Invited Commentary
June 26, 2024
Patient-Reported Outcome Measures in NSQIP—The PROMise Land
Catherine B. Jensen, Lesly A. Dossett, Susan C. Pitt
JAMA Surg. Published online June 26, 2024. doi:10.1001/jamasurg.2024.1758
Integrating patient-reported outcome measures (PROMs) into a well-established, trusted platform like the National Surgical Quality Improvement Program (NSQIP) holds tremendous promise for improving outcomes that truly matter to patients.1,2 Over 3 years, Temple and colleagues3 used sequential, additive implementation strategies to demonstrate the feasibility of collecting PROMs from at least 30% of the patient sample. Nearly 90% of the 65 diverse participating US hospitals achieved the goal collection rate by using text and email reminders personalized with hospital and surgeon names. Although no hospital-level factors influenced response rates, multivariable analysis of patient-level factors showed increasing age, female sex, lower American Society of Anesthesiologists class, obesity, chronic steroid use, and contact via email were associated with increased response.
The results of Temple et al3 are encouraging for wide-scale implementation of PROMs in surgical quality improvement, yet important considerations exist. First, attention to sampling strategy and nonresponse bias are crucial. For example, patient factors, including age and insurance status, were associated with nonresponse.3 The underrepresentation of patient groups, such as those with limited digital access, threatens generalizability. Ensuring a representative patient voice will require a multilevel, multifaceted approach that engages diverse populations and accounts for social determinants of health.4
To optimize national-level PROM integration, further comparative testing, packaging, and tailoring of the implementation strategies used by Temple et al3 are likely needed. Although the authors used up to 15 strategies to achieve the target response, the implementation burden and effectiveness of the individual component strategies are unknown.3 It also remains unclear if 30% response is adequate to inform quality-related metrics and policies.
Accuracy of PROM interpretation should also be considered. A one-time collection of PROMs, without a baseline assessment, limits the ability to determine if and how surgery affected the measured outcome. Variations in the timing of measurement will also influence results. For example, pain likely changes during the 30- to 90-day collection period and may be more informative for patient care before the 30-day start point.5 Understanding the clinical significance of PROM results is also key as the minimal clinically important difference may not be established for all measures and may vary across procedures or populations.6
Despite these considerations, the study by Temple et al3 provides an incredible window into the power of integrating PROMs with traditional clinical outcomes like surgical site infection. The complementary data sources are essential for driving meaningful patient-centered improvement in surgical care delivery. However, use of PROMs for benchmarking or performance evaluation should be cautiously approached—comparisons between clinicians and institutions require case-mix adjustment, which are not currently available through NSQIP.7 Although we may not have fully reached the surgical “PROMise land,” Temple and colleagues3 deserve a standing ovation for taking a colossal step forward.
