/clinical/,/clinical/cckm-tools/,/clinical/cckm-tools/content/,/clinical/cckm-tools/content/questionnaires/,/clinical/cckm-tools/content/questionnaires/related/,

/clinical/cckm-tools/content/questionnaires/related/name-97193-en.cckm

201606168

page

100

UWHC,UWMF,

Clinical Hub,UW Health Clinical Tool Search,UW Health Clinical Tool Search,Questionnaires,Related

TRACK - A Caregiver-Completed Questionnaire for Preschool-Aged Children

TRACK - A Caregiver-Completed Questionnaire for Preschool-Aged Children - Clinical Hub, UW Health Clinical Tool Search, UW Health Clinical Tool Search, Questionnaires, Related


o
validated asthma control tool is available for young children
with recurrent wheezing or respiratory symptoms consistent
with asthma. Although discordance between parental and
school-aged children’s reports of symptoms has been shown,
19
the reliance on a caregiver proxy for symptom assessment is
necessary in younger patients with asthma who are not as cog-
nitively developed.
We now describe the development and validation of the Test for
Respiratory and Asthma Control in Kids (TRACK), a question-
naire for caregivers that encompasses both the risk and impairment
domains of respiratory control in preschool-aged children with
either physician-diagnosed asthma or recurrent respiratory epi-
GlaxoSmithKline, MedPoint, Novartis, Schering-Plough, Sepracor, and Merck; and
has served on the speakers’ bureau for Alcon, Aventis, Genentech, AstraZeneca,
Boehringer, GlaxoSmithKline, MedPoint, Novartis, Pfizer, Schering-Plough, Sepra-
cor, and Merck. M. Mellon has served as a consultant for AstraZeneca and has served
as a speaker for AstraZeneca and Schering-Plough. K. Lampl is employed by
AstraZeneca.
Received for publication September 29, 2008; revised January 23, 2009; accepted for
publication January 26, 2009.
Available online March 17, 2009.
Reprint requests: Kevin R. Murphy, MD, Boys Town National Research Hospital,
Allergy, Asthma & Pediatric Pulmonology, 14080 Hospital Road, Omaha, NE
68010. E-mail: murphyk@boystown.org.
0091-6749/$36.00
� 2009 American Academy of Allergy, Asthma & Immunology
Test for Respiratory and Asthm
A caregiver-completed questio
children
Kevin R. Murphy, MD,
a
Robert S. Zeiger, MD, PhD,
b
Mark K
Michael Schatz, MD,
b
Kathy Lampl, MD,
e
Jennifer T. Hanlon
San Diego and Sacramento, Calif, Lincoln, RI, and Wilmington, Del
Background: A validated questionnaire is needed to monitor
respiratory control in preschool-aged children.
Objective: We sought to develop and validate a caregiver-
completed questionnaire that measures respiratory control in
young children.
Methods: A 33-item questionnaire that included asthma
impairment and risk items was administered to 486 caregivers
of children aged younger than 5 years with a current, recent, or
past history of respiratory symptoms. Stepwise regression was
used to select a subset of items with the greatest discriminant
validity in relation to guidelines-defined asthma control in a
random two-thirds development sample. Reliability, validity,
and ability to screen for respiratory control problems were
tested in development and validation samples (remaining one-
third sample).
Results: The content of the 5 items selected, the Test for
Respiratory and Asthma Control in Kids (TRACK),
included frequency of respiratory symptoms (wheeze, cough,
shortness of breath), activity limitation, and nighttime
awakenings in the past 4 weeks; rescue medication use in the
past 3 months; and oral corticosteroid use in the previous year.
Reliability was greater than 0.70 in both samples. ANOVA
showed that mean scores differed significantly (P < .001) in the
expected direction across both samples for 3 levels of guidelines-
From
a
the Boys Town National Research Hospital, Omaha;
b
Southern California Perma-
nente Medical Group, San Diego;
c
QualityMetric Inc, Lincoln;
d
Capital Allergy and
Respiratory Disease Center, Sacramento; and
e
AstraZeneca LP, Wilmington.
Supported by AstraZeneca LP.
Disclosure of potential conflict of interest: K. R. Murphy has received consulting
honoraria from AstraZeneca, Schering-Plough, Merck, and Dey and has received
research support from AstraZeneca, GlaxoSmithKline, Merck, Schering-Plough, and
Novartis. R. S. Zeiger has served as a consultant for AstraZeneca. B. Chipps has
received research support from Aventis, Genentech, AstraZeneca, GlaxoSmithKline,
Novartis, Schering-Plough, Sepracor, and Merck; has received grants for educational
activites from Alcon, Aventis, Genentech, AstraZeneca, GlaxoSmithKline, and
Novartis; has served as an advisor for Alcon, Aventis, Genentech, AstraZeneca,
doi:10.1016/j.jaci.2009.01.058
a Control in Kids (TRACK):
nnaire for preschool-aged
sinski, MA,
c
Bradley Chipps, MD,
d
Michael Mellon, MD,
b
, MPH,
c
and Sulabha Ramachandran, PhD
e
Omaha, Neb,
based respiratory control, physician-recommended change in
therapy, and symptom status. In the development and validation
samples, screening analyses revealed areas under the receiver
operating characteristic curve of 0.88 and 0.82, respectively;
control status was correctly classified in 81% and 78% of cases.
Conclusion: TRACK is a valid, easy-to-administer, caregiver-
completed questionnaire of respiratory control in preschool-
aged children with symptoms consistent with asthma. (J Allergy
Clin Immunol 2009;123:833-39.)
Key words: Asthma, respiratory symptoms, asthma control, TRACK,
validated questionnaire, asthma guidelines, young children
Asthma in children is a leading cause of activity limitation,
hospitalizations, and medical costs,
1-7
with 80% of children aged
5 to 18 years having symptoms before 5 years of age.
8
Compared
with older children with asthma, young children have greater
risk, as evidenced by increased health care use and mortality,
1-7
and experience less favorable responses to asthma management
strategies.
9-11
The Expert Panel Report-3 (EPR-3)
12
of the Na-
tional Asthma Education and Prevention Program (NAEPP)
and the Global Initiative for Asthma
13
now emphasize 2 major
domains of asthma control: risk and impairment.
12
Asthma risk
addresses the frequency and severity of exacerbations, progress-
ive loss of lung function, and medication side effects. Asthma
impairment refers to current pulmonary function, symptom bur-
den, nighttime awakenings, activity limitations, and school/work
absences.
Determining asthma control has been facilitated with the
development of specific control assessments in children aged 4
to 11 years,
14
children aged 5 to 17 years,
15
and adolescents
and adults.
16-18
All existing control tools were developed before
the issuance of the new guidelines that recommend the inclu-
sion of a risk domain to the usually assessed impairment do-
main in the determination of asthma control. Furthermore, no
sodes consistent with asthma.
833

