. 2020 Apr 29;21(1):41. doi: 10.1186/s10194-020-01110-9

Table 5.

Logistic regression modeling triptan insufficient response with backward selection results. P-values in bold indicate statistically significant factors (N = 1033)

	Odds ratio estimates
Factor	Effect	Point estimate	95% Wald confidence intervals	p-value
Age^a		0.999	0.988, 1.011	0.930
Sex^a	Female vs Male	0.920	0.663, 1.277	0.612
Currently taking OTC medications for migraines^a	Yes vs No	1.465	1.000, 2.147	0.050
When do you usually take your prescribed acute medication?^a	After pain has started and I have an idea of how severe it is vs At first sign of a migraine (before pain starts)	2.433	1.258, 4.703	0.008
	When pain starts vs At first sign of a migraine (before pain starts)	1.726	1.252, 2.380	0.001
On average, per month over the last 3 months: No. of migraine-related headache days^b		0.926	0.856, 1.002	0.057
MSQ domain: Preventive^a		0.980	0.968, 0.992	0.001
MSQ domain: Emotional^a		0.981	0.970, 0.992	0.001
EQ-5D utility scores^a		0.106	0.032, 0.355	< 0.001
Total headache days (including migraine, tension, rebound, etc.) per month in last 3 months^b		1.095	1.023, 1.174	0.010
How severe have the patient’s migraine attacks been over the last 3 months?^b		1.025	1.003, 1.048	0.027

^aPatient reported. ^bPhysician reported

The results in the table are from a backward logistic regression with a significance level of 0.1 for a variable to stay in the model. A total of 380 patients (380/1413; 26.8%) were excluded from the analyses due to missingness of variables

EQ-5D = EuroQol 5-Dimensions questionnaire; MSQ = Migraine-Specific Quality of Life Questionnaire; OTC = over-the-counter