Multicenter questionnaire study investigating characteristics of adults with unexplained chronic cough versus explained chronic cough

Abstract

Objective:

Previously, we reported that older women taking increased numbers of cough medications with increased number and frequency of medical encounters and normal or near-normal lung function more likely had unexplained chronic cough (UCC) versus asthma and/or chronic obstructive pulmonary disease. This study sought to identify clinical risk factors that could differentiate UCC from explained chronic cough (ECC).

Methods:

A validated electronic questionnaire was distributed to patients with chronic cough (CC) at seven cough centers throughout the United States. The mean ± standard error, frequencies of continuous variables (one-way analysis), and cross-tabulation frequencies for categorical variables (two-way analysis) were calculated. Univariate comparisons between UCC and ECC were performed by using the t-test and nonparametric one-way analysis. Significant determinants of UCC and cough severity were assessed by using multiple logistic regression.

Results:

A total of 150 patients were enrolled, of whom 29 of 150 were classified as having UCC, and 121 of 150 were classified as having ECC. No significant differences for family history, age, gender, and race, or seasonality differentiated UCC from ECC. Multiple logistic regression revealed the absence of postnasal drip significantly differentiated UCC from ECC (odds ratio 4.8 [95% Confidence Level, 1.6–15.3]). The severity of CC was worse for patients with UCC, patients with chronic bronchitis and/or emphysema, hypertension, ex-smoking history, high body mass index, female gender, education level, and reactivity to more environmental irritants (perfume, p = 0.006; household cleaners, p = 0.01; air fresheners, p = 0.03; cold air, p = 0.007; cigarette smoke, p = 0.05).

Conclusion:

Patients with UCC more frequently presented with specific demographic features, comorbid characteristics, and more severe cough induced by environmental irritants compared with ECC. These clinical characteristics may be useful for identifying risk factors that can accelerate the diagnosis of UCC.

Coughing is a fundamental protective reflex that clears the airways of foreign bodies, irritants, and pathogens. The cough mechanism is essential for maintaining the integrity of the respiratory system by preventing the accumulation of harmful substances.^1–3 However, when cough becomes chronic, defined as lasting ≥ 8 weeks, it can significantly affect a patient’s quality of life, which leads to issues such as sleep disturbances, urinary incontinence, and, in severe cases, rib fractures and syncope.^4,5 These symptoms, consequently, can lead to anxiety, depression, and social embarrassment in patients with chronic cough (CC).^4,5 CC is estimated to affect ∼8–10% of the adult population worldwide and is more commonly reported in Europe and North America compared with Asia and Africa.^6,7 This geographic variation may be due to several factors, including environmental exposure to allergens and pollutants, health-care access, and differences in diagnostic criteria and reporting practices.^6,7

It is generally accepted that the neurogenic inflammation that causes sensory nerves in the airway mucosa to become more reactive to nonspecific irritants such as smoke or fragrances due to transient receptor potential channel activation, particularly Transient Response Vanilloid 1 and Transient Response Ankyrin 1, plays a critical role in the cough reflex.^8–10 Activation of these channels leads to the release of proinflammatory mediators and neuropeptides, e.g., substance P, which can enhance nerve sensitivity, which leads to a persistent cough.¹¹ Central sensitization, which refers to changes in the brainstem and higher cortical areas that increase responsiveness of these neural pathways involved in the cough reflex, are also involved in CC.¹²

Unexplained CC (UCC) is defined as persistent cough that lasts >8 weeks when a defined underlying cause cannot be identified.⁵ It has been reported that adenosine triphosphate activation of purinergic receptors plays an important role in cough that leads to the development of P2X purinoceptor 3 antagonists as targeted therapies for UCC.¹³ Several of these agents have been investigated for UCC, and one (gefipixant), has recently been approved for CC by the European Medical Agency.^13,14

Despite the advancements achieved in understanding the pathogenesis and treatment of UCC, this condition remains a significant health-care burden for patients largely due to the prolonged duration from onset to diagnosis and ineffectiveness of treatment and impaired quality of life. A major difficulty in establishing a diagnosis of UCC is that multiple comorbid conditions must be excluded, which results in numerous physician encounters and many ineffective therapies associated with significant health-care costs.⁵ This emphasizes the urgent need for improved algorithms that can accelerate the diagnosis of UCC and reduce patient morbidity. Ideally, this would require defining specific clinical characteristics and/or laboratory biomarkers that distinguish UCC from other respiratory conditions associated with CC.