confirmatory factor analysis (CFA) methods were used to confirm the
J ALLERGY CLIN IMMUNOL834 MURPHY ET AL
Abbreviations used
C-ACT: Childhood Asthma Control Test
CFA: Confirmatory factor analysis
EFA: Exploratory factor analysis
EPR-3: Expert Panel Report-3
NAEPP: National Asthma Education and Prevention Program
ROC: Receiver operating characteristic
TRACK: Test for Respiratory and Asthma Control in Kids
METHODS
Draft questionnaire development
Development of the draft questionnaire was based on input from a working
group, caregiver and physician interviews, and subsequent qualitative
research (see the Methods section in this article’s Online Repository at
www.jacionline.org). Items for the draft questionnaire were based on content
generated from 2 focus groups, each consisting of 8 primary caregivers of pre-
school-aged children with recurring respiratory problems or asthma. Physical
signs of respiratory problems most often reported (>75% of respondents) were
wheeze, cough, depressed abdomen, shallow breathing, fatigue, mood
changes, susceptibility to illness, dark rings around eyes, clinginess, and sleep
disruption. Significant activity limitation (eg, missed school, interference with
usual play) was often reported as a consequence of such physical signs, and
failure of over-the-counter drugs to reduce these physical signs was the
main reason for unscheduled physician/emergency department visits. Based
on this information, a 33-item draft questionnaire was developed for further
testing in the TRACK development study described below. This draft ques-
tionnaire included 18 items on the frequency and severity of respiratory symp-
toms, 8 on the effect of these symptoms on the child’s life, and 7 on respiratory
medication and health care use (see the Methods section in this article’s Online
Repository at www.jacionline.org).
Study design and sample
A cross-sectional nonrandomized study was conducted to identify and
validate a subset of questions from the 33-item draft questionnaire to include
in the final TRACK instrument. Study participants were recruited from 11
asthma specialist sites and 6 primary care sites across the United States.
Institutional review boards for each site approved the study; all caregivers
provided written informed consent.
Caregivers of children younger than 5 years who met the study inclusion/
exclusion criteria at the time of a scheduled or unscheduled visit to the
physician were invited to participate in the study. Caregivers were eligible to
participate if they were at least 18 years old and if their young child had a
history of 2 or more episodes of wheezing, shortness of breath, or cough that
lasted more than 24 hours and either a diagnosis of asthma or receipt of
treatment for respiratory symptoms with a prescription bronchodilator.
Exclusions to participation were patient respiratory conditions not consistent
with asthma, other significant chronic disorders or congenital abnormalities,
or enrollment in a clinical trial.
Data collection
Data were obtained from eligible caregivers and physicians regarding the
child’s respiratory/asthma control problems. During the child’s visit to the
physician, caregivers completed screening questions on symptom frequency
and the 33-item draft questionnaire (see Fig E1 in this article’s Online Repos-
itory at www.jacionline.org). Answers to the screening questions were used to
stratify the sample into categories of current symptomatic (episodes of wheez-
ing, shortness of breath, or coughing in the past 4 weeks), recent past sympto-
matic (episodes of wheezing, shortness of breath, or coughing in the past year
but not within the past 4 weeks), or asymptomatic (without symptoms
for > 1 year).
Separately, physicians blinded to the caregiver questionnaire responses
completed a 15-item questionnaire; their answers were used to establish
criterion and construct validity of the items in the caregiver-completed
questionnaire. Five specific items that were based on the risk and impairment
factors from the NAEPP EPR-3’s asthma control table for 0- to 4-year-old
children
12
assessed the frequency of wheeze, nighttime awakenings from res-
piratory symptoms, activity limitation caused by respiratory symptoms, use of
rescue medications for respiratory symptom flares, and oral corticosteroid use
in the past year (see Table E1 in this article’s Online Repository at
www.jacionline.org). Based on guidelines-defined asthma control (impair-
ment and risk), responses to each item were scored on a 3-point scale (1,
well controlled; 2, not well controlled; 3, very poorly controlled). ‘‘Very
poorly controlled’’ was assigned if a score of 3 was selected for any question.
‘‘Not well controlled’’ was assigned if a score of 2 was selected for 1 or more
questions. ‘‘Well controlled’’ was assigned if a score of 1 was selected for all 5
questions. The criterion measure of controlled versus uncontrolled respiratory
status was based on the dichotomized responses to the physician question-
naire: controlled (well controlled rating) and uncontrolled (not well controlled
or very poorly controlled ratings). One question that asked physicians whether
therapy needed to be stepped up, be stepped down, or remain the same was
used in analyses of construct validity.
Data analysis
Data from the cross-sectional study were used to develop and then validate
the TRACK instrument. Two thirds (n 5 321) of caregivers who completed the
questionnaires were randomly selected for the item development sample and
one third (n 5 165) for the validation sample.
Item selection and scoring. Stepwise logistic regression with
the backwards selection method was used to identify the subset of the 33
items that showed the greatest validity in discriminating between the
guidelines-based controlled and uncontrolled ratings. All items were entered
in the model. Based on an a priori–defined significance level of P < .05,
items that were deemed nonsignificant (ie, P � .05) were then iteratively
eliminated. The subset of items identified was further evaluated for clinical
validity and refined to eliminate redundancy in item content. Each item was
scored on a 5-point scale; the total score was the sum of individual scores on
each item. These scores were then transformed to a 0- to 100-point scale
for the final TRACK instrument (0, 5, 10, 15, or 20 points for each item
response), with higher scores indicating better respiratory control.
Reliability and validity. Tests for questionnaire reliability and
validity were conducted separately in both the development and validation
samples. Internal consistency reliability was evaluated with the Cronbach a
statistic. Tests of validity were designed to evaluate how well TRACK
discriminated among groups of children who differed in terms of respiratory
control. This standard method of construct validation follows the logic of
‘‘known groups’’ validity.
20
For these tests, children were categorized into
groups known to differ in respiratory control derived from 3 criteria measures.
The first criterion measure was based on the previously described NAEPP
EPR-3 definition of control. As previously described, responses to the first 5
items on the physician questionnaire were used to categorize children into 3
categories of control (well controlled, not well controlled, and very poorly
controlled). The second criterion measure was whether the physician changed
the child’s therapy as a result of the visit. Children were categorized into the
following 3 groups: stepped-down therapy, no change in therapy, or stepped-
up therapy. The third criterion measure was symptom status, which was de-
rived from the study screening criteria (ie, current symptomatic, recent past
symptomatic, or asymptomatic).
We hypothesized that the groups of children classified as well controlled
would have higher TRACK scores than the groups of children classified as not
well controlled or very poorly controlled. Similarly, we hypothesized that
children categorized as stepped-down therapy or no change in therapy would
have higher TRACK scores than children categorized as stepped-up therapy.
Lastly, we hypothesized that TRACK scores would be lowest for the
symptomatic group and highest for the asymptomatic group. ANOVA methods
were used to test these hypotheses. Exploratory factor analysis (EFA) and
APRIL 2009

