Can Patient-Reported Outcomes Measurement Information System® (PROMIS) measures accurately enhance understanding of acceptable symptoms and functioning in primary care?

Abstract

Background: Value-based healthcare models will require prioritization of the patient's voice in their own care toward better outcomes. The Patient-Reported Outcomes Measurement Information System® (PROMIS) gives patients a voice and leads providers to actionable treatments across a broad range of diagnoses. However, better interpretation of PROMIS measures is needed. The purpose of this study was to evaluate the accuracy of PROMIS Physical Function (PF), Self-Efficacy for Managing Symptoms (SE), Pain Interference (PI), Fatigue, and Depression measures to discriminate patient acceptable symptom state (PASS) in primary care, determining if that accuracy is stable over time and/or retained when PROMIS score thresholds are set at either ½ or 1 SD worse than the reference population mean. Methods: Primary care patients completed the five PROMIS measures and answered the PASS yes/no question at intake (n = 360), 3-14 days follow-up (n = 230), and 45-60 days follow-up (n = 227). Thresholds (optimal, ½ SD, and 1 SD worse than reference values) for PROMIS T-scores associated with PASS were determined through receiveroperator curve analysis. Accuracy was calculated at the three time points for each threshold value. Logistic regression analyses were used to determine combinations of PROMIS measures that best predicted PASS. Results: PROMIS PF, SE, PI, and Fatigue optimal score thresholds (maximizing sensitivity and specificity) yielded area under the curve values of 0.77-0.85, with accuracies ranging from 71.7% to 79.1%. Accuracy increased minimally (1.9% to 5.5%) from intake to follow-ups. Thresholds of 1 SD worse than the mean for PROMIS PF and PI measures and ½ SD worse for SE and Fatigue overall retained accuracy versus optimal (+ 1.3% to − 3.6%). Regression models retained SE, PI, and Fatigue as independent predictors of PASS, and minimally increased accuracy to 83.1?%. Conclusions: This study establishes actionable PROMIS score thresholds that are stable over time and anchored to patient self-reported health status, increasing interpretability of PF, SE, PI, and Fatigue scores. The findings support the use of these PROMIS measures in primary care toward improving provider-patient communication, prioritizing patient concerns, and optimizing clinical decision making.

Clinimetric Criteria for Patient-Reported Outcome Measures

“…PROMs are self-reported scales and indices specifically developed to improve the detection of the patients' subjective experience [13][14][15][16][17][18][19][20][21][22][23][24][25][26][27]. Originally designed to measure treatment outcomes, PROMs have a number of applications.…”

“…Such integration appears to be particularly important in randomized controlled trials and for measuring the effects of mental care [12], where there has been growing interest in the use of rating scales in the clinical process of psychological and psychiatric assessments. The importance of patient-reported outcome measures (PROMs), self-rated subjective experiences of symptom burden and psychological well-being in relation to health conditions and/or treatments, has gained increased recognition [13][14][15][16][17][18][19][20][21][22][23][24][25][26][27].…”

Using mobile apps in social work behavioral health care service: The case for China

“…Social workers in China may use mobile apps as screening and assessment tools to identify the need for behavioral health care treatment and encourage help-seeking. A few validated mobile apps, for example, already exist in the market for screening and assessment of PTSD and trauma (Olff, 2015), depression and anxiety symptoms (Bakker and Rickard, 2018), as well as self-reported physical-, mental-, and social-health status (Jacobson et al, 2020). In addition, the portability of smartphones enables social workers to provide direct interventions and ask clients to practice coping skills in everyday settings (Intille, 2004) and to access interventions in the moment of need at any time and location, such as during high-risk or triggering situations (Luxton et al, 2011).…”

Detecting the patient’s need for help with machine learning based on expressions

“…On the other hand, comprehensive modular questionnaire systems have been proposed and implemented, for example relying on International Classification of Functioning, Health and Disability, and Patient-Reported Outcomes Measurement Information System (PROMIS) [ 7 , 8 ]. Despite the possibility to offer increasingly specifically tailored question sets and to create links between them [ 9 , 10 ], a general challenge is to interpret the gained specific answers in greater agglomerated entities to make analytic conclusions and predictions in a broader context of the person’s health and wellbeing, such as in a long-term care planning and clinical decision making [ 11 ].…”