Introduction
The Leicester Cough Questionnaire (LCQ) was designed to assess the effect of cough on patients’ quality of life [1,2]. It comprises three domains: physical (eight items), psychological (seven items), and social (four items), totaling 19 items. The LCQ includes heterogeneous items distributed among the domains. The physical domain features items that can indicate the underlying etiology of cough, such as the presence and severity of phlegm [3,4], hypersensitivity to irritants [5], and sleep disruption due to nocturnal cough [6], while the psychological and social domains focus on patients’ perceptions of cough severity, including anxiety [7] and concern about annoying others [8].
Although psychological and social factors are fundamental to quality of life, these domains are frequently neglected in the management of chronic cough. Examining the interactions among physical, psychological, and social factors could provide improved understanding of coughing mechanisms. Nevertheless, the specific impacts and underlying mechanisms among these domains have not been fully investigated. Methodological advancements in causal inference have markedly enhanced our comprehension of causal relationships in observational research [9,10]. Mediation analysis, in particular, is utilized to determine whether an intermediary variable exists within the pathway connecting an independent and dependent variable [11]. Mediation analysis separates the total influence of exposure on the outcome into direct and indirect effects through the mediator and describes the associations between exposure, mediator, and outcome [12]. A direct effect is defined as the influence of the exposure on an outcome without involvement of mediators. An indirect effect is the influence of exposure transmitted through a mediator affecting the outcome. This analytic approach enriches our understanding of exposure-mediator-outcome dynamics. To elucidate the progression of cough-related symptoms within the physical, psychological, and social contexts, this study sought to evaluate the associations among chronic cough traits across these dimensions using the LCQ, assess their causal relationships, and measure corresponding mediation effects.
Materials and Methods
1. Study patients and data collection
Individuals aged ≥18 years who presented with chronic cough were enrolled from 16 respiratory centers in the Republic of Korea between March 2016 and February 2018. Chronic cough was defined as a cough persisting for more than 8 weeks [13]. Enrollment was limited to patients who responded to the LCQ at baseline. Individuals with chest radiographic abnormalities were excluded from the study. Potential etiologies of chronic cough were determined by pulmonary specialists at each site, following the Korean cough management guidelines [13,14]. Demographic data, symptom duration, identified etiology of cough, and LCQ scores were documented at initial consultation. Study findings were validated in a separate cohort from a multicenter prospective investigation for the Korean LCQ validation. Data were analyzed for research purposes as of April 26, 2024.
Ethical approval was granted by the Institutional Review Board of the Inje University Ilsan Paik Hospital (approval number: 2017-12-025), and the study adhered to the principles of the Declaration of Helsinki. The requirement for informed consent was waived due to the retrospective design of the research. At no point did the authors access personally identifiable participant data during or following the data collection process.
2. Leicester cough questionnaire
The LCQ is a validated instrument designed to assess cough-specific quality of life [1,2]. Each of the 19 items, distributed across three domains, was rated on a 7-point Likert scale, and the mean scores from each domain were summed to obtain the total score, resulting in a range from 3 to 21 points. Higher total scores corresponded to a better quality of life. The physical domain comprised eight items: chest/stomach pain (LCQ1), presence of bothersome phlegm (LCQ2), tiredness (LCQ3), hypersensitivity to irritants (LCQ9), sleep disturbance (LCQ10), frequency of coughing bouts (LCQ11), hoarseness of voice (LCQ14), and loss of energy (LCQ15). The psychological domain consisted of seven items: feeling in control of cough control (LCQ4), embarrassment (LCQ5), anxiety (LCQ6), frustration (LCQ12), feeling of being fed-up (LCQ13), concerns about serious illness (LCQ16), and worries regarding others’ perceptions (LCQ17). The social domain was assessed by four items: interference with daily work (LCQ7), hindrance to overall enjoyment of life (LCQ8), interruption of conversations or phone calls (LCQ18), and causing annoyance to partner, family, or friends (LCQ19).
