International Perthes Study Group October 6 7 2017
International Perthes Study Group October 6 -7, 2017 Patient Reported Outcomes Measurement Information System (PROMIS) for Patients with Perthes Disease Hiroko Matsumoto, Ph. Dc Joshua E. Hyman, MD Columbia University Medical Center
2 What is PROMIS?
What is PROMIS? PROMIS Patient-Reported Outcomes Measurement Information System § NIH-funded psychometric evaluation that has grown significantly over last several years: (http: //www. hhs. gov/asl/testify/t 051208 a. html) – $90 -million since 2004. § PROMIS measures the following metrics: (NIHPromis. org) – Patient-reported outcomes: § Subjective, e. g. how a patient perceives about their symptoms and function (http: //www. nihpromis. org/measures/domainframework 1)
What PROMIS? PROMIS is Mission Statement A comprehensive instrument to assess pain and pain behaviors § “PROMIS creates an opportunity for clinicians and patients to develop a common language around self-reported health status and in the process allow better assessment of the unique information from patients thereby enabling improved quality of care and quality of clinical research. ” (NIHPromis. org)
What is PROMIS? Patient-Reported Outcomes Measurement Information System § Universal domains – Domain = Specific areas of importance regarding health and wellbeing – Cuts across different diseases § E. g. , Sleep disturbance, depressive symptoms, ability to participate in usual role, global health § Generic measures versus disease specific measures Source: Health. Measures
6 PROMIS Pediatric Self and Proxy: 21 Domains * *Profile domains: measuring the most important PROMIS concepts Source: Health. Measures
What is PROMIS? Patient-Reported Outcomes Measurement Information System § Item banks – Collection of items all measuring the same thing – Can be administered in different ways (CAT, short forms) always producing a score on the same metric Item Bank Short Form CAT Computer Adaptive Test Source: Health. Measures
Static versus Dynamic Short From § Static: Questions to follow do not change based on patient responses. § Questions presented in same sequence, regardless of prior answers.
PROMIS Example of Short Form: Pain Interference
Static versus Dynamic § Dynamic option: Questions are presented in changing sequence, with following sequence based on prior answers. § Purpose: – Minimizes # questions; – Greater measurement precision; § Presented as Computer Adaptive Test (CAT). CAT Computer Adaptive Test
PROMIS Example of CAT Form: Pain Interference It was hard to stay standing when I had pain
Longitudinal tracking of HRQo. L is a strength of PROMIS T-score allows adult and pediatric scores to be compared directly Longitudinal studies Mean: 50 80 e. g. Raw score of (Peds pain interference) 65 40 60 e. g. Raw score of (Adult pain interference) T-score of: 60 68% of population T-score of 60 = 15. 8% of population at your age and gender perceive to have higher HRQo. L than you
PROMIS § Psychometrically sound § Brief § Applicable in a variety of settings § Applicable across groups § Cover the full range of the domain § Individual domain stands alone § Available for use across ages § Score on one metric Source: Health. Measures
14 Data Collection by REDCap
15 PROMIS Validity Study in Perthes Disease Funded by POSNA
16 Methods Study Design and setting: • Prospective Validation Study • 13 institutions from International Perthes Study Group Inclusion Criteria: • • Diagnosis of Perthes Ages 8 -17 Exclusion Criteria: • Surgical treatment within 6 months prior to enrollment
17 What is Validity? • The degree to which any measurement approach or instrument succeeds in describing or quantifying what it is designed to measure
18 Assessing Validities Types Translational Validity Methods Face Validity Content Validity Criterion Validity Predictive Validity Concurrent Validity Postdictive Validity Construct Validity Internal Construct Validity Convergent Validity (Convergent Evidence) Discriminant Validity (Discriminant Evidence) Factorial Evidence (observed variables vs. latent construct) External Construct Validity Nomological Validity
19 Construct Validity • Is the measure consistent with theoretical concept being measured? – All tests of validity ultimately designed to support/refute the instrument’s construct validity
