Multilevel factors associated with nonresponse to patient-reported outcome measures in routine radiation oncology care

Jason B. Liu, Yu-Jen Chen, Maria O. Edelen, Andrea L. Pusic, Neil E. Martin, Chengbo Zeng · 2026

Read Multilevel factors associated with nonresponse to patient-reported outcome measures in routine radiation oncology care by Jason B. Liu, Yu-Jen Chen, Maria O. Edelen, Andrea L. Pusic, Neil E. Martin, Chengbo Zeng (2026) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/pmc-13585823.txt.

Introduction

Jason Liu and co-authors study why cancer patients skip outcome surveys in a 2026 paper on radiation oncology care. Clinics now collect patient-reported outcome measures as routine care, but missing replies can skew what the data seem to say. The team focused on a short global health survey sent through the patient portal.

They covered 12,214 patients, 71 providers, and five clinics over one year. Only about 35 percent of patients ever completed the survey, and rates ranged nearly fivefold across clinics. Recent surgery raised the odds of nonresponse, while more time since diagnosis lowered them. Clinic habits, such as how early the program launched and past collection rates, mattered more than patient traits once all levels were modeled.

Catalog record

CREATORJason B. Liu, Yu-Jen Chen, Maria O. Edelen, Andrea L. Pusic, Neil E. Martin, Chengbo Zeng
YEAR2026
COLLECTIONPAPERS
FORMATTXT
FILE SIZE45 KB
VERSION1.0
LICENSECC BY 4.0
RIGHTSCC BY
SOURCEEurope PMC

Source: https://europepmc.org/articles/PMC13585823

Browse the full library · About the Tower of Babel

Full text

The opening of the work, from the library's hosted text:

