Biases in estimates from the RNZCGP computer research group

This study aimed to determine whether conclusions drawn in studies using data from the computer research group of the Royal New Zealand College of General Practitioners (RNZCGP) could be extrapolated to other New Zealand general practices.

Retrospectively collected data on doctor, practice, and consultation variables form the study database. The control group comprised a random sample of 106 New Zealand general practitioners. The study group were 67 general practitioners participating in the RNZCGP computer research group. Comparisons between groups were based on doctor and practice variables, patient demography and morbidity, and number and type of service items observed.

Study group doctors were more likely to have received post graduate training in general practice (p < 0.01) and saw more patients entitled to government subsidised health care (p < 0.01). The geographical distribution of the study group was skewed with more located in the south of New Zealand. Wide variability on most other study parameters was seen among doctors in both groups although overall patient morbidity was similar for both groups in the 8612 consultations analysed. Computer research group general practitioners reported lower rates of patient referral and laboratory investigations, and higher immunisation rates than the control group (p < 0.01).

Although the profile of study group general practitioners was different from that of the control group, data collected by both groups provided a similar reflection of the morbidity and services of New Zealand general practice. Adjustments will be needed for extrapolating the results of research from the RNZCGP computer research group where the focus of the investigation is referrals from primary care, investigations, or immunisations.