Katherine B Bevans1, Isaac L Ahuvia2, Taye M Hallock3, Rochelle Mendonca4, Stephanie Roth3, Christopher B Forrest5,6, Courtney Blackwell2, Jessica Kramer7, Lauren Wakschlag2. 1. College of Public Health, Temple University, 1913 N Broad Street, Philadelphia, PA, 19122-6092, USA. katherine.bevans@temple.edu. 2. Northwestern University, Chicago, USA. 3. College of Public Health, Temple University, 1913 N Broad Street, Philadelphia, PA, 19122-6092, USA. 4. Columbia University, New York, USA. 5. Children's Hospital of Philadelphia, Philadelphia, USA. 6. University of Pennsylvania Perelman School of Medicine, Philadelphia, USA. 7. University of Florida, Gainesville, USA.
Abstract
PURPOSE: To identify and evaluate methods for assessing pediatric patient-reported outcome (PRO) data quality at the individual level. METHODS: We conducted a systematic literature review to identify methods for detecting invalid responses to PRO measures. Eight data quality indicators were applied to child-report data collected from 1780 children ages 8-11 years. We grouped children with similar data quality patterns and tested for between-group differences in factors hypothesized to influence self-report capacity. RESULTS: We identified 126 articles that described 494 instances in which special measures or statistical techniques were applied to evaluate data quality at the individual level. We identified 22 data quality indicator subtypes: 9 direct methods (require administration of special items) and 13 archival techniques (statistical procedures applied to PRO data post hoc). Application of archival techniques to child-report PRO data revealed 3 distinct patterns (or classes) of the data quality indicators. Compared to class 1 (56%), classes 2 (36%) and 3 (8%) had greater variation in their PRO item responses. Three archival indicators were especially useful for differentiating plausible item response variation (class 2) from statistically unlikely response patterns (class 3). Neurodevelopmental conditions, which are associated with a range of cognitive processing challenges, were more common among children in class 3. CONCLUSION: A multi-indicator approach is needed to identify invalid PRO responses. Once identified, assessment environments and measurement tools should be adapted to best support these individuals' self-report capacity. Individual-level data quality indicators can be used to gauge the effectiveness of these accommodations.
PURPOSE: To identify and evaluate methods for assessing pediatric patient-reported outcome (PRO) data quality at the individual level. METHODS: We conducted a systematic literature review to identify methods for detecting invalid responses to PRO measures. Eight data quality indicators were applied to child-report data collected from 1780 children ages 8-11 years. We grouped children with similar data quality patterns and tested for between-group differences in factors hypothesized to influence self-report capacity. RESULTS: We identified 126 articles that described 494 instances in which special measures or statistical techniques were applied to evaluate data quality at the individual level. We identified 22 data quality indicator subtypes: 9 direct methods (require administration of special items) and 13 archival techniques (statistical procedures applied to PRO data post hoc). Application of archival techniques to child-report PRO data revealed 3 distinct patterns (or classes) of the data quality indicators. Compared to class 1 (56%), classes 2 (36%) and 3 (8%) had greater variation in their PRO item responses. Three archival indicators were especially useful for differentiating plausible item response variation (class 2) from statistically unlikely response patterns (class 3). Neurodevelopmental conditions, which are associated with a range of cognitive processing challenges, were more common among children in class 3. CONCLUSION: A multi-indicator approach is needed to identify invalid PRO responses. Once identified, assessment environments and measurement tools should be adapted to best support these individuals' self-report capacity. Individual-level data quality indicators can be used to gauge the effectiveness of these accommodations.
Entities:
Keywords:
Data quality; Patient-reported outcome measures; Pediatric; Self-report capacity
Authors: Kathryn Eilene Lasch; Patrick Marquis; Marc Vigneux; Linda Abetz; Benoit Arnould; Martha Bayliss; Bruce Crawford; Kathleen Rosa Journal: Qual Life Res Date: 2010-05-30 Impact factor: 4.147
Authors: L Coombes; K Bristowe; C Ellis-Smith; J Aworinde; L K Fraser; J Downing; M Bluebond-Langner; L Chambers; F E M Murtagh; R Harding Journal: Qual Life Res Date: 2021-03-18 Impact factor: 4.147
Authors: Courtney K Blackwell; Jin-Shei Lai; Michael Kallen; Katherine B Bevans; Matthew M Davis; Lauren S Wakschlag; David Cella Journal: J Pediatr Psychol Date: 2022-05-13
Authors: Steven Hirschfeld; Elizabeth Goodman; Shari Barkin; Elaine Faustman; Neal Halfon; Anne W Riley Journal: Front Pediatr Date: 2021-06-10 Impact factor: 3.418