Vortragssitzung
Patient reported Outcome Measures (PROMs)
Vorträge
Exploring treatment effect heterogeneity of a PROMs alert intervention: A causal forest application to a randomized controlled trial
Benedikt Langenberger
Einleitung / Introduction
Patient reported outcome measures (PROMs) help gain insights into knee or hip arthroplasty patient's perceiving of health and function. In addition, PROMs may also be used as intervention themselves. This study aimed to evaluate I) the overall treatment effect and, of special interest for practical implementation, II) subgroup-specific treatment effect heterogeneity of an electronic PROMs (ePROMs) intervention in knee or hip arthroplasty patients using a recently developed machine learning method.
Methode / Method
We used data of 6,807 hip and knee arthroplasty patients from the PROMoting Quality study, a multi-centered randomized controlled trial (RCT) that included 9 German hospitals. We applied a novel machine learning method, the causal forest, to estimate the average treatment effect (ATE) of the intervention and the conditional average treatment effect (CATE) for certain subgroups to explore treatment effect heterogeneity. As outcomes, we used pre- and one-year post-operative differences in EQ-5D-5L, EQ-VAS and Hip disability and Osteoarthritis Outcome Score - Physical Function Short Form (HOOS-PS) / Knee injury and Osteoarthritis Outcome Score - Physical Function Short Form (KOOS-PS). Subgroups were built with respect to pre-operative outcome PROM scores, body mass index, PROMIS depression score, age, gender, smoking behavior, employment status and the presence of a blood pressure disease.
Ergebnisse / Results
The average treatment effect for EQ-5D-5L was 0.012 (95% confidence interval (CI): 0.000 to 0.024) for hip and 0.010 (95% CI: 0.004 to 0.016) for knee arthroplasty patients. EQ-VAS improved by 1.861 (95% CI: 0.999 to 2.723) for hip and 2.218 (95% CI: 0.822 to 3.613) for knee arthroplasty patients. HOOS-PS scores declined statistically significantly by -1.002 (95% CI: -1.713 to -0.291) and KOOS-PS by -0.972 (95% CI: -1.868 to -0.077). Subgroup analysis revealed treatment effect heterogeneity in certain subgroups. Female, older, non-smoking, unemployed patients and patients with blood pressure disease benefitted more from the intervention as their counterparts, which sometimes did not benefit at all from the intervention.
Zusammenfassung / Conclusion
This study found that a PROMs intervention was overall effective for both knee and hip arthroplasty patients on all PROMs used for outcome determination. CATE estimates indicated that the intervention was especially effective in certain subgroups, and not in others. Therefore, this study helps facilitate decision making and targeting the intervention towards certain patient groups in which it is especially effective.
AutorInnen
David Kuklinski
Validität und Praktikabilität relevanter Messinstrumente bei Psoriasis: Papier vs. Tablet
Marina Otten
Einleitung / Introduction
Patient-reported outcomes (PRO) spielen in der Dermatologie sowohl in der klinischen Forschung als auch in der Patientenversorgung eine große Rolle. Ziel dieser Arbeit ist die Validierung sowie die Ermittlung der Praktikabilität und Akzeptanz von vier in Papierform bereits validierten elektronischen PROs für Psoriasis: Dermatologischer-Lebensqualitäts-Index (DLQI), Behandlungsnutzen (PBI-S), Self-administered Psoriasis Area and Severity Index (SA-PASI), Juckreiz Numerische Rating Skala (NRS).
Methode / Method
Eine prospektive, randomisierte Studie im Cross-over-Design wurde durchgeführt: Erhebung der PROs jeweils auf Papier und Tablet. Soziodemographische und medizinische Parameter, das Nutzungsverhalten digitaler Medien im Allgemeinen, der EQ-5D zur Erfassung der Lebensqualität, die Praktikabilität und Akzeptanz der elektronischen PROs wurden erhoben. Die statistische Auswertung fand mit SPSS statt.
Ergebnisse / Results
102 Patienten nahmen teil, im Durchschnitt 48,04 Jahre alt, 69,6% waren männlich. 95,9% nutzten täglich das Smartphone und 40,5% das Tablet. Alle elektronischen PROs erwiesen sich als äquivalent zur Papierversion. Die Bestimmung der betroffenen Körperoberfläche mit dem VAS unterscheidet sich jedoch zwischen Tablet und Papier (ICC: 0,78). Die Praktikabilität wurde für die elektronischen PROs höher eingestuft als für die Papierversionen, z.B. die Lesbarkeit des (e)DLQI (ø= 0,7 vs. 0,5 auf einer Skala 0-5; p= 0.03). Hier schnitt auch lediglich der elektronische SA-PASI schlechter ab als die Papierversion (ø= 0,88 vs. 1,15; p= 0.037). 87,9% präferierten die elektronische Erfassung von PROs über die Erfassung auf Papier.
Zusammenfassung / Conclusion
Die Ergebnisse zeigen, dass elektronische PROs im Bereich der Psoriasis valide und praktikabel sind. Daher sollte die Verwendung elektronischer PROs in Studien und im klinischen Alltag vorangetrieben werden, um die Erfassung, Speicherung und Nutzung von Daten zu erleichtern.