Multilevel factors associated with nonresponse to patient-reported outcome measures in routine radiation oncology care Authors: Jason B. Liu, Yu-Jen Chen, Maria O. Edelen, Andrea L. Pusic, Neil E. Martin, Chengbo Zeng Published: 2026 ABSTRACT PurposeNonresponse to routinely collected patient-reported outcome measures (PROMs) threatens the representativeness of aggregated data. We characterized patient-, provider-, and clinic-level factors associated with PROMIS Global-10 nonresponse in routine radiation oncology care.MethodsIn this retrospective cohort study, all adults seen at five Mass General Brigham radiation oncology clinics over one year were included. The primary outcome was patient-level nonresponse, defined as never completing the portal-administered Global-10 versus completing it at least once. Using iterative mixed-effects logistic regression, we modeled patient-, provider-, and clinic-level factors.ResultsAmong 12,214 patients, 71 providers, and five clinics, patient- and appointment-level response rates were 35.4% and 31.7%, with patient-level response ranging nearly fivefold across clinics (12.8% to 66.2%). In Model 1, male sex, lower education, not working, and recent surgery had higher odds of nonresponse, and longer time since diagnosis lower odds. After provider- and clinic-level factors were added, patient sex, education, and employment became nonsignificant, whereas recent surgery (adjusted odds ratio [aOR] 1.97) and longer time since diagnosis (aOR 0.46 for > 12 months) persisted. A provider’s historical collection rate was protective but attenuated at the clinic level. There, a later program launch (aOR 0.29) and higher historical collection rate (aOR 0.79) correlated with lower nonresponse, whereas academic versus community setting did not.ConclusionsNonresponse to routinely collected PROMs is a multilevel phenomenon driven substantially by clinic-level implementation factors, not patient characteristics alone. Because response rate is only a proxy for representativeness, PROMs programs and PRO-based performance measures should prioritize representative collection over volume.Supplementary InformationThe online version contains supplementary material available at https://doi.org/10.1007/s11136-026-04383-w. Introduction Patient-reported outcome measures (PROMs) are increasingly embedded in routine clinical care to foster patient-centeredness, quality and performance improvement, and value-based care transformation. [1–5] As health systems scale PROMs collection through the electronic health record (EHR), these data are also being used to generate real-world evidence and support population-level efforts. Using PROMs data in the aggregate places a premium on data quality that extends beyond the individual patient encounter. [6]. A core threat to the utility of aggregated PROMs data is nonresponse. [7,8] When data are systematically incomplete, analyses based on observed responses may not represent the underlying patient population, leading to biased estimates that can mask or amplify disparities. [9,10] In routine clinical workflows, nonresponse can arise from patient-level barriers, such as language discordance or limited digital access, but it can also reflect clinic- and provider-level variation in how PROMs are operationalized, monitored, and incorporated into care. [8,11–13] Most efforts to address nonresponse focus on raising overall response rates (aka completion rates), but this alone does not necessarily improve data representativeness, because the strategies that increase responses may not preferentially reach the patients most likely to be missing. [14,15] Improving representativeness therefore requires first understanding where in the care delivery system nonresponse concentrates and which patients are disproportionately absent from the resulting data. Implementation science offers an organizing framework for understanding why PROMs initiatives succeed or fail in routine care and for designing strategies to address the barriers involved. [16] Prior work shows that the determinants of these initiatives span multiple domains and levels, from individual patients to providers and the broader care delivery system, [18] and emphasizes implementation outcomes such as adoption, fidelity, and sustainment. [19] Viewed through implementation science theory, missing PROMs data are less a statistical nuisance than a modifiable implementation outcome that reflects the underlying system’s design and workflow performance. Characterizing nonresponse as a multilevel phenomenon is therefore central to ensuring the representativeness of aggregated PROMs data. Mass General Brigham (MGB) radiation oncology clinics routinely collect the Patient-Reported Outcomes Measurement Information System (PROMIS®) Global-10 across all cancer types. [20] Every patient scheduled to see a physician or advanced practice provider (APP) is electronically administered the Global-10 before each visit, creating an opportunity to study nonresponse in a real-world setting. Drawing on established multilevel implementation science frameworks, [2122] We characterized patterns of Global-10 nonresponse across the clinic, provider, and patient levels. By identifying factors at each level, we sought to delineate whether nonresponse is shaped not only by patient characteristics but also by the providers and clinics responsible for collection. We posit that such data can strengthen the case for making data representativeness, rather than response volume alone, a core quality target for evidence generated from routinely collected PROMs. Methods Study design, setting, and participants All adults seen at five radiation oncology clinics within the MGB health system between October 2023 and September 2024 were included in this retrospective cohort study. Data were obtained from the EHR and other readily available institutional administrative data sources (e.g., credentialing information). This study was approved by the Dana-Farber/Harvard Cancer Center Institutional Review Board (24–225) and exempted from oversight. As part of routine care, the PROMIS Global-10 version 1.0/1.1 was electronically administered to all patients via the EHR’s patient portal seven days before any visit with a provider (i.e., physician or APP). [20,23] Patients could complete the Global-10 at any point within the 7-day window and those who did not complete the Global-10 in advance were provided electronic tablets (i.e., iPads) at the time of check-in while waiting for their appointment. Patients were required to complete all 10 Global-10 items, thus there is no item-level missingness. Per institutional policy, the Global-10 was made available in English, Spanish, Portuguese, Haitian Creole, traditional Chinese, Russian, and Arabic. [11] Responses and scores were immediately available in the EHR for provider use. Definition of nonresponse The primary outcome for this analysis was patient-level nonresponse, defined as whether the patient completed the Global-10 one or more times during the study period (respondent) versus none (nonrespondent). Because a patient could have one or more appointments during the study period and therefore could be administered the Global-10 more than once, we also calculated appointment-level response rates as secondary outcomes. Multilevel factors: clinics, providers, and patients Following published multilevel implementation frameworks, we conceptualized potential clinic, provider, and patient factors that might contribute to nonresponse using readily available data. [16,19,21,22,24] Supplemental Table 1 outlines the factors in detail along with corresponding implementation domains. Briefly, clinic-level factors included: clinic type, year in which PROMs collection was implemented, weekly patient volume, and weekly response rates. Provider-level factors included: sex, years since graduation from their terminal clinical degree, and role. Weekly patient volume and weekly responses rates were also calculated by provider. Patient-level demographic characteristics included: age, sex, race/ethnicity, highest level of education attainment, insurance type, employment status, preferred language, comorbidities based on the NCI version of the Charlson comorbidity index (CCI), [25] cancer type, months since cancer diagnosis, and whether the patient had a surgical procedure within the previous six months. Statistical analysis Clinic-, provider-, and patient-level characteristics were summarized using medians and interquartile range (IQRs) for continuous variables and frequencies and percentages for categorical variables. We calculated both patient- and appointment-level response rates for each clinic during the study period. Differences in clinic-, provider-, and patient-level characteristics between nonrespondents and respondents were assessed, and effect sizes (ES), as quantified with Cramér’s V or Cohen’s d. We considered ES of 0.2, 0.5, and 0.8

Read the complete text