OBJECTIVES: This study's objective was to determine the prevalence of body shape changes and metabolic abnormalities in an ambulant population with HIV infection. Three different definitions of lipodystrophy were used to assess these changes. Patients' anthropometric measures and dual-energy X-ray absorptiometry (DEXA) scans were compared in order to estimate fat distribution in this population. We sought to evaluate potential predictors for lipodystrophy according to each of the three definitions. METHODS: We performed a cross-sectional study in the outpatient clinic of a tertiary referral hospital in Melbourne, Australia. We enrolled a total of 167 HIV-infected ambulatory patients over 3 months in mid-1998. Data on 159 males, 149 of whom were receiving triple combination antiretroviral therapy, were evaluated. Anthropometric measures, clinical examination, self-report of body shape changes, biochemical measures and DEXA scan were used to assess lipodystrophy and risk factors for cardiovascular disease. Patients described body shape changes in the face, trunk, arms and legs. Laboratory parameters measured included fasting triglyceride (TG), cholesterol, high-density lipoproteins (HDL), glucose, insulin, CD4 cell count and plasma HIV RNA. Current and past antiretroviral therapies were ascertained. RESULTS: According to one proposed Australian national definition of lipodystrophy (LDNC), the prevalence of lipodystrophy in this population was 65%. This definition included an objective assessment with major and minor criteria. Patient-defined lipodystrophy (LDP), which involved a subjective assessment of thinning arms and legs and central adiposity, occurred in 19%. Patient-defined lipoatrophy (LAP), which involved a subjective assessment of thinning arms and legs without central adiposity, occurred in 21.3%. No change in body habitus was noted by 37% of the cohort. Hypercholesterolaemia was recorded in 44%, hypertriglyceridaemia in 52% and elevated insulin levels in 23%. Anthropometry was predictive of the per cent total body fat recorded by DEXA scan, but produced consistently lower values. In multivariate analysis, LDP and LAP were significantly associated with stavudine (d4T) use, while LAP was also associated with zidovudine (ZDV) treatment. There were no treatment associations with LDNC. Protease inhibitor (PI) exposure was associated with metabolic changes but not patient perceived body shape changes, while d4T and ZDV exposure was associated with increased triglycerides and reduced peripheral fat stores. CONCLUSIONS: The prevalence of body shape changes in a single population varied depending on the definition applied. The LDNC definition overestimated body shape abnormalities in comparison with patient perception. LAP was associated with significantly lower fat stores measured by anthropometry and DEXA scan than those identified under the LDNC definition. In contrast to LDNC, LAP was associated with d4T exposure, nucleoside reverse transcriptase inhibitor (NRTI) and ZDV duration of use, but not PI use. Until a consensus definition for lipodystrophy is developed, including agreement on objective measurement and thresholds for abnormality, careful description of the individual components of the syndrome is required to enable cohort comparisons so that predictors of the syndrome can be assessed more accurately and outcome studies made feasible.