Do marginal changes in PHI membership accurately predict marginal changes in PHI use in Western Australia?

Annual trends in the rate of utilisation of PHI in three different clinical categories were compared with published trends in PHI membership to assess the degree to which PHI membership predicts PHI use in Western Australia.

The WA Data Linkage System was used to extract all hospital morbidity records in Western Australia from 1981 to 2001. The adjusted annual incidence rate ratio of hospitalisation as a privately insured patient versus a public (Medicare) patient was estimated using Poisson regression in each clinical category across three age groups in each year. The rate ratios were graphed as segmented trend lines and compared with published data for trends in PHI membership.

The most significant changes in the use of PHI versus the public system occurred between 1981 and 1984 overall clinical categories. These changes were consistent with those documented for PHI membership. From 1992 onwards, significant changes in the trend were observed in the surgical clinical category, compared with the medical and obstetric clinical categories. Further, the trend observed in the surgical clinical category at this time was inconsistent with that documented for PHI membership. Between 2000 and 2001, only the surgical clinical category showed a similar change in trend as that documented for PHI membership.

Between 1981 and 1991 the timing and direction of changes in PHI membership were found to be congruent with that of PHI use in all three clinical categories. However, between 2000 and 2001 trends in PHI membership were only congruent with trends in PHI use in the surgical clinical category. We conclude that investigating marginal changes in PHI membership represents an incomplete method for assessing the effectiveness of policies aimed at reducing the pressure on the public system.