Globally, cough is the most common symptom in people of all ages seeking medical attention.1,2 However, in children, few randomized controlled trials (RCTs) have evaluated its general management,3 despite the fact that Chronic cough is a major cause of morbidity,3 recurring visits to the doctor,4 impaired quality of life,4,5 and, in some cases, increased likelihood of a serious underlying lung disorder.6,7
There is little data to validate the diagnostic threshold used to define chronic cough by duration. International guidelines define chronic cough as a daily cough for at least 4 weeks,8 and according to British guidelines for at least 8 weeks.9
Early diagnosis of underlying lung disorders in children with chronic cough (e.g., asthma, inhaled foreign body, bronchiectasis) improves clinical outcomes. To the authors’ knowledge, no RCT has evaluated an algorithm for managing chronic cough in its early phase in children.
Therefore, the main objective was to determine whether the implementation of an evidence-based cough management algorithm10 in early phases of chronic cough in children seen in non-specialized clinics would improve the proportion of children with resolution of cough compared to standard care. The secondary objective was to identify independent predictors of success or failure of the intervention.
| Methods |
A multicenter RCT was conducted within a prospective cohort study of children with any acute respiratory illness with cough as the presenting symptom. Children were eligible if they were younger than 15 years and had presented with cough (of any severity) as a symptom.
Patients were excluded due to underlying chronic medical condition (including lung disorders, but not asthma), immunosuppressive disease, use of immunomodulatory medications for more than 2 weeks in the previous 30 days, severe symptoms requiring hospitalization, current or planned participation in another study. , or lack of English proficiency to complete the study requirements.
Randomization was stratified by reason for presentation (acute respiratory illness with cough or presentation for another reason with incidentally noted acute respiratory illness with cough), study site, and total duration of cough at day 28 of enrollment (4 weeks to < 6 weeks or ≥ 6 weeks).
On day 28, children with persistent cough in the intervention group were reviewed by a pediatrician within the next 2 weeks and cough was managed according to an evidence-based chronic cough management algorithm composed of two pathways, one for children with nonspecific cough and the second for children with specific cough points (Figures 1, 2).
Parents were contacted 2 weeks after the initial appointment with the pediatrician and, if the cough had not resolved, a second check-up (by the same doctor) was arranged. Children who required more than two revisions were subsequently referred to a tertiary pediatric respiratory clinic.
Children in the control group did not receive specific treatment in the study. Parents were advised to seek medical attention if they were concerned or to continue self-managing their child’s cough. If they were still coughing on day 56, they were offered a review by a study pediatrician, and from that point on, management according to the proposed algorithm.
Demographic, clinical, and socioeconomic data were collected from all children at baseline and weekly for 8 weeks (56 days) after enrollment, regardless of randomization at day 28. 11
The primary outcome was resolution of cough by day 56, defined as absence of cough for at least 3 days and nights beginning on day 28, and determined by weekly parental report of a 3-day history of cough, or by a greater than 75% reduction in the average daytime and nighttime cough score (based on a validated cough rating scale12) at day 56 compared to day 28.
Children were classified as having persistent cough if the cough did not have a 3-day break or cough score reduction during the entire 8-week study period, and as having cough of unknown outcome if there was at least 1 week of absence. follow-up and a 3-day break in cough was not reported at other time points. The secondary outcome was whether epidemiological, cultural, clinical or socioeconomic predictors could identify the success or failure of the intervention.
| Results |
Between July 7, 2015 and October 31, 2018, 1018 children were screened and 509 were enrolled in the cohort study. By day 28, cough had resolved in 234 (46%) children, cough status was unknown in 158 (31%), and 117 (23%) of the 509 children with chronic cough were randomized.
Two children were excluded for protocol violations, so 115 children were included in the intention-to-treat (ITT) analysis, with 57 assigned to the intervention group and 58 to the control group. Of the total, 45 children (39%) were indigenous, the median age was 1.6 years (IQR 1.0 – 4.5), and 59 (51%) were boys.
In the intervention group, 41 (72%) of the 57 children were seen by the study pediatrician at a median of 9 days (IQR 6–14) after randomization.
For the remaining 16 children, parents of 11 declined consultation, three children were lost to follow-up, and one resolved their cough before their appointment. Additionally, one child moved to a remote area and was unable to attend within the required time frame.
Among the control group, 32 children still had chronic cough on day 56. Of these 32 children, 21 attended the recommended study review.
Of the remaining 11 children, three were lost to follow-up after day 56, two repeatedly canceled their appointments and in both cases the child eventually stopped coughing, the parents of three other children decided not to accept the appointment, and three parents sought specialists. independently for the management of their children’s cough elsewhere.
By day 56, 33 (58%) of 57 children in the intervention group had resolution of cough compared to 23 (40%) of 58 in the control group. Resolution of cough by day 56 was unknown in 12 (21%) children in the intervention group and in 13 (22%) in the control group.
Using ITT (conservatively assuming that children whose cough status at day 56 was unknown had persistent cough), the absolute risk difference (ARbs) was 18.3% (95% CI 0.3–36.2) and the number of treatments necessary to obtain a benefit was five (95% CI: 3-364).
The adjusted OR to predict resolution of cough was 1.5 (95% CI 1.3 - 1.6), in favor of the intervention group. Repeating the analysis with an unknown cough status at day 56 classified as resolved, the adjusted OR was 1.4 (95% CI 1.0 - 2.0) in favor of the intervention group. In the per protocol analysis, the adjusted OR for resolution of cough was 1.7 (95% CI 1, - 2.2) in favor of the intervention group.
| Discussion |
The effectiveness of an evidence-based cough management algorithm in achieving cough resolution was evaluated in 115 children shortly after transitioning from the acute to chronic cough stage of their illness.
