Table 2.
Reason for referral | GP age per 10 years, OR (95% CI) | GP sex Male = 0 Female = 1, OR (95% CI) | Patient age per 10 years, OR (95% CI) | Patient sex Male = 0 Female = 1, OR (95% CI) | Speciality in family medicine No = 0 Yes = 1, OR (95% CI) | Travel time to nearest hospital per 60 mins, OR (95% CI) | GPs’ referral rate per 1% increase, OR (95% CI) |
---|---|---|---|---|---|---|---|
Medically necessarya | 1.06 (0.77 to 1.45) | 1.26 (0.74 to 2.15) | 1.10b (1.02 to 1.18) | 0.70 (0.48 to 1.02) | 0.73 (0.34 to 1.58) | 1.20 (0.95 to 1.53) | 0.97 (0.92 to 1.02) |
To avoid overlooking anythingc | – | 0.90 (0.50 to 1.64) | – | 1.10 (0.83 to 1.48) | – | 1.20 (0.87 to 1.66) | 1.06b,d (1.01 to 1.11) |
Perceived deficient medical knowledgea | 0.85 (0.68 to 1.08) | 2.22b (1.47 to 3.36) | 0.94 (0.87 to 1.02) | 0.82 (0.58 to 1.16) | 0.997 (0.59 to 1.68) | 1.22 (0.97 to 1.53) | 1.05 (0.99 to 1.11) |
To reassure the patienta | 0.87 (0.63 to 1.20) | 1.97b (1.11 to 3.50) | 1.04 (0.96 to 1.12) | 1.06 (0.74 to 1.51) | 1.17 (0.60 to 2.27) | 1.25 (0.90 to 1.74) | 1.03 (0.97 to 1.09) |
Patient preferencea | 0.54b (0.40 to 0.74) | 1.36 (0.71 to 2.60) | 1.01 (0.95 to 1.08) | 1.10 (0.80 to 1.52) | 3.18b (1.54 to 6.57) | 1.46b (1.002 to 1.90) | 1.03 (0.96 to 1.11) |
Perceived easily accessible specialist | 0.74 (0.50 to 1.11) | 0.29b (0.09 to 0.995) | – | – | – | – | 1.10e (0.997 to 1.22) |
Analysed by multivariable ordered logistic regression, reporting standard errors that allowed for clustering at GP level. Adjusted for patient age, patient sex, GP age, GP sex, speciality in family medicine, practice type, travel time to nearest hospital, country where medical degree was obtained, and GPs’ referral rates in analyses of ‘perceived deficient medical knowledge’, ‘to reassure the patient’, and ‘patient preference’. Adjusted for the former background variables except practice type in ‘medically necessary’, and adjusted for background variables presented in the row for ‘perceived easily accessible specialist’ (see Method).
P<0.05.
Analysed by multivariable ordered logistic regression, by the command Gologit2, reporting standard errors that allowed for clustering at GP level, and adjusted for background variables presented in the row (see Method).
OR = 1.06 comparing the three highest agreement levels (2+3+4) with the lowest level (0). OR = 1.009 (95% CI = 0.97 to 1.05) level 1+2 versus 3+4, and OR = 1.004 (95% CI = 0.94 to 1.08) level 1+2+3 versus 4.
P = 0.056. OR = odds ratio.