Table 3 Associations of visual impairment with the number of unhealthy behaviours using different cut-off and imputed data. Sensitivity analyses.

From: Unhealthy behaviours and risk of visual impairment: The CONSTANCES population-based cohort

No. of unhealthy behaviours

N

Model 1a

Model 2b

OR (95% CI)

P

OR (95% CI)

P

MedDiet score (low/very low vs. intermediate/high)

38 903

 

0.0003

 

0.0004

  0

 

1.00 (reference)

 

1.00 (reference)

 

  1

 

1.17 (0.86–1.61)

 

1.17 (0.85–1.61)

 

  2

 

1.88 (1.31–2.70)

 

1.89 (1.31–2.72)

 

  3

 

2.12 (1.12–4.03)

 

2.14 (1.12–4.09)

 

Multiple imputations for missing data

52 035

 

<0.0001c

 

0.0003c

  0

 

1.00 (reference)

 

1.00 (reference)

 

  1

 

1.23 (0.89–1.70)

 

1.21 (0.87–1.69)

 

  2

 

1.53 (1.08–2.17)

 

1.47 (1.03–2.10)

 

  3

 

2.52 (1.59–4.00)

 

2.37 (1.49–3.79)

 
  1. aMixed logistic regression model adjusted for age, sex, education level and monthly income with random intercept for inclusion center.
  2. bMixed logistic regression model adjusted for age, sex, education level, monthly income, diabetes, hypertension, hypercholesterolemia and body mass index with random intercept for inclusion center.
  3. cp for trend.