Compared to the control group, children in the intervention group were 1.5 times more likely to have stopped coughing 56 days after the initial consultation with a health service.
Independent predictors of cough outcomes that showed strong associations were study site, age at enrollment, private health insurance, number of rooms in the home, and admission to a neonatal intensive care unit for treatment of respiratory problems after birth. These factors were also associated with unknown cough status at day 56.
To the authors’ knowledge, this was the first RCT of chronic cough that recruited children from the community and not from specialist clinics. He was also the first to evaluate the outcomes of children with cough soon after it became chronic, a situation that is more likely to be captured in a primary care setting. Third, it was the second study to evaluate the use of a chronic cough management algorithm in pediatric patients.
The first RCT10 that evaluated the effectiveness of this algorithm included only children referred to clinics specialized in respiratory pathology and the median duration of cough at enrollment was longer (16 weeks).
Their reported DRabs of 24.7%10 at week 6 was not very different from that of the present study (18.3%) after 4 weeks, suggesting that this algorithm could also be used within the community, helping to decrease the substantial burden of cough on children, their families and the health system.4,17
However, implementation of the algorithm in primary care is likely to require additional education so that healthcare professionals can recognize so-called cough specific points and their clinical significance.
The prevalence of the different causes of chronic cough is dependent on age and environment.18 Therefore, it is not surprising that fewer children in this RCT had a serious underlying disorder. However, the predominant causes for chronic cough were prolonged bacterial bronchitis and asthma.19,20 These conditions are quickly identified in cough management algorithms15 and their immediate and appropriate management has probably contributed to the effectiveness observed in children. with established chronic cough.
This RCT must be interpreted considering several factors. First, subgroup analyzes were not feasible. Additionally, regression models suggested that study site, age, socioeconomic status, and NICU admission for respiratory problems were associated with outcomes for known or unknown cough.
However, it is possible that several other important factors (daycare attendance, previous history of chronic cough, indigenous population and season of the year) are associated with these results. It is unknown why socioeconomic status may influence outcomes, but it could be associated with intervention uptake.
Compared with children enrolled in emergency departments, children enrolled in primary care were less likely to resolve cough by day 56.
The observed associations could be a factor in the increased burden of chronic cough and underlying chronic causes in the indigenous communities served by these primary care practices. This was seen in the authors’ previous study,19 which showed differences in medical interventions for cough in the emergency department compared to primary care, and in the relationships researchers had with communities.
Child age at enrollment was identified as an independent predictor of cough outcome, with children younger than 2 years being less likely to resolve cough and to have unknown cough status at day 56.
It is possible that these young children, with higher rates of respiratory infection, could have developed infection from other causes, since resolution of cough was defined as 3 days without cough or 75% reduction for 3 days.
The strengths of this RCT were its multicenter approach that increased the generalizability of the findings, and its inclusion within a larger cohort study that allowed monitoring of cough duration from disease onset to ensure that children met the criteria for persistent cough for at least 4 weeks.
The proportion of children enrolled with confirmed chronic cough at day 28 (117 [23%] of 509) was within the range of 20 to 27% observed in other studies using the same cohort methods in similar or the same settings.20,26 This finding contrasts with a systematic review that suggested that only 10% of children presenting to primary care or the emergency department will continue to have cough 3 weeks after an acute respiratory infection.27
However, the population of the present study had several key differences from those of the studies that formed the review, including the assessment of cough at 4 weeks and not at 3 weeks,27 the enrollment of a high proportion of Aboriginal children and islanders and those with risk factors such as a history of chronic cough, premature birth, admission to the NICU at birth for respiratory problems, recurrent infections, and hospitalization in the previous 12 months for respiratory illness.
The most important limitation of this study is not having achieved the required sample size to adequately address biases, confounding factors, and potential subgroup differences in the results of the regression models. However, the DRabs was in the range of that reported in the pivotal study of the chronic cough algorithm in children,10 and the finding was consistent. Furthermore, time and budget limitations forced the protocol to be adapted, although the differences between groups were similar to those of larger RCTs.10
This study reflects what is likely to occur with respect to patient engagement and follow-up if the intervention is incorporated into the clinical care of children in the early stages of chronic cough.
Evaluation and management of the child was performed by trained pediatricians within 2 weeks of identification of chronic cough, which is unlikely to occur in many jurisdictions given the availability of specialized care or the potential costs of treatment for families, or both.
Therefore, the algorithm should be implemented and evaluated in primary care and at follow-up checks for children discharged from the emergency department or hospital. Finally, because this study included children from communities at higher risk for respiratory diseases, the results may not be generalizable.
Generalizability was also influenced by the choice of 4 weeks as the end point for defining chronic cough according to most national guidelines (including Australia and the USA),8 but not the British guidelines.
However, the observed effects were consistent with the first study of the algorithm in children, and in addition, the majority of children were diagnosed with conditions that responded well to appropriate and early interventions, and therefore, the 4 weeks were considered as a ideal time point at which management and prevention of chronic cough should be implemented.
Despite limitations, this study contributes to evidence8 that cough management algorithms improve cough resolution in children when implemented soon after the transition phase from acute to chronic cough.
Early identification and management of chronic cough is crucial for improving quality of life and early diagnosis of underlying chronic lung disorders. More research is needed on how to implement the findings in primary care settings and to address any differences in potentially important subgroups.