3. Statistical analysis
All statistical analyses were conducted using R software version 3.6.0 (R Foundation for Statistical Computing, Vienna, Austria). Continuous variables are summarized as mean±standard deviation or median with interquartile range, while categorical variables are expressed as number (percentage). A correlation matrix was generated utilizing Spearman’s rank correlation, and the strength of correlations was categorized based on the Spearman coefficient: >0.7 as very strong, 0.6-0.7 as strong, 0.3-0.6 as moderate, and <0.3 as weak. In the correlation networks, each LCQ item corresponded to a node, with the node color indicating the characteristic domain: physical (red), psychological (green), and social (blue). Node size corresponded to the mean score of each item. Statistically significant correlations between nodes (p<0.05) were depicted as edges. The darkness of an edge indicated the correlation strength (Spearman’s coefficient). The igraph package was employed for network visualization. Mediation analysis, conducted with the mediation package, assessed whether the relationship between two domains was mediated by a third domain and quantified the mediation effect. The counterfactual framework facilitated the partitioning of the total effect into direct and indirect components even when inter-domain interactions existed. The mediated proportion quantifies how much of the total effect of exposure on outcome is attributable to the indirect pathway, calculated as the ratio of indirect effect to total effect, where the total effect is the sum of direct and indirect effects. This metric provides insight into the extent that the relationship between two domains is accounted for by mediation through the third domain. Potential confounders, including age and sex, were adjusted when evaluating interactions among domains. A conceptual illustration of the mediation analysis is provided in Figure 1.
Results
1. Characteristics of the study patients
This study included a total of 255 patients. Baseline characteristics are detailed in Table 1. The mean age was 47.7±14.3 years, and females comprised 63.1% of the cohort. The average total LCQ score was 11.2±3.1. Scores for the physical, psychological, and social domains were 4.1±0.9, 3.5±1.2, and 3.6±1.3, respectively. Figure 2 displays the median and interquartile ranges for each LCQ item. The highest scores were observed for chest/stomach pain (LCQ1), frustration (LCQ12), and voice hoarseness (LCQ14), while feeling in control of cough (LCQ4) had the lowest score.
2. Network analysis of the LCQ items
Figure 3 illustrates the correlation patterns among items of the LCQ. In general, correlations between domains were weaker for items in the physical domain (red circles) compared to those in the psychological (green circles) or social (blue circles) domains (Figure 3A, B). Among the physical domain items, none showed very strong correlations (r>0.7) with items from the psychological or social domains. Only two physical domain items demonstrated strong correlations between domains (r=0.6-0.7): (1) tiredness (LCQ3) with embarrassment (LCQ5) (r=0.676), interference with daily work (LCQ7) (r=0.640), or interference with overall enjoyment (LCQ8) (r=0.640); and (2) cough bout frequency (LCQ11) with feeling of being fed-up (LCQ13) (r=0.602). The correlations among items within the physical domain ranged from weak to moderate, indicating that the physical aspects of chronic cough are relatively distinct. Notably, hypersensitivity to irritants (LCQ9) was only weakly associated with other items.
Conversely, items in the psychological and social domains exhibited moderate to very strong correlations between domains (Figure 3C). There were very strong correlations (r>0.7) observed between (1) interference with daily work (LCQ7) and embarrassment (LCQ5) (r=0.751); (2) feeling of being fed-up (LCQ13) and interference with overall life enjoyment (LCQ8) (r=0.704) or annoyance to acquaintances (LCQ19) (r=0.751); and (3) anxiety (LCQ6) and interference with overall life enjoyment (LCQ8) (r=0.754). Correlations within and between psychological and social domain items were moderate to very strong, consistent with strong interrelationships among these aspects.
3. Mediation analysis
Table 2 presents the total, direct, and indirect effects of each domain on the others, along with the proportion mediated. The interactions among the physical, psychological, and social domains reached statistical significance (all p<0.05). The highest total effect estimated was from the psychological domain to the social domain. The second highest total effect was observed from the physical to social domains; however, 76.1% of this effect was mediated by the psychological domain. Likewise, the social domain accounted for 67.1% of the total effect from the physical to psychological domain. In contrast, the physical domain explained only a small part of the total effect from the psychological domain to the social domain, and reciprocally. The estimated proportions mediated by the physical domain for the psychological-to-social and social-to-psychological effects were 12.8% and 18.0%, respectively. Summaries of mediation models describing effects among physical, psychological, and social domains are provided in Figure 4.
4. External validation
The findings from the mediation analysis were subsequently validated in an independent cohort consisting of 203 patients diagnosed with chronic cough. The mean patient age was 49.3±14.0 years, with females constituting 60.6% of the validation cohort. The data obtained from the external cohort confirmed the primary analysis, exhibiting a similar pattern. Notably, the estimated total effect was most pronounced in pathways originating from the psychological domain towards the social domain. The mediation effect of the psychological domain accounted for 64.8% of the total effect exerted by the physical domain on the social domain, while 73.9% of the total effect of the physical domain on the psychological domain was mediated by the social domain. Conversely, the mediation proportions attributed to the physical domain in the pathways from psychological-to-social domain and vice versa were 18.5% and 14.3%, respectively. Detailed results are provided in Supplementary Table S1.
