[
Original Research Signs and Symptoms of Chest Diseases
]
Children With Chronic Cough When Is Watchful Waiting Appropriate? Development of Likelihood Ratios for Assessing Children With Chronic Cough Anne B. Chang, PhD; Peter P. Van Asperen, MD; Nicholas Glasgow, MD; Colin F. Robertson, MD; Craig M. Mellis, MD; I. Brent Masters, PhD; Louis I. Landau, MD; Laurel Teoh, MD; Irene Tjhung, MD; Helen L. Petsky, PhD; and Peter S. Morris, PhD
Chronic cough is associated with poor quality of life and may signify a serious underlying disease. Differentiating nonspecific cough (when watchful waiting can be safely undertaken) from specific cough (treatment and further investigations are beneficial) would be clinically useful. In 326 children, we aimed to (1) determine how well cough pointers (used in guidelines) differentiate specific from nonspecific cough and (2) describe the clinical profile of children whose cough resolved without medications (spontaneous resolution).
BACKGROUND:
A dataset from a multicenter study involving children newly referred for chronic cough (median duration, 3-4 months) was used to determine the sensitivity, specificity, predictive values, and likelihood ratios (LRs) of cough pointers (symptoms, signs, and simple investigations [chest radiography, spirometry]) recommended in guidelines.
METHODS:
The pretest probability of specific cough was 88%. The absence of false-positive results meant that most pointers had strongly positive LRs. The most sensitive pointer (wet cough) had a positive LR of 26.2 (95% CI, 3.8-181.5). Although the absence of other individual pointers did not change the pretest probability much (negative LR close to 1), the absence of all pointers had a strongly negative LR of 0 (95% CI, 0-0.03). Children in the resolved spontaneously group were significantly more likely to be older, to be non-Indigenous, and to have a dry cough and a normal chest radiograph.
RESULTS:
Children with chronic dry cough without any cough pointers can be safely managed using the watchful waiting approach. The high pretest probability and high positive LRs of cough pointers support the use of individual cough pointers to identify high risk of specific cough in pediatric chronic cough guidelines. CONCLUSIONS:
TRIAL REGISTRY: Australian New Zealand Clinical Trials Registry; No.: 12607000526471; CHEST 2015; 147(3):745-753 URL: www.anzctr.org.au
Manuscript received August 31, 2014; revision accepted November 10, 2014; originally published Online First December 11, 2014. ABBREVIATIONS: LR 5 likelihood ratio; PBB 5 protracted bacterial bronchitis; QoL 5 quality of life AFFILIATIONS: From the Child Health Division (Drs Chang, Teoh, Tjhung, and Morris), Menzies School of Health Research, Charles Darwin University, Darwin, NT; Queensland Children’s Respiratory Centre (Drs Chang, Masters, and Petsky), Queensland Children’s Medical Research Institute, Royal Children’s Hospital, Brisbane, QLD; Discipline of Paediatrics and Child Health (Dr Van Asperen), Sydney Medical School, and Central Clinical School (Dr Mellis), University of Sydney, Sydney, NSW; Department of Respiratory Medicine (Dr Van Asperen), The Children’s Hospital at Westmead, Sydney Children’s Hospital Network,
journal.publications.chestnet.org
Westmead, NSW; Medical School (Dr Glasgow), Australian National University, Canberra, ACT; Department of Respiratory Medicine (Dr Robertson), Murdoch Children’s Research Institute, Royal Children’s Hospital, Melbourne, VIC; Medical Workforce (Dr Landau), Health Department of Western Australia and The University of Western Australia, Perth, WA; The Canberra Hospital (Dr Teoh), Canberra, ACT; and Thursday Island Primary Health Care Centre (Dr Tjhung), Thursday Island, QLD, Australia. FUNDING/SUPPORT: The study was funded by an Australian National Health and Medical Research Council (NHMRC) [Project Grant 490321] and supported by an NHMRC Centre for Research Excellence in Lung Health of Aboriginal and Torres Strait Islander Children [Grant 1040830].
745
Chronic cough is a common reason why parents seek a consultation in pediatric practice. In the management of chronic cough in children, identification of the underlying cause (ie, obtaining the correct diagnosis), as opposed to empirical treatment, is considered important.1-3 Several factors underpin the importance of obtaining the correct diagnosis with consequent cough resolution upon appropriate treatment. First, chronic cough may signify the presence of a serious underlying airways disease.4,5 Early diagnosis of these conditions (eg, bronchiectasis, aspiration lung disease) would lead to earlier treatment and prevent further ongoing lung injury.5,6 Second, parents of children with chronic cough have substantial stress and impaired quality of life (QoL) that improves when the cough abates.4,7,8 Third, many parents seek repeated consultations for their child’s cough until the cough resolves. Our single-center7 and multicenter studies4 showed that 75% to 80% of children had more than five consultations prior to presentation to a respiratory specialist for cough. Earlier resolution of cough through identification and treatment of the etiology would reduce the economic burden associated with recurrent consultations. Finally, obtaining a correct diagnosis potentially reduces use of unnecessary investigations and medications with consequent severe adverse events.9,10 The initial step in obtaining a diagnosis requires differentiating those who require treatment (ie, specific cough) from those whose cough is likely to resolve without treatment. In the latter group, a watchful waiting approach has been suggested.1,3 To differentiate these groups, identification of cough pointers (symptoms, signs, and simple investigations [e-Table 1]) is used in
Materials and Methods Data collected from our multicenter cohort study 4 formed the dataset of this article. Briefly, 346 children aged , 18 years newly referred for chronic cough (duration, . 4 weeks) to any of the participating sites were enrolled. These sites were clinics at five major hospitals (Brisbane, Sydney, Melbourne, Canberra, Darwin) and three regional-remote clinics in Orange, New South Wales; Anangu-Pitjantjatjara Lands, Central Australia; and Thursday Island, Queensland. We excluded children with a known chronic respiratory illness previously diagnosed by a respiratory physician or those who have had diagnosed conditions confirmed by objective tests (eg, asthma, cystic fibrosis, bronchiectasis) prior to Dr Chang is supported by an NHMRC practitioner fellowship [Grant 1058213]. CORRESPONDENCE TO: Anne B. Chang, PhD, Queensland Children’s Respiratory Centre, Royal Children’s Hospital, Herston, QLD 4029, Australia; e-mail: [email protected] © 2015 AMERICAN COLLEGE OF CHEST PHYSICIANS. Reproduction of this article is prohibited without written permission from the American College of Chest Physicians. See online for more details. DOI: 10.1378/chest.14-2155
746 Original Research
many chronic cough guidelines (eg, Australian,1 British,2 United States3). However, published evidence is scarce on how well cough pointers differentiate specific from nonspecific cough and on the profile of children whose chronic cough is likely to resolve without treatment. Obtaining evidence for the utility of these clinical findings and investigations will increase the robustness of these guidelines necessary in current standards of evidencebased medicine.11 Additionally, when evaluating children with chronic cough, development of categoryorientated likelihood ratios (LRs)12 for cough pointers (e-Table 1) would help clinicians to differentiate children for whom watchful waiting can be safely undertaken from those who would benefit from treatment and further investigations. In this study involving 326 children newly referred for chronic cough,4 we evaluated data relating cough pointers to specific cough and the three most common etiologies from our recent multicenter study. In the field of chronic cough, our multicenter study4,13 had unique features of research rigor,14 including the following: (1) Rules for assigning etiology of cough were defined a priori; (2) the diagnoses, particularly those of protracted bacterial bronchitis (PBB) and asthma, were temporally related to treatment within a strict time frame; (3) validated cough outcome measures were used; and (4) the children were followed for 6 months to ensure that misdiagnosis had not occurred. We hypothesized that cough pointers are clinically valid (have high positive LRs) and can be used to differentiate specific cough from cough that resolves without treatment (spontaneous resolution).
referral. Children with asthma diagnosed by a nonrespiratory physician were not excluded because asthma is commonly misdiagnosed in our setting. Written informed consent was obtained from parents, and the study was approved by the ethics committees of all participating sites. Enrolled children were managed in accordance with a standardized evidence-based cough algorithm13 until the study’s end point, which was defined as either a primary diagnosis established with cough resolved or the presence of exit criteria (hospitalization related to etiology of the cough or 12 months postenrollment, whichever occurred earliest). The ascribed etiology was defined a priori and based on a temporal and prompt response to the treatment as defined by a validated cough score (verbal categorical descriptive score).15 Chest radiograph and spirometry findings were interpreted by the attending respiratory physician as a dichotomous variable (normal/abnormal). Statistics Because the data had a skewed distribution, medians and interquartile ranges were used for continuous measures. The x2 test was used for categorical data. Kruskal-Wallis analysis was used for group comparisons. SPSS software (IBM Corporation) was used, and a two-tailed P ⱕ .05 was considered significant. For calculation of the OR and 95% CI, where
[
147#3 CHEST MARCH 2015
]
cells are 0, 0.5 was added to each cell using Microsoft Excel. For the OR, the variable was considered significant when the lower limit of the 95% CI value approximated to ⱖ 1.0 (to one decimal point). Sensitivity, specificity, positive and negative predictive values, and positive and negative LRs for the presence of symptoms and signs (specific cough pointers) and for investigations (spirometry and chest radiography) were calculated using an online program (http://vassarstats.net/clin1. html). LRs are presented in light of current standards of assessing the
diagnostic values of symptoms and tests12,16 of the key cough pointers (e-Table 1) used in Australian,1 British,2 and US3 chronic cough guidelines for differentiating specific from nonspecific cough. These LRs (specific cough-all causes) serve as category-orientated LRs12 to assist clinicians in determining which children might safely be managed with watchful waiting (as opposed to those in whom specific treatment is recommended). Calculations were not performed for cough pointers absent in the entire cohort.
Results
significant (Table 2). For these three diagnoses, the specificity of all pointers was high (. 98%), but sensitivity was generally low other than wet cough for PBB and dry cough for asthma (data not shown).
