# paper.pdf ## Metadata - PDFFormatVersion=1.6 - Language=en - IsLinearized=false - IsAcroFormPresent=false - IsXFAPresent=false - IsCollectionPresent=false - IsSignaturesPresent=false - Author=Xufeng Fu; Xiaohua Li; Yuanmei Gao; Nan Wu; Ting Ao; Wentao Ni - CreationDate=D:20260429233905+05'30' - Creator=Servigistics Arbortext Advanced Print Publisher 11.1.4546/W-x64 - Custom.EPSprocessor=PStill version 1.84.42 - Keywords=healthy sleep characteristics; pneumonia; prospective cohort study; risk reduction; UK biobank - ModDate=D:20260430094318+05'30' - Producer=PDFlib+PDI 9.0.7p3 (C++/Win64) - Subject=DOI: 10.1093/ofid/ofag233; Open Forum Infectious Diseases, 2026-04-21.; Abstract: Background: Sleep disturbances are hypothesized to increase susceptibility to pneumonia; however, most evidence arises from studies examining isolated sleep characteristics. The relationship between a multidimensional healthy sleep score and pneumonia risk remains unclear. Herein, the association between a multidimensional healthy sleep score and risk of pneumonia was evaluated. Methods: This prospective analysis encompassed 361 589 participants from the United Kingdom (UK) Biobank (mean age: 56.1 years; 46.1% male). Five sleep characteristics—sleep duration, chronotype, insomnia symptoms, snoring, and daytime sleepiness—were used to construct a healthy sleep score (range: 0–5). Participants were categorized into poor (0–1), intermediate (2–3), and healthy (4–5) sleep score categories. Cox proportional hazards models were employed to estimate hazard ratios (HRs) and 95% confidence intervals (CIs) for pneumonia incidence. Results: Over a median follow-up period of 13.2 years, there were 20 116 new cases of pneumonia. A healthy sleep score was associated with a 26% lower risk of pneumonia (adjusted HR: 0.74, 95% CI: 0.69–0.80). This association exhibited a graded pattern, with higher sleep scores associated with progressively lower risk. The absence of frequent daytime sleepiness (HR: 0.79, 95% CI: 0.74–0.85) and insomnia symptoms (HR: 0.88, 95% CI: 0.86–0.91) exhibited the strongest component-specific inverse associations. Associations were stronger among participants younger than 60 years and females (P for interaction <0.001 for both). Conclusions: A multidimensional healthy sleep score is associated with a lower risk of pneumonia. These findings support the potential relevance of multidimensional sleep assessment to respiratory infection susceptibility and warrant further investigation. - Title=Multidimensional Healthy Sleep Characteristics and Pneumonia Risk: A Large-scale, Population-based Prospective Cohort Study - dc:format=application/pdf - dc:identifier=10.1093/ofid/ofag233 - dc:lang=en - dc:publisher=Oxford University Press - dc:title=Multidimensional Healthy Sleep Characteristics and Pneumonia Risk: A Large-scale, Population-based Prospective Cohort Study - dc:creator=Xufeng Fu; Xiaohua Li; Yuanmei Gao; Nan Wu; Ting Ao; Wentao Ni,Xiaohua Li,Yuanmei Gao,Nan Wu,Ting Ao,Wentao Ni - dc:description=DOI: 10.1093/ofid/ofag233; Open Forum Infectious Diseases, 2026-04-21.; Abstract: Background: Sleep disturbances are hypothesized to increase susceptibility to pneumonia; however, most evidence arises from studies examining isolated sleep characteristics. The relationship between a multidimensional healthy sleep score and pneumonia risk remains unclear. Herein, the association between a multidimensional healthy sleep score and risk of pneumonia was evaluated. Methods: This prospective analysis encompassed 361 589 participants from the United Kingdom (UK) Biobank (mean age: 56.1 years; 46.1% male). Five sleep characteristics—sleep duration, chronotype, insomnia symptoms, snoring, and daytime sleepiness—were used to construct a healthy sleep score (range: 0–5). Participants were categorized into poor (0–1), intermediate (2–3), and healthy (4–5) sleep score categories. Cox proportional hazards models were employed to estimate hazard ratios (HRs) and 95% confidence intervals (CIs) for pneumonia incidence. Results: Over a median follow-up period of 13.2 years, there were 20 116 new cases of pneumonia. A healthy sleep score was associated with a 26% lower risk of pneumonia (adjusted HR: 0.74, 95% CI: 0.69–0.80). This association exhibited a graded pattern, with higher sleep scores associated with progressively lower risk. The absence of frequent daytime sleepiness (HR: 0.79, 95% CI: 0.74–0.85) and insomnia symptoms (HR: 0.88, 95% CI: 0.86–0.91) exhibited the strongest component-specific inverse