A B S T R A C T PurposeParents of children with incurable cancer make complex and difficult decisions about remaining treatment options. We compared the self-reported rationale, good parent definition, and desired clinical staff behaviors of parents who recently decided for phase I (P1) chemotherapy with parents who chose a do not resuscitate (DNR) or terminal care (TC) option. Patients and MethodsSixty-two parents of 58 children were asked for the basis of their decision, their definition of a good parent, and what staff behaviors supported their good parent role. After semantic content analysis, results were compared in the P1 versus DNR/TC groups. These categories were mutually exclusive but did not necessarily represent an either/or decision. ResultsThirty-one decisions were for P1 chemotherapy and 27 for DNR/TC. Median survival time after study enrollment was greater in the P1 group (0.4 v 0.1 years). Most P1 group parents reported having felt compelled to continue cancer-directed therapy (71% v 7%), whereas those who opted for DNR/TC cited quality of life (QOL; 74% v 3%) and patient wishes (67% v 13%). Decision factors common to both groups were medical facts, doing right, and others' opinions. Both groups believed that a good parent did right, provided support and presence, and sacrificed for the child. The groups desired similar support from clinicians and expressed gratitude. ConclusionDespite similar definitions of a good parent and desired staff behaviors, parents in the P1 group reported having felt compelled to continue cancer-directed therapy, whereas QOL and patient wishes were emphasized in decisions for DNR/TC.

IMPORTANCE Adult patients are considered the best reporters of their own health-related quality of life (HRQOL). Self-report in pediatrics has been challenged by a limited array of valid measures. Caregiver report is therefore often used as a proxy for child report.OBJECTIVES To examine the degree of alignment between child and caregiver proxy report for Patient-Reported Outcomes Measurement Information System (PROMIS) HRQOL domains among children with cancer and to identify factors associated with better child and caregiver-proxy congruence. DESIGN, SETTING, AND PARTICIPANTSIn this multicenter cohort study, children with a first cancer diagnosis and their caregivers completed surveys at 2 time points: within 72 hours preceding treatment initiation (T1) and during follow-up (T2), when symptom burden was expected to be higher (eg, 7-17 days later for chemotherapy). Data were collected from October 26, 2016, to October 5, 2018, at 9 pediatric oncology hospitals. Five hundred eighty children (aged 7-18 years) and their caregivers were approached; 482 child-caregiver dyads completed surveys at T1 (response rate 83%), and 403 completed surveys at T2 (84% of T1 participants). Data were analyzed from July 1, 2019, to April 22, 2020.EXPOSURES Participants received up-front cancer treatment, including chemotherapy and radiotherapy. MAIN OUTCOMES AND MEASURESCongruence between child self-report and caregiver-proxy report of PROMIS pediatric domains of mobility (physical functioning), pain interference, fatigue, depressive symptoms, anxiety, and psychological stress. RESULTSOf the 482 dyads included in the analysis, 262 children (54%) were male (mean [SD] age, 12.9 [3.4] years), 80 (17%) were Black, and 71 (15%) were Hispanic. Intraclass correlations between child self-report and caregiver proxy report showed moderate agreement for mobility (0.57 [95% CI, 0.50-0.63]) and poor agreement for symptoms (range, 0.32 [95% CI, 0.24-0.41] for fatigue to 0.42 [95% CI, 0.34-0.50] for psychological stress). Children reported lower symptom burden and higher mobility than caregivers reported. In a multivariable model adjusted for child and parent sociodemographic factors and the caregiver's own self-reported HRQOL, caregivers reported the child's mobility score 6.00 points worse than the child's self-report at T2 (95% CI, −7.45 to −4.51), exceeding the PROMIS minimally important difference of 3 points. Caregivers overestimated the child's self-reported symptom levels, ranging from 5.79 (95% CI, 3.99-7.60) points for psychological stress to 13.69 (95% CI,) points for fatigue. The caregiver's own self-reported HRQOL was associated with the magnitude of difference between child and caregiver scores for all domains except mobility; for example, for fatigue, the magnitude of difference between child and caregiver-proxy scores increased by 0.21 (95% CI, 0.13-0.30) points for each 1-point increase in the caregiver's own fatigue score.CONCLUSIONS AND RELEVANCE This study found that caregivers consistently overestimated symptoms and underestimat...