Case-Control Study of the Association Between Kava Use and Ischaemic Heart Disease in Aboriginal Communities in Eastern Arnhem Land (Northern Territory) Australia

摘要

In a case-control study, cases comprised 83 people admitted to hospital for the first time during 1992-1997 from the region with a medical officeru27s confirmed diagnosis of IHD (ICD9 Codes 4100-4149). Of these, 25 were admitted on more than one occasion. Up to four randomly selected controls (n=302) were matched with each case for age, sex, and home locality. NT registries indicated that a further 20 people with no record of hospital admission died with IHD during 1992-1997. These were matched with 75 controls.Comprehensive data were not available to identify IHD morbidity before 1992.Methods to measure exposure to kava use, alcohol, tobacco, petrol sniffing, cannabis use, and other possible confounding factors, data analysis techniques, and ethics approvals have been described elsewhere. RESULTS Adjusting for confounders, odds ratios (OR) for kava use before or during 1992-97 changed from 1.41 (95% CI 0.73 to 2.73, p=0.303) to 1.51 (0.75 to 3.05, p=0.247) (table 1). There was no residual confounding effect of age in the multivariate model (OR=1.50, 0.74 to 3.04), (x2=0.23, likelihood ratio test, p=0.635). There was no association with kava use in just those communities where kava had been used for up to 15 years (adjusted OR=1.75, 0.82 to 3.74, p=0.140) or when those admitted on more than one occasion (n=25) were compared with their matched controls (n=132) (adjusted OR=2.24, 0.65 to 7.68, p=0.191). Twenty who died from IHD without hospital admission and 75 matched controls were combined with 83 known admissions and 302 matched controls. No association with kava use was found (adjusted OR=1.44, 0.78 to 2.66, p=0.245) so the results of the analysis of IHD admissions alone were probably not influenced by survival bias.While the expected association between IHD and tobacco use was not found in the univariate analysis (table 1), it appeared when 36 cases were compared with 158 controls who had no record of kava, alcohol, cannabis, or petrol use (OR=3.96, 1.08 to 14.49, p=0.021).