Missing data precluded analyses in 20 children; thus, only 326 children (from the original 346) were included for this dataset. Of the 326 children, 40 did not receive any treatment specific for their cough, and their cough eventually resolved spontaneously with the watchful waiting13 approach (Table 1). This means that, in this cohort of children with chronic cough for an average of 3 to 4 months, the pretest probability of specific cough was 88%. Compared with children who received treatment specific for their cough (n 5 286) and fulfilled the a priori-defined criteria for specified etiologies, children in the resolved spontaneously group (n 5 40) were significantly older, were more likely to have a dry cough and to be non-Indigenous, and had normal chest radiograph findings (Table 1). There was no significant difference between the groups for smoke exposure, duration of cough, prior use of asthma medications, how the children were referred, number of physician visits, spirometry abnormality, Pediatric Quality of Life Inventory version 4.0 (PedsQL 4.0),18 and parent coughspecific QoL17 scores at baseline. Details of the three most common diagnoses (PBB, asthma, bronchiectasis) showed similar distribution, other than Indigenous children being the most likely to have bronchiectasis (Table 1). These data were previously analyzed statistically4 and are not repeated here. Comparing all-cause specific cough with cough resolving spontaneously, significant ORs were found for wet cough, wheeze or reversible airway obstruction, and chest radiograph abnormality (Table 2). However, the frequency of the other cough pointers was low in the specific cough-all causes group (Table 2). Only two children in the resolved spontaneously group had a specific cough pointer, whereas at least one cough pointer was present in all children in the specific cough-all causes group. For the diagnosis of PBB, ORs for wet cough and abnormal chest radiograph findings were significant. For asthma, the presence of wheeze or reversible airway obstruction was significant. For bronchiectasis, wet cough, feeding difficulties, recurrent PBB, recurrent pneumonia, and chest radiographic abnormality were
journal.publications.chestnet.org
Although the absence of other individual cough pointers did not change the pretest probability very much (negative LRs close to 1), the absence of all the pointers had a negative LR of 0 (95% CI, 0-0.03). In the current study, the presence of any one of the specific cough pointers was a highly sensitive measure (sensitivity, 100%). The absence of all cough pointers ruled out the possibility of specific cough with a negative predictive value of 100% (one-sided 97.5% CI, 91%-100%) (Table 3).
Discussion To our knowledge, this study is the first to examine the value of cough pointers from a multicenter study in differentiating children with specific cough (where treatment and further investigations are required) from those whose cough resolved without medications (spontaneous resolution). We developed categoryorientated LRs for specific cough for the key symptoms, signs, and simple investigations recommended in the major pediatric chronic cough guidelines.1-3 We found that all the individual cough pointers had high or infinite positive LRs, but none had strongly negative LRs. When combined with a high pretest probability (present in the current cohort), this means that individual cough pointers can be used to identify children at risk for specific cough. However, only the absence of all the cough pointers considered together reduced the risk of specific cough to almost zero. In this study, none of the 286 children who eventually received a diagnosis of specific cough had an absence of all the specific cough pointers at their initial assessment, meaning that the negative predictive value was 100% (one-sided 97.5% CI, 91%-100%). The importance of defining LRs for symptoms, signs, and simple investigations has been outlined by others in light of current standards for evidence-based practice.12,16,19 High positive LRs (. 10) or low negative LRs (, 0.1) are sometimes accepted as conclusive.12
747
748 Original Research
[
147#3 CHEST MARCH 2015
]
33 (11.5)
0
7 (17.9) 4 (10.3) 1 (4.2)
10-15
15-20
. 20
0
Spirometry abnormality
8 (2.8)
58 (20.5)
P Value
.023
.484
.001b
.185
.323
.271
.607
.146
b
.970
.660
.0001b
.196
.206
.012b
2 (1.5)
31 (22.6)
76.3 (64.3-86.4)
3.7 (2.7-4.9)
3.0 (2.0-3.0)
5
16 (11.3)
15 (11.3)
33 (24.8)
37 (27.8)
32 (24.1)
9 (6.7)
67 (49.6)
5 (9.6)
9 (17.3)
72.8 (62.5-85.1)
3.6 (2.8-4.9)
3.0 (2.0-4.0)
0
2 (3.8)
5 (9.6)
12 (23.1)
20 (38.5)
13 (25.0)
4 (7.8)
24 (47.1)
23 (45.1)
4 (7.3)
19 (13.8)
59 (43.7)
443 (87.8)
17 (32.7)
7 (45)
12 (6-32)
38 (14)
95 (70.4)
44 (31.1)
138 (0)
20 (10-40)
82 (56)
Asthma (n 5 52) 2.9 (1.4-5.8)
PBB (n 5 138) 2.4 (1.2-4.8)
1 (3.6)
10 (37.0)
72.5 (54.6-82.4)
4.3 (2.4-5.5)
3.0 (1.0-4.0)
2
1 (3.7)
4 (14.8)
5 (18.5)
8 (29.6)
9 (33.3)
10 (34.3)
11 (37.9)
8 (27.5)
9 (31)
20 (71.4)
10 (34.5)
27 (2)
27 (7.3-52.00)
17 (12)
3.9 (2.2-6.4)
Bronchiectasis (n 5 29)
Details of the Three Most Common Etiologies
Data are presented as median (interquartile range), No., or No. (%). GP 5 general practitioner; PBB 5 protracted bacterial bronchitis; PC-QOL 5 parent cough-specific quality of life; PedsQL 5 Pediatric Quality of Life Inventory. aBecause the cough in this group of children resolved without any specific treatment, it is likely that these children had postviral cough. bSignificance at P ⱕ .05. cEight children were missing data.
0
PedsQL18
Chest radiograph abnormality
3.8 (2.7-5.0)
4.3 (2.7-5.5) 82.3 (65.3-89.3)
PC-QOL17
76 (64.3-85.8)
3.0 (2.0-4.0)
3.0 (2.0-3.0)
9
24 (8.7)
31 (11.2)
60 (21.7)
91 (32.9)
71 (25.2)
26 (8.2)
139 (49.6)
115 (41.1)
Cough score15
1
14 (35.9)
5-10
Data not filled
13 (33.3)
,5
No. physician visits in 12 mo for coughing illness
0
21 (52.1)
GP-initiated
Specialist-initiated
17 (37.5)
Requested
Referral typec
Indigenous
207 (75.3)
30 (75)
Asthma medications used
90 (31.7)
187 (99)
1 (39)
16 (8.00-36.0)
173 (113)
11 (28.2)
12 (7.0-24.0)
Duration of cough, wks
Wet (dry) cough
20 (20)
Male (female) sex
3.1 (1.3-6.3)
Specific Cough-All Causes (n 5 286)
Smoke exposure
5.2 (2.1-8.2)
Age, y
Resolved Without Medications (ie, Spontaneously Resolved)a (n 5 40)
] Comparison Between Children Whose Cough Spontaneously Resolved With All Other Diagnosed Conditions at the First Appointment
Demographics (at First Visit)
TABLE 1
journal.publications.chestnet.org
749
0 0 0
Recurrent pneumonia
Chest radiograph abnormal
Spirometry abnormal
8
58
10
1
5
21
16
a
2.4 (0.14-42.7)
20.3 (1.2-335.4)a
3.0 (0.2-52.4)
0.4 (0.02-10.4)
0.7 (0.08-6.1)
1.8 (0.1-32.4)
NA
1.0 (0.05-19.3)
3.0 (0.2-52.4)
6.4 (0.4-108.1)
4.8 (0.3-82.2)
0.4 (0.02-10.4)
2.4 (0.14-42.7)
2.7 (0.15-47.6)
21.2 (1.3-349.8)a
73.7 (10-544.2)
OR (95% CI)
11.4 (0.7-195.2)
17
1.5 (0.07-31.0)
23.4 (1.4-391.0)a
31 2
2.7 (0.1-50.5)
NA
2.7 (0.1-50.5)
1.5 (0.07-31.0)
NA
NA
NA
7.9 (0.5-135.9)
4.5 (0.3-81.3)
NA
2.7 (0.1-50.5)
3.3 (0.2-60.6)
4
0
4
2
0
0
0
12
7
0
4
5
(Infinite)
138 a
OR (95% CI)
No.
PBB (n 5 138)
5
9
0
1
0
0
0
0
0
0
0
1
0
3
41
7
No.
9.3 (0.5-174.3)
17.7 (1.0-313.5)a
NA
2.3 (0.1-59.2)
NA
NA
NA
NA
NA
NA
NA
2.3 (0.1-59.2)
NA
5.7 (0.3-113.7)
312.2 (17.7-5,509)a
6.1 (0.7-51.5)
OR (95% CI)
Asthma (n 5 52)
1
10
6
0
1
3
0
2
2
5
6
0
2
1
0
27
No.
4.3 (0.2-109.7)
41.6 (2.3-750)a
22.8 (1.2-424.3)a
NA
4.3 (0.2-109.7)
10.8 (0.5-218.9)
NA
7.5 (0.3-161.5)
7.5 (0.3-161.5)
18.4 (1.0-349.7)a
22.8 (1.2-424.3)a
NA
7.5 (0.3-161.5)
4.3 (0.2-109.7)
NA
527 (45.4-6,102)a
OR (95% CI)
Bronchiectasis (n 5 29)
Diagnostic groups are primary diagnosis. Some children within a diagnostic group had more than one diagnosed condition. ORs of all conditions were compared with the resolved spontaneously group. The nonsignificance of other pointers are likely related to low frequency of the pointer and the small sample size. NA 5 not applicable. See Table 1 legend for expansion of other abbreviation. aStatistically significant findings.
0
Chest pain
6
0
0 0
Digital clubbing
Suspected medications
1
0
Wet cough not resolved after 4 wk
Crepitations present
3 0
0
Recurrent PBB
Hyperinflated or pectus carinatum
10
0
Feeding difficulties
8 1
0
9
0
0
Exertional dyspnea
60
Differential airway sounds
0
Wheeze or reversible airway obstruction
187
No.
