Model 1 | Model 2 | |||
---|---|---|---|---|
1987 | 2001 | 1987 | 2001 | |
Variation | ||||
Practice level variance | 0.063 | 0.061 | 0.029 | 0.035 |
Patient level variance | 0.939 | 0.948 | 0.788 | 0.756 |
Patient characteristics | ||||
Intercepta (mean NEQ score, standardised) | −0.006 (0.028) | 0.019 (0.028) | 0.072 (0.028) | −0.111 (0.031) |
Medical need variables | ||||
Self-reported healtha (0 = [very] good, 1 = moderate/poor) | 0.047 (0.034) | 0.141b (0.028) | ||
One or more health complaintsa (1 = yes) | −0.053 (0.032) | 0.050 (0.038) | ||
Chronic illness (1 = yes) | −0.033 (0.026) | 0.009 (0.024) | ||
Visit to GP last 2 months (1 = yes) | 0.089c (0.025) | 0.123b (0.022) | ||
Enabling factors | ||||
Socioeconomic status (1 = low, 2 = medium, 3 = high) | −0.103c (0.023) | −0.087b (0.018) | ||
Public insurance (1 = yes) | 0.150c (0.028) | 0.107b (0.025) | ||
Factors independent of illness | ||||
Sexa (1 = male) | 0.190b (0.024) | 0.324b (0.022) | ||
Educational level (1 = no/low, 3 = high) | −0.238b (0.029) | −0.196b (0.023) | ||
Age (continuous)c | 0.023b (0.002) | 0.027b (0.002) | ||
Non-Western background (1 = yes) | 0.337b (0.103) | <0.570b (0.070) | ||
GP attributes | ||||
Duration of residency (continuous) | 0.002 (0.002) | −0.006 (0.005) | ||
Age (continuous) | 0.004 (0.002) | 0.010 (0.007) | ||
Sex (1 = male) | 0.010 (0.052) | −0.011 (0.043) | ||
Hours working (continuous) | 0.117 (0.110) | 0.226 (0.132) | ||
Practice attributes | ||||
Type of practice (0 = single-handed, 1 = group) | 0.031 (0.057) | 0.008 (0.064) | ||
Urbanicity (1 = most urban to 5 = rural) | 0.013 (0.016) | 0.005 (0.017) | ||
Total GPs in practice (continuous) | −0.066 (0.034) | 0.018 (0.037) | ||
Total listed patients (continuous) | <0.001 (0.000) | <−0.001 (0.000) |
↵a Significant effect between both samples: the difference in standardised mean score was significant at the level of P<0.001; the difference in effect of male sex between the two studies was significant at the level of P<0.01, whereas the difference in effects of self-reported bad health and reporting more health complaints were significant at the level of P<0.05.
↵b In order to capture a possible non-linear relationship between age and patients' attitudes, age was modelled as a separate polynomial effect.
↵c Significant effect of independent variable on patients' scores as measured with the Nijmegen Expectation Questionnaire (NEQ).