Driving factors for the utilisation of healthcare services by people with osteoarthritis in Portugal: results from a nationwide population-based study

Table 3 Final Multinomial Regression Model

		HighUsers vs. LowUsers			GPUsers vs. LowUsers
Variables added	Determinants	OR	95%CI	p-value	OR	95%CI	p-value
Predisposing Characteristics	Age	0.96	0.95–0.99	0.001	0.99	0.96–1.01	0.172
	Male sex^a	1.05	0.69–1.58	0.826	1.12	0.72–1.76	0.613
	Geographic Location^b
	North	1.545	0.93–2.56	0.091	1.23	0.70–2.15	0.475
	Centre	1.50	0.89–2.51	0.130	2.11	1.21–3.68	0.008
	Islands	0.43	0.24–0.77	0.005	0.42	0.21–0.83	0.013
	South	1.49	0.76–2.94	0.250	1.82	0.89–3.75	0.102
Step 1	AUC^g	0.58	0.54–0.61	0.001	0.61	0.57–0.65	0.001
Step 1	McFadden Pseudo-R² = 0.026
Enabling Factors	NHS only^c	0.65	0.43–0.98	0.042	1.34	0.81–2.22	0.249
	< 4 Years of Education^d	0.90	0.56–1.45	0.900	1.50	0.93–2,43	0.096
	Employed^e	0.55	0.31–0.97	0.038	0.81	0.44–1.50	0.512
Step 2	AUC^g	0.60 (+ 0.02)	0.56–0.64	0.001	0.63 (+ 0.02)	0.59–0.67	0.001
Step 2	McFadden Pseudo-R² = 0.040 (+ 0.014)
Need Variables	Number of Comorbidities	1.12	1.03–1.21	0.011	1.22	1.11–1.33	< 0.001
	HRQoL (EQ-5D-3L index score)	0.33	0.14–0.79	0.013	0.61	0.23–1.57	0.303
	Physical function (HAQ score)	1.59	1.10–2.23	0.013	1.03	0.69–1.53	0.902
	Anxiety (HADS-A)	1.02	0.97–1.07	0.432	1.09	1.03–1.14	0.002
	Regular Physical Exercise^f	0.57	0.37–0.88	0.010	0.55	0.34–0.89	0.014
Step 3: Final model	AUC^g	0.68 (+ 0.08)	0.64–0.71	0.001	0.69 (+ 0.07)	0.65–0.73	0.001
Step 3: Final model	McFadden Pseudo-R² = 0.098 (+ 0.058)

NHS National Health System, GP General Practitioner, HRQoL Health Related Quality of Life, EQ-5D-3L EuroQol with five dimensions and three levels, HAQ Health Assessment Questionnaire, HADS-A Hospital Anxiety and Depression Scale – Anxiety subscale. Reference Categories: ^aFemale; ^bLisbon and Tagus valley; ^c Healthcare insurance along with NHS; ^d ≥ 4 years of education; ^eNon employed or retired; ^f Don’t perform regular physical exercise; ^gArea Under the ROC Curve (95% CI) – reference cluster is LowUsers. Differences in discriminatory capacity (AUC) and in variance of the model regarding the previous step is shown in brackets. χ²(28) = 180.328, p < 0.001
This procedure excluded all the participants with missing data. Sample included in the analysis: Total: n = 838 (85,69% of the initial sample), HighUsers: n = 295 (86.0% of the initial cluster sample); GPUsers: n = 232 (85,29% of the initial cluster sample); LowUsers: n = 311 (85.67% of the initial cluster sample)

ISSN: 1472-6963