AutorInnen
Maike Wagner
Matthias Augustin
What matters when interpreting patient-reported outcomes? - Adjusted minimal clinically important differences for hip and knee replacement patients
Irene Salvi, University of St. Gallen
Einleitung / Introduction
One main goal of hip and knee replacements is to improve patients’ health-related quality of life (HRQoL). Thus, to properly assess treatment quality it is crucial to understand which hip and knee replacements are successful in terms of changes in patient-reported HRQoL. A meaningful evaluation of treatment success can be achieved by calculating minimal clinically important differences (MCIDs) for changes in patient-reported outcome measures (PROMs). Furthermore, evidence shows that considering patient characteristics is essential for calculating accurate MCIDs. This study aims at providing MCID estimates adjusted for patient characteristics for EQ-5D-3L index score improvements after hip and knee re-placements at 6-months follow-up. It adds to existing literature by relying on a large national sample and a precise clustering algorithm.
Methode / Method
We conducted a retrospective observational study using the publicly available National Health Service (NHS) PROMs dataset for primary hip and knee replacements. The dataset contains patients’ demographic, clinical, and PROM information on 252,331 hip replacements and 279,668 knee replacements, from all NHS-funded providers in England between 2013 and 2020. To identify the best pre-operative predictor for EQ-5D-3L changes, multi-variate linear regression was run. Subsequently, based on regression results, clusters of patients were created. Finally, unstratified and stratified MCIDs for EQ-5D-3L improvements were estimated using anchor-based receiver operator curves.
Ergebnisse / Results
The best pre-operative predictor for EQ-5D-3L index score improvement at 6-months follow-up was the pre-operative EQ-5D-3L index score. Stratifying patients based on this factor led to increased precision in the MCID calculations. It further showed that patients with worse pre-operative scores have larger MCIDs and, correspondingly, patients with better pre-operative scores have smaller MCIDs. On the contrary, MCIDs were equally large for both genders. MCIDs calculated for knee replacement patients were consistently lower than those calculated for hip replacement patients.
Zusammenfassung / Conclusion
Our research shows that when interpreting PROM results, MCIDs should be used and adjusted for patients’ pre-operative EQ-5D-3L index score to accurately assess the success of hip and knee replacements. Failing to do so would lead to a misclassification of surgery success. MCIDs can on the other hand be accurately calculated independently of gender. Our findings should be considered by health policy makers when using PROMs for quality-related payments and benchmarking initiatives, and by patients and providers to avoid making sub-optimal hospital or treatment choices.
AutorInnen
Irene Salvi, University of St. Gallen
David Kuklinski, University of St. Gallen
Justus Vogel, University of St. Gallen
Anja Bischof, University of St. Gallen
Alexander Geissler, University of St. Gallen
Results of a randomized-controlled trial using patient-reported outcome-based alerts to optimize recovery in routine orthopedic care
Viktoria Steinbeck, Technische Universität Berlin
Einleitung / Introduction
Patient reported outcome measures (PROMs) are increasingly used for research and health technology assessment; however, their clinical value in routine care and patient monitoring is still under investigation. At the same time, digital technologies make it easier to collect PROMs and integrate them into care analytics in real time. As PROMs usage is expanding, determining their ability to not just capture outcome quality, but potentially improve outcomes by enhancing early adverse event detection is important. Whereas critical recovery detection via PROMs has shown promising results, there is no evidence from the field of orthopedics yet.
Methode / Method
A two-arm double-blinded patient-level randomized control trial was conducted across nine German orthopedic hospital departments to assess a PROM-based alert intervention among hip and knee replacement patients. Based on pre-specified PROMs thresholds, an automated digital alert was used in the intervention group to signal critical recovery paths to hospital study nurse at one-, three- and six-months post-surgery. These then contacted patients and referred them to physicians if deemed necessary. An intention to treat analysis (ITT) was run separately for hip and knee replacement patients utilizing the PROM-score improvement (EQ-5D-5L, EQ-VAS, HOOS-PS, KOOS-PS, PROMIS-fatigue and depression) from baseline to 12 months post-surgery as primary outcome. The t-test was used to asses for significant differences between intervention and control group. Subgroup analyses based on pre-surgery PROM clusters, gender and education were run as well as a mixed random effects model and a compliance corrected analysis as sensitivity analyses.
Ergebnisse / Results
The study covers 3,697 hip replacement patients (1854 in the intervention group and 1843 in the control group) and 3,110 knee replacement patients (1564 in the intervention group and 1546 in the control group). Baseline variables were well balanced between study groups. In the ITT analysis all PROM-improvements were significantly higher in the intervention group among hip patients and the KOOS and PROMIs-fatigue among knee patients. The only difference in the compliance corrected analysis was that the EQ-5D improvement among hip patients was no longer significant. Subgroups with the worst pre-surgery PROM-scores (HOOS-PS, PROMIS-fatigue and PROMIs depression) and females accounted for the main differences between intervention and control groups.
Zusammenfassung / Conclusion
The PROM-based alert intervention is effective in creating larger health improvements among hip patients, whereas it is effective in leading to larger improvements in two health domains among knee patients. Specifically relevant target groups for this intervention could be females and those with worse pre-surgery PROM-scores.