Failure to thrive
1
Wet cough
No. Resolved Without Medications (ie, Spontaneously Resolved) (n 5 40)
Specific Cough-All Causes (n 5 286)
] OR of Symptoms, Signs, and Chest Radiograph and Spirometry Abnormalities of the Various Groups Compared With Spontaneous Resolution
Specific Pointers (at Initial Visit)
TABLE 2
TABLE 3
] Sensitivity, Speci¿city, and Negative and Positive LRs for the Symptoms, Signs, and Chest Radiograph and Spirometry Abnormalities for the Speci¿c Cough-All Causes Group
Cough Pointers (at Initial Visit)
Specific Cough-All Causes (N 5 286)
Wet cough
TABLE 3
] (continued)
Cough Pointers (at Initial Visit) Negative PV
Specific Cough-All Causes (N 5 286) 0.13 (0.09-0.17)
Positive LR
Infinity
Negative LR
0.94 (0.92-0.97)
Recurrent PBB
Sensitivity
0.65 (0.60-0.71)
Sensitivity
0.07 (0.05-0.11)
Specificity
0.98 (0.85-1.0)
Specificity
1.0 (0.89-1.0)
Positive PV
0.99 (0.97-1.0)
Positive PV
Negative PV
0.28 (0.21-0.37)
Negative PV
Positive LR Negative LR
26.15 (3.77-181.48) 0.36 (0.30-0.42)
Wheeze or reversible airway obstruction
1 (0.81-1.0) 0.13 (0.09-0.18)
Positive LR
Infinity
Negative LR
0.93 (0.90-0.96)
Wet cough not resolved after 4 wk
Sensitivity
0.21 (0.17-0.26)
Sensitivity
0.03 (0.02-0.07)
Specificity
1.0 (0.89-1.0)
Specificity
1.0 (0.89-1.0)
Positive PV
1.0 (0.92-1.0)
Positive PV
1.0 (0.65-1.0)
Negative PV
0.15 (0.11-0.20)
Negative PV
0.13 (0.09-0.16)
Positive LR Negative LR
Infinity 0.79 (0.74-0.84)
Exertional dyspnea Sensitivity
Positive LR
Infinity
Negative LR
0.97 (0.94-0.99)
Digital clubbing 0.03 (0.02-0.06)
Sensitivity
0.01 (0.00-0.03)
Specificity
1.0 (0.9-1.0)
Specificity
1.0 (0.89-1.0)
Positive PV
1.0 (0.63-1.0)
Positive PV
1.0 (0.31-1.0)
Negative PV
0.1 (0.1-0.2)
Negative PV
0.12 (0.09-0.17)
Positive LR Negative LR
Infinity 1.0 (0.9-1.0)
Differential airway sounds
Positive LR
Infinity
Negative LR
0.99 (0.98-1.0)
Crepitations present
Sensitivity
0.03 (0.01-0.06)
Sensitivity
0.02 (0.01-0.05)
Specificity
1.0 (0.89-1.0)
Specificity
1.0 (0.89-1.0)
Positive PV
1.0 (0.60-1.0)
Positive PV
1.0 (0.52-1.0)
Negative PV
0.13 (0.09-0.17)
Negative PV
0.13 (0.09-0.17)
Positive LR Negative LR
Infinity 0.97 (0.95-0.99)
Failure to thrive
Positive LR
Infinity
Negative LR
0.98 (0.96-0.99)
Hyperinflated or pectus carinatum
Sensitivity
0 (0.0-0.02)
Specificity
1.0 (0.89-1.0)
Sensitivity
0.02 (0.00-0.04)
Positive PV
1.0 (0.05-1.0)
Specificity
0.98 (0.85-1.0)
0.13 (0.09-0.17)
Positive PV
0.83 (0.36-0.99)
Negative PV
0.12 (0.09-0.16)
Positive LR
0.70 (0.08-5.83)
Negative PV Positive LR Negative LR
Infinity 1.0 (0.99-1.00)
Negative LR
Feeding difficulties
1.0 (0.99-1.02)
Chest pain
Sensitivity
0.06 (0.03-0.09)
Specificity
1.0 (0.89-1.0)
Sensitivity
0 (0.0-0.02)
1 (0.76-1.0)
Specificity
1.0 (0.89-1.0)
Positive PV
1.0 (0.05-1.0)
Positive PV
(Continued)
(Continued)
750 Original Research
[
147#3 CHEST MARCH 2015
]
TABLE 3
] (continued)
Cough Pointers (at Initial Visit) Negative PV
Specific Cough-All Causes (N 5 286) 0.13 (0.09-0.17)
Positive LR
Infinity
Negative LR
1.0 (0.99-1.00)
Recurrent pneumonia Sensitivity
0.03 (0.02-0.07)
Specificity
1.0 (0.89-1.0)
Positive PV
1.0 (0.66-1.0)
Negative PV
0.12 (0.09-0.17)
Positive LR Negative LR
Infinity 0.96 (0.94-0.99)
Abnormal chest radiograph Sensitivity
0.20 (0.16-0.26)
Specificity
1.0 (0.89-1.0)
Positive PV Negative PV
1.0 (0.92-1.0) 0.15 (0.11-0.2)
Positive LR
Infinity
Negative LR
0.8 (0.75-0.85)
Abnormal spirometry Sensitivity
0.03 (0.01-0.06)
Specificity
1.0 (0.89-1.0)
Positive PV
1.0 (0.60-1.0)
Negative PV
0.13 (0.09-0.17)
Positive LR Negative LR
Infinity 0.97 (0.95-0.99)
Any specific cough pointer Sensitivity
1.0 (0.98-1.0)
Specificity
0.95 (0.82-0.99)
Positive PV
0.99 (0.97-1.0)
Negative PV Positive LR Negative LR
1.0 (0.89-1.0) 20 (5.18-77.21) 0 (0-0.03a)
Data are presented with 95% CI. All factors were compared with the resolved spontaneously group as the reference. LR 5 likelihood ratio; PV 5 predictive value. See Table 1 legend for expansion of other abbreviation. aThe upper bound of the 95% CI was calculated using the lower bound of the 95% CI of the sensitivity.
More extreme LRs result in major shifts between pretest and posttest probability of disease. High positive LRs can rule in disease, whereas a highly negative LR will rule out disease. Use of LRs is increasingly advocated in clinical medicine because LRs can be translated to individual patients in contrast to traditional diagnostic tests (sensitivity, specificity, and negative and positive predictive values) that reflect population characteristics20
journal.publications.chestnet.org
(e-Appendix 1). The current findings, the first to systematically examine these pointers from a multicenter study, substantially contribute to the robustness of evidence in the management of pediatric chronic cough. Additionally, with correct interpretation, the provision of LRs potentially helps clinicians to identify children with chronic cough who can be safely watched and those who would benefit from treatment and further investigations. Because clinicians use these pointers (e-Table 1, Tables 2, 3) to help with determining the etiology of chronic cough, it is not surprising that the positive LRs were high. For PBB and asthma, the diagnosis and determination of etiology is imperfect. However, without a gold standard for comparison, it is not possible to avoid a circular argument. Nevertheless, the novelty and appropriateness of the data for describing LRs within the current sampling frame are several fold. First, the multicenter design of the study adds precision, reduces selection bias, and increases generalizability. Second, the multicenter study4 that provided the dataset for this article had research rigor as outlined in the introduction. Ascribing etiologies for the cough has an inherent high risk of bias related to the placebo and period effects that may be seen in children with chronic cough.14 Studies on cough can limit this risk of bias by limiting the time frame where response to treatment is considered.14 In the randomized controlled trial8 that formed most of the dataset for this study, all but two children (1.65%) in the delayed arm were still coughing at the 2-week point after study recruitment. Because we used a 2-week time frame for the temporal relationship associating treatment with diagnosis, the likelihood of spontaneous resolution (as opposed to treatment effect) is small. This is supported by all the children having been coughing continuously for 3 to 4 months at enrollment (median, 18 weeks; interquartile range, 10-42 weeks). The 2-week time frame was based on systematic reviews of interventions for chronic cough in children.1,3 Third, although the pointers were used to aid in the diagnosis, this does not necessarily result in the final diagnosis unless the cough resolved with treatment specific for that etiology, or there was objective confirmation of the etiology (eg, cystic fibrosis). Pointers were analyzed as a dichotomous (present/ absent) rather than as a polytomous variable because these clinical findings are not independent.12 Based on the LRs (Table 3), wet cough as one of the discriminatory points in a flowchart8 is logical. We had previously described that a dry cough (compared with wet cough) is significantly more likely to resolve spontaneously.21,22
751
This study confirms that spontaneous resolution of chronic cough in young children is most likely to occur when the cough is dry and there are no associated specific cough pointers. Our previous study22 was small (N 5 100) and limited to a single center, and the approach was nonstandardized. In the current study, the finding of the importance of wet cough is even more striking. Chronic wet cough in children signifies the presence of airway secretions23 that are usually infected and associated with airway neutrophilia.24,25 Children with a chronic wet cough that responds to antibiotic treatment within a 2-week time frame have PBB.1 Moreover, when the cough does not respond to 4 weeks of antibiotics, the odds of having bronchiectasis is significantly increased (adjusted OR, 5.9; 95% CI, 1.2-28.5).26 Despite the novelty of data presented here, some controversies and study limitations require discussion. The definition of the duration of chronic cough is controversial: Some guidelines (eg, Australian,1 United States,3 Chinese,27 Belgian28) use . 4 weeks, whereas those of the British Thoracic Society2 use . 8 weeks. The guidelines1,3 that use the shorter time frame do not advocate use of medications for all children at that time point, recognizing that the cough settles without any specific treatment in a number of children (ie, assumed to resolve spontaneously). The shorter time frame is recommended for reasons outlined in previous publications.1,4 One such reason is to ensure that all children with chronic cough are carefully assessed and not quickly dismissed for a postviral cough. This is particularly important in children because chronic cough may be due to a serious underlying condition, and earlier diagnosis and treatment result in less damage (eg, retained foreign body,5 bronchiectasis6). Indeed, a serious potentially progressive underlying respiratory illness (bronchiectasis, aspiration lung disease, or cystic fibrosis) was documented in 18% of the children in our multicenter study.4 A study by Karakoç et al5 in 174 children with foreign body aspiration in the airways noted that “the risk of long-term complications increased with increasing elapsed time from aspiration to diagnosis,” and 60% of those receiving a diagnosis 30 days postaspiration had complications. Additionally, although chronic cough may seem unimportant to physicians, we believe that a 4-week reduction (the difference between 4-week vs 8-week cutoffs) in the duration of cough8 is clinically important in light of the associated stress,7 impaired generic health-related and cough-specific QoL,4,8 and recurrent physician visits endured by parents of children with chronic cough.4,7 Our earlier study found that at the time of first consulta752 Original Research