associations. Associations were stronger among participants younger than 60 years and females (P for interaction <0.001 for both). Conclusions: A multidimensional healthy sleep score is associated with a lower risk of pneumonia. These findings support the potential relevance of multidimensional sleep assessment to respiratory infection susceptibility and warrant further investigation. - dc:subject=healthy sleep characteristics,pneumonia,prospective cohort study,risk reduction,UK biobank - dc:rights=© The Author(s) 2026. Published by Oxford University Press on behalf of Infectious Diseases Society of America. - pdf:keywords=healthy sleep characteristics; pneumonia; prospective cohort study; risk reduction; UK biobank - pdf:producer=PDFlib+PDI 9.0.7p3 (C++/Win64) - pdfuaid:part=1 - pdfx:epsprocessor=PStill version 1.84.42 - prism:aggregationtype=Journal - prism:copyright=© The Author(s) 2026. Published by Oxford University Press on behalf of Infectious Diseases Society of America. - prism:doi=10.1093/ofid/ofag233 - prism:elssn=2328-8957 - prism:endingpage=9 - prism:publicationname=Open Forum Infectious Diseases - prism:startingpage=1 - prism:url=https://doi.org/10.1093/ofid/ofag233 - xmp:createdate=2026-04-29T23:39:05+05:30 - xmp:creatortool=Servigistics Arbortext Advanced Print Publisher 11.1.4546/W-x64 - xmp:metadatadate=2026-04-30T09:43:18+05:30 - xmp:modifydate=2026-04-30T09:43:18+05:30 - xmpmm:documentid=uuid:05A13565-EBCF-1566-3121-599A8576B755 - xmpmm:instanceid=uuid:b5a4a2cd-2fad-4c17-878b-8c2615cd45ee - xmprights:marked=True - pdfaextension:schemas=http://www.aiim.org/pdfua/ns/id/pdfuaidPDF/UA Universal Accessibility SchemainternalIndicates, which part of ISO 14289 standard is followedpartInteger ## Contents ### Page 1 Open Forum Infectious DiseasesM A J O R A R T I C L E # Multidimensional Healthy Sleep Characteristics and Pneumonia Risk: A Large-scale, Population-based Prospective Cohort Study Xufeng Fu, Xiaohua Li,, Yuanmei Gao,, Nan Wu, Ting Ao, and Wentao Ni, Department of Respiratory and Critical Care Medicine, The Seventh Affiliated Hospital of Sun Yat-sen University, Shenzhen, China, Department of Respiratory and Critical Care Medicine, Fujian Medical University Affiliated Fuzhou First Hospital, Fuzhou, China, Department of Respiratory and Critical Care Medicine, Key Laboratory for Major Obstetric Diseases of Guangdong Province, Guangdong Provincial Clinical Medical Research Center for Obstetrics and Gynecology, Guangdong-Hong Kong-Macao Greater Bay Area Higher Education Joint Laboratory of Maternal-Fetal Medicine, The Third Affiliated Hospital of Guangzhou Medical University, Guangzhou, China, Department of Respiratory and Critical Care Medicine, Human Genome Research Center, Peking University People’s Hospital, Beijing, China, and Department of Infectious Diseases, Beijing Luhe Hospital, Capital Medical University, Beijing, China Background. Sleep disturbances are hypothesized to increase susceptibility to pneumonia; however, most evidence arises from studies examining isolated sleep characteristics. The relationship between a multidimensional healthy sleep score and pneumonia risk remains unclear. Herein, the association between a multidimensional healthy sleep score and risk of pneumonia was evaluated. Methods. This prospective analysis encompassed 361 589 participants from the United Kingdom (UK) Biobank (mean age: 56.1 years; 46.1% male). Five sleep characteristics—sleep duration, chronotype, insomnia symptoms, snoring, and daytime sleepiness—were used to construct a healthy sleep score (range: 0–5). Participants were categorized into poor (0–1), intermediate (2–3), and healthy (4–5) sleep score categories. Cox proportional hazards models were employed to estimate hazard ratios (HRs) and 95% confidence intervals (CIs) for pneumonia incidence. Results. Over a median follow-up period of 13.2 years, there were 20 116 new cases of pneumonia. A healthy sleep score was associated with a 26% lower risk of pneumonia (adjusted HR: 0.74, 95% CI: 0.69–0.80). This association exhibited a graded pattern, with higher sleep scores associated with progressively lower risk. The absence of frequent daytime sleepiness (HR: 0.79, 95% CI: 0.74–0.85) and insomnia symptoms (HR: 0.88, 95% CI: 0.86–0.91) exhibited the strongest component-specific inverse associations. Associations were stronger among participants younger than 60 years and females (P for interaction <0.001 for both). Conclusions. A multidimensional healthy sleep score is associated with a lower risk of pneumonia. These