Migraine day frequency in migraine prevention: longitudinal modelling approaches

Table 3 CM regression output for negative binomial, beta-binomial and Poisson

Negative binomial (Dispersion parameter 0.1323)						Beta-binomial (ICC 0.1370)						Poisson
	Predicted MD Frequency^a	95% CI	FRR	95% CI	P value		Predicted MD Frequency^a	95% CI	Coefficient	95% CI	P value		Predicted MD Frequency^a	95% CI	FRR	95% CI	P value
Week 0	18.111	17.052, 19.171)	–	–	–	Week 0	17.111	(16.156, 18.066)	–	–	–	Week 0	18.298	(17.373, 19.223)	–	–	–
Week 4	15.418	(14.579, 16.257)	0.783	(0.754, 0.812)	< 0.001	Week 4	15.843	(15.028, 16.657)	−0.256	(− 0.321, -0.192)	< 0.001	Week 4	15.577	(14.770, 16.385)	0.798	(0.773, 0.824)	< 0.001
Week 8	14.538	(13.759, 15.317)	0.721	(0.694, 0.749)	< 0.001	Week 8	15.256	(14.484, 16.027)	−0.359	(−0.426, -0.293)	< 0.001	Week 8	14.688	13.919, 15.457)	0.739	(0.715, 0.764)	< 0.001
Week 12	13.997	(13.242, 14.753)	0.696	(0.670, 0.724)	< 0.001	Week 12	14.894	(14.146, 15.641)	−0.408	(−0.475, -0.341)	< 0.001	Week 12	14.142	13.397, 14.887)	0.715	(0.692, 0.739)	< 0.001
Treatment (Erenumab vs Placebo)			0.828	(0.767, 0.894)	< 0.001	Treatment (Erenumab vs Placebo)			−0.3600	(−0.430, -0.290)	< 0.001	Treatment (Erenumab vs Placebo)			0.831	(0.770, 0.896)	< 0.001
RSME		0.082				RMSE		0.081				RMSE		0.152
MAE		0.330				MAE		0.339				MAE		0.654

Regression output analysis was based on the whole sample of patients (1872 observations)
CI confidence interval, FRR frequency rate ratio, ICC intraclass correlation coefficients, MAE mean absolute error, MMD monthly migraine day, RMSE root mean squared errors
^a In the placebo arm

ISSN: 1471-2288