Skip to main content

Table 3 Final adjusted and trimmed model for headache days per montha

From: Fluctuations in episodic and chronic migraine status over the course of 1 year: implications for diagnosis, treatment and clinical trial design

Random-Effect Variance Parameter Estimates

 

Estimate

Standard Error

 

 Intercept

0.52

0.01

 

 Linear trend

0.15

0.01

 

 Quadratic trend

0.09

0.01

 

 Scale

0.08

0.005

 

Fixed Effects b

 Label

RR (95% CI) c

DF

t

P value

 Intercept

1.16 (1.02–1.32)

16,650

2.28

0.0224

 Sex: male vs. female

0.9 (0.87–0.93)

16,650

6.62

<0.0001

 Education: college graduate vs. non–college graduate

0.96 (0.93–0.99)

16,650

2.92

0.0035

 Race: Caucasian vs. nonwhite

1.19 (1.14–1.23)

16,650

8.78

<0.0001

 Household income: at or above national median vs. those below

0.96 (0.94–0.99)

16,650

2.42

0.0155

 Comorbid count

1.07 (1.05–1.09)

16,650

8.06

<0.0001

 Disability: MIDAS category TIC

1.47 (1.46–1.49)

16,650

73.56

<0.0001

 Age at onset, y

0.998 (0.997–0.999)

16,650

3.20

0.0014

 Duration of illness

1.00 (1.001.00)

16,650

0.83

0.4090

 Linear trend

0.74 (0.73–0.76)

29,304

28.63

<0.0001

 Quadratic trend

1.09 (1.07–1.11)

29,304

8.35

<0.0001

 Cubic trend

0.99 (0.971.01)

29,304

1.34

0.1797

 Headache status: CM vs. EM

3.95 (3.84–4.07)

29,304

90.71

<0.0001

 Headache status by linear trend interaction

1.26 (1.2–1.33)

29,304

9.34

<0.0001

 Headache status by quadratic trend interaction

0.99 (0.941.03)

29,304

0.51

0.6434

 Headache status by cubic trend interaction

1.00 (0.961.04)

29,304

0.17

0.8627

  1. CM chronic migraine, DF degrees of freedom, EM episodic migraine, MIDAS Migraine Disability Assessment questionnaire, RR rate ratio, t t-statistic; TIC time invariant covariate, TVC time-varying covariate
  2. aIn persons with EM and CM and adjusting for demographic variables, headache related disability and comorbidity
  3. bAll covariates were specified as TICs, though TVCs could have been specified for several including MIDAS and comorbidity count, which may not have been constant over time
  4. cStatistically significant RR are bolded