Acute Otitis Media Severity of Symptom Scale (AOM-SOS) Development and Validation

Nader Shaikh, MD Alejandro Hoberman, MD Jack Paradise, MD Howard Rockette, PhD* General Academic Pediatrics Children’s Hospital of Pittsburgh *Graduate School of Public Health

Why measure symptoms?

Symptoms are important  Bring children to medical attention  AOM treated to improve symptoms Need for symptom measurement in AOM trials  Current research limited due to lack of validated symptom measurement strategy  Use of surrogate outcomes (bacteriologic eradication) problematic

Potential uses

 Comparison of treatment modalities   Natural history of symptoms in AOM Relationship between bacteriology and symptoms  Relationship between otoscopy and symptoms

Conceptual model

Middle ear infection Physiologic changes Symptoms Activity limitation Quality of life Intended scope of the AOM-SOS

Scale Development

  List of 28 symptoms Reduction using triangulation Literature review Expert opinion • • Parent interview (n=33) Which of 28 symptoms present How much symptom affects child

Validation study # 1

Objective  Establish the reliability, validity and responsiveness of SOS by comparing it with otoscopy    Methods  Cohort of children 6-24 mo with/without AOM (n = 327) 2003 and 2004 respiratory seasons Followed for one season (~3 visits/child, 949 visits) At each visit Examined by validated otoscopist SOS completed

Internal Reliability

Definition  Are items in scale measuring the same concept?

Methods    Correlation of items with each other Assessed by Cronbach’s α Cronbach’s α >0.7 indicates good reliability Results  α = 0.83  Excellent inter-item correlation given short scale and heterogeneous population

Content Validity

Definition  Does each item measure what it is supposed to?

 Are items associated with AOM (or URI)?

Methods  Examined association between items and otoscopic diagnosis adjusting for URI

Content validity

Ear pain Tugging Irritability/crying Difficulty sleeping Eating less Less playful AOM vs. No AOM

p < .001

p < .001

p < .001

p < .001

p < .001

p < .001

AOM vs. OME

p < .001

p < .001

p < .001

p < .001

p < .004

p = .01

OME vs. Normal

p = .14

p = .02

p =.049

p = .09

p = .08

p = .04

*All p values adjusted for URI

Construct Validity

Does the scale measure what it is supposed to?

Otoscopic diagnosis Mean AOM-SOS score AOM N = 211

4.4

OME N = 154

1.6

NORMAL N = 584

0.9

Standard Deviation

4.6

2.8

1.9

AOM vs. Normal AOM vs. OME OME vs. Normal p < .001 p < .001 p < .03

Responsiveness

Definition  Can the scale detect change?

Methods  Examined changes in score in children seen twice in 3-week period   Responsiveness measured by standardized response mean (SRM) SRM > 0.5  good responsiveness

Responsiveness – Change in score within 3-week period Change P value SRM Unchanged No AOM



No AOM N = 24 Mean Std AOM



AOM N = 38 Mean Std

-0.1

.57

4.8

0 -1.8

.08

0.28

6.4

Changed No AOM



AOM N = 23 Mean Std AOM



No AOM N = 108 Mean Std

+4.9

<.001

4.9

1.00

-3.4

<.001

0.69

4.9

Study # 2 – Further Validation

Needed to further evaluate:   Day-to-day responsiveness Construct validity     Design   Cohort study 3 mo to 3 yr with AOM treated with antibiotics Otoscopic exam on days 1 and 5 SOS administered days 1 through 5 (q 12 hrs) Reference measures also administered 56 children enrolled

Construct validity Correlation of SOS with reference measures on day 1

Reference Measures Pain Visual analog scale Post operative pain scale † Functional status IIR ¥

†

Chambers ¥ Stein AOM-SOS r = .71

r = .83

r = .88

p value < .05

< .05

< .05

Responsiveness – Change in score

AOM-SOS score 7 6 5 4 1 0 3 2 Day 1 Day 2 Day 3 Day 4 Day 5 Standardized response mean (day 1 to 5) = 0.76

Conclusions

 Developed scale for measurement of AOM symptoms  Demonstrated: Good measurement properties Correlates with otoscopy Correlates with validated measures of pain and functional status  Anticipated use in NIH funded RCT to evaluate the efficacy of antibiotics versus placebo in young children with AOM

Acknowledgements

Consultants  Gordon Guyatt, MD – McMaster U.  Galen Switzer, PhD – U. of Pittsburgh Study Team  Diana Kearney, Study Coordinator  Kathleen Colborn, Data Manager  Lisa Zoffel, CRNP  Stephanie Konieczka, RN