construct validity of TRACK in relation to our conceptual model of respiratory
specialist sites; 41.6% were completed at primary care sites. Sample
J ALLERGY CLIN IMMUNOL
VOLUME 123, NUMBER 4
MURPHY ET AL 835
control. These methods and the results of these analyses are available in the
Methods section in this article’s Online Repository (see Tables E2 and E3 in
this article’s Online Repository at www.jacionline.org).
Screening accuracy. The screening accuracy of TRACK as a tool
to identify children with respiratory control problems was evaluated using
logistic regression methods and receiver operating characteristic (ROC) curve
analyses. The criterion measure of respiratory control was based on the
responses to the physician questionnaire. Because the ability to detect ‘‘any’’
control problems was the objective of this analysis, a dichotomous variable of
respiratory control problems was derived by grouping children categorized as
uncontrolled (very poorly controlled and not well controlled) versus those
categorized as well controlled. In addition, sensitivity, specificity, positive and
negative predictive values, percentage correctly classified, and the area under
the ROC curve were computed across various cutoffs along the TRACK score
distribution in the total sample as a means of finding the optimal cutoff score
for screening purposes.
RESULTS
Total sample
Characteristics of the total sample (N 5 486) of caregivers who
completed the draft 33-item questionnaire and their children are
provided in Table I. Most caregivers were younger than 45 years of
age (95.7%), female (88.9%), and white (71.8%). Most children
were aged 3 to 5 years (82.1%); had 1 or more episodes of wheez-
ing, coughing, or shortness of breath in the past 4 weeks (61.9%);
and had symptoms that were either not well controlled or very
poorly controlled (73.7%). Most (61.9%) children had a diagnosis
of asthma; the remaining 38.1% had used a prescription bronchodi-
lator but did not have an asthma diagnosis. More than half of the
caregiver questionnaires (58.4%) were completed at asthma
characteristics were generally similar between the 2 types of sites.
Item selection
Based on the results of analyses performed in the randomly
selected development sample (n 5 321), 6 items were identified
as meeting the model selection entry criteria for predicting
respiratory control (controlled versus uncontrolled). The items
included 4 impairment questions on symptom burden and activity
limitations over a 4-week period, 1 impairment question on rescue
medication use over a 3-month period, and 1 risk question on oral
corticosteroid use over the past 12 months.
The 6 items were further evaluated for relevance, meaning,
language clarity, administrative burden, and feasibility. Based on
these additional considerations, 2 items (Q1 and Q13) were
deleted based on content redundancy, and 1 item was replaced
(Q20) with a more comprehensive item (Q22). Additionally, an
item (Q9) about the frequency of nighttime awakenings caused by
respiratory problems was added because it is a component of
asthma control in the NAEPP EPR-3 guidelines and was a
relevant concept during qualitative interviews with physicians
and caregivers. The measurement properties of the 5-item scale
were retained (see Table E4 in this article’s Online Repository at
www.jacionline.org); the area under the ROC curves for both
scales was 0.88. Additional logistic regression analysis showed
the ability of each of the 5 items to independently contribute to
the discrimination between uncontrolled and controlled asthma
(Table II). This 5-item measure was selected as the final TRACK
instrument and was tested further in the item development and
validation samples.
TABLE I. Caregiver and patient demographics and baseline clinical
Characteristic
Total (N 5 486),
no. (%)
Female 432 (88.9)
Caregiver age (y)*
18-24 41 (8.4)
25-34 241 (49.6)
35-44 183 (37.7)
�45 19 (3.9)
Caregiver education*
Not a high school graduate 13 (2.7)
High school graduate or some college 165 (34.0)
College graduate or higher 301 (61.9)
Caregiver ethnicity*
White 349 (71.8)
African American 54 (11.1)
Hispanic 41 (8.4)
Other 37 (7.6)
Child age (y)
0-2 87 (17.9)
3-5 399 (82.1)
Caregiver-assessed child symptom status
Symptomatic (last 4 wk) 301 (61.9)
Asymptomatic (recent) 153 (31.5)
Asymptomatic (>1 y) 32 (6.6)
Caregiver-assessed child control status
Well controlled 127 (26.1)
Not well controlled 197 (40.5)
Very poorly controlled 162 (33.3)
*Percentages total less than 100 because some caregivers did not answer every question
characteristics
Specialist site (n 5 284),
no. (%)
Primary care site (n 5 202),
no. (%)
251 (88.4) 181 (89.6)
21 (7.4) 20 (9.9)
131 (46.1) 110 (54.5)
120 (42.3) 63 (31.2)
10 (3.5) 9 (4.5)
8 (2.8) 5 (2.5)
83 (29.2) 82 (40.6)
187 (65.8) 114 (56.4)
206 (72.5) 143 (70.8)
24 (8.5) 30 (14.9)
24 (8.5) 14 (6.9)
26 (9.2) 11 (5.4)
52 (18.2) 37 (18.3)
232 (81.8) 165 (81.7)
178 (62.7) 123 (60.9)
92 (32.4) 61 (30.2)
14 (4.9) 18 (8.9)
77 (27.1) 50 (24.8)
120 (42.3) 77 (38.1)
87 (30.6) 75 (37.1)
.