tion for the child’s chronic cough, the low PedsQL (a generic health-related QoL)17 score (reflecting poorer QoL) is in the realm of children with cardiac disease, diabetes, and chronic GI conditions.4 When the child’s chronic cough resolves, scores reflecting stress and QoL significantly improve to known population norms.7,8 The watchful waiting approach used in the multicenter study4 that formed the dataset for the current study includes addressing parental concerns, expectations, exposure to tobacco smoke, and the child’s activities.8 Addressing these issues may have contributed to the spontaneous resolution of the chronic cough in addition to the period effect. In regular clinical practice, the comparative control group used for this study (ie, resolved without medications) may be different, and the LR values may change. However, until there are better data available, we believe that the same approach used in this study can be incorporated into usual clinical practice in an outpatient setting. The dataset used for this study was largely based on a hospital cohort of young children, and thus, the LRs calculated may change in primary care settings and in older children. In primary care, abnormal chest signs like wheeze and crepitations often are unreliable in young children (k 5 0.39 for agreement among primary care physicians; 95% CI, 0.26-0.5329). However, wet cough in young children is reliably reported by most parents (k 5 0.75 between parents and clinicians; 95% CI, 0.58-0.9323), easily learned by medical students (k 5 0.86 between one medical student and a senior clinician; 95% CI, 0.68-1.0030), and has excellent intrarater (k 5 1.0) and interrater agreement between physicians (k 5 0.88; 95% CI, 0.82-0.9423). Older children (aged . 12 years) are more likely to be similar to adults where a larger amount of airway secretions need to be present for the cough to sound wet, and hence, the concept of dry and wet cough is less valid. In summary, we describe LRs for cough pointers signifying specific cough based on a multicenter study involving 326 children. The study has provided evidence for the use of cough pointers in pediatric chronic cough guidelines. Young children with chronic dry cough without any other cough pointer can be safely managed initially using the watchful waiting approach. Children with any cough pointers present are at high risk of specific cough and should be tested and treated. Treatments are based on the cough etiology and aim to prevent possible ongoing irreversible lung damage, improve generic health and cough-specific QoL,8 and reduce parent stress.7
[
147#3 CHEST MARCH 2015
]
Acknowledgments Author contributions: A. B. C. had full access to all of the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis. A. B. C. contributed to the study concept and design, study coordination, drafting and preparation of the manuscript, and critical review of the manuscript; P. P. V. A., C. F. R., I. B. M., and P. S. M. contributed to the design of the multicenter study,4 data collection and analysis, data interpretation, and critical review of the manuscript; N. G., C. M. M., and L. I. L. contributed to the design of the multicenter study,4 data interpretation, and critical review of the manuscript; and L. T., I. T., and H. L. P. contributed to the data collection and interpretation and review of the manuscript. Financial/nonfinancial disclosures: The authors have reported to CHEST that no potential conflicts of interest exist with any companies/organizations whose products or services may be discussed in this article. Role of sponsors: The views expressed in this article are those of the authors and do not reflect the views of the Australian National Health and Medical Research Council. The sponsors had no role in the design of the study, the collection and analysis of the data, or the preparation of the manuscript. Other contributions: The authors thank the parents and children who participated in this study and the research nurses and coordinators for facilitating the study: E. Bailey, B Nursing; S. Anderson-James, B Nursing; J. Kappers, B Nursing; M. Larkin, B Nursing; K. McKay, PhD; T. Gonzalez, B Nursing; G. McCallum, MPH; A. Revell, B Nursing; and K. Schulz, MPH. They also thank P. Francis, MD, and H. Buntain, PhD, for their assistance in recruiting the children into the study. Additional information: The e-Appendix and e-Table can be found in the Supplemental Materials section of the online article.
References 1. Chang AB, Landau LI, Van Asperen PP, et al; Thoracic Society of Australia and New Zealand. Cough in children: definitions and clinical evaluation. Med J Aust. 2006;184(8):398-403. 2. Shields MD, Bush A, Everard ML, et al. BTS guidelines recommendations for the assessment and management of cough in children. Thorax. 2008;63(suppl 3): iii1-iii15. 3. Chang AB, Glomb WB. Guidelines for evaluating chronic cough in pediatrics: ACCP evidence-based clinical practice guidelines. Chest. 2006;129(1_suppl): 260S-283S.
journal.publications.chestnet.org
4. Chang AB, Robertson CF, Van Asperen PP, et al. A multicenter study on chronic cough in children: burden and etiologies based on a standardized management pathway. Chest. 2012;142(4): 943-950. 5. Karakoç F, Karadağ B, Akbenlioğlu C, et al. Foreign body aspiration: what is the outcome? Pediatr Pulmonol. 2002;34(1): 30-36. 6. Bastardo CM, Sonnappa S, Stanojevic S, et al. Non-cystic fibrosis bronchiectasis in childhood: longitudinal growth and lung function. Thorax. 2009;64(3):246-251. 7. Marchant JM, Newcombe PA, Juniper EF, Sheffield JK, Stathis SL, Chang AB. What is the burden of chronic cough for families? Chest. 2008;134(2):303-309. 8. Chang AB, Robertson CF, van Asperen PP et al. A cough algorithm for chronic cough in children: a multicenter, randomized controlled study. Pediatrics. 2013;131(5):e1576-e1583. 9. Dart RC, Paul IM, Bond GR, et al. Pediatric fatalities associated with over the counter (nonprescription) cough and cold medications. Ann Emerg Med. 2009;53(4):411-417. 10. Thomson F, Masters IB, Chang AB. Persistent cough in children and the overuse of medications. J Paediatr Child Health. 2002;38(6):578-581. 11. Greenhalgh T, Howick J, Maskrey N; Evidence Based Medicine Renaissance Group. Evidence based medicine: a movement in crisis? BMJ. 2014;348:g3725. 12. Moosapour H, Raza M, Rambod M, Soltani A. Conceptualization of categoryoriented likelihood ratio: a useful tool for clinical diagnostic reasoning. BMC Med Educ. 2011;11:94. 13. Chang AB, Robertson CF, van Asperen PP, et al. Can a management pathway for chronic cough in children improve clinical outcomes: protocol for a multicentre evaluation. Trials. 2010;11(1):103. 14. Chang AB. Therapy for cough: where does it fall short? Expert Rev Respir Med. 2011;5(4):503-513. 15. Chang AB, Newman RG, Carlin JB, Phelan PD, Robertson CF. Subjective scoring of cough in children: parentcompleted vs child-completed diary cards vs an objective method. Eur Respir J. 1998;11(2):462-466. 16. Van den Bruel A, Thompson MJ, HajHassan T, et al. Diagnostic value of laboratory tests in identifying serious infections in febrile children: systematic review. BMJ. 2011;342:d3082. 17. Newcombe PA, Sheffield JK, Juniper EF, Petsky HL, Willis C, Chang AB. Validation of a parent-proxy quality of life questionnaire for paediatric chronic cough (PC-QOL). Thorax. 2010;65(9):819-823.
18. Varni JW, Seid M, Kurtin PS. PedsQL 4.0: reliability and validity of the Pediatric Quality of Life Inventory version 4.0 generic core scales in healthy and patient populations. Med Care. 2001;39(8): 800-812. 19. Grimes DA, Schulz KF. Refining clinical diagnosis with likelihood ratios. Lancet. 2005;365(9469):1500-1505. 20. Attia J. Moving beyond sensitivity and specificity: using likelihood ratios to help interpret diagnostic tests. Australian Prescriber. 2003;26(5):111-113. 21. Marchant JM, Masters IB, Taylor SM, Cox NC, Seymour GJ, Chang AB. Evaluation and outcome of young children with chronic cough. Chest. 2006;129(5):1132-1141. 22. Marchant JM, Masters IB, Taylor SM, Chang AB. Utility of signs and symptoms of chronic cough in predicting specific cause in children. Thorax. 2006;61(8): 694-698. 23. Chang AB, Gaffney JT, Eastburn MM, Faoagali J, Cox NC, Masters IB. Cough quality in children: a comparison of subjective vs bronchoscopic findings. Respir Res. 2005;6:3. 24. Chang AB, Faoagali J, Cox NC, et al. A bronchoscopic scoring system for airway secretions—airway cellularity and microbiological validation. Pediatr Pulmonol. 2006;41(9):887-892. 25. Wurzel DF, Marchant JM, Clark JE, et al. Wet cough in children: infective and inflammatory characteristics in bronchoalveolar lavage fluid. Pediatr Pulmonol. 2014;49(6):561-568. 26. Goyal V, Grimwood K, Marchant J, Masters IB, Chang AB. Does failed chronic wet cough response to antibiotics predict bronchiectasis? Arch Dis Child. 2014;99(6):522-525. 27. Clinical Research Coordination Group of Chronic Cough; The Subspecialty Group of Respiratory Diseases; The Society of Pediatrics; Chinese Medical Association; Editorial Board, Chinese Journal of Pediatrics. Guideline for diagnosis and treatment of chronic cough in Chinese children [in Chinese]. Zhonghua Er Ke Za Zhi. 2014;52(3):184-188. 28. Leconte S, Paulus D, Degryse J. Prolonged cough in children: a summary of the Belgian primary care clinical guideline. Prim Care Respir J. 2008;17(4):206-211. 29. Hay AD, Wilson A, Fahey T, Peters TJ. The inter-observer agreement of examining pre-school children with acute cough: a nested study. BMC Fam Pract. 2004;5(1):4. 30. Morey MJ, Cheng AC, McCallum GB, Chang AB. Accuracy of cough reporting by carers of Indigenous children. J Paediatr Child Health. 2013;49(3): E199-E203.
