Lung sound analysis in infants with risk factors for asthma development

Abstract Background and objectives Using a lung sound analysis, the prognosis of asthma was investigated in infants with risk factors for asthma development by a long‐term observation. Methods A total of 268 infants were included (median age: 8 months old). The lung sound parameters (the ratio of the third and fourth area to the total area under the curve [A3/AT and B4/AT], and the ratio of power and frequency at 50% and 75% of the highest frequency [RPF50 and RPF75]) were evaluated at the first visit. At 3 years old, using a questionnaire, we examined the relationship between the lung sound parameters and risk factors of asthma development. Results Among the 268 infants, 175 infants were in good health and 93 had a history of acute respiratory infection (ARI) within 7 days at the first visit. Among the 3‐ to 12‐month‐old infants with an ARI, the A3/AT, B4/AT values in those with a history of asthma/asthmatic bronchitis, atopic dermatitis, and atopy were smaller than in the infants without such histories. Among the 13‐ to 24‐month‐old infants with an ARI, the A3/AT and B4/AT values in those with a wheezing history were larger than in the infants without such a history. Conclusions The characteristics of the lung sounds in infants with risk factors for asthma development were demonstrated over long‐term follow‐up. Lung sound analyses may be useful for assessing the airway condition of infants.


| Study subjects
In an ongoing, multi-institutional prospective study (Diagnosis of Infantile Asthma Using Lung Sound Analysis; DIAL) 14 To target infants in good health, the following exclusion criteria were used: infants with severe diseases of the lung, heart, and other organs and a fever and/or respiratory symptoms. At the first visit, none of the subjects had wheezing on auscultation. According to our previous reports, 16 infants who had an acute respiratory infection (ARI) within the 7 days prior to their first visit were identified as infants with an ARI. As in a previous study, 16 since the proportion of subjects with and without an ARI varied significantly with age, we divided the subjects into two age groups. Subjects who were 3 to 24 months old at the first visit were evenly divided into two age groups based on the midline of the age (3-12 and 13-24 months old) for a stratified analysis (

| Breath sound analyses
All participants underwent the collection of lung sounds as previously The slope indicates the roll-off of the middle spectrum curve (ÀdB/octave). 18 A T , RPF 75 , and RPF 50 also decrease. 11,12 In addition to the above, the frequency limiting 99% of the power spectrum (F 99 ) was also measured in accordance with the methods of previous reports. [18][19][20] In this study, three lung sound samples from 10 or more samples were obtained. In each institute, two or more physicians, who were licensed pediatricians, discussed the selection of sound samples without noises and with the same sound spectrum size for each individual.
After deciding on the zero point, personal lung sounds were automatically calculated using an in-house calculation software program. 9,13 We used the median values of the three samples as the measured values for each subject.
T A B L E 2 Questionnaire results of the infants aged 3 to 12 months and 13 to 24 months of age with and without ARI

| Statistical analyses
The statistical analyses were conducted using the SPSS software pro-  Table 2 shows the questionnaire results of the infants with and without an ARI in the two age groups. In this report, the zero level and the zero point were used to calculate the AUC in the sound spectrum, and all pediatricians who participated in this study agreed with this technique.
To confirm any overlap between the infants with positive responses to wheezing-related items (Question 2, 3, or 5) and those with positive responses to atopy-related items (Question 10 or 11), we performed Fisher's exact test. The numbers of infants with both wheezing and atopy, with wheezing but without atopy, without wheezing but with atopy, and without either wheezing or atopy were 28, 41, 60, and 140, respectively. The P value of Fisher's exact test was 0.072.

| Differences in the breath sound parameters in each item of questionnaire in the younger group
In the ARI-positive group, the B 4 /A T values in the infants with a history of asthma were significantly lower than those without a history of asthma or asthmatic bronchitis ( Table 3). The A 3 /A T and B 4 /A T values in the infants with atopic dermatitis were significantly lower than those without atopic dermatitis ( Table 3)

