New Insights Into Refractory Chronic Cough and Unexplained Chronic Cough: A 6-Year Ambispective Cohort Study

Chronic cough (CC) in adults is broadly defined as a cough lasting for more than 8 weeks. Based on current cough guidelines,1, 2 its causes can be identified in most patients, and they respond well to subsequent cause-specific treatments that finally lead to cough resolution. However, in 12%–42% of CC patients, etiologically targeted treatments may fail, or no underlying etiology is found after thorough assessments.3 Refractory chronic cough (RCC) and unexplained chronic cough (UCC) are used to define these conditions, respectively.4, 5 They are more prevalent at specialist cough clinics in countries with a protocoled referral system.6 These patients self-reported a cough easily triggered by innocuous stimuli which would not be tussive in the healthy population, an exaggerated cough response to noxious stimuli and irritation or itching sensation in the throat., These symptoms leave a severe negative impact on their quality of life (QoL) and stressful socio-economic pressure.3

The demographic characteristics of patients with CC appear to differ slightly between China and other countries. The female predominance in CC was commonly reported in some worldwide surveys, especially in older females, who were thought to have higher cough sensitivity.7 However, the latest national cross-section study on CC reported a male predominance in China.8 The 2 global randomized placebo-controlled trials (COUGH-1 and COUGH-2) regarding the efficacy of gefapixant (antagonist of P2X3 receptors) on RCC partially provided the preliminary characteristics of the participants, including the common comorbidity such as asthma, gastro-oesophageal reflux disease (GERD) and allergic rhinitis (upper airway cough syndrome, UACS).9, 10 However, specific knowledge about the profile of RCC/UCC in China and the sex differences in this condition are lacking. The reason why patients with abnormal test results failed to respond to the subsequent etiologically targeted treatment also remained unclear. RCC has an underlying cause but does not respond well to currently available treatments; however, there may be treatments that achieve greater efficacy in the future. This makes RCC clinically more amenable to resolution than UCC, which does not yet have a definitive etiological diagnosis and has no therapeutic direction. Defining the etiology and discovering effective therapeutic measures is currently a challenge in the clinical management of UCC. In UCC with at least 1 positive laboratory finding (UCCpos), abnormal findings may not be causally related to cough, making it difficult to effectively predict etiological treatment response, whereas the underlying etiology of UCC without any abnormal laboratory findings (UCCneg) may currently be poorly perceived by clinicians. Therefore, in this study, we retrospectively collected clinical information from RCC and UCC patients attending our cough clinic during the past 6 years and traced the outcomes prospectively to fully reveal the sex specificity of RCC/UCC and to find out differences among RCC, UCCpos, and UCCneg in order to provide clues for early identification of these 3 phenotypes in the clinical setting and to facilitate the search for effective treatment.

This is the first cohort study that reports the clinical nature and long-term outcomes of RCC and UCC in a large series of consecutive patients. Most of these patients were referred from primary care or other healthcare facilities and had previously undergone repeated laboratory investigations and ineffective treatments. In addition to additional examinations, such as induced sputum cytology, histamine bronchial provocation, and MII-pH, the failed empirical trials also helped efficiently rule out suspected causes such as CRC. The diagnosis of RCC or UCC can therefore be established within an average of 6 weeks in our clinic.

This cohort had a self-reported long duration of severe cough. The onset of cough was often ascribed to presumed acute URTI, but most coughs do not have recognizable causes (especially UCC). The prominent clinical feature was a paroxysmal dry cough, which was more severe during the daytime, and triggered or aggravated by talking, irritating, or pungent odor, eating, and cold air. Up to 95.9% of patients reported laryngeal discomfort. Other concomitant manifestations included epigastric discomfort, chest tightness, nasal symptoms, chest pain, and urinary incontinence (only females). These characteristics are comparable with those described in previous studies.23, 24, 25