Previously, we reported clinical characteristics that differentiate UCC from asthma and/or chronic obstructive pulmonary disease.¹⁵ Patients were more likely to have UCC if they were older, women, previously prescribed an increased number of cough medications, required a greater number and frequency of medical encounters for CC, and had normal or near-normal lung function.¹⁵ Thus, the aim of this multicenter questionnaire study was to build on these findings by identifying clinical risk factors that could differentiate UCC from explained CC (ECC).

METHODS

Study Design

This was a cross-sectional observational questionnaire study aimed at assessing the severity and influence of various health conditions and characteristics on CC. The study was conducted in conjunction with seven cough centers distributed geographically across the United States. The study received an exemption from each site’s institutional review board because the questionnaire was de-identified. By answering the questionnaire, the subjects were providing consent to participate.

Inclusion and Exclusion Criteria

Patients ages ≥18 years who presented with a history of CC, defined as a cough that lasted >8 weeks, were recruited for the study. Exclusion criteria included current smokers or smokers who quit within the past 12 months, currently taking an angiotensin-converting enzyme (ACE) inhibitor medication, and having a history of interstitial lung disease or uncontrolled heart failure. All the patients had been previously extensively evaluated with diagnostic testing at the investigator’s cough center to characterize their CC as ECC or UCC. This included but was not limited to spirometry, methacholine provocation, vocal cord videostroboscopy, computed tomography studies, and esophagogastroduodenoscopy to exclude underlying causes.

Data Collection

The participants completed a electronic data capture questionnaire that collected demographic and clinical data, including age, gender, body mass index (BMI), and smoking history in 2021–2022. In addition, the participants provided a medical history and completed the Hull Airway Reflux Questionnaire (HARQ) to assess CC severity. This validated patient-reported outcome measure consists of 14 questions about symptoms or triggers on a scale of 0–70.¹⁶ Respondents were classified as having UCC if they had no physician diagnosis explaining their CC.

Statistical Analyses

Data were analyzed by using GraphPad Prism 10 software (GraphPad Software, San Diego, CA). For continuous data, the Wilcoxon nonparametric rank sum test was used to compare groups when data did not follow a normal distribution, whereas one-way and two-way analyses of variance were used for normally distributed data. The normality of data distributions was assessed by using the Shapiro-Wilk test. For skewed data, the Kruskal-Wallis test was applied. Continuous data, depending on the distribution, were reported as either the mean ± standard deviation (SD) or median (interquartile range [IQR]). To explore potential predictors of CC severity, multiple logistic regression analyses were conducted. Variables such as age, gender, BMI, smoking status, and comorbid conditions were included in the models. The regression results are expressed as odds ratios (OR) with 95% confidence intervals (CI), and statistical significance was set at p ≤ 0.05. All statistical tests were two-tailed. Sensitivity analyses were performed when appropriate to ensure robustness of the findings.

RESULTS

Demographics

The study included 150 respondents: 53% were women (80/150), 32% were men (48/150), and 14.7% (20/150) did not disclose their gender. The mean age was 58.3 years for the women and 53.5 years for the men. The median (IQR) BMI was 28.7 kg/m² (4.9–33.7 kg/m²) for the women and 27.7 kg/m² (23.7–33.4 kg/m²) for the men. A significantly higher BMI (overweight, obese, or extremely obese) was found in 23.3% of the men (n = 35) and in 36.7% of the women (n = 55) with CC, which was statistically significant (p < 0.05) (Fig. 1). The patients were most commonly seen for their CC by an allergist/immunologist or their family medicine/internal medicine physician for CC, followed by an otolaryngologist then pulmonologist and gastroenterologist. Two-thirds of the patients with UCC or ECC had used at least one over-the-counter medication of any kind before seeing a physician.

Figure 1.

A significantly higher BMI (overweight, obese, and extremely obese) was found in 23.3% of the men (n = 35) and 36.7% of the women (n = 55) with UCC, which was statistically significant when compared with a normal BMI in men (*p < 0.05, ****p < 0.0001) and women (####p < 0.0001). Calculated BMI was made by using reported pounds and inches by gender (underweight, <18.5 kg/m²; normal, 18.5–24.9 kg/m²; overweight, 25–29.9 kg/m²; obese, 30–39.9 kg/m²; extremely obese, >40 kg/m²). BMI = Body mass index; UCC = unexplained chronic cough.

