characterizing patients with patient acceptable symptom ... · chi-square analysis (2x2 table) 2....
TRANSCRIPT
Health Measures Conference 2017
Characterizing Patients with Patient Acceptable Symptom State (PASS) after Selected Foot and Ankle Orthopedic
Procedures using Patient Reported Outcomes
Jeff Houck, PT, PhD, George Fox University
Micheal Anderson, MD, University of Rochester
A.Samuel Flemister, MD, University of Rochester
Irvine Oh, MD, University of Rochester
Benjamin DiGiovanni, MD, University of Rochester
John Ketz, MD, University of Rochester
Judy Baumhauer, MD University of Rochester
Health Measures Conference 2017
Background• Patient Outcomes
• Patient Acceptable Symptom State (PASS)• Well versus healed?
• Pubmed (77 studies) Cataracts, Arthritis (RA, OA, AS), generic musculoskeletal problems, and others
• PASS versus MCID
• Success/Failure • Patients definition of success dependent on expectations
• related however distinct
Health Measures Conference 2017
Generic PRO’s(PROMIS Scales & Global Rating of Normal Function)
• PROMIS• T-score is abstract• MCID versus PASS • Benchmarks that are useful
• Global Rating of Normal Function (GRNF)• Used widely in psychology • Common question on legacy instruments (e.g. FAAM)
Health Measures Conference 2017
Purpose/Objectives
Determine
1. …association of patient outcomes PASS and Success
2. …ability of generic PRO’s (PROMIS and GRNF) to discriminate between positive and negative patient outcomes (aggregate)
3. …ability of generic PRO’s to predict positive and negative patient outcomes using preoperative assessment (PROMIS only) and at follow up
4. …proportion patients identified by accurate thresholds for positive/negative outcomes.
Health Measures Conference 2017
PARAMETERS NUMBER MEAN SD RANGEAGE 90 53.0 16.5 15-80GENDER
FEMALEMALE
49/90 (54.4%)41/90 (45.6%)
FOLLOW-UP (MONTHS) 90 17.0 5.9 7.1-28.8
PROCEDURECPT 28899- subtalar arthrodesis
CPT 28296- bunion correction
CPT 27698- reconstruction ankle ligaments
CPT 27691- deep tendon transfer
17/90 (18.9%)15/90 (16.7%)29/90 (32.2%)29/90 (32.2%)
OUTCOMESPASS (YES %)SUCCESSFULGRNFPROMIS PFPROMIS PIPROMIS D
66/90 (73.3%)75/90 (83.0%)90909090
7.0 2.346.4 9.551.3 10.145.1 9.4
0-1027.2-73.338.7-74.134.2-69.5
Sample
90 Patients reached by phone (7.1-28.8 months post surgery)
4/10 most common F&A surgeries
Pre-operative PROMIS data available for 64 patients
Health Measures Conference 2017
Patient OutcomesConcept Dichotomous Question
Success/Failure Which of the following applies to you?
_____ I consider my surgery a success
_____ I consider my surgery a failure
Patient Acceptable
Symptom State
(PASS)
Taking into account all the activity you have during your daily life, your
level of pain, and also your functional impairment, do you consider that
the current state of your foot and ankle is satisfactory?
_____ Yes
_____ N0
Health Measures Conference 2017
Generic Patient Reported Outcomes (PRO’s)
PRO Rating/Instrument
Global Rating of
Normal Function
(GRNF)
How would you rate the function of your foot and ankle on a scale of 0-10 with 10 being normal, excellent function and 0 being the inability to perform any of your usual daily activities which may include sports?
Patient Reported
Outcome
Instrumentation
System (PROMIS)
Pain Interference CAT
Physical Function CAT
Depression CAT
Health Measures Conference 2017
Analyses • Determine
1. …association of patient outcomes PASS and Success
Chi-square analysis (2x2 table)
2. …ability of generic PRO’s (PROMIS and GRNF) to discriminate between positive and negative patient outcomes (aggregate)
Two Way ANOVA (PROMIS Scales) & Mann-Whitney U test (GRNF)
3. …ability of generic PRO’s to predict positive and negative patient outcomes using preoperative assessment (PROMIS only) and at follow up
ROC analysis, thresholds nearest 95 % sensitivity/specificity
4. …proportion of patients identified by accurate thresholds for positive/negative outcomes for each generic measure.