findings support the potential relevance of multidimensional sleep assessment to respiratory infection susceptibility and warrant further investigation. Keywords. healthy sleep characteristics; pneumonia; prospective cohort study; risk reduction; UK biobank. Received 26 November 2025; accepted 10 April 2026; published online 21 April 2026 Correspondence: Wentao Ni, MD, PhD, Department of Respiratory and Critical Care Medicine, Human Genome Research Center, Peking University People’s Hospital, No. 11, Xizhimen South Street, Xicheng District, Beijing, Beijing 100044, China (). Open Forum Infectious Diseases®© The Author(s) 2026. Published by Oxford University Press on behalf of Infectious Diseases Society of America. This is an Open Access article distributed under the terms of the Creative Commons Attribution-NonCommercial-NoDerivs licence (), which permits non-commercial reproduction and distribution of the work, in any medium, provided the original work is not altered or transformed in any way, and that the work is properly cited. For commercial re-use, please contact reprints@oup.-com for reprints and translation rights for reprints. All other permissions can be obtained through our RightsLink service via the Permissions link on the article page on our site—for fur-ther information please contact journals.permissions@oup.com. Pneumonia remains a leading cause of mortality and a substan-tial healthcare burden. According to the Global Burden of Disease study, lower respiratory infections—primarily pneu-monia—accounted for an estimated 2.18 million deaths world-wide in 2021 []. In the United States, community-acquired pneumonia accounts for ∼1.4 million emergency department visits, 740 000 hospitalizations, and 41 000 deaths per year []. Therefore, identifying modifiable risk factors is an impor-tant public health concern. Sleep plays a vital role in immune regulation. Meta-analyses of epidemiological studies and human virus-challenge experi-ments have linked short sleep duration to greater susceptibility to respiratory infections []. Insufficient sleep during the im-munization period may also attenuate antibody responses to vaccination []. More broadly, sleep disturbance can alter in-nate and adaptive immune function and inflammatory signal-ing, including cytokine activity, leukocyte trafficking, and the formation of immunologic memory [,]. Because pneumonia develops in the setting of impaired host defense against lower respiratory pathogens, these observations provide a biologically plausible basis for studying sleep in relation to pneumonia risk. However, sleep is multidimensional, and studies focusing on individual sleep dimensions or related sleep disorders—such as sleep duration [,], insomnia symptoms [], or obstructive sleep apnea (OSA) []—may overlook how multiple sleep-related characteristics co-occur and jointly shape infection susceptibility. Composite sleep health scores have therefore been proposed to better capture overall sleep in epidemiologic re-search, although most applications to date have been in cardio-metabolic settings []. Whether a multidimensional sleep ### Page 2 profile is similarly relevant to pneumonia remains uncertain, particularly because candidate measures span behaviors, symptoms, daytime consequences, and pragmatic proxies of sleep-disordered breathing such as self-reported snoring. In this study, we examined the association between a multidimensional healthy sleep score and incident pneumonia in the UK Biobank. We hypothesized that a more favorable sleep profile would be as-sociated with a lower risk of pneumonia, and further assessed graded associations, associations for individual sleep dimensions, and potential effect modification across prespecified subgroups. ## METHODS ### Data Source and Population We performed a prospective cohort study using data from the UK Biobank. The UK Biobank is a large-scale population-based study that enrolled more than 500 000 individuals aged 40–69 years from 22 assessment centers across the UK between 2006 and 2010 []. All participants provided written informed con-sent at recruitment. Ethical approval for UK Biobank was grant-ed by the North West Multi-centre Research Ethics Committee (REC reference: 11/NW/0382), and this research was conducted under UK Biobank application number 688295. The study was performed in accordance with the Declaration of Helsinki. From the initial 502 128 participants, we applied exclusion criteria in 2 stages. In the first stage, we excluded participants who: (1) withdrew consent (n = 1297), (2) had a history of can-cer at baseline (n = 41 676), (3) had undergone organ transplant (n = 689), (4) had immunosuppressive diseases (n = 14 315), or (5) were using oral corticosteroids at baseline (n = 5826). During data preprocessing, we further excluded participants with missing data on all 5 sleep components (n = 80 819) and those with extreme sleep durations (<3 or >12 hours, n = 262). After applying all exclusion criteria, 361 589 participants were included in the final analysis (). ### Exposure Assessment We constructed a composite sleep score (range 0–5) from 5 self-reported sleep characteristics collected at UK Biobank base-line. Each component was scored as 1 (more favorable) or 0 (reference): (1) sleep duration (Field 1160, “About how many hours sleep do you get in every 24 hours? [please include naps]”): 7–8 hours/day versus <7 or >8 hours/day; (2) chrono-type (Field 1180, “Do you consider yourself to be?”): morning or more morning than evening versus more evening than morn-ing or definitely evening; (3) insomnia symptoms (Field 1200, “Do you have trouble falling asleep at night or do you wake up in the middle of the night?”): never/rarely or sometimes versus usually; (4) self-reported snoring (Field 1210, “Does your partner or a close relative or friend complain about your snoring?”), a pragmatic proxy for sleep-disordered breathing: no versus yes; and (5) daytime sleepiness (Field 1220, “How likely are you to doze off or fall asleep during the daytime when you don’t mean to?”): never/rarely or sometimes versus often or all of the time. For insomnia symptoms, “sometimes” was grouped with “never/rarely” to isolate frequent symptoms (“usually”), consis-tent with prior UK Biobank composite sleep score studies [,]. This scoring framework—including component definitions, dichotomization, and equal weighting—followed prior UK Biobank studies of cardiovascular and arrhythmic outcomes [,]. Because sleep duration was recorded in whole hours,>8 hours/day corresponded to reports of 9 hours or more. Participants were categorized into poor (0–1), intermediate (2–3), and healthy (4–5) sleep score categories. ### Outcome Ascertainment The primary outcome was all-cause pneumonia (ICD-10 codes: J12–J18). Secondary outcomes included: (1) pathogen-specific pneumonia (bacterial [J15], viral [J12], and organism unspeci-fied [J18]); (2) other acute lower respiratory infections (ALRIs; J20–J22); and (3) respiratory failure (J96). Outcomes were identified through an association with national medical records (including primary care and hospital inpatient data), death registry records, and self-reported conditions collect-ed at baseline or subsequent assessment visits. Follow-up began at baseline and continued until the earliest occurrence of an incident event, death, loss to follow-up, or administrative censoring at the end of follow-up. ### Covariates Covariates were selected a priori as potential confounders of the sleep–pneumonia association and grouped into 3 domains: socio-demographic (age, sex, ethnicity, Townsend deprivation index, education level); lifestyle (body mass index [BMI], smoking status [current vs not current (former/never)], alcohol consumption [regular versus nonregular], physical activity [metabolic equiva-lent of task (MET)-min/week], and dietary approaches to stop hy-pertension [DASH] score); and health-related factors (chronic obstructive pulmonary disease [COPD], asthma, and multimor-bidity [0, 1, or ≥2 among hypertension, cardiovascular disease, di-abetes, and dementia]). Influenza and pneumococcal vaccination status, while potentially important confounders, were not collect-ed as part of the UK Biobank baseline assessment for the full co-hort. As such, these variables were not included as covariates. ### Statistical Analysis Missingness ranged from 0% (outcomes, age, sex, and comorbid-ities) to 23.1% (physical activity); sleep variables had 0.3%–18.3% missing data (). We used multiple imputa-tions by chained equations under a missing-at-random assump-tion (predictive mean matching for continuous, logistic/ multinomial regression for categorical variables; 20 datasets, 15 it-erations). The imputation model included all analytical variables and outcome indicators; quality was assessed by comparing ### Page 3 observed versus imputed distributions (). Continuous data are presented as mean ± standard deviation or median (interquartile range [IQR]), and categorical variables as frequencies (percentages). Baseline comparisons