Characteristics of CC

Respondents experienced daily or almost daily CC for an average (range) of 4.9 years (4.3–5.7 years). The presence of asthma was the most prevalent comorbid condition reported by 40% of the respondents, followed by gastroesophageal reflux disease (GERD) (39.3%), allergic rhinitis (38%), and hypertension (33%), whereas 10% had no known comorbidities (Table 1). The most common family history conditions reported were hypertension (54.7%), cancer (41.3%), smoking tobacco products (38.7%), asthma (34.7%), and GERD (26.0%) (Table 1).

Table 1.

Prevalence of comorbid conditions or the use of ACE inhibitors associated with chronic cough

ACE = Angiotensin-converting-enzyme; GERD = gastroesophageal reflux disease; CB = chronic bronchitis; COPD = chronic obstructive pulmonary disease.

Understandably, 28.6% (14/49) discontinued ACE inhibitors due to CC. Among those who discontinued ACE inhibitors, 35.7% (5/14) reported no improvement in their cough, whereas 57% (8/14) experienced some improvement (Table 1).

Most respondents (65.6% [86/131]) had used nonprescription over-the-counter medications and 53.7% (29/54) indicated that these medications did not relieve their cough, whereas 37% (20/54) reported some improvement. There was no significant difference between the use of nonprescription treatments by patients with UCC (75%) and patients with ECC (62.7%) (p = 0.19).

Of the 132 respondents who responded to questions with regard to smoking history, 24.2% (32/132) previously smoked. The average (IQR) starting age for smoking was 18 years (16.7–20.1 years), and the average (IQR) cessation age was 34 years (30–38 years). The median (IQR) pack-year history was 8.2 (7.6–14.9) pack years. Among those who quit smoking, 59% (19/32) reported stopping for reasons other than CC, whereas only 4 subjects (12.5%) reported some improvement after cessation.

UCC versus ECC

Of the 150 respondents with CC, 19% (29/150) were classified as having UCC, whereas the remaining 81% (121/150) had ECC. The demographic features between the two groups were similar, with no significant difference in age (UCC: 54.0 ± 6.7 years versus ECC: 57.1 ± 6.7 years; p = 0.37) or gender distribution (UCC: 64.3% women versus ECC: 61.2% women; p = 0.84). The average duration of CC was slightly shorter in the patients with UCC (4.5 years) compared with the patients with ECC (5.1 years), although this difference was not statistically significant (p = 0.44). Among the previous smokers, the patients with UCC had a higher but nonsignificant median pack-year smoking history versus the patients with ECC (17.1 versus 10.1; p = 0.15).

Significant differences were observed in comorbidities between patients with UCC and those with ECC. Allergic rhinitis was less common (17.2% versus 43%; p = 0.01), whereas immunodeficiency was more common in those with UCC (13.8% versus 4.1%; p = 0.04). A diagnosis of vocal cord dysfunction (VCD) (also known as idiopathic laryngeal obstruction) was only endorsed by 18 respondents with ECC (12%) and did not emerge as a distinguishing comorbid characteristic that differentiated ECC from UCC.

A family history of CC was reported only in the patients with ECC (15% versus 0%; p = 0.03) and a family history of hypertension was more common in ECC (72% versus 50.4%; p = 0.03), whereas a family history of immunodeficiency was only reported by the patients with UCC (7% versus 0%; p = 0.004).

Determinants of Cough Severity: HARQ

Overall cough severity scores were significantly higher in the patients with UCC compared with the patients with ECC, with an OR of 4.8 (95% CI, 1.3–17.7). Other factors associated with more severe cough symptoms for respondents with UCC compared with ECC included a history of chronic bronchitis/emphysema (OR 9.1 [95% CI, 1.4–999]), hypertension (OR 3.4 [95% CI, 1.2–10.2]), past cigarette smoking (OR 6.6 [95% CI, 1.7–24.9]), female gender (OR 4.7 [95% CI, 1.5–15.1]), and higher BMI (OR 1.1 [95% CI, 1.0–1.2]). The receiver operator characteristic curve for the logistic regression model indicated that these factors had good accuracy for predicting severe cough symptoms in this population (Fig. 2).

Figure 2.

ROC curve for logistic regression model for determinants of CC severity symptoms greater in UCC versus ECC, including being female, having a higher BMI and comorbid chronic bronchitis and/or asthma, hypertension or previous cigarette smoking any time before the last 1 year. ROC = Receiver operator characteristic; CC = chronic cough; UCC = unexplained chronic cough; ECC = explained chronic cough; BMI = body mass index.

