Skip to main content

Table 3 Covariates in nonlinear mixed-effects models

From: Using multivariate nonlinear mixed-effects model to investigate factors influencing symptom improvement after high tibial osteotomy in combination with bone marrow concentrate injection for medial compartment knee osteoarthritis: a prospective, open-label study

A. VAS

Nonlinear mixed-model with covariates

\(\Delta (t)=(a+\sum_{i}{\beta }_{i}{C}_{i})+\left[{a}_{0}-(a+\sum_{i}{\beta }_{i}{C}_{i})\right]{e}^{-rt}\)

Parameter

Estimate

[95% CI]

P value

\(a\)

-9.32

[-20.75, 2.11]

-

\({\beta }_{Sex}\)

5.95

[0.99, 10.91]

0.009

\({\beta }_{KL grade}\)

4.69

[1.34, 8.04]

0.003

\({\beta }_{FTA}\)

0.45

[0.02, 0.88]

0.03

\({\beta }_{PTS}\)

-1.25

[-1.88, -0.63]

< 0.0001

\({\beta }_{VAS baseline}\)

-0.80

[-0.88, -0.72]

< 0.0001

\({a}_{0}\)

0.67

[-1.43, 1.70]

-

\(r\)

0.54

[0.36, 0.72]

-

AICc; BIC

693.07; 704.25

  

B. WOMAC

Nonlinear mixed-model with covariates

\(\Delta (t)=(a+\sum_{i}{\beta }_{i}{C}_{i})+\left[{a}_{0}-(a+\sum_{i}{\beta }_{i}{C}_{i})\right]{e}^{-rt}\)

Parameter

Estimate

[95% CI]

P value

\(a\)

-14.53

[-26.68, -2.38]

-

\({\beta }_{BMI}\)

0.28

[0.01, 0.55]

0.04

\({\beta }_{KL grade}\)

3.27

[0.98, 5.56]

0.003

\({\beta }_{WOMAC baseline}\)

-0.87

[-0.88, -0.86]

< 0.0001

\({a}_{0}\)

0.13

[-0.81, 1.07]

-

\(r\)

0.47

[0.27, 0.67]

-

AICc; BIC

573.85; 583.44

 

C. KOOS

Nonlinear mixed-model with covariates

\(\Delta (t)=(a+\sum_{i}{\beta }_{i}{C}_{i})+\left[{a}_{0}-(a+\sum_{i}{\beta }_{i}{C}_{i})\right]{e}^{-rt}\)

Parameter

Estimate

[95% CI]

P value

\(a\)

364.13

[211.94, 516.32]

-

\({\beta }_{Age}\)

1.58

[0.03, 3.13]

0.04

\({\beta }_{BMI}\)

-2.13

[-3.30, -0.06]

0.02

\({\beta }_{KL grade}\)

-10.06

[-20.08, -0.04]

0.04

\({\beta }_{FTA}\)

-5.51

[-7.85, -3.17]

< 0.0001

\({\beta }_{KOOS baseline}\)

-0.91

[-1.07, -0.75]

< 0.0001

\({a}_{0}\)

-8.18

[-16.80, 0.44]

-

\(r\)

0.40

[0.25, 0.55]

-

AICc; BIC

988.82; 1000.77

 
  1. Sex: male = 0, female = 1
  2. Femorotibial angle: varus = negative value, vulgus = positive value
  3. Abbreviations: \({\varvec{a}}\) asymptote, \({{\varvec{a}}}_{0}\) intercept, AICc Akaike information criterion with correction for small sample sizes, BIC Bayesian information criterion, BMI body mass index, \({{\varvec{C}}}_{{\varvec{i}}}\) the ith covariate, CV coefficient of variation, FTA femorotibial angle, KL Kellgren-Lawrence, KOOS Knee injury and Osteoarthritis Outcome Score, PTS posterior tibial slope, r natural logarithm of the rate constant, SE standard error, t follow-up time, VAS Visual analogue Scale, WOMAC Western Ontario and McMaster Universities Arthritis Index, \({{\varvec{\beta}}}_{{\varvec{i}}}\) correlation coefficient of the ith covariate, ∆ difference between patients’ postintervention and baseline scores