used χ2 tests, analysis of variance, or Kruskal–Wallis tests as appropriate, pooled using Rubin’s rules. In the primary analysis, we employed Cox proportional haz-ards regression to calculate hazard ratios (HRs) and 95% con-fidence intervals (CIs), with follow-up time in years as the time scale. The proportional hazards assumption was evaluated using log–log survival curves and Schoenfeld residual tests. We fitted 3 sequential models with increasing levels of adjustment: Model 0 adjusted for age and sex; Model 1 additionally adjusted for sociodemographic factors (ethnicity, Townsend deprivation index, and education); and Model 2 further adjusted for life-style factors (BMI, smoking, alcohol, physical activity, and DASH score) and health-related factors (COPD, asthma, and multimorbidity). For clinical interpretation, we estimated Kaplan–Meier cu-mulative risks by sleep score category [] and calculated abso-lute risk differences (vs poor sleep) with 95% CIs at 2, 5, and 10 years. The number needed to benefit (NNTB) was defined as 1/| risk difference| for healthy versus poor sleep, representing the number associated with one fewer pneumonia event. Graded associations were examined using restricted cubic splines with 3 knots at the 25th, 50th, and 75th percentiles. Because scores 0 (n = 555) and 1 (n = 7863) were uncommon, they were combined; score 2, the lowest remaining category with adequate sample size, served as the reference. Models were adjusted for Model 2 covariates, with HRs predicted at co-variate means (continuous) and modes (categorical); nonline-arity was tested by likelihood ratio comparison of linear and spline models []. Component-specific associations were estimated in separate Cox models for each sleep characteristic, adjusted for Model 2 covariates but not the other components; these were descriptive rather than estimate of independent effects. Sleep duration was modeled as <7, 7–8, and >8 hours/day (>8 corresponding to≥9 hours given whole-hour recording), and BMI was omitted from the snoring model due to collinearity. To assess effect modification, we conducted prespecified subgroup analyses for the primary and secondary outcomes across 7 variables: age (<60 vs ≥ 60 years), sex (female vs male), BMI (<30 vs ≥ 30 kg/m2), DASH adherence (low, medi-um, or high based on tertiles), smoking status (current vs not current), physical activity (<600 vs ≥600 MET-min/week based on WHO recommendations), and multimorbidity (0 vs 1 vs ≥ 2 chronic conditions). Interaction P values were calculated using Wald χ2 tests. We conducted 2 sensitivity analyses to evaluate the robust-ness of our results. To reduce potential reverse causality, partic-ipants who developed respiratory illnesses within the first 2 years of follow-up were excluded, leaving 359 443 participants for analysis. Second, to assess the effect of imputation, we re-peated all analyses restricted to participants without missing data on any covariate (complete case analysis, n = 280 731). Finally, to evaluate the impact of unmeasured confounding, we calculated E-values for all analyses (main, component, sub-group, and sensitivity analyses), representing the minimum strength of association that an unmeasured confounder would require with both the exposure and outcome to explain away the observed effect []. For protective associations (HR <1), we applied the E-value formula to the inverse of the HR. When the 95% CI included the null value (HR = 1), the E-value for the CI bound was set to 1. All analyses were performed using R, version 4.3.3. Unless oth-erwise specified, estimates from multiply imputed datasets were combined using Rubin’s rules, which account for both within-imputation and between-imputation variance []. We applied the Barnard–Rubin adjustment to calculate the degrees of freedom for improved small-sample inference []. Statistical significance was defined as a two-sided P value of <.05. ## RESULTS ### Participant Characteristics Among the 361 589 participants (mean age: 56.1 years; 46.1% male), 2.3% (n = 8418) were in the poor, 39.1% (n = 141 447) in-termediate, and 58.5% (n = 211 724) healthy sleep score categories (). The distribution of sleep scores indicated that most par-ticipants scored 3 (28.1%) or 4 (37.0%) (). Participants in the poor sleep score category exhibited a higher BMI (30.1 vs 26.8 kg/m2), greater prevalence of current smoking (17.9% vs 8.5%), lower physical activity levels (1414 vs 1928 MET-min/week), lower DASH diet scores (21.5 vs 23.0), and a higher prevalence of COPD (0.7% vs 0.2%), asthma (16.6% vs 10.4%), and multimorbidity (41.1% vs 24.0% with ≥1 condition) compared to those in the healthy sleep score category (). During a median follow-up of 13.2 years (IQR: 12.3–14.2), we documented 20 116 incident cases of all-cause pneumonia (over-all: 5.6%). Among pneumonia subtypes, cases with unspecified organisms were the most common (n = 18 198, 5.0%), followed by viral (n = 2447, 0.7%) and bacterial pneumonia (n = 1108, 0.3%). Additional secondary outcomes included other ALRIs (n = 24 482, 6.8%) and respiratory failure (n = 5709, 1.6%). Crude incidence rates demonstrated graded patterns across sleep score categories for all outcomes (). ### Main Associations illustrates the associations between sleep score catego-ries and lower respiratory tract outcomes (complete results are presented in). In the fully adjusted model (Model 2), compared with the poor sleep score category, the healthy sleep score category was associated with a lower risk ### Page 4 of all-cause pneumonia (HR: 0.74, 95% CI: 0.69–0.80; E-value: 2.05 [lower CI bound: 1.83]), with the intermediate category showing an intermediate association (HR: 0.83, 95% CI: 0.77–0.89; E-value: 1.72 [lower CI bound: 1.50]). Consistent gradients were evident in Models 0 and 1 (). ### Page 5 For pathogen-specific pneumonia subtypes, a healthy sleep score showed the strongest inverse association for bacterial pneumonia (HR: 0.60, 95% CI: 0.45–0.80; E-value: 2.70), fol-lowed by viral pneumonia (HR: 0.70, 95% CI: 0.57–0.85; E-value: 2.22) and pneumonia with unspecified organisms (HR: 0.75, 95% CI: 0.70–0.82; E-value: 1.99). Inverse associations were also observed for other ALRIs (HR: 0.81, 95% CI: 0.75– 0.87; E-value: 1.78) and respiratory failure (HR: 0.73, 95% CI: 0.64–0.83; E-value: 2.07). The proportional hazards assumption was met for all outcomes ( and). Kaplan–Meier curves demonstrated graded patterns with cumulative risks separated by sleep score category over the 15-year period (). For all-cause pneu-monia, the 5-year cumulative risk rates were 1.73% (poor), 1.19% (intermediate), and 0.89% (healthy), with an absolute risk reduction of 0.84% between healthy and poor sleep score categories (NNTB: 120, 95% CI: 90–180). At 10 years, risk rates increased to 5.61%, 3.71%, and 2.80%, respectively, with an absolute risk reduction of 2.81% (NNTB: 36, 95% CI: 31–44). Similar gradients were observed across all secondary outcomes (complete results are presented in). RCS analysis demonstrated graded associations between sleep score and all outcomes (). The association for all- cause pneumonia had indications of nonlinearity (P = .02), al-though the relationship was predominantly monotone. Using score 2 as the reference, participants with scores 0–1 had a higher risk (HR: 1.12, 95% CI: 1.08–1.16), whereas those with score 5 had a lower risk (HR: 0.80, 95% CI: 0.77–0.84). Similar opposite trends were observed for secondary outcomes, with score 5 always reflecting the lowest risk (detailed results are presented in). Individual sleep characteristics showed component-specific as-sociations with respiratory infection risk (). The absence of frequent daytime sleepiness (HR: 0.79, 95% CI: 0.74–0.85; E-value: 1.84 [1.64]) and the absence of frequent insomnia symp-toms (HR: 0.88, 95% CI: 0.86–0.91; E-value: 1.51 [1.42]) showed the strongest inverse associations for all-cause pneumonia. Morning chronotype was also inversely associated with pneumo-nia risk (HR: 0.93, 95% CI: 0.91–0.96; E-value: 1.35 [1.25]). Sleep ### Page 6 duration exhibited a U-shaped relationship: compared with a sleep duration of 7–8 hours, those of <7 hours (HR: 1.15, 95% CI: 1.12– 1.19) and >8 hours (HR: 1.29, 95% CI: 1.23–1.35) were associated with increased risk. Snoring demonstrated no significant associa-tion (HR: 1.00, 95% CI: 0.97–1.03). Similar patterns were observed across secondary outcomes (). The inverse association between a healthy sleep score and all- cause pneumonia was largely consistent across the 7 prespecified subgroups. Significant effect modification was observed by age (P-interaction <.001) and sex (P-interaction <.001), with stronger inverse associations among participants aged <60 years and among females (). No clear evidence of interaction was found for BMI, smoking status, physical activity, DASH adher-ence, or multimorbidity. Complete subgroup analyses for all 6 out-comes are presented in and, with interaction heatmaps illustrated in. ### Sensitivity Analyses The main findings proved robust across sensitivity