To describe patterns of health-service usage and the resulting costs in 1992-1993 for Australian men.
A prospective survey, stratified by phase of illness.
Hospital and community-based care.
A total ...of 128 homosexual men: 20 in phase 1 (CD4+ count > or = 500 x 10(6)/l), 31 in phase 2 (CD4+ count < 500 and > or = 200 x 10(6)/l), 30 in phase 3 (CD4+ count < 200 x 10(6)/l), and 47 in phase 4 (AIDS).
Mean monthly service usage rates and costs.
Health-service utilization increased and became more hospital-based as illness worsened; the main exception was use of antiretroviral drugs, which peaked in phases 2 and 3. Hospital admission was rare before diagnosis of AIDS. Hospital bed-days per patient per month averaged 3.3 for AIDS patients until the final 3 months of life increasing to 15.8 in the 3 months before death. Mean monthly costs (in 1992-1993 Australian dollars) were $331 95% confidence interval (CI), 218-455 in phase 1, $667 (95% CI, 540-836) in phase 2, $1372 (95% CI, 1044-1776) in phase 3, and $4615 (95% CI, 3456-5985) for AIDS patients until the last 3 months of life and $13,308 (95% CI, 10,538-16,516) in the 3 months before death. Drugs comprised 57% of total costs in phase 1, but only 30% of costs for patients with AIDS, whereas hospital bed-days comprised 10% of phase 1 costs and 60% of AIDS costs.
Health-care utilization and resulting costs increased with severity of illness, and were particularly high for AIDS patients in the 3 months before death. Service-utilization patterns and components of costs varied between each phase.
The correct answer to the question: ?Should we use cluster or individual randomisation?? will depend on the: type of intervention being tested, level of intervention delivery, and risk of ...contamination. If a cluster design is chosen (eg. at the level of GP, practice or community), increasing the number of clusters is more efficient in terms of sample size than increasing the number of participants per cluster. The decision whether to use individual or cluster randomisation should be considered early in the design of a study. Sample size calculations and analytical techniques need to be carried out using methods appropriate to the chosen design in order to obtain a valid result when assessing the effect of an intervention in primary health care. (author abstract)