Responsiveness and the Minimal Clinically Important Difference for HM-PRO in Patients with Hematological Malignancies

Background: The responsiveness of a patient-reported outcome (PRO) measure is an important measurement property which is required to show that the instrument is capable of detecting change over time or after a therapeutic intervention. Minimal clinically important difference (MCID) on the other hand is the smallest change in the scores that is meaningful from a patient's perspective. Evaluating both responsiveness and MCID of a newly developed PRO instrument is a measurement necessity. A novel hematological malignancy (HM) specific PRO tool, HM-PRO, has been developed for use in daily clinical practice as well as clinical research. The HM-PRO is a composite measure consisting of two scales: Part A-measuring the impact on patients' quality of life (QoL); and Part B-measuring the effect of signs and symptoms experienced by the patients. The aims of this study were to evaluate the responsiveness and estimate the MCID of the newly developed HM-PRO.

Methods: In a prospective multi-center cross-sectional study, a total of 299 patients: male 167 (55.9%); mean age 64.9 (±11.5) years, range 17.9-88.5 years; mean time since diagnosis 4.9 (±4.8) years, range 0.003 - 27.4 years; with different HM's (acute lymphoblastic leukemia n=8, acute myeloid leukemia n=18, aggressive non Hodgkin lymphoma n=21, chronic lymphocytic leukemia n=24, chronic myeloid leukemia n=11, Hodgkin lymphoma n=12, indolent non Hodgkin lymphoma n=8, myelodysplastic syndrome n=-65, multiple myeloma=85, myeloproliferative neoplasm n=47); in different disease states (stable-139, remission-82, progressing-78) were recruited. Patients were asked to complete the HM-PRO at baseline (t0) and after three months (t1). Data analysis was performed using IBM SPSS 23 statistical software. Paired t-test, effect size and standard response mean (SRM) analysis were performed to test HM-PRO's responsiveness as the ability to detect change. An anchor question (5-point Likert scale from 'much worse to much better') measuring change in HRQoL from patients' perspective was used. Distribution-based approach was used to estimate the MCID, using standard deviation (SD) and standard error of measurement (SEM).

Results: Both parts of HM-PRO were assessed individually for responsiveness. The paired sample t-test was significant in the 'slightly better' group for Part A (mean score change=6.8, SD=13.7, t=3.343, df=44, p=0.002; Cohen's d= 0.35; SRM =0.50), and Part B (mean score change=3.68, SD=8.7, t=2.7, df=42, p=0.009; Cohen's d=0.31; SRM=0.42) showing significant decrease in score over time and with moderate effect size. This was followed by 'much worse' and 'much better'. In contrast, patient group with 'about the same' and 'slightly worse' HRQoL did not show significant change in scores over time for both Part A (mean score change=1.7, SD=12.7, t=1.7, df=154, p=0.09; Cohen's d=0.08; SRM=0.13) and Part B (mean score change=0.57, SD=8.93, t=0.80, df=153, p=0.42; Cohen's d=0.04; SRM=0.06) (Table 1). In order to establish MCID, statistical characteristics of the sample of baseline patient response were studied for both Part A and Part B. The analysis of Part A showed a SEM of 6.2 and ½ SD of 11.2, and for Part B showed a SEM of 5.9 and ½ SD of 7.3 (Table 2).

Conclusion: This study has established the responsiveness and MCID for both Part A and Part B of HM-PRO. The HM-PRO is capable of detecting small but clinically important changes in patients' HRQoL over time. The MCID for Part A based on SEM was 6.2 and for PART B 5.9 points. It would therefore be prudent, for practical reasons, to propose a MCID of '6' for the HM-PRO.

View largeDownload slide

View largeDownload slide

Close modal