Introduction

Pneumonia is a form of acute respiratory infection caused by viruses, bacteria, and fungi and a major cause of morbidity and mortality globally. According to the World Health Organization (WHO)’s 2019 Global Health Estimates, pneumonia and other lower respiratory infections were ranked as the fourth leading cause of death and disability-adjusted life years1. In Korea, deaths due to pneumonia have continued to increase since 2000, and in 2019, the death rate for pneumonia stood at 45.1 per 100,000 population, ranking as the third leading cause of deaths2. In a recent study in Korea, the hospitalization rate due to bacterial pneumonia was reported to be 161.5 per 10,000 population over the age of 653. Several risk factors, such as diabetes, cardiovascular disease, smoking, asthma, and chronic obstructive pulmonary disease (COPD), have been shown to increase the likelihood of the development of pneumonia4.

Chronic rhinosinusitis (CRS) is a chronic upper airway inflammation that can greatly influence the health-related quality of life of patients and the socioeconomic burden5. Several epidemiological studies have shown that CRS is associated with lower airway diseases such as asthma6, bronchiectasis7, and COPD8. This interdependence of the upper and lower respiratory tracts has led to the concept of the “unified airway”, which considers the upper and lower airways to be a single functional unit9.

Despite the correlations between the upper and lower airways, few studies have evaluated the association of CRS and pneumonia. A retrospective study using X-ray findings reported that the frequency of sinusitis was found to be 84% among patients with pneumonia10. A previous study in the U.S. showed an association between community-acquired pneumonia and CRS with sphenoid involvement (odds ratio [OR] = 19.76, 95% confidence interval [CI] = 8.78–44.47)11. Furthermore, population-based studies evaluating the risk of pneumonia in patients with CRS are lacking.

This study aimed to compare the incidence of pneumonia between patients with CRS and those without CRS (control) using a national sample cohort from the Korean population. An understanding of the association between CRS and pneumonia will contribute to disease prevention.

Results

Table 1 shows the general characteristics of the participants. Age, sex, income, region of residence, and 1-y pre-index date (ID) pneumonia showed no difference between the CRS and control groups due to matching (standardized difference = 0.00). Total cholesterol level, systolic blood pressure (SBP)/diastolic blood pressure (DBP), fasting glucose level, obesity, smoking status, alcohol consumption, and Charlson Comorbidity Index (CCI) scores were similar in the CRS total group and controls (standardized difference < 0.20), while the prevalence rates of asthma and COPD were different between the two groups (standardized difference ≥ 0.20).

Table 1 General characteristics of participants.
Full size table

The adjusted estimated value (EV) of the incidence of post-ID pneumonia did not reach statistical significance at 1-y (P = 0.059) and 2-y (P = 0.310) post-ID. However, the adjusted EV showed statistical significance at 3-y post-ID (EV = 0.017, 95% CI = 0.002–0.031, P = 0.030) (Table 2).

Table 2 Simple and multiple linear regression model (estimated value [95% confidence intervals]) for post index date of pneumonia (post-ID pneumonia) periods in total chronic rhinosinusitis compared to control group and subgroup analysis according to age and sex.
Full size table

In the subgroup analyses according to age and sex, statistical significance was seen in the younger age group (< 60 years old) at 3-y post-ID (EV = 0.025, 95% CI = 0.011–0.038, P < 0.001) and the female group at the 1-y (EV = 0.037, 95% CI = 0.012–0.062, P = 0.004) and 3-y (EV = 0.022, 95% CI = 0.000–0.044, P = 0.046) post-ID periods (Table 2).

When analyzing the CRS with nasal polyps (CRSwNP) (Table 3) and CRS without nasal polyps (CRSsNP) (Table 4) groups separately, an association between CRS and pneumonia was found in the 1-y (EV = 0.035, 95% CI = 0.010–0.060, P = 0.007) and 3-y (EV = 0.037, 95% CI = 0.014–0.061, P = 0.002) post-ID periods in the CRSsNP group only.

Table 3 Simple and multiple linear regression model (estimated value [95% confidence intervals]) for post index date of pneumonia (post-ID pneumonia) periods in chronic rhinosinusitis with nasal polyp compared to control group and subgroup analysis according to age and sex.
Full size table
Table 4 Simple and multiple linear regression model (estimated value [95% confidence intervals]) for post index date of pneumonia (post-ID pneumonia) periods in chronic rhinosinusitis without nasal polyp compared to control group and subgroup analysis according to age and sex.
Full size table

Discussion

This study showed the association between CRS and pneumonia. This association was seen only in the younger age, female sex, and CRSsNP subgroups.