Q9: During the past 4 weeks, how often did your child’s breathing problems 1.38 0.18 2.91 .004
J ALLERGY CLIN IMMUNOL836 MURPHY ET AL
Reliability
The internal consistency reliability of the 5-item TRACK
instrument was similar in the development (n 5 321) and
validation (n 5 165) samples (Cronbach a coefficients 5 0.75
and 0.71, respectively). In both samples, the Cronbach a coef-
ficients were greater than the minimum accepted threshold for
reliability of 0.70.
21
Criterion validation
Based on the first criterion for respiratory control, determined
by using the guidelines-based physician questionnaire, mean
direction across the 3 groups of children (well controlled, not
well controlled, or very poorly controlled) in the development
(F 5 105.1, P < .001) and validation (F 5 58.7, P < .001) samples
(Table III). As hypothesized, mean TRACK scores were highest
among those with a well-controlled rating and lowest among
those with a very poorly controlled rating.
Using the second criterion for respiratory control, based on
physician-recommended changes to therapy, mean TRACK
scores differed significantly in the hypothesized direction across
the 3 groups of children (stepped-down therapy, no change in
therapy, or stepped-up therapy) in the development (F 5 30.4,
(wheezing, coughing, shortness of breath) wake him or her up at night?
A9: Not at all, once or twice, once every week, 2 or 3 times a week, 4 or more
times a week
Q30: During the past 3 months, how often did you need to treat your child’s
breathing problems (wheezing, coughing, shortness of breath) with rescue or
quick-relief medications (Albuterol, Ventolin, Proventil, Maxair, proAIR,
Xopenex, or Primatene Mist)?
1.61 0.16 4.75 .000
A30: Not at all, once or twice, once every week, 2 or 3 times a week, 4 or more
times a week
Q33: During the past 12 months, how often did your child need to take oral
corticosteroids (prednisone, prednisolone, Orapred, Prelone, Decadron) for
breathing problems not controlled by other medications?
1.42 0.15 3.21 .001
A33: Never, once, twice, 3 times, 4 or more times
*The dependent variable was uncontrolled asthma based on physician questionnaire, and independent variables were each of the 5 questions.
TABLE III. Evaluation of the discriminant validity of TRACK: physicians’ ratings of control
Control rating
Well controlled Not well controlled Very poorly controlled
Statistical test
F, P value
Development sample* n 5 102 n 5 132 n 5 80
TRACK score, mean (SD) 85.1 (13.1) 64.0 (18.2) 47.4 (21.3) 105.1, <.001
Validation sample n 5 59 n 5 60 n 5 46
TRACK score, mean (SD) 81.9 (15.4) 67.3 (14.7) 45.9 (20.9) 58.7, <.001
*n 5 314. Data are missing for 7 patients.
TABLE IV. Evaluation of the discriminant validity of TRACK: recommended change in therapy
Therapy recommendation
Stepped-down therapy No change in therapy Stepped-up therapy
Statistical test
F, P value
Development sample* n 5 13 n 5 196 n 5 105
TRACK score, mean (SD) 76.2 (17.8) 72.9 (20.3) 53.7 (22.2) 30.4, <.001
Validation sample n 5 11 n 5 106 n 5 48
TRACK score, mean (SD) 59.4 (28.2) 73.9 (19.0) 52.0 (19.2) 21.2, <.001
*n 5 314. Data are missing for 7 patients.
TABLE II. Summary of logistic regression and item-selection analys
Item text and response options
Q8: During the past 4 weeks, how often was your child bothered by breathing
problems, such as wheezing, coughing, or shortness of breath?
A8: Not at all, once or twice, once every week, 2 or 3 times a week, 4 or mo
times a week
Q22: During the past 4 weeks, to what extent did your child’s breathing problem
such as wheezing, cough, or shortness of breath, interfere with his or her abil
to play, go to school, or engage in usual activities that a child should be doing
his or her age?
A22: Not at all, slightly, moderately, quite a lot, extremely
TRACK scores differed significantly in the hypothesized
is for the 5 final TRACK items*
Odds ratio SE z Score P
2.21 0.39 4.52 .000
re
s,
ity
at
1.70 0.33 2.75 .006
APRIL 2009
P < .001) and validation (F 5 21.2, P < .001) samples (Table IV).

J ALLERGY CLIN IMMUNOL MURPHY ET AL 837
As hypothesized, mean TRACK scores were lowest in the group
for whom a step up in therapy was recommended.
Mean TRACK scores also differed significantly in the hypoth-
esized direction across the 3 groups of children who differed in
symptom status according to the third criterion for respiratory
control (frequency of symptoms at screening) in the development
(F 5 52.8, P < .001) and validation (F 5 39.3, P < .001) samples
(Table V). As hypothesized, mean TRACK scores were highest
for those with the least frequent respiratory symptoms and lowest
for those with the most frequent respiratory symptoms.
Screening accuracy
In the development and validation samples, respectively,
analyses showed that the screening ability of the entire scale
across all scores (0-100) provided an area under the ROC curve of
0.88 and 0.82; control status was correctly classified in 80.6% and
78.3% of cases. Table VI (development and validation samples)
summarizes the performance of the TRACK instrument in screen-
ing for children with respiratory control problems at various cut-
offs. Each TRACK score level represents a cutoff that separates
children whose symptoms are well controlled from children
whose symptoms are not well controlled or very poorly con-
trolled. Statistics are presented beginning with score level 65.
Score levels of less than 65 are not presented because they yielded
poor classification statistics. In both samples, lower cutoff scores
were associated with lower sensitivity and higher specificity. Con-
versely, higher cutoff scores were associated with relatively
higher sensitivity and lower specificity, indicating that at higher
uncontrolled cases (higher sensitivity) but also identified more
controlled cases as uncontrolled (low specificity). A cutoff score
of less than 80 provided the best balance between sensitivity and
specificity and the best discrimination between patients with con-
trolled and uncontrolled symptoms based on the highest area un-
der the ROC curve in both development and validation samples.
DISCUSSION
The objective of this study was to develop an instrument to
measure respiratory and asthma control in preschool-aged chil-
dren with symptoms consistent with asthma that included the
impairment and risk domains of control recommended by the
most recent NAEPP guidelines.
12
Regular assessment and moni-
toring of respiratory symptoms and activity limitations are essen-
tial in the care of asthma. However, standardized methods for
measuring control in clinical settings are not available for
preschool-aged children. The tools that do exist focus primarily
on the adult population,
17,18,22
children and adolescents aged
1 to 18 years,
23
or children 4 to 11 years of age.
14
None of these
tools considers both impairment and risk in defining asthma
control.
The development of TRACK began with a comprehensive
qualitative study designed to describe respiratory and asthma
control issues from the perspectives of specialists, pediatricians,
and caregivers of preschool-aged children. Content used to draft
the questionnaire also reflected the NAEPP EPR-3 asthma
management guidelines.
12
A content analysis was performed to
TABLE V. Evaluation of the discriminant validity of TRACK: sympto
Sym
Symptomatic for
last 4 wk
Sym
Development sample* n 5 201 n 5
TRACK score, mean (SD) 58.2 (21.2) 80.1
Validation sample n 5 95 n 5
TRACK score, mean (SD) 56.2 (20.4) 78.0
*n 5 314. Data are missing for 7 patients.
TABLE VI. Evaluation of cutoff scores: 5-item TRACK scale
Cutoff
scores
Odds
ratio Sensitivity Specificity
Posi
predi
val
Development sample
<90 17.3 95.3 46.1 78
<85 13.8 89.2 62.8 83
<80 17.8 83.0 78.4 88
<75 19.1 73.6 87.3 92
<70 24.8 64.6 93.1 95
<65 23.4 54.7 95.1 95
Validation sample
<90 35.5 97.2 50.9 78
<85 12.2 90.6 55.9 78
<80 6.5 78.3 64.4 79
<75 5.9 68.9 72.9 82
<70 5.5 63.2 76.3 82
<65 8.0 55.7 86.4 88
VOLUME 123, NUMBER 4
cutoff scores, the TRACK instrument was better at detecting
m status
ptom frequency
ptomatic for >4wk
and #1y
Asymptomatic
for >1y
Statistical test
F, P value
96 n 5 17
(17.4) 90.0 (11.0) 52.8, <.001
55 n 5 15
(15.4) 90.7 (10.7) 39.3, <.001
tive
ctive
ue
Negative
predictive
value
False-positive
rate
Percent
correctly
classified
Area under
ROC
.6 82.5 53.9 79.3 0.71
.3 73.6 37.2 80.6 0.76
.9 68.9 21.6 81.5 0.81
.3 61.4 12.7 78.0 0.80
.1 55.9 6.9 73.9 0.79
.9 50.3 4.9 67.8 0.75
.0 90.9 49.1 80.6 0.74
.7 76.7 44.1 78.2 0.73
.8 62.3 35.6 78.3 0.74
.8 56.6 27.1 70.3 0.71
.7 53.6 23.7 67.8 0.70
.1 52.0 13.6 66.7 0.71
ensure that essential elements of control were included in the draft