. The A 3 /A T , B 4 /A T , and
RPF 75 values in those with atopy were significantly lower than those without an ARI (Table 3).
Furthermore, although no significant difference was found (Bonferroni's multiple comparison test), the RPF 75 value of the ARIpositive and atopy-positive infants was lower than that of the ARInegative and atopy-negative infants (#; P = 0.038, Table 3). In  contrast, the RPF 75 value of the ARI-positive and atopy-negative infants was higher than that of the ARI-negative and atopy-negative infants ( §; P = 0.035, Table 3).
No marked differences were observed in the other spectrum curve indices of the infants between the question-positive and the question-negative groups.
3.3 | Differences in the breath sound parameters in each item of questionnaire in the older group In the ARI-positive group, the A 3 /A T and B 4 /A T values in the infants with a wheezing history were significantly higher than those without a wheezing history (P = 0.009 and P = 0.006, respectively, Table 3).
No marked differences were observed in the other spectrum curve indices of the infants between the question-positive and question-negative groups.

| DISCUSSION
A lung sound analysis has been evaluated as a reliable, noninvasive respiratory function test. 6,7,9 With recent technological advances, data collection using mobile phones 21 and automatic analyses by artificial intelligence 22,23 have also been reported. Furthermore, the target patients have expanded to include newborn babies. 24 In this report, we used a newly revised technique to conduct a sound spectrum analysis. 15 Previously, to analyze the spectrum images, small differences in the power of background noises are seen among patients. These slight differences may depend on the influence of wide-ranging outside noises and/or the difference in the pressure with which each examiner held the microphone. 15 When the original zero level is lower than À90 dBm, the high-pitched area may be underestimated. These factors can adversely affect the accuracy of the lung sound analysis. To resolve this problem, the zero level was visually corrected based on the lung sound spectra in each sample before the zero point (the frequency at 0 dB) was decided by at least two examiners. Using the new technique, we were able to measure the ratio of the AUC more accurately than in the past.
Through the present long-term follow-up study, we showed that the risk factors for asthma development were clearly associated with the results of the lung sound analysis. These findings were similar to those of two previous reports concerning first-visit infants in the healthy period 14 and after an ARI. 16  However, we obtained different findings in the older group, with the A 3 /A T and B 4 /A T values in the infants with a wheezing history being higher than those without a wheezing history in the ARIpositive group. It is reasonable to assume that an ARI, which induces acute inflammation and/or edema of the airway mucosa, may introduce some additional sounds into the normal array of breath sounds.
It has been reported that an increase in the middle-pitched area of spectrum curves induces an increase in the spectrum curve indices. 14 Our results suggest the possibility of an increase in the middle-pitched area of the spectrum curves in older infants with a wheezing history who have an ARI.
Given the above findings, we believe that there may be two kinds of abnormal additional sounds related to ARIs and an atopic state: an increase in the high-pitched area or an increase in the middle-pitched area of the sound spectrum. These inaudible additional sounds are thought to be generated by an independent mechanism. One reason for this suspicion is that no relationships were noted between the presence of wheezing and atopy, as in the previous study. 16 While an age-dependent difference does appear to exist, ≤12 months of atopyrelated sounds and >12 months of infection-related sounds, the lack of any marked difference in the spectrum curve index values in the older group may be due to the number of subjects in the older group being too small. We speculate that these differences are dependent on pure bronchial constriction (the high-pitched sound) 9,12 and the edema and/or remodeling in the bronchial wall and peri-wall region due to airway inflammation (the middle-pitched sound), but we lack any supporting evidence.
Some phenotypes associated with recurrent wheezing and/or infantile asthma were reported in previous studies. 26 Recurrent wheezing after viral infection has been noted in infants, and the pathophysiology of the virus infection-induced recurrent wheezing/ infantile asthma may be different from that of atopic asthma. 27 Although it is often difficult to distinguish atopic asthma and infectious asthma in infants, prospective studies using palivizumab, an antirespiratory syncytial virus monoclonal antibody, have also suggested that these phenotypes are independent. 28 Our results suggest the existence of this phenotype-dependent airway condition in early infants. We will examine this point further in future studies.
As a limitation of our study, the number of cases in some groups was small. A stratified analysis based on age should be performed because the ratio of ARI differed by patient age. 16 This may be related to the social situation in Japan, as the ratio of children attending nursery schools clearly increases after 1 year of age, along with the frequency of respiratory tract infections. Furthermore, we were unable to indicate which part of the airway was causing these additional lung sounds. 12 To consider the presence of pathological differences in infection-or atopy-dependent additional lung sounds, our results may indicate the difference depending on the airway condition and/or the site of region. This seems to be an interesting point, and we intend to explore it further in the future.