The concept of “united airway disease” or “one linked airway disease” has emerged, and allergic rhinitis, sinusitis, and asthma are expressed in different parts of the respiratory tract as one pathological process that is inflammatory in nature12. In addition, the concept has expanded beyond asthma, and an association with CRS has been reported in other lung diseases, including bronchiectasis7,13 and COPD8. In a systematic review, the pooled prevalence of CRS in adults with bronchiectasis was 62%, and CRS was associated with a greater degree of bronchiectasis severity, poorer quality of life, reduction in olfactory detection, and elevated levels of inflammatory markers7. A large retrospective study in a U.S. population found that almost half of the patients with bronchiectasis had comorbid CRS13. In a Taiwanese population-based study, COPD was associated with CRSsNP8.

On the other hand, there have been few studies on the association between CRS and pneumonia. Previous studies have mainly reported the incidence of sinusitis and pneumonia in intensive care unit (ICU) patients14,15,16. A meta-analysis on ventilator-associated sinusitis showed that 41% of patients with ICU-derived sinusitis also developed hospital-acquired pneumonia17. They showed that the same pathogen was isolated from culture specimens of the lung and sinuses in 59% of patients17. However, few studies have evaluated the association of CRS and pneumonia in the general population.

We hypothesize that there are several possible mechanisms underlying the association between CRS and pneumonia. First, the nasal cavities constitute the first barrier of the whole airway system that has immunological functions against bacteria, viruses, and other pathogens. Several studies have indicated that CRS is associated with a failure of the mechanical and immunological barriers across the nasal mucosa18,19. Furthermore, patients with CRS have been shown to have impaired mucociliary function20, which increases susceptibility to respiratory infections21. Second, there is biological plausibility for a relationship between CRS and pneumonia. Haemophilus influenzae, Moraxella catarrhalis, and Streptococcus pneumoniae are the primary respiratory tract pathogens involved in both acute and chronic rhinosinusitis22. Staphylococcus aureus, S. epidermidis, and anaerobic Gram-negative bacteria predominate in CRS23. Moreover, these are the main causative pathogens of community-acquired and hospital-acquired pneumonia24. Third, the effectiveness of the pneumococcal vaccination may be reduced in patients with CRS. CRS is known to be one of the most common manifestations of humoral immunodeficiency25. Specific antibody deficiency (SAD) is a primary immune deficiency associated with a qualitative defect in antibody function. SAD is defined as a poor antibody response to polysaccharide antipneumococcal vaccine, such as Pneumovax®, in the presence of recurrent or chronic sinopulmonary infections26. A meta-analysis reported a high prevalence of humoral immune deficiency in CRS, with 8–34% of subjects diagnosed with SAD27. Last, an increase in cytokines induced by CRS-related inflammation may promote systemic inflammation, including in the lung. However, it is necessary to determine whether CRS directly causes pneumonia, or whether both diseases are simply different manifestations of common pathophysiology.

The association between CRS and pneumonia was seen only in the younger age, female sex, and CRSsNP subgroups. Since the risk of pneumonia increases with age and the presence of comorbidities24,28, the effects of other risk factors, such as CRS, may be emphasized in the younger age group. A study showing increased odds of having community-acquired pneumonia in CRS with severe sphenoid involvement attributed this association to the posterior location with the possibility of postnasal drainage into the lower respiratory tract11. One of the explanations for the linkage between the upper and lower airways includes postnasal drainage of inflammatory material into the lower airway. A previous study evaluating sex-specific differences in CRS patients showed that women reported more problems with postnasal drainage than men29. While CRSwNP is characterized by eosinophil-related inflammation, Asian CRSsNP patients show prominent neutrophil-related inflammation30. Although neutrophils are essential components of the immune response to pulmonary infection, they can also be harmful to host tissue. Excessive neutrophil activation results in severe tissue damage as a result of the exacerbated release of toxic agents, including proteinases, cationic polypeptides, cytokines, and reactive oxygen species31. However, further studies may be required to clarify the pathophysiology behind this finding.

The present study has some limitations. First, the incidence of pneumonia was relatively low and there may be potential for diagnostic misclassification. However, in Korea, the National Health Insurance system covers almost the entire population and is characterized by cost effectiveness and easy access to medical institutions. Given the distinct clinical characteristics of lower respiratory tract infection, misclassification of patients with signs and symptoms of pneumonia into the nonpneumonic diagnostic category is unlikely32. Second, there was a possibility of unadjusted covariates because there were no data about pneumococcal vaccination, details of medication use such as corticosteroid or antibiotics, immunosuppression status, atopy, and allergies. However, we adjusted for several confounding factors, including asthma, COPD, and other comorbidities that could affect the risk of pneumonia. We used CCI, which is widely used to classify various comorbid conditions and underlying diseases, including diabetes, renal failure, any malignancy, and AIDS/HIV33. Third, the incidence rates of community-acquired and hospital-acquired pneumonia cannot be investigated separately. Nevertheless, the strength of this study is that it is a population-based study that revealed the relationship between CRS and pneumonia.

In conclusion, CRS patients were more likely to have pneumonia. Physicians should be aware of the possible risk of pneumonia in patients with CRS.