J ALLERGY CLIN IMMUNOL
APRIL 2009
838 MURPHY ET AL
questionnaire before implementation in the present study. To be
practical, however, the draft questionnaire tested in this study
needed to be shortened considerably, from 33 items to 5 to 10
items, while maintaining content validity. The final 5 items se-
lected for the TRACK survey are consistent with the current
NAEPP EPR-3 asthma management guidelines for both the im-
pairment domain of control assessment (ie, asthma symptoms,
use of rescue medications, nighttime awakenings, and the effect
of asthma on everyday functioning) and the risk domain of control
assessment (ie, oral corticosteroid courses in the past 12
months).
12
As such, the TRACK instrument supports the premise
that respiratory and asthma control is a multidimensional
construct.
The TRACK instrument is the first respiratory control ques-
tionnaire to include an item to measure risk.
12,14
Capturing risk in
a control tool for young children is novel and important given the
frequency of exacerbations in this very young cohort with fre-
quent intermittent episodes. Although the components of impair-
ment assessed in the Childhood Asthma Control Test (C-ACT)
and TRACK are similar (eg, nighttime awakenings, interference
with activity), the wording of the questions and the populations
studied are different. Because of the challenges in diagnosing
asthma in preschool-aged children, TRACK was specifically de-
veloped for caregivers of children aged younger than 5 years with
respiratory symptoms consistent with asthma but not necessarily
with a specific diagnosis of asthma because health care providers
are often hesitant to give young children diagnoses of asthma. The
TRACK cohort consisted of young children with at least 2 epi-
sodes of respiratory symptoms consistent with asthma plus a phy-
sician’s diagnosis of asthma (61.9%) or use of a bronchodilator
(38.1%). In contrast, the C-ACT includes both child-completed
and caregiver-completed items and was developed or validated
for generally school-aged (mean age [SD], 8 [2.4] years) children
with asthma.
14
In this study, a scale constructed from the 5 items selected for
TRACK was shown to be reliable and valid. In screening for
control problems, the final instrument showed a good area under
the ROC curve relative to the NAEPP EPR-3–based ratings of
asthma control. The ROC value of 0.88 is consistent with the ROC
values of 0.791 and 0.862 reported for the development and
validation samples, respectively, of the Childhood Asthma Con-
trol Test.
14
The TRACK instrument is not a diagnostic screening
tool. However, this study suggests that TRACK correctly clas-
sifies respiratory control levels in approximately 80% of children
younger than 5 years with a history of 2 or more episodes of
respiratory symptoms suggestive of asthma. Assessments of
TRACK’s screening accuracy for poor respiratory control suggest
that children with a TRACK score of less than 80 are likely to
have uncontrolled asthma. This score provided the optimum bal-
ance of sensitivity and specificity in both the development and
validation samples. The sensitivity and specificity of TRACK at
the less-than-80 cutoff are consistent with those reported for the
Asthma Control Test at a cutoff of 19 or less (sensitivity, 69.2;
specificity, 76.2).
17
For control status, the percentage correctly
classified of approximately 80% is comparable with that reported
at a cutoff score of 19 for the Asthma Control Test (74% correctly
classified) and C-ACT (72% in development sample and 83% in
confirmatory sample).
14,17
Therefore clinicians should be aware
that control status potentially can be misclassified in approxi-
mately 20% of patients using currently available asthma control
tools, including TRACK. To strengthen TRACK’s clinical utility
and improve the capture of true-positive results and, especially,
false-negative results, further validation studies are needed to
assess test-retest reliability, responsiveness to change in therapy,
and the minimum important difference for improvement or wors-
ening of asthma control. These instruments, however, are a useful
starting point for identifying most patients who have suboptimal
respiratory or asthma control. Therefore, in patients with a
TRACK score of less than 80, one should consider the need for
further evaluation, treatment adjustments, or both.
Recognizing that TRACK might be used for many different
purposes in the future, we have provided screening statistics
across various score levels of TRACK. For example, clinicians
involved in a disease management program who plan to use
TRACK as an initial screening tool might decide to choose a
cutoff associated with a high degree of sensitivity (higher TRACK
score cutoff). Such a cutoff would ensure that most children with
respiratory or asthma control problems are identified for inclusion
in the program. More resource-intensive screening (to identify
false-positive results) could then be applied to those who were
selected through the initial screening with TRACK.
The reliability of the 5-item TRACK instrument was tested by
using internal consistency methods. The Cronbach a coefficients
of the 5-item TRACK instrument were relatively low in the
development (0.75) and validation (0.71) samples. These results
were not surprising because the 5 items of TRACK were not
designed to be internally consistent and because they represent 3
different dimensions of control. A more appropriate method for
assessing the reliability of TRACK is to assess the test-retest
reliability. A follow-up longitudinal study is currently being
conducted to evaluate the stability of TRACK scores over time
and to determine TRACK’s responsiveness to change in
control.
Given the increasingly limited physician-child-caregiver inter-
action time and that objective measures of lung function are not
readily available for young children, an accurate, reliable, and
easy-to-use control tool is essential in the management of
respiratory symptoms of young children. A brief standardized
instrument for preschool-aged children should be considered for
collecting basic information about respiratory and asthma control
before a visit (eg, in the waiting room) to a health care provider,
thereby saving time for more in-depth and focused discussions.
TRACK was specifically designed to raise awareness of respira-
tory control problems in young children for both the health care
provider and caregiver and might be most useful in clinical
practice by facilitating a dialogue between the health care
provider and caregiver to identify strategies to better manage
the child’s respiratory symptoms. Furthermore, use of TRACK
over time might prove useful in monitoring trends in respiratory
and asthma control for use in clinical care and research. However,
further research is necessary before the role of TRACK in clinical
practice and research can be established clearly.
Recruitment of young children and their caregivers into clinical
trials can be challenging. The number of patients in the present
study, however, is as large as or larger than the number of patients
in published development/validation studies of patient-reported
asthma control tools for children aged 4 to 11 years,
14
children
and adolescents aged 5 years and older,
15
and adolescents and
adults.
17,24
The number of children who were severely affected
or younger than 2 years was small in the current study, and there-
fore, these subgroups will require additional study. Another lim-
itation is that the study population was English-speaking only,