Discussion
This study investigated the intricate intercorrelations among LCQ physical, psychological, and social domain items by employing both network analysis and mediation analysis. Through network analysis, robust associations were identified between psychological and social influences on coughing, whereas the physical components exhibited weaker interrelationships with psychological or social features. The mediation analysis provided further insight by partitioning total effects into direct and indirect pathways. Findings indicated that the prominent total effects of the physical domain on psychological and social domains were largely mediated by another domain within the model. In addition, the physical domain demonstrated a negligible mediating effect in the interactions between psychological and social domains.
Understanding the multidimensional effects of cough on quality of life is essential for accurate cough severity measurement [15]. Nevertheless, the introduction of the LCQ into clinical practice may be restricted without a thorough comprehension of the complex interrelationships among questionnaire items. This study elucidates the distinct contributions of physical, psychological, and social domains in shaping cough-related symptoms. In particular, items in the physical domain displayed comparatively greater independence, as reflected by weaker intra-domain and inter-domain correlations relative to those observed for the psychological and social domains. For example, ‘hypersensitivity to irritants’ within the physical domain demonstrated only minimal correlations with other items. This observation implies that individual physical symptoms may not adequately reflect the total severity within the domain, and that intervening on a single physical manifestation is unlikely to result in broad improvements across other components or overall quality of life. Thus, a more nuanced and multidimensional approach may be needed to fully characterize the range of physical impairment in patients with chronic cough. Alternatively, the relatively weaker associations among physical symptoms may stem from intrinsic variability in clinical presentation or reduced sensitivity in capturing overlaps with psychological and social constructs. Further research should aim to clarify these possibilities.
In contrast, items within the psychological and social domains exhibited robust interrelationships. Social limitations resulting from chronic cough seem to intensify psychological stress [16]. Additionally, psychological influences, such as anxiety or frustration, may have a more pronounced impact on negative social outcomes than physical discomfort alone. These findings underscore that patients experiencing marked psychological or social impacts may benefit from focused assessment of their cough perception, complemented by educational and counseling strategies to address related psychosocial concerns. Supporting these conclusions, French et al. [17] found that health-related challenges in individuals with chronic cough predominantly pertained to psychosocial dysfunction rather than physical complications in a prospective study evaluating 28 patients before and after intervention. Therefore, integrating assessments of psychosocial perceptions is pivotal for optimal chronic cough management.
The mediation analysis further quantified the causal pathways of cough traits across domains, demonstrating that the direct effects of the physical domain on the psychological and social domains were less pronounced than the indirect effects. This highlights the substantial mediating influence of the psychological and social domains in modulating the impact of the physical component. In contrast, the physical domain exhibited minimal mediation effects within the interplay between psychological and social domains, which showed reciprocal influences. Our findings illustrate the pivotal role of psychosocial components in elucidating the mechanisms driving cough-related symptoms.
There are limitations to our study. Restricting our sample to Korean patients may constrain the external validity of the results, as cultural variations could modulate the psychological and social responses to coughing. Although we performed validation using an independent patient cohort, additional research is necessary to assess the applicability of these findings to broader populations. Despite mediation analysis being conceptually anchored in causal inference, the cross-sectional nature of this study prohibits drawing definitive causal conclusions. Therefore, the reported direct and indirect effects should be interpreted as associative rather than causal. While statistical adjustments were made, the influence of unmeasured confounders cannot be fully dismissed. Thus, our results indicate potential pathways linking domain-specific impairments without confirming explicit causal relationships. Nonetheless, we believe that these analyses provide valuable insights into the complex interactions among cough domains, informing efforts to enhance patient management.
In conclusion, utilizing network analysis, we identified intricate interconnections among the physical, psychological, and social domains of the LCQ. Results indicated that the physical components appeared relatively distinct and weakly associated, whereas the psychological and social domains were highly interconnected. Mediation analysis demonstrated the prominent indirect influence of psychosocial components on the physical domain, emphasizing the value of addressing psychosocial aspects in managing chronic cough. Interpreting LCQ scores by domain may help clinicians recognize central psychosocial determinants of symptom burden and inform targeted, multidisciplinary treatment strategies. Further investigations are warranted to clarify the clinical significance of these observations.
PDF Links
PubReader
ePub Link
XML Download
Data Sharing Statement
Full text via DOI
Supplement
Print
Download Citation