Materials and methods

Study population

We used data from the Korean National Health Insurance Service-Health Screening Cohort for this study, and a comprehensive explanation for this cohort is provided elsewhere34. This study was performed in accordance with relevant guidelines/regulations and approved by the Institutional Review Board (IRB) of Hallym University (IRB No: 2019-10-023). The need for written informed consent was waived due to its minimal risk involved in the retrospective analysis.

Participant selection

Among the 514,866 participants with 615,488,428 medical claim codes, participants were selected as the CRS total group (n = 8,560) according to the definition in our previous study35. Participants without a history of CRS were selected as a control group (n = 506,306). Participants in the CRS total group who diagnosed with CRS in 2002 were removed (n = 1,366) to calculate the history of pneumonia from the date of first diagnosis of CRS (ID) to the date 1 year prior (pre-ID pneumonia for 1 year). In the control group, participants who died before 2003 or had no records after 2003 (n = 34) and those diagnosed with J32 (chronic rhinosinusitis) or J33 (nasal polyps) using the ICD-10 code at least once (n = 124,993) were excluded. If the participants were treated for the first time from 2013 to 2015, they were excluded from the CRS total group to count the diagnosis of pneumonia for 3 years following the ID (n = 801). The participants with CRS were 1:4 matched with the controls for age, sex, income, and region of residence. The CRS total and control groups were additionally matched by pneumonia incidence for 1 year pre-ID (0 times; ≥ 1 time). To minimize selection bias, the controls were selected with random number order. The ID of the controls was set as the ID of their matched participant with CRS. Therefore, each CRS participant and his or her matched controls had the same ID. During the matching process, 355,707 control participants were excluded. Finally, 6,393 of the total CRS participants were 1:4 matched with 25,572 control participants. Additionally, the final selected participants were classified into CRSwNP (n = 3,304) with matched controls (n = 13,216) and CRSsNP (n = 3,089) with matched controls (n = 12,356) (Fig. 1).

Figure 1
figure 1

A schematic illustration of the participant selection process that was used in the present study. Of a total of 514,866 participants, 6,393 of the total CRS participants were matched with 25,572 control participants for age, sex, income, region of residence, and pneumonia for 1 year prior to the index date. Additionally, the final participants were further classified into two groups: CRS with nasal polyps (n = 3,304) with matched controls (n = 13,216) and CRS without nasal polyps (n = 3,089) with matched controls (n = 12,356).

Full size image

Chronic rhinosinusitis (Exposure)

CRS with/without nasal polyps were defined using ICD-10 codes (J32, chronic sinusitis; J33, nasal polyp) and head and neck computed tomography evaluations following our previous study35, 36.

Pneumonia (Outcome)

Pneumonia was identified based on ICD-10 codes (J12 to J18). Patients with these codes, who underwent chest X-ray (claim codes: G2101-G2105, G2111, G2112, G2121, G2201, G2301-G2305, G2322) or chest CT (claim codes: HA424, HA434, HA444, HA454, HA464, HA474) were defined as having pneumonia. Pneumonia for 1 year period pre-ID was categorized as 0 times and ≥ 1 time. The number of pneumonia diagnoses was counted from the ID to the date after the first year period (1-y post-ID pneumonia), second year period (2-y post-ID pneumonia), and third year period (3-y post-ID pneumonia).

Covariates

Age groups were divided by 5-year intervals into 10 groups. Income groups were classified into 5 classes (class 1 [lowest income] – 5 [highest income]). The region of residence was grouped into urban and rural areas according to our previous study37. Tobacco smoking, alcohol consumption, and obesity were categorized in the same way as in our previous study38. Total cholesterol (mg/dL), SBP (mmHg), DBP (mmHg), and fasting blood glucose (mg/dL) were measured. The CCI39 score ranged from 0 (no comorbidities) to 29 (multiple comorbidities) except for COPD. Asthma and COPD were defined following our previous studies40, 41.

Statistical analyses

The general characteristics of the CRS total and control groups were compared using standardized differences. Simple and multiple linear regressions were analyzed to calculate estimated values and 95% CIs for 1-y post-ID pneumonia, 2-y post-ID pneumonia, and 3-y post-ID pneumonia in the CRS total, CRSwNP, and CRSsNP groups compared to each matched control group. Both simple and multiple linear regressions were stratified by age, sex, income, and region of residence. In multiple linear regression, we used 2 models for analyses. Model 1 was adjusted for obesity, smoking status, alcohol consumption, total cholesterol, SBP, DBP, fasting blood glucose, CCI score, asthma history, and COPD history. Model 2 was adjusted for model 1 plus 1-y pre-ID pneumonia. For the subgroup analyses, we divided participants by age (< 60 years old and ≥ 60 years old) and sex (males and females). Simple and multiple linear regressions were calculated. All analyses were two tailed, and significance was defined as P-values less than 0.05. SAS version 9.4 (SAS Institute Inc., Cary, NC, USA) was used for statistical analyses.