analyses. Exclusion of participants who developed respiratory outcomes within 2 years (n = 2146) yielded similar results (HR: 0.73, 95% CI: 0.67–0.79;,). Complete case analysis (n = 280 731) also produced compara-ble estimates (HR: 0.75, 95% CI: 0.69–0.82;,). ## DISCUSSION In this large prospective cohort study of 361 589 participants from the UK Biobank with a median follow-up of 13.2 years, we found that a higher multidimensional healthy sleep score was associated with a lower risk of pneumonia. Specifically, in-dividuals in the healthy sleep score category exhibited a 26% lower risk of all-cause pneumonia than those in the poor sleep score category. These inverse associations were consistently ob-served across pathogen-specific pneumonia subtypes, other ALRIs, and respiratory failure. Previous studies have primarily examined individual sleep characteristics in relation to respiratory infections. For exam-ple, a cross-sectional population-based study linked self- reported insufficient sleep with a higher occurrence of ### Page 7 ### Page 8 respiratory infections, including pneumonia []. A prospective cohort study further reported a U-shaped association between sleep duration and pneumonia incidence, with higher risks at both short and long durations [], consistent with our component-specific findings. In addition, a nationwide cohort study found that patients with clinically diagnosed insomnia had a substantially increased risk of pneumonia [], broadly aligning with our finding that the absence of frequent insomnia symptoms was inversely associated with pneumonia risk. Data relating chronotype, daytime sleepiness, and self-reported snoring to pneumonia are more limited. In our study, morning chronotype and the absence of frequent daytime sleepiness were each associated with lower pneumonia risk, whereas self- reported snoring was not. Although clinically defined OSA has been associated with pneumonia risk [,], snoring alone is an imprecise proxy for sleep-disordered breathing and cannot distinguish benign snoring from clinically significant OSA. Sleep is inherently multidimensional, with individual sleep characteristics often coexisting and potentially exerting synergis-tic effects. For example, insomnia symptoms frequently co-occur with short sleep duration and daytime sleepiness. To our knowledge, this is the first study to examine the combined association of multiple sleep dimensions—including chronotype, sleep duration, insomnia symptoms, snoring, and daytime sleepiness—with pneumonia risk. We observed a graded inverse association between higher sleep scores and pneumonia inci-dence, and similar gradients were seen across related respiratory outcomes. Notably, these inverse associations appeared stronger among participants younger than 60 years and among females. These subgroup findings should be interpreted cautiously, but may be consistent with evidence that younger adults exhibit greater immunological responsiveness to sleep variation before the onset of age-related immunosenescence, and that association between sleep disturbance and inflammatory responses are, in some settings, stronger in women [,]. Several mechanisms may underlie the observed inverse asso-ciations. Experimental and epidemiologic studies indicate that short sleep and sleep disturbance alter innate and adaptive im-munity, including natural killer cell activity, T-cell function, and inflammatory signaling [,,,]. These processes are central to pulmonary host defense, particularly early pathogen recognition, cytokine coordination, and alveolar macrophage and neutrophil responses [,], which may help explain the strongest inverse association observed for bacterial pneumonia. Evening chronotype is not itself circadian disruption; rather, risk may arise when a later chronotype is misaligned with external timing demands, producing social jetlag or other circadian mis-alignment that can adversely affect immune timing []. The U-shaped association for sleep duration suggests distinct mechanisms at each extreme: short sleep may directly perturb immune regulation, whereas long sleep may partly reflect pre- existing conditions such as undiagnosed cardiopulmonary disease, depression, or severe sleep apnea []. Given that these dimensions may affect infection susceptibility through distinct yet complementary immune pathways, the graded inverse asso-ciation with the overall sleep score is biologically plausible. Taken together, our findings suggest that multidimensional sleep assessment may complement established