Skip to main content

Table 3 MCID estimated by the coefficients of linear-mixed effect model, considering baseline General Health scores, and applied to the LIGALONGO dataset using different imputation methods for missing data

From: Practical issues encountered while determining Minimal Clinically Important Difference in Patient-Reported Outcomes

 

MCID for improvement

MCID for deterioration

All range baseline GHa

Low baseline GHb

Medium baseline GHb

High baseline GHb

All range baseline GHa

Low baseline GHb

Medium baseline GHb

High baseline GHb

Complete casesi

3 [1; 5]

12 [8; 15]

2 [−2; 5]

−7 [− 12; − 3]

− 10 [− 15; − 5]

− 4 [− 10; 3]

−16 [− 27; − 5]

− 17 [− 27; − 7]

Imputation by the meanic

3 [1; 5]

14 [10; 18]

− 1 [− 7; 5]

−13 [− 19; − 6]

− 11 [− 16; − 6]

5 [− 3; 13]

− 8 [− 20; 4]

− 19 [− 32; − 5]

Simple MICE imputationd

3 [1; 5]

12 [8; 15]

2 [− 2; 5]

−7 [− 11; − 4]

−8 [− 13; − 4]

−2 [− 8; 5]

− 15 [− 24; − 5]

− 17 [− 26; − 7]

Complex MICE imputatione

3 [1; 6]

10 [7; 14]

1 [− 2; 4]

− 6 [− 10; − 3]

− 8 [− 13; − 4]

− 3 [− 8; 3]

−12 [− 18; − 7]

− 19 [− 25; − 13]

Simple MICE imputation (available)f

4 [1; 6]

13 [10; 16]

0 [− 3; 3]

− 6 [− 10; − 2]

−9 [− 14; − 4]

0 [− 5; 5]

−12 [− 18; − 7]

− 19 [− 24; − 13]

Complex MICE imputation (available)f

1 [− 1; 4]

10 [7; 13]

− 1 [− 4; 2]

−8 [− 11; − 4]

− 10 [− 15; − 5]

− 3 [− 8; 2]

−14 [− 19; − 8]

− 20 [− 26; − 15]

  1. The presented scores were obtained with a linear-mixed effects regression, with a random effect on the individual (random intercept model) to estimate the mean change score in each category of patients (little better/ same/ little worse), and the baseline GH score as a fixed effect (+/− an interaction between baseline GH score and the anchor). The models including an interaction term are signaled with i
  2. MICE Multivariate Imputation by Chained Equations, GH General Health
  3. aMinimal Clincally Important Difference estimated by the Mean GH-score difference between visits 1 and 5 or 1 and 7
  4. bMinimal Clincally Important Difference estimated by the Mean GH-score difference between visits 1 and 5 or 1 and 7, according to each group of baseline GH score. Baseline GH scores are classified as low [0,65], medium (65,82] or high (82,100). The mean GH differences correspond to the fixed-effect associated coefficients of the patients who answered respectively “little better”, “little worse” and “same” at the anchor question. Values in bracket are Confidence Interval at a 95% level
  5. cMissing scores were imputed by the mean-score, and missing anchors were imputed on the base of a weighted-probability
  6. dMissing scores were imputed using personal mean matching, anchor was imputed using a polytomous regression, both using clinical and demographic variables, and GH scores
  7. eMissing scores were imputed using personal mean matching, anchor was imputed using a polytomous regression, both using clinical and demographic variables, and all SF-36 scores
  8. fThe same MICE methods were applied, using only available information