Lack of Concordance in Symptomatic Adverse Event Reporting by Children, Clinicians, and Caregivers: Implications for Cancer Clinical Trials

Abstract

Purpose: To examine concordance in symptomatic adverse event (AE) grading using the Common Terminology Criteria for Adverse Events (CTCAE 4.0) for clinicians and its patient-reported outcome (PRO) versions for children (Ped-PRO-CTCAE) and caregivers (Ped-PRO-CTCAE [Caregiver]).

Methods: Children age 7-18 years with a first cancer diagnosis, their clinicians, and caregivers completed CTCAE-based measures before starting a treatment course (T1) and after the treatment (T2). Grades (0-3) were assigned by each reporter for 15 core AEs spanning physical and mental health. Mean grades were compared between reporters using two-sample t-tests; agreement was estimated using weighted kappa (κ) statistics. Multivariable mixed regression models were used to evaluate associations of clinical factors with AE reporting concordance. Significance was set at α = .05 (two-sided).

Results: There were 438 child-clinician-caregiver triads with complete data at either T1 or T2. For children, the mean age was 13 years (standard deviation = 3.4), 53.7% were male, 32.6% non-White, and 56.4% had leukemia/lymphoma. At T1, clinician mean AE grades were significantly lower (ie, better) than children for all AEs and remained significantly lower at T2 except for constipation, nausea, anorexia, neuropathy, and anxiety. Caregiver mean AE grades were similar to children at T1 and significantly higher (ie, worse) at T2 for nausea, vomiting, anorexia, pain, fatigue, anxiety, and depression. Agreement for child-clinician grading was poor-to-fair at T1 (κ range, 0.08-0.34) and T2 (0.11-0.35), and for child-caregiver, was fair-to-good at T1 (0.34-0.65) and T2 (0.24-0.60). No factors were consistently associated with reporter concordance across AEs.

Conclusion: Compared with children, symptomatic AEs were consistently under-reported by clinicians with low agreement and over-reported by caregivers with low-moderate agreement. Direct reporting by children using Ped-PRO-CTCAE or similar measures should be routinely incorporated for toxicity assessment in clinical trials.

FIG 1.

Mean symptomatic AE grades by the reporter and time point. ^aBlue = values significantly lower than those of the child. Orange = values significantly higher than those of the child. ^bChild or caregiver values significantly different from those of the clinician. ^cChild values significantly different from those of the caregiver. AE, adverse event; SD, standard deviation.

FIG 1.

Mean symptomatic AE grades by the reporter and time point. ^aBlue = values significantly lower than those of the child. Orange = values significantly higher than those of the child. ^bChild or caregiver values significantly different from those of the clinician. ^cChild values significantly different from those of the caregiver. AE, adverse event; SD, standard deviation.

FIG 2.

Mean symptomatic AE grade by the reporter at post-treatment time point 2 (T2) when the AE was present by child self-report. Data points represent mean grades and 95% CIs for each of the 15 core symptomatic AEs as reported by children, clinicians, and caregivers using the Ped-PRO-CTCAE, CTCAE, and Ped-PRO-CTCAE (Caregiver), respectively. Data in this figure include only reporting triads where the child self-reported the symptomatic AE at T2. AE, adverse event; CTCAE, Common Terminology Criteria for Adverse Events; PRO, patient-reported outcome.

FIG 3.

Comparison of symptomatic AE grades by children and clinicians. Green shading represents perfect agreement between clinician and child grades along the diagonal. Blue shading below the diagonal represents that clinician grades were lower than child grades. Orange shading above the diagonal represents that clinician grades were higher than child grades. Values within cells represent the number of child-clinician dyads who reported the indicated combination of symptomatic AE grades at T2. AE, adverse event.

FIG 3.

Comparison of symptomatic AE grades by children and clinicians. Green shading represents perfect agreement between clinician and child grades along the diagonal. Blue shading below the diagonal represents that clinician grades were lower than child grades. Orange shading above the diagonal represents that clinician grades were higher than child grades. Values within cells represent the number of child-clinician dyads who reported the indicated combination of symptomatic AE grades at T2. AE, adverse event.

FIG 4.

Agreement in symptomatic AE grades by the reporting pair and time point. ^aWeighted kappa criteria for agreement: κ < 0.20 = poor (blue), 0.20 ≤ κ < 0.40 = fair (green), 0.40 ≤ κ < 0.60 = moderate (yellow), 0.60 ≤ κ < 0.80 = good (orange), and κ ≥ 0.80 = very good (none). AE, adverse event; T1, pretreatment; T2, post-treatment.

FIG 5.

Weighted kappa agreement in symptomatic AE grades by reporting dyads at post-treatment time point 2 (T2). Data points represent weighted kappa values and 95% CIs, indicating agreement in grading of the 15 core symptomatic AEs at T2 by children, clinicians, and caregivers using the Ped-PRO-CTCAE, CTCAE, and Ped-PRO-CTCAE (Caregiver), respectively. Data in this figure include all reporting triads at T2. AE, adverse event; CTCAE, Common Terminology Criteria for Adverse Events; PRO, patient-reported outcome.

FIG A1.

Percentage of children reporting symptomatic AE grades higher than the clinician report. For each symptomatic AE, bars indicate the percentage of children who reported an AE grade higher than what was reported by the clinician. Blue bars indicate this percentage among children regardless of clinician grade assigned; red bars indicate this percentage limited to children whose clinician assigned a grade of 0 (ie, no symptom). Numbers above bars denote the number of children in that group (data for proportions shown in Table A1). AE, adverse event.