Triggers of UCC

The absence of postnasal drip was the most distinguishing clinical feature of UCC (OR 5.2 [95% CI, 1.7–16.0]; p = 0.008), with 44.8% of the patients with UCC reporting no postnasal drip compared with 16.5% of the patients with ECC (p = 0.004) (Table 5). Conversely, severe postnasal drip was more prevalent among the patients with ECC versus those with UCC (21.4% versus 6.9%) (Tables 4 and 5).

Table 5.

Significant determinants of UCC based on the Hull Airway Reflux Questionnaire multiple logistic regression analysis*#

UCC = Unexplained chronic cough; OR = odds ratio; CI = confidence interval.

*The most significant distinguishing feature of UCC based on the Hull Cough Hypersensitivity Questionnaire is the absence of postnasal drip.

#OR estimates for determination of UCC vs explained chronic cough.

Table 4.

Distinguishing features of unexplained vs explained chronic cough based on HARQ: univariate analysis

HARQ = Hull Airway Reflux Questionnaire; GERD = gastroesophageal reflux disease.

Bold indicates statistically significant.

Seasonality also distinguished UCC from ECC. In the patients with ECC, cough severity was associated with worsening in the winter (51.4%) and fall (48.6%). In contrast, the patients with UCC reported less seasonal variation, with 51% reporting no seasonal effect (Table 2).

Table 2.

Seasonality and exposure triggers for explained (ECC) versus unexplained (UCC) chronic cough (ECC = Yes responses: UCC = No responses)

In addition, cough triggered by eating certain foods was reported significantly less frequently by the respondents with UCC versus the respondents with ECC (10.34% versus 24.27%; p = 0.0259). Consistent with these findings, other factors, such as the severity of GERD and chest tightness, were not significant predictors of UCC (Table 2).

Common environmental triggers for ECC included cold air (38%), cigarette smoke (36.67%), and perfume (36%), whereas the patients with UCC reported a broader range of triggers, including car exhaust (18.67%), hair spray (18%), and spicy food (15.33%). Notably, 13.33% of the patients with UCC did not identify any specific triggers, which emphasizes the unpredictability of cough triggers for patients with UCC.

DISCUSSION

CC is a common and often debilitating condition that can significantly impact patients' quality of life and their ability to work, as demonstrated in this study (Table 2). The study aimed to differentiate clinical characteristics between UCC and ECC by examining demographic data, comorbidities, symptom profiles, and treatment responses in a cohort of 150 respondents with a diagnosis of CC. Our findings reveal important distinctions between these two entities, which have potential implications for facilitating the diagnosis of UCC. In our cohort, UCC was found in 19% of the respondents, which is higher than what has been reported in the general population in which the estimated prevalence of UCC ranges from 0.4% to 1.3% in the general adult population and is 1.4% to 8.7% of CC cases for adults presenting to primary care and cough clinics.^17–23 The higher prevalence of UCC in this study compared with the normal population reflects the expertise of the participating experts who all operate CC clinics and were recruiting well-characterized patients with CC is a strength of this study.

Asthma, chronic bronchitis, and allergic rhinitis are known common causes of CC, which were diagnoses assigned to the patients with ECC.^24–27 The lower prevalence of allergic rhinitis (17.2%) and the absence of postnasal drip in UCC suggests that allergic and/or eosinophilic airway inflammation plays a less central role for this condition. This distinction is critical for clinicians when considering prescribing oral or topical glucocorticoids, which are less effective in UCC but, nonetheless, prescribed repeatedly for these patients.^28,29 When considering treatment options for patients with CC, considering the absence of these conditions could be important clinical clues for the diagnosis of UCC.

Notably, 39.3% of the respondents with CC reported having GERD (Table 1) and 30.7% of the respondents with ECC were assigned a physician diagnosis of GERD as a cause of their CC (Table 3), which reaffirms the need for a thorough evaluation and empiric treatment with a proton-pump inhibitor for 4 to 6 weeks to distinguish UCC from ECC. Surprisingly, very few of the respondents with ECC endorsed a diagnosis of VCD (12%) as a comorbid condition, although, in a previous study, we identified shortness of breath as the most common presenting symptoms of VCD (Table 3).³⁰

Table 3.

Patients with explained vs patients with unexplained chronic cough*

GERD = Gastroesophageal reflux disease; CR = chronic rhinitis; VCD = vocal cord dysfunction; COPD = chronic obstructive pulmonary disease.

