Variation in the use of medicines by ethnicity during 2006/07 in New Zealand: a preliminary analysis.

AIM: To describe variations in dispensing of specific medication groups by ethnicity in New Zealand, adjusting for health need. METHOD: Preliminary linkage of dispensings of prescription medicines in 2006/07 to age/disease burden proxies of health need for Maori, Pacific peoples (Pasifika)--who are mostly of Samoan, Tongan, Niuean, or Cook Islands descent--in New Zealand, and non-Maori/non-Pasifika. These disease burden proxies combine differences in prevalence, age, morbidity, and mortality. Variations were disaggregated by patients being first dispensed medicines ('access') versus subsequent dispensings ('persistence'). RESULTS: Initially, overall age-adjusted incidence of 'scripts' (prescriptions dispensed) to Maori was similar to that of non-Maori. There were differences in therapeutic coverage between Maori and Pasifika, for example greater use of asthma medicines in Maori. However, further adjustments linking with disease burden showed marked variance for a number of diseases. Differences in dispensing included areas of high health need such as heart disease, infections, diabetes, mental health and respiratory disease. Maori had 19-37% lower dispensings overall than non-Maori, with a net difference of nearly 1 million scripts. Maori were both less likely to access medicines, and then after first dispensing had fewer subsequent scripts. Patterns for Pasifika appeared similar, although needs-adjusted analysis is awaited for this population. CONCLUSIONS: Once adjusting for need, there was variable but sizeable differences in medicines dispensed to Maori compared with non-Maori, and likely differences for Pasifika populations. There are however important limitations to this preliminary analysis. Crude and age-standardised metrics may be poor predictors of needs-adjusted gaps in medicines use. In this analysis, solely age-standardised rates tended to underestimate differences once adjusting for burden of disease; future analyses of prescribing patterns should consider better adjusting for disease burden.