The sex distribution of CC and RCC/UCC is skewed toward females in some worldwide studies,7, 9, 26, 27 whereas in China, this tends to be even with variability between studies.7, 28, 29 The fact that patients can be free to visit reputable specialist clinics in China may lead to a bias in the proportion of patients among healthcare institutions, but promote patients to visit clinics. This may explain why RCC/UCC patients in this study were approximately 20 years younger than those in a large epidemiological survey.7 This cohort had a marginal predominance of older females. The particular prevalence in menopausal females is due in part to the increased CD4⁺ T-cells in the respiratory tract and the hormone-induced anatomical brain changes.7, 30 Exposure to occupational and environmental cough triggers (male smokers are more common) may be responsible for the considerable number of younger male coughers in China.31 In this study, women reported more nocturnal cough with better results in lung function. Over half of the males coughed during meal-times and were more likely to have GERC and higher eosinophil levels in the blood. Currently, it is difficult to assume a definite association between eosinophils and sex, given the impact of several intrinsic, extrinsic, and technical factors.32 The total LCQ scores between sexes were comparable, in line with the result of our previous study18; however, psychological domain scores in women were significantly higher with QoL better, which was contradictory with the traditional thinking.18 The predominance of elderly women might be postulated, as higher scores in psychological and social dimensions associated with aging have previously been described.33 Further confirmation work is needed. Persistent troublesome cough seems to have a greater negative psychological impact on predominantly young and middle-aged men (the main working and social networking population), which could, to some extent, explain why UCCneg, with up to 65% female prevalence, had fewer current smokers and scored higher in LCQ psychological and social domains than UCCpos. Consistent with most studies, cough sensitivity was higher in females.31, 34, 35 Since angiotensin-converting enzyme inhibitor-related cough has been excluded, more prevalent hypertension in females may be associated with aging and not likely to be related to enhanced cough sensitivity. There was no sex difference in HARQ scores, whereas previous data by Morice et al.36 demonstrated that females had higher scores in a Caucasian population. In addition, the HARQ scores were lower compared to those of the previous report9 (mean of approximately 23 vs 40, respectively), and 90% of patients were above the upper limit of normal. The shorter duration of the cough history in this study was likely to be a contributor to the lower score, since higher HARQ scores are associated with a longer duration of cough.

RCC patients in this study were more likely to display sensitivity to cold air or have abnormal reflux than UCC. Stimulation of a thermosensor of noxious cold, Transient Receptor Potential Cation Channel A1 (TRPA1), has been demonstrated to be enhanced in CC patients37 and may be associated with cold-air-induced cough.38, 39 We speculate that the more frequent abnormal reflux seen in RCC may sensitize TRPA1 and thus enhance the sensitivity to cold air. Although airway reflux was observed in over one-fifth of UCC patients, as demonstrated by MII-pH measurements, intensive treatment failure excluded GERC from the consideration. The greater FeNO results and a higher proportion of sinusitis in UCC may indicate a potential association between intractable airway reflux and upper/lower airway inflammation.

Unlike previous reports which had a predominance of RCC (62%),10 our cohort was mostly UCC, accounting for over 60%. This might stem from the difference in diagnostic criteria. Patients with positive clinical examinations pointing to a possible etiology, but who subsequently were unresponsive to targeted treatment have been termed as RCC by some investigators.40, 41 In contrast, we suggest that the diagnosis should be made based on clinical history associated with abnormal laboratory findings and effective targeted treatment; positive findings in patients refractory to targeted treatment may be a bystander rather than the confirmative relevance to the etiology of cough. We have therefore developed a clear classification of RCC and UCC in this study. The majority of RCC had GERC with non-acid reflux shown by impedance. These patients had abnormal reflux-associated findings and poor response to low doses of omeprazole + domperidone but benefited from double doses of omeprazole or a combination of neuromodulators such as baclofen and gabapentin.15 The fact that this study was conducted at a single and specialized center may also contribute to the high proportion of refractory GERC. UACS associated with chronic sinusitis constituted the second leading cause of RCC. Due to the lack of effective treatments for GERD or sinusitis,42, 43 GERC, and UACS are more likely to be refractory. In general, CVA, EB, and AC are sensitive to steroids and less likely to be refractory. The reason for refractory cough in 5 patients of AC and 5 of EB may be unavoidable exposure to allergens or non-allergic mechanisms.