^*Twenty-three patients did not have an underlying diagnosis associated with their cough.

The proportion of respondents self-medicating with nonprescription therapies to treat their CC was not different between ECC or UCC. This finding further illustrates the prolonged patient journey who experience CC in general, which results in unnecessary health-care costs. Although this clinical characteristic was previously reported for patients with UCC, it is does not seem to be a distinguishing feature from other causes of CC, which emphasizes the need for determining risk identification factors for UCC that would identify appropriate patients most likely to benefit from UCC-specific therapies, which would shorten the time from diagnosis to resolution of symptoms.^13,15,31

Other potential important clinical clues that could facilitate a UCC diagnosis is that in those patients who were ex-smokers, smoking cessation was ineffective at improving their cough. In addition, UCC symptoms were not seasonal and were provoked by a range of irritant chemical and environmental triggers compared with ECC (Table 2). As noted, only 12% of the respondents with ECC indicated that they had a confirmed diagnosis of VCD, which is also triggered by chemical irritants. This finding is strengthened by the fact that all the patients were under the care of CC experts and previously had extensive evaluations to rule out comorbid conditions, e.g., VCD. Thus, this emphasizes the importance of recognizing that irritants are triggers independent of a CC diagnosis. It is possible to speculate similar neurogenic pathways between CC and VCD that may not be discernible by current diagnostic techniques, e.g., videostroboscopy. Therefore, use of biofeedback techniques recommended for VCD may also be useful for some cases of patients with CC that is triggered predominantly by chemical irritants.

Although this study provides valuable insights into the distinctions between UCC and ECC, several limitations should be acknowledged. First, the study’s reliance on self-reported data when using a electronic data capture questionnaire data base may introduce recall bias, particularly in the assessment of symptom severity and triggers. However, using validated questionnaires, e.g., HARQ, which has been validated for assessing cough severity, mitigates the issue with regard to recall bias. Furthermore, all the subjects were recruited and monitored closely by the CC center investigators and, therefore, more cognizant of their long diagnostic journey. It is worth noting that the Leicester cough questionnaire had not been validated during the time this study was conducted and was, therefore, not used.³² However, most of the questions contained in the Leicester cough questionnaire were also captured by the HARQ. Finally, the absolute number of patients with UCC was small (n = 29). Applying these results to larger well-characterized UCC populations would help validate our findings.

Second, the cross-sectional design of the study limits the ability to draw definitive conclusions because reported comorbid conditions could not be verified. For example, it was not possible to know the specific immunodeficiency diagnosis that was reported by the respondents. Further longitudinal studies are needed to explore the presence or absence of specific symptoms and triggers of UCC and their associations with comorbid conditions, e.g., immunodeficiency, found in this study. These studies should incorporate a translational component with the purpose of identifying a diagnostic biomarker(s) for UCC that could be used to predict responses to cough-specific therapies such as P2X purinoreceptor 3 antagonists in patients with UCC.

Third, the study population was limited to the U.S. population, which may affect the generalizability of the findings. However, the cough centers that participated in this study were strategically selected to minimize this effect, thereby making the findings of this study more impactful. However, future studies should aim to include more diverse populations to assess whether the observed differences between UCC and ECC hold true across different settings and populations.

CONCLUSION

The findings of this study may help shape a common patient phenotype that presents with CC and ultimately have a final diagnosis of UCC. For example, a woman who is overweight and an ex-smoker with a perennial CC, without a medical history of Allergic rhinitis, postnasal drainage, or chronic bronchitis and/or asthma, triggered by a broad range of environmental irritants should heighten a clinician’s suspicion for the diagnosis of UCC. By considering these clinical features, the diagnosis of UCC could be made earlier, thereby reducing unnecessary office visits, ineffective medications, and the overall burden of CC. When cough-specific therapies, such as purinergic receptor antagonists, become available, the relevance of these clinical clues should be confirmed by assessing treatment outcomes for patients selected with the specific clinical characteristics identified in this study (Table 6).

Table 6.

Multiple logistic regression analysis for determinants of cough severity in UCC vs ECC*

UCC = Unexplained chronic cough; ECC = explained chronic cough; OR = odds ratio; CL = confidence level; BMI = body mass index.

*The severity of cough symptoms is greater in UCC vs ECC for women and for higher BMI, and those with comorbid chronic bronchitis, asthma, hypertension, or history of cigarette smoking any time before the last year.