mostly white, and mostly college educated. Efforts are underway
to translate and test TRACK in Spanish.
In conclusion, the results of this study provide evidence of the
reliability and validity of TRACK. In addition, TRACK might be
very useful in identifying young children with respiratory control
problems, which is important in this population because objective
measures are either too difficult for children to perform or have
not been standardized. The 5-question TRACK tool is easy to
administer and score and addresses both the impairment and risk
domains of asthma control consistent with current asthma guide-
5. Weiss KB, Sullivan SD, Lyttle CS. Trends in the cost of illness for asthma in the
United States, 1985–1994. J Allergy Clin Immunol 2000;106:493-9.
6. Weiss KB, Sullivan SD. The health economics of asthma and rhinitis. I. Assessing
the economic impact. J Allergy Clin Immunol 2001;107:3-8.
7. Sullivan SD. Asthma in the United States: recent trends and current status. J Manag
Care Pharm 2003;9:3-7.
8. American Lung Association Epidemiology & Statistics Unit Research and
Program Services. Trends in asthma morbidity and mortality 2005. Available at
http://www.lungusa.org/atf/cf/{7A8D42C2-FCCA-4604-8ADE-7F5D5E762256}/
ASTHMA1.PDF. Accessed June 21, 2005.
9. Boulet LP, Phillips R, O’Byrne P, Becker A. Evaluation of asthma control by phy-
sicians and patients: comparison with current guidelines. Can Respir J 2002;9:
J ALLERGY CLIN IMMUNOL
VOLUME 123, NUMBER 4
MURPHY ET AL 839
lines. The establishment of a cutoff score for identifying children
with respiratory control problems makes TRACK a valuable aid
to health care providers and caregivers in monitoring young
children with symptoms consistent with asthma. This study
provides the first evidence for the validation of TRACK to assess
respiratory and asthma control in children aged younger than
5 years.
We thank Joseph Spahn (Denver, Colo) for his invaluable contributions to
the design of the study and interpretation of the data. We also acknowledge the
efforts of the physicians who participated in this study: William E. Berger
(Mission Viejo, Calif), Michael Blaiss (Germantown, Tenn), Randall Brown
(Atlanta, Ga), Don Bukstein (Madison, Wis), John Calcagno (Gresham, Ore),
Stuart Cohen (San Diego, Calif), Marc Goldstein (Philadelphia, Pa), William
Hitchcock (La Jolla, Calif), Michael Lawrence (Taunton, Mass), Todd Mahr
(La Crosse, Wis), John Matz (Baltimore, Md), Nancy K. Ostrom (San Diego,
Calif), Michael Pichichero (Rochester, NY), Shelley Senders (Cleveland,
Ohio), Shailen Shah (Collegeville, Pa), Timothy Sullivan (Norwich, Conn),
and Steven Weinstein (Huntington Beach, Calif). Finally, we thank Marissa
Buttaro, MPH, from Scientific Connexions (Newtown, Pa) for editorial
assistance funded by AstraZeneca LP.
Clinical implications: Clinicians now have a validated, easy-to-
administer, caregiver-completed questionnaire to evaluate
respiratory control in preschool-aged children. This tool aids
clinicians in evaluating respiratory control in young children
with symptoms consistent with asthma.
REFERENCES
1. Mannino DM, Homa DM, Akinbami LJ, Moorman JE, Gwynn C, Redd SC. Surveil-
lance for asthma—United States, 1980-1999. MMWR Surveill Summ 2002;51:1-13.
2. Centers for Disease Control and Prevention. Asthma prevalence and control char-
acteristics by race/ethnicity—United States, 2002. MMWR Morb Mortal Wkly
Rep 2004;53):145-8.
3. Centers for Disease Control and Prevention. Asthma mortality and hospitalization
among children and young adults: United States, 1980-1993. JAMA 1996;275:1535-7.
4. Newacheck PW, Halfon N. Prevalence, impact, and trends in childhood disability
due to asthma. Arch Pediatr Adolesc Med 2000;154:287-93.
377-8.
10. Rabe KF, Adahi M, Lai CKW, et al. Worldwide severity and control of asthma in
children and adults: the global asthma insights and reality surveys. J Allergy Clin
Immunol 2004;114:40-7.
11. Halterman JS, McConnochie KM, Conn KM, Yoos HL, Kaczorowski JM, Holzha-
uer RJ, et al. A potential pitfall in provider assessments of the quality of asthma
control. Ambul Pediatr 2003;3:102-5.
12. NHLBI/NAEPP. Expert panel report 3: guidelines for the diagnosis and manage-
ment of asthma—full report 2007. Bethesda (MD): National Institutes of Health/
National Heart, Lung, and Blood Institute; 2007. Publication no. 08-4051.
13. Global Strategy for Asthma Management and Prevention, Global Initiative for
Asthma (GINA) 2007. Available at: http://www.ginasthma.org. Accessed July
31, 2008.
14. Liu AH, Zeiger R, Sorkness C, Mahr T, Ostrom N, Burgess S, et al. Development
and cross-sectional validation of the Childhood Asthma Control Test. J Allergy
Clin Immunol 2007;119:817-25.
15. Skinner EA, Diette GB, Algatt-Bergstrom PJ, Nguyen TTH, Clark RD, Markson
LE, et al. The Asthma Therapy Assessment Questionnaire (ATAQ) for Children
and Adolescents. Dis Manage 2004;7:305-13.
16. Juniper EF, O’Byrne PM, Ferrie PJ, King DR, Roberts JN. Measuring asthma con-
trol. Clinic questionnaire or daily diary? Am J Respir Crit Care Med 2000;162:
1330-4.
17. Nathan RA, Sorkness CA, Kosinski M, Schatz M, Li JT, Marcus P, et al. Develop-
ment of the asthma control test: a survey for assessing asthma control. J Allergy
Clin Immunol 2004;113:59-65.
18. Vollmer WM, Markson LE, O’Connor E, Frazier EA, Berger M, Buist AS. Asso-
ciation of asthma control with health care utilization: a prospective evaluation. Am
J Respir Crit Care Med 2002;165:195-9.
19. Lara M, Duan N, Sherbourne C, Lewis MA, Landon C, Halfon N, et al. Differences
between child and parent reports of symptoms among Latino children with asthma.
Pediatrics 1998;102:e68.
20. Kerlinger FN. Foundations of Behavioral Research. New York: Holt, Rinehart and
Winston; 1973.
21. Nunnally JC, Bernstein IH. Psychometric theory. New York: McGraw-Hill, Inc;
1994.
22. Juniper EF, O’Byrne PM, Guyatt GH, Ferrie PJ, King DR. Development and
validation of a questionnaire to measure asthma control. Eur Respir J 1999;
14:902-7.
23. Zorc JJ, Pawlowski NA, Allen JL, Bryant Stephens T, Winston M, et al. Develop-
ment and validation of an instrument to measure asthma symptom control in chil-
dren. J Asthma 2006;43:753-8.
24. Schatz M, Sorkness CA, Li JT, Marcus P, Murray JJ, Nathan RA, et al. Asthma
Control Test: reliability, validity, and responsiveness in patients not previously fol-
lowed by asthma specialists. J Allergy Clin Immunol 2006;117:549-56.