753
Original Research Signs and Symptoms of Chest Diseases
]
Children With Chronic Cough When Is Watchful Waiting Appropriate? Development of Likelihood Ratios for Assessing Children With Chronic Cough Anne B. Chang, PhD; Peter P. Van Asperen, MD; Nicholas Glasgow, MD; Colin F. Robertson, MD; Craig M. Mellis, MD; I. Brent Masters, PhD; Louis I. Landau, MD; Laurel Teoh, MD; Irene Tjhung, MD; Helen L. Petsky, PhD; and Peter S. Morris, PhD
Chronic cough is associated with poor quality of life and may signify a serious underlying disease. Differentiating nonspecific cough (when watchful waiting can be safely undertaken) from specific cough (treatment and further investigations are beneficial) would be clinically useful. In 326 children, we aimed to (1) determine how well cough pointers (used in guidelines) differentiate specific from nonspecific cough and (2) describe the clinical profile of children whose cough resolved without medications (spontaneous resolution).
BACKGROUND:
A dataset from a multicenter study involving children newly referred for chronic cough (median duration, 3-4 months) was used to determine the sensitivity, specificity, predictive values, and likelihood ratios (LRs) of cough pointers (symptoms, signs, and simple investigations [chest radiography, spirometry]) recommended in guidelines.
METHODS:
The pretest probability of specific cough was 88%. The absence of false-positive results meant that most pointers had strongly positive LRs. The most sensitive pointer (wet cough) had a positive LR of 26.2 (95% CI, 3.8-181.5). Although the absence of other individual pointers did not change the pretest probability much (negative LR close to 1), the absence of all pointers had a strongly negative LR of 0 (95% CI, 0-0.03). Children in the resolved spontaneously group were significantly more likely to be older, to be non-Indigenous, and to have a dry cough and a normal chest radiograph.
RESULTS:
Children with chronic dry cough without any cough pointers can be safely managed using the watchful waiting approach. The high pretest probability and high positive LRs of cough pointers support the use of individual cough pointers to identify high risk of specific cough in pediatric chronic cough guidelines. CONCLUSIONS:
TRIAL REGISTRY: Australian New Zealand Clinical Trials Registry; No.: 12607000526471; CHEST 2015; 147(3):745-753 URL: www.anzctr.org.au
Manuscript received August 31, 2014; revision accepted November 10, 2014; originally published Online First December 11, 2014. ABBREVIATIONS: LR 5 likelihood ratio; PBB 5 protracted bacterial bronchitis; QoL 5 quality of life AFFILIATIONS: From the Child Health Division (Drs Chang, Teoh, Tjhung, and Morris), Menzies School of Health Research, Charles Darwin University, Darwin, NT; Queensland Children’s Respiratory Centre (Drs Chang, Masters, and Petsky), Queensland Children’s Medical Research Institute, Royal Children’s Hospital, Brisbane, QLD; Discipline of Paediatrics and Child Health (Dr Van Asperen), Sydney Medical School, and Central Clinical School (Dr Mellis), University of Sydney, Sydney, NSW; Department of Respiratory Medicine (Dr Van Asperen), The Children’s Hospital at Westmead, Sydney Children’s Hospital Network,
journal.publications.chestnet.org
Westmead, NSW; Medical School (Dr Glasgow), Australian National University, Canberra, ACT; Department of Respiratory Medicine (Dr Robertson), Murdoch Children’s Research Institute, Royal Children’s Hospital, Melbourne, VIC; Medical Workforce (Dr Landau), Health Department of Western Australia and The University of Western Australia, Perth, WA; The Canberra Hospital (Dr Teoh), Canberra, ACT; and Thursday Island Primary Health Care Centre (Dr Tjhung), Thursday Island, QLD, Australia. FUNDING/SUPPORT: The study was funded by an Australian National Health and Medical Research Council (NHMRC) [Project Grant 490321] and supported by an NHMRC Centre for Research Excellence in Lung Health of Aboriginal and Torres Strait Islander Children [Grant 1040830].
745
Chronic cough is a common reason why parents seek a consultation in pediatric practice. In the management of chronic cough in children, identification of the underlying cause (ie, obtaining the correct diagnosis), as opposed to empirical treatment, is considered important.1-3 Several factors underpin the importance of obtaining the correct diagnosis with consequent cough resolution upon appropriate treatment. First, chronic cough may signify the presence of a serious underlying airways disease.4,5 Early diagnosis of these conditions (eg, bronchiectasis, aspiration lung disease) would lead to earlier treatment and prevent further ongoing lung injury.5,6 Second, parents of children with chronic cough have substantial stress and impaired quality of life (QoL) that improves when the cough abates.4,7,8 Third, many parents seek repeated consultations for their child’s cough until the cough resolves. Our single-center7 and multicenter studies4 showed that 75% to 80% of children had more than five consultations prior to presentation to a respiratory specialist for cough. Earlier resolution of cough through identification and treatment of the etiology would reduce the economic burden associated with recurrent consultations. Finally, obtaining a correct diagnosis potentially reduces use of unnecessary investigations and medications with consequent severe adverse events.9,10 The initial step in obtaining a diagnosis requires differentiating those who require treatment (ie, specific cough) from those whose cough is likely to resolve without treatment. In the latter group, a watchful waiting approach has been suggested.1,3 To differentiate these groups, identification of cough pointers (symptoms, signs, and simple investigations [e-Table 1]) is used in
Materials and Methods Data collected from our multicenter cohort study 4 formed the dataset of this article. Briefly, 346 children aged , 18 years newly referred for chronic cough (duration, . 4 weeks) to any of the participating sites were enrolled. These sites were clinics at five major hospitals (Brisbane, Sydney, Melbourne, Canberra, Darwin) and three regional-remote clinics in Orange, New South Wales; Anangu-Pitjantjatjara Lands, Central Australia; and Thursday Island, Queensland. We excluded children with a known chronic respiratory illness previously diagnosed by a respiratory physician or those who have had diagnosed conditions confirmed by objective tests (eg, asthma, cystic fibrosis, bronchiectasis) prior to Dr Chang is supported by an NHMRC practitioner fellowship [Grant 1058213]. CORRESPONDENCE TO: Anne B. Chang, PhD, Queensland Children’s Respiratory Centre, Royal Children’s Hospital, Herston, QLD 4029, Australia; e-mail: [email protected] © 2015 AMERICAN COLLEGE OF CHEST PHYSICIANS. Reproduction of this article is prohibited without written permission from the American College of Chest Physicians. See online for more details. DOI: 10.1378/chest.14-2155
746 Original Research
many chronic cough guidelines (eg, Australian,1 British,2 United States3). However, published evidence is scarce on how well cough pointers differentiate specific from nonspecific cough and on the profile of children whose chronic cough is likely to resolve without treatment. Obtaining evidence for the utility of these clinical findings and investigations will increase the robustness of these guidelines necessary in current standards of evidencebased medicine.11 Additionally, when evaluating children with chronic cough, development of categoryorientated likelihood ratios (LRs)12 for cough pointers (e-Table 1) would help clinicians to differentiate children for whom watchful waiting can be safely undertaken from those who would benefit from treatment and further investigations. In this study involving 326 children newly referred for chronic cough,4 we evaluated data relating cough pointers to specific cough and the three most common etiologies from our recent multicenter study. In the field of chronic cough, our multicenter study4,13 had unique features of research rigor,14 including the following: (1) Rules for assigning etiology of cough were defined a priori; (2) the diagnoses, particularly those of protracted bacterial bronchitis (PBB) and asthma, were temporally related to treatment within a strict time frame; (3) validated cough outcome measures were used; and (4) the children were followed for 6 months to ensure that misdiagnosis had not occurred. We hypothesized that cough pointers are clinically valid (have high positive LRs) and can be used to differentiate specific cough from cough that resolves without treatment (spontaneous resolution).
referral. Children with asthma diagnosed by a nonrespiratory physician were not excluded because asthma is commonly misdiagnosed in our setting. Written informed consent was obtained from parents, and the study was approved by the ethics committees of all participating sites. Enrolled children were managed in accordance with a standardized evidence-based cough algorithm13 until the study’s end point, which was defined as either a primary diagnosis established with cough resolved or the presence of exit criteria (hospitalization related to etiology of the cough or 12 months postenrollment, whichever occurred earliest). The ascribed etiology was defined a priori and based on a temporal and prompt response to the treatment as defined by a validated cough score (verbal categorical descriptive score).15 Chest radiograph and spirometry findings were interpreted by the attending respiratory physician as a dichotomous variable (normal/abnormal). Statistics Because the data had a skewed distribution, medians and interquartile ranges were used for continuous measures. The x2 test was used for categorical data. Kruskal-Wallis analysis was used for group comparisons. SPSS software (IBM Corporation) was used, and a two-tailed P ⱕ .05 was considered significant. For calculation of the OR and 95% CI, where
[
147#3 CHEST MARCH 2015
]
cells are 0, 0.5 was added to each cell using Microsoft Excel. For the OR, the variable was considered significant when the lower limit of the 95% CI value approximated to ⱖ 1.0 (to one decimal point). Sensitivity, specificity, positive and negative predictive values, and positive and negative LRs for the presence of symptoms and signs (specific cough pointers) and for investigations (spirometry and chest radiography) were calculated using an online program (http://vassarstats.net/clin1. html). LRs are presented in light of current standards of assessing the
diagnostic values of symptoms and tests12,16 of the key cough pointers (e-Table 1) used in Australian,1 British,2 and US3 chronic cough guidelines for differentiating specific from nonspecific cough. These LRs (specific cough-all causes) serve as category-orientated LRs12 to assist clinicians in determining which children might safely be managed with watchful waiting (as opposed to those in whom specific treatment is recommended). Calculations were not performed for cough pointers absent in the entire cohort.
Results
significant (Table 2). For these three diagnoses, the specificity of all pointers was high (. 98%), but sensitivity was generally low other than wet cough for PBB and dry cough for asthma (data not shown).
