1. Diagnosing Asthma in Symptomatic children using lung function: Evidence from a Birth Cohort Study
Clare Murray1, Philip Foden1, Lesley Lowe1, Hannah Durrington1, Adnan Custovic2, Angela Simpson1.
1. University of Manchester; 2. Imperial College

2. Introduction No gold standard test for making a diagnosis of asthma.
History; signs and symptoms; atopic/family history.
Lung function tests – spirometry, bronchodilator reversibility, PEFR, FeNO, airway hyper-reactivity.
In UK, recent publication of national draft guideline (NICE)
Aim – “to determine the most clinical and cost-effective way of effectively diagnosing people with asthma”

3. NICE – Asthma Diagnosis Clinical guideline (5-16 years) Proposal
Initial clinical assessment
Algorithm based of four tests of lung function, each used as a dichotomous variable
FEV1/FVC ratio less 70% or LLN
Bronchodilator reversibility > 12%
FeNO > 35 ppb
PEFR variability >20%

4. Objective tests for Children aged 5-16 years

5. Aim
Within a population-based birth cohort (Manchester Asthma and Allergy study – MAAS) we investigated the value of FEV1/FVC ratio, Bronchodilator reversibility and FeNO in diagnosing asthma in young people

6. Methods Study participants – age 14-16 year follow up
Clinical data collected with validated respiratory questionnaires
Measured:
Spirometry (FEV1, FVC, FEV1/FVC {% predicted and LLN});
BDR (positive if FEV1 increased by >12% following 400mcg salbutamol);
FeNO (ppb; Niox, Sweden)

7. Asthma Definitions
Current asthma as a positive answer to all 3 questions:
Has a doctor ever told you that your child has asthma?
Has your child had wheezing or whistling in the chest in the last 12 months?
Has your child had any treatment for asthma in the last 12 months?
Non-asthmatic controls – negative response to all 3 questions above
Indeterminates – Answered positively to <3 questions (excluded from analysis)

8. Results 772 children attended follow up age 14-16 years
630 completed spirometry
189 reported some symptoms in the previous 12 months
26 were using regular ICS (excluded)
Investigated 163 with recent symptoms not on regular ICS

9. Symptomatic population
Results
Whole population
(n=630)
Symptomatic population
(n=163)
Age
mean (95% CI; Range)
15.55
( ; 13-16)
15.44
( ; 13-16)
Male gender
N (%)
325 (51.6%)
89 (54.6%)
FEV1
(Mean; 95% CI)
3.63 ( )
3.74 ( )
FEV1 % predicted
98.7 ( )
101.6 ( )
FEV1/FVC %
88.3 ( )
87.8 ( )
BDR %
4.9 ( )
4.9 ( )
FeNO ppb
(GM; 95% CI)
20.0 ( )
22.5 ( )

10. Results
163 with recent symptoms (cough, wheeze and breathlessness) not taking regular ICS
34 Current asthma (doctor diagnosis, prescribed asthma treatment, wheeze in last 12 months)
55 Non-asthma controls (no doctor diagnosis, no asthma treatment, no wheeze)
74 indeterminate (excluded from further analysis)

11. Predictive values for individual tests
AUC
Cut-off
Sensitivity
Specificity
PPV
NPV
FEV1/FVC
0.616
<70% 0%
(0/34)
96.4% (53/55)
(0/2)
60.9% (53/87)
LLN
11.8% (4/34)
92.7% (51/55)
50.5% (4/8)
63.0% (51/81)
*<85.5%
55.9%
69.1%
52.8%
71.7%
FeNO
0.618
> 35 ppb
44.1% (15/34)
83.6% (46/55)
62.5% (15/24)
70.8% (46/65)
*> 37 ppb
44.1%
85.5%
65.2%
71.2%
BDR%
0.594
> 12%
8.8% (3/34)
42.9% (3/7)
62.2% (51/82)
*> 3.2%
79.4%
41.8%
45.8%
76.7%
*best cut-offs according to Youden’s statistic

12. Multivariable model for asthma
Odds ratio (95% CI)
p-value
FeNO
(per ppb increase)
1.02 ( )
0.006
FEV1/FVC
(per % decrease)
1.06 (0.98,1.14)
0.17
BDR - % change in FEV1 after salbutamol
(per % increase)
1.00 ( )
0.94
Amongst those with symptoms, not on regular ICS
(AUROC 0.68)

13. Overlap of positive tests
For 89 children with reported symptoms – current asthma n=34; without asthma n=55

14. Summary Results
Only 2 children were positive for all 3 tests – both had “current asthma”
62% of those with a positive FeNO had “current asthma”
Only 8 children had obstructive lung function (<LLN) – 4 of these had “current asthma”
Half of the children with “current asthma” were negative on all 3 tests

15. Conclusions
Applying 3 tests of lung function in symptomatic children failed to detect “current asthma” in 50% of children.
Overall there was poor agreement between our standard definition of asthma and lung function results
These findings challenge the cut-off values and the inclusion of BDR in making a diagnosis of asthma
Proposed algorithms for diagnosing asthma need to be tested prospectively

16. Thanks to all the MAAS team and all our parents and participants