Chi-square analysis 3x2 tables (predicted Positive outcome, negative outcome, and ambiguous)
Health Measures Conference 2017
1. …association of patient outcomes PASS and Success
Success Failure Total Chi Square, p value
PASS
No 11 (12.2%) 13 (14.4%) 26.7%
Yes 64 (71.1%) 2 (2.2 %) 73.3% 33.1, p < 0.001
Health Measures Conference 2017
2. ...discriminate between positive/negative patient outcomes
0
10
20
30
40
50
60
70
PROMISPF PROMISPI PROMISDepression
t-score
PROMISScales
AverageFollowUpAveragePROMISScoresforPASSYes/No
PASSNO
PASSYES
0
10
20
30
40
50
60
70
PROMISPF PROMISPI PROMISDepressiont-score
PROMISScales
AverageFollowUpAveragePROMISScoresforSuccessYes/No
SuccessNO
SuccessYES
9.7
-8.0
-7.610.6
-9.9
-7.3
Health Measures Conference 2017
3. …to predict positive/negative patient outcomes
0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
1
0 0.2 0.4 0.6 0.8 1
Sensitivity
1-Specificity
RecieverOperatorCurvesofPre-OperativeandFollowUpPROMISScoresPredictingPASS
PFFollowUp
PIFollowUp
DepressionFollowUp
PFPre-Operative
PIPre-Operative
DepressionPre-Operative
0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
1
0 0.2 0.4 0.6 0.8 1
Sensitivity
1-Specificity
RecieverOperatorCurvesforPre-OperativeandFollowUpPROMISScoresPredicting"Success"
PFFollowUp
PIFollowUp
DepressionFollowUp
PFPre-Operative
PIPre-Operative
DepressionPre-Operative
Health Measures Conference 2017
Thresholds for Positive/Negative Patient Outcomes
PARAMETER PASS YES PASS NO SENSITIVITY SPECIFICITY
GRNF >7 <5 97.1 % 93.8 %
PF >50.0 <33.5 95.7% 95.8 %
PI <46.9 >65.6 95.7% 95.8 %
DEPRESSION <44.0 >60.3 95.6 % 91.3 %
PARAMETER SUCCESS FAILURE SENSITIVITY SPECIFICITY
GRNF >6* <5 94.7 % 73.3 %*
PF >45.3 <33.2 96.0% 92.3 %
PI <47.3 >64.5 94.7 % 92.3 %
DEPRESSION <45.8 >60.3 95.6 % 85.7 %
*note using >7 as the cut off yields a specificity of 100%
Health Measures Conference 2017
4. …proportion of patients identified
PROMIS PI SUCCESS FAILURE TOTAL CHI
SQUARE,
P VALUE
>64.5 FAILURE
47.3-64.5AMBIGUOUS
<47.3 SUCCESS
4 (4.4%)
39 (43.3%)
32 (35.6%)
6 (6.7 %)
8 (8.9%)
1 (1.1%)
11.1%
52.2%
36.7%
17.9, p< 0.001
GRNF SUCCESS FAILURE TOTAL CHI
SQUARE, P
VALUE
<5 FAILURE
5-7 AMBIGUOUS
>7 SUCCESS
4 (4.4%)
14 (15.6%)
57 (63.3%)
8 (8.9%)
3 (3.3%)
4 (4.4%)
13.3%
18.9%
67.8%
26.1, p < 0.001
All proportions (categorizations) based on generic PRO’s were significant 43-81.1% of patients were accurately identified (95% Sn/Sp)Positive outcomes (PASS yes or Success) were more successfully predicted than negative outcomes
Health Measures Conference 2017
Summary
• The PASS and Success definitions of patient outcome are similar but not the same
• Generic PRO’s assessed discriminate groups based on patient outcome
• PROMIS and GRNF were predictive of current patient status (Follow Up), however, pre-operative scores did NOT predict Follow Up data.
• High proportions (43-81%) were identified by ONLY taking into account PRO assessments.
Health Measures Conference 2017
Notable Limitations
• Sample was too small to derive and validate thresholds
• Number of surgical failures was small – so models are likely overfit
• Specific diagnoses may give different results
• Representative sample (contacted versus not-contacted patients – same mean age and % female) ?
Health Measures Conference 2017
Questions