Missing data precluded analyses in 20 children; thus, only 326 children (from the original 346) were included for this dataset. Of the 326 children, 40 did not receive any treatment specific for their cough, and their cough eventually resolved spontaneously with the watchful waiting13 approach (Table 1). This means that, in this cohort of children with chronic cough for an average of 3 to 4 months, the pretest probability of specific cough was 88%. Compared with children who received treatment specific for their cough (n 5 286) and fulfilled the a priori-defined criteria for specified etiologies, children in the resolved spontaneously group (n 5 40) were significantly older, were more likely to have a dry cough and to be non-Indigenous, and had normal chest radiograph findings (Table 1). There was no significant difference between the groups for smoke exposure, duration of cough, prior use of asthma medications, how the children were referred, number of physician visits, spirometry abnormality, Pediatric Quality of Life Inventory version 4.0 (PedsQL 4.0),18 and parent coughspecific QoL17 scores at baseline. Details of the three most common diagnoses (PBB, asthma, bronchiectasis) showed similar distribution, other than Indigenous children being the most likely to have bronchiectasis (Table 1). These data were previously analyzed statistically4 and are not repeated here. Comparing all-cause specific cough with cough resolving spontaneously, significant ORs were found for wet cough, wheeze or reversible airway obstruction, and chest radiograph abnormality (Table 2). However, the frequency of the other cough pointers was low in the specific cough-all causes group (Table 2). Only two children in the resolved spontaneously group had a specific cough pointer, whereas at least one cough pointer was present in all children in the specific cough-all causes group. For the diagnosis of PBB, ORs for wet cough and abnormal chest radiograph findings were significant. For asthma, the presence of wheeze or reversible airway obstruction was significant. For bronchiectasis, wet cough, feeding difficulties, recurrent PBB, recurrent pneumonia, and chest radiographic abnormality were
journal.publications.chestnet.org
Although the absence of other individual cough pointers did not change the pretest probability very much (negative LRs close to 1), the absence of all the pointers had a negative LR of 0 (95% CI, 0-0.03). In the current study, the presence of any one of the specific cough pointers was a highly sensitive measure (sensitivity, 100%). The absence of all cough pointers ruled out the possibility of specific cough with a negative predictive value of 100% (one-sided 97.5% CI, 91%-100%) (Table 3).
Discussion To our knowledge, this study is the first to examine the value of cough pointers from a multicenter study in differentiating children with specific cough (where treatment and further investigations are required) from those whose cough resolved without medications (spontaneous resolution). We developed categoryorientated LRs for specific cough for the key symptoms, signs, and simple investigations recommended in the major pediatric chronic cough guidelines.1-3 We found that all the individual cough pointers had high or infinite positive LRs, but none had strongly negative LRs. When combined with a high pretest probability (present in the current cohort), this means that individual cough pointers can be used to identify children at risk for specific cough. However, only the absence of all the cough pointers considered together reduced the risk of specific cough to almost zero. In this study, none of the 286 children who eventually received a diagnosis of specific cough had an absence of all the specific cough pointers at their initial assessment, meaning that the negative predictive value was 100% (one-sided 97.5% CI, 91%-100%). The importance of defining LRs for symptoms, signs, and simple investigations has been outlined by others in light of current standards for evidence-based practice.12,16,19 High positive LRs (. 10) or low negative LRs (, 0.1) are sometimes accepted as conclusive.12
747
748 Original Research
[
147#3 CHEST MARCH 2015
]
33 (11.5)
0
7 (17.9) 4 (10.3) 1 (4.2)
10-15
15-20
. 20
0
Spirometry abnormality
8 (2.8)
58 (20.5)
P Value
.023
.484
.001b
.185
.323
.271
.607
.146
b
.970
.660
.0001b
.196
.206
.012b
2 (1.5)
31 (22.6)
76.3 (64.3-86.4)
3.7 (2.7-4.9)
3.0 (2.0-3.0)
5
16 (11.3)
15 (11.3)
33 (24.8)
37 (27.8)
32 (24.1)
9 (6.7)
67 (49.6)
5 (9.6)
9 (17.3)
72.8 (62.5-85.1)
3.6 (2.8-4.9)
3.0 (2.0-4.0)
0
2 (3.8)
5 (9.6)
12 (23.1)
20 (38.5)
13 (25.0)
4 (7.8)
24 (47.1)
23 (45.1)
4 (7.3)
19 (13.8)
59 (43.7)
443 (87.8)
17 (32.7)
7 (45)
12 (6-32)
38 (14)
95 (70.4)
44 (31.1)
138 (0)
20 (10-40)
82 (56)
Asthma (n 5 52) 2.9 (1.4-5.8)
PBB (n 5 138) 2.4 (1.2-4.8)
1 (3.6)
10 (37.0)
72.5 (54.6-82.4)
4.3 (2.4-5.5)
3.0 (1.0-4.0)
2
1 (3.7)
4 (14.8)
5 (18.5)
8 (29.6)
9 (33.3)
10 (34.3)
11 (37.9)
8 (27.5)
9 (31)
20 (71.4)
10 (34.5)
27 (2)
27 (7.3-52.00)
17 (12)
3.9 (2.2-6.4)
Bronchiectasis (n 5 29)
Details of the Three Most Common Etiologies
Data are presented as median (interquartile range), No., or No. (%). GP 5 general practitioner; PBB 5 protracted bacterial bronchitis; PC-QOL 5 parent cough-specific quality of life; PedsQL 5 Pediatric Quality of Life Inventory. aBecause the cough in this group of children resolved without any specific treatment, it is likely that these children had postviral cough. bSignificance at P ⱕ .05. cEight children were missing data.
0
PedsQL18
Chest radiograph abnormality
3.8 (2.7-5.0)
4.3 (2.7-5.5) 82.3 (65.3-89.3)
PC-QOL17
76 (64.3-85.8)
3.0 (2.0-4.0)
3.0 (2.0-3.0)
9
24 (8.7)
31 (11.2)
60 (21.7)
91 (32.9)
71 (25.2)
26 (8.2)
139 (49.6)
115 (41.1)
Cough score15
1
14 (35.9)
5-10
Data not filled
13 (33.3)
,5
No. physician visits in 12 mo for coughing illness
0
21 (52.1)
GP-initiated
Specialist-initiated
17 (37.5)
Requested
Referral typec
Indigenous
207 (75.3)
30 (75)
Asthma medications used
90 (31.7)
187 (99)
1 (39)
16 (8.00-36.0)
173 (113)
11 (28.2)
12 (7.0-24.0)
Duration of cough, wks
Wet (dry) cough
20 (20)
Male (female) sex
3.1 (1.3-6.3)
Specific Cough-All Causes (n 5 286)
Smoke exposure
5.2 (2.1-8.2)
Age, y
Resolved Without Medications (ie, Spontaneously Resolved)a (n 5 40)
] Comparison Between Children Whose Cough Spontaneously Resolved With All Other Diagnosed Conditions at the First Appointment
Demographics (at First Visit)
TABLE 1
journal.publications.chestnet.org
749
0 0 0
Recurrent pneumonia
Chest radiograph abnormal
Spirometry abnormal
8
58
10
1
5
21
16
a
2.4 (0.14-42.7)
20.3 (1.2-335.4)a
3.0 (0.2-52.4)
0.4 (0.02-10.4)
0.7 (0.08-6.1)
1.8 (0.1-32.4)
NA
1.0 (0.05-19.3)
3.0 (0.2-52.4)
6.4 (0.4-108.1)
4.8 (0.3-82.2)
0.4 (0.02-10.4)
2.4 (0.14-42.7)
2.7 (0.15-47.6)
21.2 (1.3-349.8)a
73.7 (10-544.2)
OR (95% CI)
11.4 (0.7-195.2)
17
1.5 (0.07-31.0)
23.4 (1.4-391.0)a
31 2
2.7 (0.1-50.5)
NA
2.7 (0.1-50.5)
1.5 (0.07-31.0)
NA
NA
NA
7.9 (0.5-135.9)
4.5 (0.3-81.3)
NA
2.7 (0.1-50.5)
3.3 (0.2-60.6)
4
0
4
2
0
0
0
12
7
0
4
5
(Infinite)
138 a
OR (95% CI)
No.
PBB (n 5 138)
5
9
0
1
0
0
0
0
0
0
0
1
0
3
41
7
No.
9.3 (0.5-174.3)
17.7 (1.0-313.5)a
NA
2.3 (0.1-59.2)
NA
NA
NA
NA
NA
NA
NA
2.3 (0.1-59.2)
NA
5.7 (0.3-113.7)
312.2 (17.7-5,509)a
6.1 (0.7-51.5)
OR (95% CI)
Asthma (n 5 52)
1
10
6
0
1
3
0
2
2
5
6
0
2
1
0
27
No.
4.3 (0.2-109.7)
41.6 (2.3-750)a
22.8 (1.2-424.3)a
NA
4.3 (0.2-109.7)
10.8 (0.5-218.9)
NA
7.5 (0.3-161.5)
7.5 (0.3-161.5)
18.4 (1.0-349.7)a
22.8 (1.2-424.3)a
NA
7.5 (0.3-161.5)
4.3 (0.2-109.7)
NA
527 (45.4-6,102)a
OR (95% CI)
Bronchiectasis (n 5 29)
Diagnostic groups are primary diagnosis. Some children within a diagnostic group had more than one diagnosed condition. ORs of all conditions were compared with the resolved spontaneously group. The nonsignificance of other pointers are likely related to low frequency of the pointer and the small sample size. NA 5 not applicable. See Table 1 legend for expansion of other abbreviation. aStatistically significant findings.
0
Chest pain
6
0
0 0
Digital clubbing
Suspected medications
1
0
Wet cough not resolved after 4 wk
Crepitations present
3 0
0
Recurrent PBB
Hyperinflated or pectus carinatum
10
0
Feeding difficulties
8 1
0
9
0
0
Exertional dyspnea
60
Differential airway sounds
0
Wheeze or reversible airway obstruction
187
No.
