Improved statistical analysis of pre- and post-treatment patient-reported outcome measures (PROMs): the applicability of piecewise linear regression splines

• Published in: Quality of life research Vol. 24; no. 3; pp. 567 - 573
• Main Authors: Greene, M. E., Rolfson, O., Garellick, G., Gordon, M., Nemes, S.
• Format: Journal Article
• Language: English
• Published: Cham Springer 01.03.2015; Springer International Publishing; Springer Nature B.V
• Subjects: Aged; Arthritis; Arthroplasty, Replacement, Hip; Clinical outcomes; Female; Hip joint; Humans; Intervention; Joint replacement surgery; Joint surgery; Linear Models; Male; Medicin och hälsovetenskap; Medicine; Medicine & Public Health; Middle Aged; Orthopedics; Ortopedi; Osteoarthritis; Osteoarthritis, Hip - surgery; Outcome Assessment (Health Care) - statistics & numerical data; Patient Outcome Assessment; Patients; Public Health; Quality of Life; Quality of Life Research; QUANTITATIVE METHODS; Questionnaires; Regression analysis; Response rates; Self Report; Sociology; Statistical analysis; Surveys and Questionnaires; EQ-5D; Segmented regression; Change point; Total hip arthroplasty
• ISSN: 0962-9343; 1573-2649; 1573-2649
• DOI: 10.1007/s11136-014-0808-3

Purpose Patient-reported health-related quality-of-life (HRQoL) measures such as the EuroQol 5 dimension (EQ-5D) index are commonplace when assessing healthcare providers or efficiency of medical techniques. HRQoL measures are generally bounded, and the magnitude of possible improvement depends on the pre-treatment HRQoL value. This paper aimed to assess and illustrated the possibility of modelling the relationship between preand post-treatment HRQoL measures with piecewise linear splines. Methods The method was illustrated using a longitudinal dataset of 36,625 patients with one EQ-5D index before and one a year after total hip arthroplasty. We considered four models: intercept only model, single line regression, and segmented regression with 1 and 2 change points. The post-operative EQ-5D index served as the outcome, while the preoperative EQ-5D index was the predictor. Results We found that a two-line regression best described the data with the lines meeting at 0.159 on the preoperative EQ-5D index scale. In the low preoperative group (with an initial preoperative index from —0.594 to 0.159), the predicted post-operative scores ranged from 0.368 to 0.765, with post-operative scores increasing 0.528 points for each unit in the preoperative score. In the high preoperative group (initial range from 0.159 to 1), the predicted post-operative scores ranged from 0.765 to 0.855, increasing 0.106 points for each unit in the preoperative score. Conclusions Piecewise linear regression is a straightforward approach to analyse baseline and follow-up HRQoL measures such as the EQ-5D index. It can provide a reasonable approximation of the shape of the underlying relationship where the threshold and slopes prove informative and meaningful.