Appropriate statistical testing of quality of life scores from children with asthma and their caregivers
| Autor: | James Legleiter, Robert William Prasaad Steiner, John A Myers, Robert J. Esterhay, Yu-Ting Chen |
|---|---|
| Jazyk: | angličtina |
| Rok vydání: | 2012 |
| Předmět: |
Wilcoxon signed-rank test
business.industry media_common.quotation_subject Nonparametric statistics medicine.disease Statistical power Test (assessment) Quality of life Statistics medicine Open Access Medical Statistics business Normality Statistical hypothesis testing media_common Clinical psychology Asthma |
| Zdroj: | Open Access Medical Statistics. |
| ISSN: | 2230-3251 |
| Popis: | John A Myers1, Robert William Prasaad Steiner1, James Legleiter2, Yu-Ting Chen1, Robert J Esterhay11University of Louisville School of Public Health and Information Sciences, 2Asthma Disease Management, Passport Health Plan, Louisville, KY, USAObjective: Scores on the Mini-Pediatric Asthma Quality of Life Questionnaire (mPAQLQ) and Pediatric Asthma Caregiver’s Quality of Life Questionnaire (PACQLQ) assess disease-specific quality of life (QoL). Paired t-tests are a commonly used analytic method to examine differences between these QoL scores. This report examines whether nonparametric approaches may be more appropriate.Study design and setting: Responses on the mPAQLQ and PACQLQ were collected in a cross-sectional design with 68 child–caregiver dyads attending a community asthma prevention program – Healthy Hoops® Kentucky. Scores and differences in scores were compared using both parametric and nonparametric approaches.Results: Scores on mPAQLQ subscales and overall QoL were not normally distributed, indicating that parametric tests may not be appropriate. Results of statistically significant differences in overall mean QoL scores between caregiver and children differed with parametric methods (paired t-tests: P ≤ 0.013) and nonparametric approaches (Wilcoxon signed-rank tests: P ≤ 0.15). Achieved statistical power with nonparametric methods was 97%–99% for differences in subscale and overall QoL scores. Probability plots for the tests for normality showed skewness at extreme values.Conclusion: The currently advocated method for using parametric methods to compare scores from the mPAQLQ and PACQLQ may be incorrect. Using parametric methods with non-normal distributions can lead to errors. This study supports the nonparametric approach (Wilcoxon signed-rank test) for testing for differences between the mPAQLQ and PACQLQ. Instructions for scoring the mPAQLQ and PACQLQ may need to be revised and updated.Keywords: pediatric asthma, quality of life, Mini-Pediatric Asthma Quality of Life Questionnaire, Caregiver’s Pediatric Asthma Quality of Life Questionnaire, nonparametric, Wilcoxon signed-rank test |
| Databáze: | OpenAIRE |
| Externí odkaz: |
|