Failure to thrive
1
Wet cough
No. Resolved Without Medications (ie, Spontaneously Resolved) (n 5 40)
Specific Cough-All Causes (n 5 286)
] OR of Symptoms, Signs, and Chest Radiograph and Spirometry Abnormalities of the Various Groups Compared With Spontaneous Resolution
Specific Pointers (at Initial Visit)
TABLE 2
TABLE 3
] Sensitivity, Speci¿city, and Negative and Positive LRs for the Symptoms, Signs, and Chest Radiograph and Spirometry Abnormalities for the Speci¿c Cough-All Causes Group
Cough Pointers (at Initial Visit)
Specific Cough-All Causes (N 5 286)
Wet cough
TABLE 3
] (continued)
Cough Pointers (at Initial Visit) Negative PV
Specific Cough-All Causes (N 5 286) 0.13 (0.09-0.17)
Positive LR
Infinity
Negative LR
0.94 (0.92-0.97)
Recurrent PBB
Sensitivity
0.65 (0.60-0.71)
Sensitivity
0.07 (0.05-0.11)
Specificity
0.98 (0.85-1.0)
Specificity
1.0 (0.89-1.0)
Positive PV
0.99 (0.97-1.0)
Positive PV
Negative PV
0.28 (0.21-0.37)
Negative PV
Positive LR Negative LR
26.15 (3.77-181.48) 0.36 (0.30-0.42)
Wheeze or reversible airway obstruction
1 (0.81-1.0) 0.13 (0.09-0.18)
Positive LR
Infinity
Negative LR
0.93 (0.90-0.96)
Wet cough not resolved after 4 wk
Sensitivity
0.21 (0.17-0.26)
Sensitivity
0.03 (0.02-0.07)
Specificity
1.0 (0.89-1.0)
Specificity
1.0 (0.89-1.0)
Positive PV
1.0 (0.92-1.0)
Positive PV
1.0 (0.65-1.0)
Negative PV
0.15 (0.11-0.20)
Negative PV
0.13 (0.09-0.16)
Positive LR Negative LR
Infinity 0.79 (0.74-0.84)
Exertional dyspnea Sensitivity
Positive LR
Infinity
Negative LR
0.97 (0.94-0.99)
Digital clubbing 0.03 (0.02-0.06)
Sensitivity
0.01 (0.00-0.03)
Specificity
1.0 (0.9-1.0)
Specificity
1.0 (0.89-1.0)
Positive PV
1.0 (0.63-1.0)
Positive PV
1.0 (0.31-1.0)
Negative PV
0.1 (0.1-0.2)
Negative PV
0.12 (0.09-0.17)
Positive LR Negative LR
Infinity 1.0 (0.9-1.0)
Differential airway sounds
Positive LR
Infinity
Negative LR
0.99 (0.98-1.0)
Crepitations present
Sensitivity
0.03 (0.01-0.06)
Sensitivity
0.02 (0.01-0.05)
Specificity
1.0 (0.89-1.0)
Specificity
1.0 (0.89-1.0)
Positive PV
1.0 (0.60-1.0)
Positive PV
1.0 (0.52-1.0)
Negative PV
0.13 (0.09-0.17)
Negative PV
0.13 (0.09-0.17)
Positive LR Negative LR
Infinity 0.97 (0.95-0.99)
Failure to thrive
Positive LR
Infinity
Negative LR
0.98 (0.96-0.99)
Hyperinflated or pectus carinatum
Sensitivity
0 (0.0-0.02)
Specificity
1.0 (0.89-1.0)
Sensitivity
0.02 (0.00-0.04)
Positive PV
1.0 (0.05-1.0)
Specificity
0.98 (0.85-1.0)
0.13 (0.09-0.17)
Positive PV
0.83 (0.36-0.99)
Negative PV
0.12 (0.09-0.16)
Positive LR
0.70 (0.08-5.83)
Negative PV Positive LR Negative LR
Infinity 1.0 (0.99-1.00)
Negative LR
Feeding difficulties
1.0 (0.99-1.02)
Chest pain
Sensitivity
0.06 (0.03-0.09)
Specificity
1.0 (0.89-1.0)
Sensitivity
0 (0.0-0.02)
1 (0.76-1.0)
Specificity
1.0 (0.89-1.0)
Positive PV
1.0 (0.05-1.0)
Positive PV
(Continued)
(Continued)
750 Original Research
[
147#3 CHEST MARCH 2015
]
TABLE 3
] (continued)
Cough Pointers (at Initial Visit) Negative PV
Specific Cough-All Causes (N 5 286) 0.13 (0.09-0.17)
Positive LR
Infinity
Negative LR
1.0 (0.99-1.00)
Recurrent pneumonia Sensitivity
0.03 (0.02-0.07)
Specificity
1.0 (0.89-1.0)
Positive PV
1.0 (0.66-1.0)
Negative PV
0.12 (0.09-0.17)
Positive LR Negative LR
Infinity 0.96 (0.94-0.99)
Abnormal chest radiograph Sensitivity
0.20 (0.16-0.26)
Specificity
1.0 (0.89-1.0)
Positive PV Negative PV
1.0 (0.92-1.0) 0.15 (0.11-0.2)
Positive LR
Infinity
Negative LR
0.8 (0.75-0.85)
Abnormal spirometry Sensitivity
0.03 (0.01-0.06)
Specificity
1.0 (0.89-1.0)
Positive PV
1.0 (0.60-1.0)
Negative PV
0.13 (0.09-0.17)
Positive LR Negative LR
Infinity 0.97 (0.95-0.99)
Any specific cough pointer Sensitivity
1.0 (0.98-1.0)
Specificity
0.95 (0.82-0.99)
Positive PV
0.99 (0.97-1.0)
Negative PV Positive LR Negative LR
1.0 (0.89-1.0) 20 (5.18-77.21) 0 (0-0.03a)
Data are presented with 95% CI. All factors were compared with the resolved spontaneously group as the reference. LR 5 likelihood ratio; PV 5 predictive value. See Table 1 legend for expansion of other abbreviation. aThe upper bound of the 95% CI was calculated using the lower bound of the 95% CI of the sensitivity.
More extreme LRs result in major shifts between pretest and posttest probability of disease. High positive LRs can rule in disease, whereas a highly negative LR will rule out disease. Use of LRs is increasingly advocated in clinical medicine because LRs can be translated to individual patients in contrast to traditional diagnostic tests (sensitivity, specificity, and negative and positive predictive values) that reflect population characteristics20
journal.publications.chestnet.org
(e-Appendix 1). The current findings, the first to systematically examine these pointers from a multicenter study, substantially contribute to the robustness of evidence in the management of pediatric chronic cough. Additionally, with correct interpretation, the provision of LRs potentially helps clinicians to identify children with chronic cough who can be safely watched and those who would benefit from treatment and further investigations. Because clinicians use these pointers (e-Table 1, Tables 2, 3) to help with determining the etiology of chronic cough, it is not surprising that the positive LRs were high. For PBB and asthma, the diagnosis and determination of etiology is imperfect. However, without a gold standard for comparison, it is not possible to avoid a circular argument. Nevertheless, the novelty and appropriateness of the data for describing LRs within the current sampling frame are several fold. First, the multicenter design of the study adds precision, reduces selection bias, and increases generalizability. Second, the multicenter study4 that provided the dataset for this article had research rigor as outlined in the introduction. Ascribing etiologies for the cough has an inherent high risk of bias related to the placebo and period effects that may be seen in children with chronic cough.14 Studies on cough can limit this risk of bias by limiting the time frame where response to treatment is considered.14 In the randomized controlled trial8 that formed most of the dataset for this study, all but two children (1.65%) in the delayed arm were still coughing at the 2-week point after study recruitment. Because we used a 2-week time frame for the temporal relationship associating treatment with diagnosis, the likelihood of spontaneous resolution (as opposed to treatment effect) is small. This is supported by all the children having been coughing continuously for 3 to 4 months at enrollment (median, 18 weeks; interquartile range, 10-42 weeks). The 2-week time frame was based on systematic reviews of interventions for chronic cough in children.1,3 Third, although the pointers were used to aid in the diagnosis, this does not necessarily result in the final diagnosis unless the cough resolved with treatment specific for that etiology, or there was objective confirmation of the etiology (eg, cystic fibrosis). Pointers were analyzed as a dichotomous (present/ absent) rather than as a polytomous variable because these clinical findings are not independent.12 Based on the LRs (Table 3), wet cough as one of the discriminatory points in a flowchart8 is logical. We had previously described that a dry cough (compared with wet cough) is significantly more likely to resolve spontaneously.21,22
751
This study confirms that spontaneous resolution of chronic cough in young children is most likely to occur when the cough is dry and there are no associated specific cough pointers. Our previous study22 was small (N 5 100) and limited to a single center, and the approach was nonstandardized. In the current study, the finding of the importance of wet cough is even more striking. Chronic wet cough in children signifies the presence of airway secretions23 that are usually infected and associated with airway neutrophilia.24,25 Children with a chronic wet cough that responds to antibiotic treatment within a 2-week time frame have PBB.1 Moreover, when the cough does not respond to 4 weeks of antibiotics, the odds of having bronchiectasis is significantly increased (adjusted OR, 5.9; 95% CI, 1.2-28.5).26 Despite the novelty of data presented here, some controversies and study limitations require discussion. The definition of the duration of chronic cough is controversial: Some guidelines (eg, Australian,1 United States,3 Chinese,27 Belgian28) use . 4 weeks, whereas those of the British Thoracic Society2 use . 8 weeks. The guidelines1,3 that use the shorter time frame do not advocate use of medications for all children at that time point, recognizing that the cough settles without any specific treatment in a number of children (ie, assumed to resolve spontaneously). The shorter time frame is recommended for reasons outlined in previous publications.1,4 One such reason is to ensure that all children with chronic cough are carefully assessed and not quickly dismissed for a postviral cough. This is particularly important in children because chronic cough may be due to a serious underlying condition, and earlier diagnosis and treatment result in less damage (eg, retained foreign body,5 bronchiectasis6). Indeed, a serious potentially progressive underlying respiratory illness (bronchiectasis, aspiration lung disease, or cystic fibrosis) was documented in 18% of the children in our multicenter study.4 A study by Karakoç et al5 in 174 children with foreign body aspiration in the airways noted that “the risk of long-term complications increased with increasing elapsed time from aspiration to diagnosis,” and 60% of those receiving a diagnosis 30 days postaspiration had complications. Additionally, although chronic cough may seem unimportant to physicians, we believe that a 4-week reduction (the difference between 4-week vs 8-week cutoffs) in the duration of cough8 is clinically important in light of the associated stress,7 impaired generic health-related and cough-specific QoL,4,8 and recurrent physician visits endured by parents of children with chronic cough.4,7 Our earlier study found that at the time of first consulta752 Original Research
