Predicting AIDS deaths and prevalence in Australia.
OBJECTIVE: To describe a method for predicting deaths from, and the prevalence of, the acquired immunodeficiency syndrome (AIDS) in Australia and to obtain ranges of projections for the next five years based on data observed to 30 September 1989 and reported by January 1990. The projections are important components of planning for future health-care needs. PATIENTS: All patients with AIDS in Australia who were known to the National Centre in HIV Epidemiology and Clinical Research. The data used in the analysis were date of diagnosis of AIDS, date of death if the patient had died, and State or Territory of diagnosis. RESULTS: For Australia as a whole, the doubling time of AIDS incidence changed from 0.83 years before mid 1987 to 4.34 years after that time. Five hundred and seventy patients were diagnosed with AIDS to 30 June 1987 of whom 487 had died by 30 September 1989 and the estimated mean survival time was 1.36 years. Of 1037 patients diagnosed with AIDS after June 1987, 352 had died and the estimated mean survival time was 2.42 years. Combining the estimated survival distribution with the projected new cases of AIDS, we forecast that there will be between 530 and 680 deaths from AIDS in Australia in the year mid 1991 to mid 1992. Estimated cumulative deaths to mid 1994 range from 3390 to 4250. We predict that there will be between 1370 and 1850 people living with AIDS during the year ending mid 1992, and that this number will increase to between 1760 and 2830 for the year ending mid 1994. We also predict that the prevalence of AIDS in New South Wales will lie between 820 and 1620 for the year ending mid 1994, and that the prevalence in Victoria will lie between 270 and 740 in the same year. CONCLUSIONS: Any value in the range of predictions for AIDS incidence we give is equally likely and there is good agreement with the data now observed from mid 1989 to mid 1990. The dramatic increase in the doubling time is the result of a number of factors predicted by epidemic theory and the availability of treatment, zidovudine in particular. Our estimates of deaths and prevalence have been influenced by the quality of the available death data. The observed number of deaths in the year mid 1989 to mid 1990 has exceeded the number forecast for that year. Substantial improvements in survival were associated with the introduction of zidovudine into clinical practice in mid 1987 but the death rate has risen rapidly again in the very recent past, possibly due to patients becoming increasingly refractory to treatment with zidovudine (before or after a diagnosis of AIDS). Our approach to forecasting can be adjusted to accommodate a transient effect of treatment as further data accumulate.