Prostate cancer is a common disease with considerable variaton in clinical behaviour and therapeutic responsiveness. Uncertainty surrounds almost all aspects of prostate cancer management and it has been difficult to conduct proper randomised controlled trials required for reliable evidence-based decision-making. Although randomised controlled trials are under way and may eventually provide unbiased data on the efficacy of the management of prostate cancer in selected patient groups under optimal circumstances, population-based studies are necessary to evaluate variations in incidence and the effectiveness of management as practised in the community at large.

This study uses three prospectively assembled population-based cohorts and one cross-sectional group of men with prostate cancer:

1. a local register of all men with prostate cancer in the Central District of Östergotland 1974-1986 (n=813)

2. the South-East Region Prostate Cancer Register 1987-1996 (n=6782)

3. the National Prostate Cancer Register 1996-1997 (n=8328)

4. A cross-sectional group of all men with prostate cancer residing in Östergotland 1999 comprising patients from cohorts 1 and 2 (n=1442)

The incidence of prostate cancer in the South-East Region increased from 613 in 1987 to 780 in 1993 and then slowly declined. The age-adjusted incidence varied from 89/100 000 to 169/100 000 between the different counties included in the National Prostate Cancer Register 1996. In counties where a large percentage of the tumours were detected when still localised, the incidence was higher and the men younger at diagnosis (both p<0.05). In the age interval 50-59 years of age the median PSA was 13 ng/ml, whereas it was 35 ng/ml in those younger than 50. This difference was significant (p<0.05) and is probably explained by larger total tumour volume among men in the youngest age group. For men with well to moderately differentiated tumours and PSA < 20 ng/ml the risk for regional and distant metastases was below 10%. Between 1987 and 1996 the proportion of men treated with radical prostatectomy in the South-East Region decreased from 11% to 2.5%. During the same period the percentage of patients receiving GnRH-analogues increased from 3.9% to 37.8% while the percentage of patients treated with orchiectomy decreased from 40.0% to 12.8%. For patients treated with radiotherapy the median PSA was 16.7 ng/ml, for those who underwent radical prostatectomy 9.3 ng/ml, for patients receiving GnRH-analogues 61 ng/ml and for those treated with bilateral orchiectomy it was 88 ng/ml. All differences in PSA levels between the treatment groups were significant (p<0.05), indicating that there is a selection process with men having less advanced cancer receiving GnRH-analogues and men with more advanced cancer undergoing bilateral orchiectomy, and similarly a selection of men with smaller tumours being treated with radical prostatectomy rather than radiotherapy. Of the men answering the questionnaire sent to all prostate cancer patients residing in Ostergotland 1999 42% had perceived pain during the previous week and 26% stated their quality of life to be 50% or less on a visual analogue scale. A high health-care availability rating and short time since diagnosis were found to significantly predict lower rating of pain on average (p<0.05). Pain on average was found to be a significant predictive factor for decreased quality of life together with high age, low healthcare availability rating and palliative treatment (p<0.05). Age ≥ 70 years, advanced stage and poor differentiation were risk factors associated with increased risk for prostate cancer death in Östergotland Central District Cohort (p<0.05). The survival curve followed a continuous exponential course throughout the period of observation.

The geographical as well as temporal variations in incidence are probably explained by differences in diagnostic activity, which also affects the age at diagnosis and distribution of stages. There are also large disparities in management within the country, which reflects the lack of evidence supporting one treatment in favour of another. How diagnostic activity and different local habits in management affect outcome in the long run is still unknown, but the results of our study regarding quality of life and survival may be used as a basis for management decision-making. Patients with localised tumours have a favourable prognosis, even without initial treatment. When deciding on therapy, however, the grade of malignancy should be taken into account as it has a great influence on disease-specific survival. For men with well to moderately differentiated tumours and PSA < 20 ng/ml, further investigation to exclude distant and regional metastases is unnecessary.