20 Convergence and Discriminant Validities • Convergent Evidence – Demonstrates that your measure correlates highly with measures of the same construct – Groups known to differ along construct have different scores on measure • Discriminant Evidence – No to low correlation with measures of different constructs
21 PROMIS Pediatric Self and Proxy: 21 Domains * *Profile domains: measuring the most important PROMIS concepts Source: Health. Measures
22 Construct Validity: Known Group Method Waldenstrom Disease stages 1. Early: (IA – IIA) 2. Late: (IIB – IIIB) 3. Healed: (IV)
23 Demographics Characteristics N = 186 Age 10. 23 ± 3. 13 Gender (Male) 151 (82. 1%) Waldenström Stages Early (IB; IIA) 24/186 (12. 9%) Late (IIB; IIIA; IIIB) 63/186 (33. 9%) Healed (IV) 99/186 (53. 2%) Ethnicity White 101/186 (54. 3%) Black 3/186 (1. 6%) Asian 3/186 (1. 6%) Hispanic 4/186 (2. 2%) Other/Unknown 75/186 (40. 3%)
24 Physical Health Domains Within Physical Domain: Increased mobility was associated with lower pain and fatigue 100 90 80 70 60 50 40 30 20 10 0 Fatigue (r = -0. 43)* Pain (r = -0. 50)* * p<0. 05 0 Convergent Validity 20 40 60 Mobility 80 100
25 Mental Health Domains Within Mental Health Domain: Increased Depression was associated with more Anxiety, and Anger 100 90 80 70 60 50 40 30 20 10 0 Anxiety (r = 0. 68)* Anger (r = 0. 51)* * p<0. 05 0 Convergent Validity 20 40 60 Depression 80 100
26 Across Domains: Increased Mobility was associated with less Anxiety, Depression and Anger 100 Anxiety (r = -0. 45)* Mental Health Domains 90 Depression (r = -0. 43)* 80 Anger (r = -0. 38)* *p<0. 05 70 60 50 40 30 20 10 0 0 Convergent Validity 20 40 Mobility 60 80 100
27 Across Domains: Increased Pain Interference was associated with more Anxiety, Depression and Anger 100 Mental Health Domains 90 80 70 60 50 40 Anxiety (r = 0. 53)* 30 Depression (r = 0. 53)* 20 Anger (r = 0. 42)* 10 *p<0. 05 0 0 Convergent Validity 20 40 60 Pain Interference 80 100
28 Across Domains: Increased Fatigue was associated with more Anxiety, Depression and Anger 100 Mental Health Domains 90 80 70 60 50 40 Anxiety (r = 0. 65)* 30 Depression (r = 0. 71)* 20 Anger (r = 0. 46)* *p<0. 05 10 0 0 Convergent Validity 20 40 60 Fatigue 80 100
29 Across Domains: Peer Relationship have no to low associations with Physical Health 100 Fatigue (r = -0. 058) Mobility (r = 0. 065) Pain Interference (r = 0. 061) Physical Health Domains 90 80 70 60 50 40 30 20 10 0 0 Discriminant Validity 20 40 60 Peer Relationships 80 100
30 Across Domains: Peer Relationship have no to low associations with Mental Health 100 Anxiety (r = -0. 040) Mental Health Domains 90 Depression (r = -0. 063) 80 Anger (r = -0. 15)* 70 60 50 40 30 20 10 0 0 Discriminant Validity 20 40 60 Peer Relations 80 100
31 All domains had the worst scores in the Early Stage group Categories/Means Early Stage N=24 Late Stage N=63 Healed N=99 Fatigue Mobility 42. 9 40. 2 39. 9 45. 5 41. 4 48. 6 Pain Interference Anxiety Depression Anger Peer Relations 49. 1 49. 6 49. 8 51. 3 48. 0 45. 7 43. 9 43. 2 51. 7 46. 0 43. 9 45. 4 44. 3 51. 2 NB: Minimally important difference (MID): Smallest change of a PRO that is perceived to be important by the patient 2 -3 point (Yost et al. 2011)
32 Conclusions • PROMIS has construct validity in measuring Qo. L of patients with Perthes disease – Demonstrate that PROMIS domains correlates highly with measures of the same construct – Low correlation with domains measuring different construct – Groups known to differ along construct have different scores on measure • However…………
33 Discussions • When compared with U. S. average, some domain scores in patients with Perthes disease were counterintuitive: • Lower scores: Fatigue, pain interference, anxiety, depression, anger • Intuitive domain scores: • Lower score in Mobility • Lower score in peer relations in Early Stage group
34 Discussions and Next Steps • PROMIS not working? • Disease Specific Instrument • Skewed group? • More patients to recruit • Counterintuitive but reality? • Responsive testing (measure changes) • Further Known Group validity testing within Healed Group – Stratify by Stulberg classification
35 Thank You! Hiroko Matsumoto hm 2174@columbia. edu
- Slides: 35