tion for the child’s chronic cough, the low PedsQL (a generic health-related QoL)17 score (reflecting poorer QoL) is in the realm of children with cardiac disease, diabetes, and chronic GI conditions.4 When the child’s chronic cough resolves, scores reflecting stress and QoL significantly improve to known population norms.7,8 The watchful waiting approach used in the multicenter study4 that formed the dataset for the current study includes addressing parental concerns, expectations, exposure to tobacco smoke, and the child’s activities.8 Addressing these issues may have contributed to the spontaneous resolution of the chronic cough in addition to the period effect. In regular clinical practice, the comparative control group used for this study (ie, resolved without medications) may be different, and the LR values may change. However, until there are better data available, we believe that the same approach used in this study can be incorporated into usual clinical practice in an outpatient setting. The dataset used for this study was largely based on a hospital cohort of young children, and thus, the LRs calculated may change in primary care settings and in older children. In primary care, abnormal chest signs like wheeze and crepitations often are unreliable in young children (k 5 0.39 for agreement among primary care physicians; 95% CI, 0.26-0.5329). However, wet cough in young children is reliably reported by most parents (k 5 0.75 between parents and clinicians; 95% CI, 0.58-0.9323), easily learned by medical students (k 5 0.86 between one medical student and a senior clinician; 95% CI, 0.68-1.0030), and has excellent intrarater (k 5 1.0) and interrater agreement between physicians (k 5 0.88; 95% CI, 0.82-0.9423). Older children (aged . 12 years) are more likely to be similar to adults where a larger amount of airway secretions need to be present for the cough to sound wet, and hence, the concept of dry and wet cough is less valid. In summary, we describe LRs for cough pointers signifying specific cough based on a multicenter study involving 326 children. The study has provided evidence for the use of cough pointers in pediatric chronic cough guidelines. Young children with chronic dry cough without any other cough pointer can be safely managed initially using the watchful waiting approach. Children with any cough pointers present are at high risk of specific cough and should be tested and treated. Treatments are based on the cough etiology and aim to prevent possible ongoing irreversible lung damage, improve generic health and cough-specific QoL,8 and reduce parent stress.7
[
147#3 CHEST MARCH 2015
]
Acknowledgments Author contributions: A. B. C. had full access to all of the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis. A. B. C. contributed to the study concept and design, study coordination, drafting and preparation of the manuscript, and critical review of the manuscript; P. P. V. A., C. F. R., I. B. M., and P. S. M. contributed to the design of the multicenter study,4 data collection and analysis, data interpretation, and critical review of the manuscript; N. G., C. M. M., and L. I. L. contributed to the design of the multicenter study,4 data interpretation, and critical review of the manuscript; and L. T., I. T., and H. L. P. contributed to the data collection and interpretation and review of the manuscript. Financial/nonfinancial disclosures: The authors have reported to CHEST that no potential conflicts of interest exist with any companies/organizations whose products or services may be discussed in this article. Role of sponsors: The views expressed in this article are those of the authors and do not reflect the views of the Australian National Health and Medical Research Council. The sponsors had no role in the design of the study, the collection and analysis of the data, or the preparation of the manuscript. Other contributions: The authors thank the parents and children who participated in this study and the research nurses and coordinators for facilitating the study: E. Bailey, B Nursing; S. Anderson-James, B Nursing; J. Kappers, B Nursing; M. Larkin, B Nursing; K. McKay, PhD; T. Gonzalez, B Nursing; G. McCallum, MPH; A. Revell, B Nursing; and K. Schulz, MPH. They also thank P. Francis, MD, and H. Buntain, PhD, for their assistance in recruiting the children into the study. Additional information: The e-Appendix and e-Table can be found in the Supplemental Materials section of the online article.
References 1. Chang AB, Landau LI, Van Asperen PP, et al; Thoracic Society of Australia and New Zealand. Cough in children: definitions and clinical evaluation. Med J Aust. 2006;184(8):398-403. 2. Shields MD, Bush A, Everard ML, et al. BTS guidelines recommendations for the assessment and management of cough in children. Thorax. 2008;63(suppl 3): iii1-iii15. 3. Chang AB, Glomb WB. Guidelines for evaluating chronic cough in pediatrics: ACCP evidence-based clinical practice guidelines. Chest. 2006;129(1_suppl): 260S-283S.
journal.publications.chestnet.org
4. Chang AB, Robertson CF, Van Asperen PP, et al. A multicenter study on chronic cough in children: burden and etiologies based on a standardized management pathway. Chest. 2012;142(4): 943-950. 5. Karakoç F, Karadağ B, Akbenlioğlu C, et al. Foreign body aspiration: what is the outcome? Pediatr Pulmonol. 2002;34(1): 30-36. 6. Bastardo CM, Sonnappa S, Stanojevic S, et al. Non-cystic fibrosis bronchiectasis in childhood: longitudinal growth and lung function. Thorax. 2009;64(3):246-251. 7. Marchant JM, Newcombe PA, Juniper EF, Sheffield JK, Stathis SL, Chang AB. What is the burden of chronic cough for families? Chest. 2008;134(2):303-309. 8. Chang AB, Robertson CF, van Asperen PP et al. A cough algorithm for chronic cough in children: a multicenter, randomized controlled study. Pediatrics. 2013;131(5):e1576-e1583. 9. Dart RC, Paul IM, Bond GR, et al. Pediatric fatalities associated with over the counter (nonprescription) cough and cold medications. Ann Emerg Med. 2009;53(4):411-417. 10. Thomson F, Masters IB, Chang AB. Persistent cough in children and the overuse of medications. J Paediatr Child Health. 2002;38(6):578-581. 11. Greenhalgh T, Howick J, Maskrey N; Evidence Based Medicine Renaissance Group. Evidence based medicine: a movement in crisis? BMJ. 2014;348:g3725. 12. Moosapour H, Raza M, Rambod M, Soltani A. Conceptualization of categoryoriented likelihood ratio: a useful tool for clinical diagnostic reasoning. BMC Med Educ. 2011;11:94. 13. Chang AB, Robertson CF, van Asperen PP, et al. Can a management pathway for chronic cough in children improve clinical outcomes: protocol for a multicentre evaluation. Trials. 2010;11(1):103. 14. Chang AB. Therapy for cough: where does it fall short? Expert Rev Respir Med. 2011;5(4):503-513. 15. Chang AB, Newman RG, Carlin JB, Phelan PD, Robertson CF. Subjective scoring of cough in children: parentcompleted vs child-completed diary cards vs an objective method. Eur Respir J. 1998;11(2):462-466. 16. Van den Bruel A, Thompson MJ, HajHassan T, et al. Diagnostic value of laboratory tests in identifying serious infections in febrile children: systematic review. BMJ. 2011;342:d3082. 17. Newcombe PA, Sheffield JK, Juniper EF, Petsky HL, Willis C, Chang AB. Validation of a parent-proxy quality of life questionnaire for paediatric chronic cough (PC-QOL). Thorax. 2010;65(9):819-823.
18. Varni JW, Seid M, Kurtin PS. PedsQL 4.0: reliability and validity of the Pediatric Quality of Life Inventory version 4.0 generic core scales in healthy and patient populations. Med Care. 2001;39(8): 800-812. 19. Grimes DA, Schulz KF. Refining clinical diagnosis with likelihood ratios. Lancet. 2005;365(9469):1500-1505. 20. Attia J. Moving beyond sensitivity and specificity: using likelihood ratios to help interpret diagnostic tests. Australian Prescriber. 2003;26(5):111-113. 21. Marchant JM, Masters IB, Taylor SM, Cox NC, Seymour GJ, Chang AB. Evaluation and outcome of young children with chronic cough. Chest. 2006;129(5):1132-1141. 22. Marchant JM, Masters IB, Taylor SM, Chang AB. Utility of signs and symptoms of chronic cough in predicting specific cause in children. Thorax. 2006;61(8): 694-698. 23. Chang AB, Gaffney JT, Eastburn MM, Faoagali J, Cox NC, Masters IB. Cough quality in children: a comparison of subjective vs bronchoscopic findings. Respir Res. 2005;6:3. 24. Chang AB, Faoagali J, Cox NC, et al. A bronchoscopic scoring system for airway secretions—airway cellularity and microbiological validation. Pediatr Pulmonol. 2006;41(9):887-892. 25. Wurzel DF, Marchant JM, Clark JE, et al. Wet cough in children: infective and inflammatory characteristics in bronchoalveolar lavage fluid. Pediatr Pulmonol. 2014;49(6):561-568. 26. Goyal V, Grimwood K, Marchant J, Masters IB, Chang AB. Does failed chronic wet cough response to antibiotics predict bronchiectasis? Arch Dis Child. 2014;99(6):522-525. 27. Clinical Research Coordination Group of Chronic Cough; The Subspecialty Group of Respiratory Diseases; The Society of Pediatrics; Chinese Medical Association; Editorial Board, Chinese Journal of Pediatrics. Guideline for diagnosis and treatment of chronic cough in Chinese children [in Chinese]. Zhonghua Er Ke Za Zhi. 2014;52(3):184-188. 28. Leconte S, Paulus D, Degryse J. Prolonged cough in children: a summary of the Belgian primary care clinical guideline. Prim Care Respir J. 2008;17(4):206-211. 29. Hay AD, Wilson A, Fahey T, Peters TJ. The inter-observer agreement of examining pre-school children with acute cough: a nested study. BMC Fam Pract. 2004;5(1):4. 30. Morey MJ, Cheng AC, McCallum GB, Chang AB. Accuracy of cough reporting by carers of Indigenous children. J Paediatr Child Health. 2013;49(3): E199-E203.
753
