnosis and according to diagnostic method and were compared between geographic areas with differences in incidence rates. To estimate the independent effects of these determinants, multivariate analyses were performed.
RESULTS: For the 80,383 patients with complete follow-up, the 10- and 20-year observed survival rates were 17.5% (95% confidence interval [CI] = 17.2%-17.9%) and 3.5% (95% CI = 3.2%-3.7%), and the relative survival rates were 41.1% (95% CI = 40.3%-41.9%) and 28.6% (95% CI = 26.5%-30.1%), respectively. Relative survival rates improved markedly over time; 10-year relative survival rates increased from 29% (95% CI = 27%-31%) among case patients diagnosed in 1960 through 1964 to 45% (95% CI = 43%-46%) among those diagnosed in 1975 through 1979. Relative survival rates leveled off after about 18 years at 18% (95% CI = 15%-20%) among patients diagnosed in 1960 through 1964 and at 31% (95% CI = 28%-34%) among those diagnosed in 1970 through 1974. An even more favorable outlook was observed in those case patients diagnosed later. In areas with a high or low incidence of prostate cancer, the 10-year relative survival rates were 45% (95% CI = 44%-47%) and 36% (95% CI = 34%-38%), respectively. In the early 1960s, the calculated loss of life expectancy after diagnosis varied from about 68% (95% CI = 61%-75%) of the expected length of life in the youngest age group to about 48% (95% CI = 46%-50%) in the oldest age group. From 1960 through 1964 to 1985 through 1988, the loss of life expectancy decreased by more than 50% in all age groups. The differences in relative survival rates between age groups were small, with a gradual decrease in age groups more than 60-64 years of age.
CONCLUSIONS: Most of the great temporal improvement and geographic variation in survival rates are quantitatively consistent, with likely increases in the rate of detection of nonlethal tumors. IMPLICATIONS: The increase in relative survival rates must be taken into consideration when evaluating the outcome of treatment of prostate cancer, since nonrandomized comparisons may be confounded by time trends. Diagnosis of nonlethal tumors raises concerns because the individual would suffer from the psychologic burden of a cancer diagnosis without any therapeutic benefit.