Potential unmet need for gout diagnosis and treatment: capture-recapture analysis of a national administrative dataset

Abstract

To estimate the degree of undercount of people diagnosed with gout in administrative datasets using capture-recapture methods. Hospitalization and drug dispensing claims (allopurinol or colchicine) data for all Aotearoa New Zealand were used to estimate the prevalence of gout in 2009 (n = 4 295 296). As a comparison, we calculated gout prevalence using a large primary care dataset using general practitioner diagnosis and prescribing records (n = 555 313). For each of these datasets, we estimated the undercount through capture-recapture analysis using a Poisson regression model. A two-list model was used, which included covariates such as age, gender, ethnic groups and New Zealand deprivation quintiles. The crude prevalence of diagnosed gout in the Aotearoa New Zealand population aged ≥ 20 years was 3.75%. The covariate-adjusted capture-recapture estimate of those not recorded but likely to have gout was 0.92%, giving an overall estimated prevalence of 4.67% (95% CI 4.49, 4.90%) for the population aged ≥ 20 years. This amounts to 80% of people with gout being identified by the algorithm for the Aotearoa New Zealand data-that is being recorded in either lists of dispensing of allopurinol or colchicine or hospital discharge. After capture-recapture, gout prevalence for all males aged ≥ 20 years was 7.3% and in older (≥ 65 years) Māori and Pacific men was >30%. Capture-recapture analysis of administrative datasets provides a readily available method for estimating an aspect of unmet need in the population-in this instance potentially 20% of those with gout not being identified and treated specifically for this condition.