RCC and UCC are exclusive diagnoses with frequent mismanagement. Nearly two-fifths of patients in this cohort had received ineffective targeted treatments before. UCC patients were less likely to be responsive to the targeted treatment, particularly those with positive tests (only a 28% response rate). Thus, the test results failed to predict treatment response in most patients. Most patients had been empirically treated as CVA, AC, and EB before the first visit to our clinic. CVA has clear diagnostic criteria and usually responds well to anti-asthmatic treatment. This is supported by the fact that only 5.7% of our patients had BHR, significantly less than that in the contemporary common CC patients in our clinic (12.4% overall but dropped from 30.1% in 2016 to 4.8% in 2021 since the 3-step empirical therapy protocol was recommended44) and that reported in a Dutch CC cohort (54.2%; P < 0.01).45 Some patients with CVA later proved unresponsive to asthma treatment, possibly indicating a change in the evolution of cough etiology. After ruling out CRC, GERC needed to be considered. The MII-pH is sensitive in detecting reflux, and the results are often abnormal, explaining why 22% of patients evolved to a GERC diagnosis. Therefore, the tentative exploratory work-up before confirming RCC/UCC is inevitable, and we suggest a formalized protocol needs to be followed to effectively manage this process. Recently, Morice46 has suggested that the term UCC might be removed as a diagnosis, given the similarity between UCC and RCC. The paradigm of cough hypersensitivity syndrome has defined them as a distinct physiological entity.47 In this study, the equivalent HARQ score also supported this idea. We do agree that the current evidence points to a single underlying entity and that with improved diagnostic techniques, rare phenotypes are easy to identify, which will lead to fewer UCC diagnoses. However, considering the issue that nearly half of the current UCC might be treated with inappropriate intervention, we prefer to keep this diagnosis.

For over 40% of patients, the cough was completely controlled or spontaneously relieved in approximately 1.5 years after the last clinical review. Markedly reduced HARQ scores at the last follow-up suggested remission of cough hypersensitivity in a quarter of patients, although we failed to demonstrate the change in capsaicin sensitivity. This may be because this study was based on real-world data and factors such as smoking may confound the objective findings. Nearly a third of patients still had a persistent cough. Only coughs in patients with refractory GERC were more likely to be solved. Several clinical characteristics, such as “long cough duration,” “urinary incontinence,” “cough when talking,” and “cold-air sensitive cough,” were likely to be associated with cough persistence. Cold-air hypersensitivity was found to be associated with cough persistence in CC before, and CC patients were reported with an incidence of cough persistence at about 20% and 47%.48, 49 These patients are likely to be the major portion of the RCC/UCC population. Yousaf et al.50 described the 7-year outcomes of UCC, of which 60% had an unchanged or worse cough. The large discrepancy from this study can best be explained by our shorter follow-up period and the administration of neuromodulators.5 All evidence indicates that the cough may continue or recur after systematic management at specialist clinics. Further study is needed to understand the variation in treatment outcomes. Future clinical management of RCC with a definite etiology should involve all etiologically specific therapeutic measures to achieve cough relief, while available nonspecific antitussive therapies such as neuromodulators should be considered for UCC.

There are some limitations in our study. First, this is a single-center study, and selection bias is unavoidable. Although, as one of the few specialist cough clinics in China, there is a high proportion of referrals from across the country which makes our patients quite representative of Chinese patients, the external validity and applicability still need to be further investigated. Furthermore, some findings, such as the sex-age distribution, do not differ that much from previous multicentre studies of CC in China. This validates the results of this study. Secondly, the follow-up outcomes were self-reported by patients rather than via objective assessment, thus being subject to a recall bias. There is a high rate of non-response bias as well. Thirdly, further details of medication adherence and combination medication proved difficult to obtain. However, this is a non-randomized observational study and represents guideline-led clinical practice, which may reflect a real-world experience.

In conclusion, most features between RCC and UCC were similar, but UCC was more commonly treated inappropriately and ineffectively. The current management of RCC and UCC only improve cough symptom in two-thirds of patients. Nearly one-third of patients still had a persistent or recurrent cough after systematic treatment attempts, especially for those with long cough duration and being sensitive to “talking” or “cold air.” Novel agents for RCC and UCC are in urgent clinical need, and the promising results of recent trials give hope for the future.

Supplementary Table S1

Prospective follow-up updates

Supplementary Table S2

Comparisons of clinical characteristics and assessment results between patients with good outcomes and those with severe cough (n = 369)