METHODS
Working group and qualitative development
Development of the draft questionnaire was based on input from a working
group, caregiver and physician interviews, and subsequent qualitative re-
search. Aworking group of pediatric asthma specialists (n 5 5) advised on the
study protocol, identification and recruitment of study sites, site training and
inclusion and exclusion criteria and provided direction on item development
for the caregiver and physician questionnaires. Results from in-depth struc-
tured telephone interviews with 5 pediatric asthma specialists, 5 pediatricians,
hypothesized that items Q1 though Q18 on the caregiver survey would be
strongly correlated (r > 0.7) with the ‘‘symptom’’ factor. We also hypothesized
that items Q19 through Q26 would be strongly correlated (r > 0.7) with the
‘‘impact’’ factor. Lastly, we hypothesized that items Q27 through Q33 would
be strongly correlated (r > 0.7) with the ‘‘exacerbation’’ factor.
RESULTS
Construct validity
The first 5 eigenvalues from the EFA are presented in Table E2
J ALLERGY CLIN IMMUNOL
APRIL 2009
839.e1 MURPHY ET AL
and 13 caregivers of children with asthma concluded that both physicians and
caregivers agreed on the need for a brief, standardized, caregiver-completed
questionnaire to assist in detecting respiratory control problems in preschool-
aged children (<5 years). The working group decided to base the criterion
measure of respiratory control for this instrument on the definitions in the
National Asthma Education and Prevention Program Expert Panel Report-3
asthma control table for 0- to 4-year-old children.
E1
Additionally, they
reviewed the results of the item-selection analysis to ensure that the final set
of items selected for the instrument appropriately represented the components
of control for children aged 0-4 years.
E1
Data collection and analysis
Data were obtained from eligible caregivers and physicians regarding the
child’s respiratory/asthma control problems. Physicians completed a 15-item
questionnaire that included 5 specific items (Table E1) based on the risk and
impairment factors in the previously described control table.
E1
During the
child’s visit to the physician, caregivers answered 10 questions about the
child’s medical and clinical history and completed the 33-item draft question-
naire (Fig E1).
Construct validity
During the qualitative phase of this study, a conceptual model for
respiratory control problems in preschool-aged children was developed.
This model was derived from feedback from clinicians and caregivers of
preschool-aged children with asthma, the literature, and the asthma manage-
ment guidelines. Our conceptual model of respiratory control included 3 major
components. The first component consists of the frequency and severity of
respiratory symptoms (cough, wheeze, and shortness of breath). The second
component of respiratory control consists of the physical and emotional effect
of respiratory symptoms. Items that comprise this component measure the
extent to which the child’s respiratory problems interfere with daily activities,
such as play and other normal routines, and the extent to which they cause
distress, irritability, and fatigue. The third component consists of exacerba-
tions, which were defined by items that assessed the frequency with which the
child used care and services to treat and control respiratory symptoms (rescue
medications, emergency department visits, unscheduled visits to the physi-
cian, hospitalizations, and oral corticosteroid use).
Exploratory factor analyses first were conducted to investigate whether
there was a general structure underlying the item responses from the caregiver
questionnaire. In addition, confirmatory factor analyses were conducted to
confirm whether the correlations among items supported common factors in
our conceptual model of respiratory control in preschool-aged children. We
(development and validation samples). In both samples, the first 5
factors had eigenvalues that were greater than unity, and thus
there were 5 potential factors underlying the pattern of associa-
tions among the caregiver item responses. The first factor (eigen-
value) accounted for the majority of variance in item responses
(>59%) in both samples, meaning that there was something com-
mon to all items in the survey, namely the concept of respiratory
control. The second and third factors explained more than 5% of
the variance in item responses, which allowed for determining the
true number of factors in a factor analytic model. Combined, the
first 3 factors accounted for 75% of the variance in item responses
in both samples. The fourth and fifth factors accounted for less
than 5% of the variance in item responses and thus are not likely
to be true factors.
Table E3 (development and validation samples) summarizes
the CFA results. In both samples, the correlations between each
item and their a priori hypothesized factor were all greater than
0.70, supporting the conceptual model developed for respiratory
control in young children. In addition, the model fit statistics sug-
gest adequate fit of the model in both samples. The Confirmatory
Fit Index and the Tucker-Lewis Index values in the development
sample were 0.89 and 0.96, respectively. The Confirmatory Fit
Index and the Tucker-Lewis Index values in the validation sample
were 0.90 and 0.95, respectively. Overall, the results of the EFA
and CFA lend support that the items developed for the draft
TRACK questionnaire appropriately measure the concepts de-
rived for the model of respiratory control.
TRACK item modification
The 6 items identified by the model and the final 5-item
TRACK instrument demonstrated similar measurement proper-
ties when evaluated for variance, sensitivity, specificity, positive
and negative predictive value, false-positive rate, percentage
correctly classified, and area under the ROC curve (Table E4).
REFERENCE
E1. E1NHLBI/NAEPP. Expert panel report 3: guidelines for the diagnosis and man-
agement of asthma—full report 2007. Bethesda (MD): National Institutes of
Health/National Heart, Lung, and Blood Institute; 2007. Publication no. 08–4051.

J ALLERGY CLIN IMMUNOL
VOLUME 123, NUMBER 4
MURPHY ET AL 839.e2
FIG E1. Draft 33-item caregiver questionnaire.

J ALLERGY CLIN IMMUNOL
APRIL 2009
839.e3 MURPHY ET AL
FIG E1. (Continued)

J ALLERGY CLIN IMMUNOL
VOLUME 123, NUMBER 4
MURPHY ET AL 839.e4
FIG E1. (Continued)

J ALLERGY CLIN IMMUNOL
APRIL 2009
839.e5 MURPHY ET AL
FIG E1. (Continued)

TABLE E1. First 5 items of physicians’ questionnaire assessing respiratory control
Control rating categories*
Respiratory control assessment (1) (2) (3)
�2 days/week >2 days/week Throughout the day
1. During the past 4 weeks, how many days a week did the child have cough or
wheeze (for example, breathing that makes a high pitched whistling or
squeaking sound from the chest)?
h 1 h 2 h 3
1 time/month >1 time/month >1 time/week
2. During the past 4 weeks, how often was the child’s sleep disrupted by cough
or wheeze?
h 1 h 2 h 3
No limitation Some limitation Extremely limited
3. During the past 4 weeks, how limited was the child in performing normal
activities by cough or wheeze?
h 1 h 2 h 3
�2 days/week >2 days/week Several times/day
4. During the past 4 weeks, how many days a week did the child use albuterol
to treat his or her respiratory symptoms, such as cough or wheeze?
h 1 h 2 h 3
0-1 time/year 2-3 times/year >3 times/year
5. In the past year, how many times did the child take oral steroids to treat
episodes of cough or wheeze?
h 1 h 2 h 3
*1, Well controlled; 2, not well controlled; 3, very poorly controlled.
J ALLERGY CLIN IMMUNOL
VOLUME 123, NUMBER 4
MURPHY ET AL 839.e6

TABLE E2. Eigenvalue analysis* of caregiver survey responses
Factors
12345
Development sample
Eigenvalue 20.8 2.3 2.0 1.3 1.1
Variance explained (%) 63 7 6 3.9 3.3
Total variance explained (%) 63 70 76 79.9 83.2
Validation sample
Eigenvalue 19.7 2.9 2.2 1.7 1.4
Variance explained (%) 59.7 8.8 6.6 4.8 4.2
Total variance explained (%) 59.7 68.5 75.1 79.9 84.1
*Exploratory factor analysis.
J ALLERGY CLIN IMMUNOL
APRIL 2009
839.e7 MURPHY ET AL

TABLE E3. Confirmatory factor analysis
Symptom factor Impact factor Utilization factor
Item no. Item-factor correlation Item no. Item-factor correlation Item no. Item-factor correlation
Development sample
Q11 0.90 Q29 0.95 Q37 0.60
Q12 0.88 Q30 0.93 Q38 0.81
Q13 0.78 Q31 0.92 Q39 0.54
Q14 0.85 Q32 0.95 Q40 0.77
Q15 0.92 Q33 0.88 Q41 0.47
Q16 0.89 Q34 0.88 Q42 0.55
Q17a 0.88 Q35 0.82 Q43 0.61
J ALLERGY CLIN IMMUNOL
VOLUME 123, NUMBER 4
MURPHY ET AL 839.e8
Q17b 0.88 Q36 0.91
Q17c 0.81
Q17d 0.77
Q18 0.79
Q19 0.95
Q20 0.87
Q21 0.81
Q22 0.94
Q23 0.93
Q24 0.85
Q25 0.81
Q26 0.79
Q27 0.81
Q28 0.76
Validation sample
Q11 0.92 Q29 0.93 Q37 0.55
Q12 0.72 Q30 0.95 Q38 0.85
Q13 0.79 Q31 0.93 Q39 0.73
Q14 0.84 Q32 0.97 Q40 0.78
Q15 0.87 Q33 0.89 Q41 0.51
Q16 0.80 Q34 0.91 Q42 0.52
Q17a 0.88 Q35 0.87 Q43 0.60
Q17b 0.79 Q36 0.88
Q17c 0.86
Q17d 0.75
Q18 0.84
Q19 0.86
Q20 0.83
Q21 0.68
Q22 0.90
Q23 0.90
Q24 0.87
Q25 0.77
Q26 0.85
Q27 0.83
Q28 0.84
Development sample: Confirmatory Fit Index 5 0.89; Tucker-Lewis Index 5 0.96.
Validation sample: Confirmatory Fit Index 5 0.90; Tucker-Lewis Index 5 0.95.

TABLE E4. Comparison of the screening accuracy of TRACK
scales: 5-item TRACK scale versus 6 items identified from
stepwise logistic regression (backward selection method)
Item 5-Item scale (TRACK) 6-Item scale
Variance explained (%) 35.4 36.2
Sensitivity 89.2 91.0
Specificity 62.8 62.7
Positive predictive value 83.3 83.6
Negative predictive value 73.6 77.1
False-positive rate 37.3 37.3
Percentage correctly classified 80.6 81.8
Area under ROC curve 0.88 0.88
J ALLERGY CLIN IMMUNOL
APRIL 2009
839.e9 MURPHY ET AL