risk factors in identifying individuals at potentially higher respiratory infec-tion risk, but they should not be interpreted as evidence of cau-sality or as proof that modifying sleep will prevent pneumonia. The results are better viewed as hypothesis-generating and may help motivate future studies that integrate objective sleep and circadian measures with respiratory infection outcomes. For individuals with clinically significant insomnia, cognitive be-havioral therapy for insomnia (CBT-I) is the recommended evidence-based, first-line treatment [,]; however, whether treating insomnia symptoms or improving other sleep charac-teristics reduces pneumonia incidence remains unknown. Likewise, the stronger associations observed in younger adults and women warrant replication and mechanistic study rather than immediate subgroup-specific recommendations. Several limitations warrant consideration. First, sleep variables were self-reported at a single baseline assessment without objec-tive measures such as actigraphy, potentially introducing nondif-ferential misclassification and precluding assessment of changes over time. Second, residual confounding cannot be excluded; no-tably, vaccination status was unavailable. Third, the predominant-ly White UK Biobank cohort (aged 40–69 years), its volunteer nature, and the low prevalence of unfavorable sleep characteristics may limit generalizability and reduce exposure variability []. Fourth, nonrandom missingness cannot be entirely ruled out de-spite multiple imputations. Fifth, the composite score equally weights 5 self-reported dimensions, which does not imply equal biological importance; other potentially relevant dimensions— such as sleep efficiency, quality, and regularity—were unavailable, and self-reported snoring is an imprecise proxy for clinically sig-nificant sleep-disordered breathing. Sixth, pathogen-specific in-formation was unavailable for some cases, restricting etiologic stratification. Despite these limitations, this study benefits from a large prospective design with extended follow-up, comprehen-sive multidimensional sleep assessment, and robust analytical methods including multiple imputation and sensitivity analyses. ## CONCLUSION In this large, prospective cohort study, we observed that more favorable sleep characteristics were associated with a lower risk of pneumonia. The graded inverse association across sleep score categories, consistent across pneumonia subtypes and re-lated respiratory outcomes, supports the potential relevance of multidimensional sleep assessment to respiratory infection sus-ceptibility. These findings are hypothesis-generating and war-rant further investigation, including prospective studies with ### Page 9 objective sleep measures, to determine whether improving sleep characteristics can reduce pneumonia incidence. ## Supplementary Data Supplementary materials are available atonline. Consisting of data provided by the authors to benefit the reader, the posted materials are not copyedited and are the sole responsibility of the authors, so questions or comments should be addressed to the corre-sponding author. ## Notes Acknowledgments. We thank the UK Biobank participants for their in-valuable contributions. This research was conducted using the UK Biobank Resource (Application Number 688295). Author contributions. Conceptualization: X.F. and W.N.; Methodology: X.F. and N.W.; Formal analysis: X.F.; Validation: N.W.; Data curation: X.F.; Investigation: X.F., X.L., Y.G., N.W., T.A., and W.N.; Writing—original draft: X.F.; Writing—review & editing: X.L., Y.G., N.W., T.A., and W.N.; Supervision: W.N.; Funding acquisition: W.N. Data availability. The data that support the findings of this study are available from UK Biobank upon application and institutional approval. Due to data use agreements, the authors are not permitted to publicly share individual-level data. Analytic code is available from the corresponding au-thor upon reasonable request. Financial support. This study was supported by Peking University People Hospital Research and Development Funds (No. RDJP2022-43). The sponsor had no role in the study design; data collection, analysis, and interpretation; writing of the report; or the decision to submit the ar-ticle for publication. Potential conflicts of interest. All authors: No reported conflicts. ## References GBD 2021 Lower Respiratory Infections and Antimicrobial Resistance Collaborators, Global, regional, and national incidence and mortality burden of non-COVID-19 lower respiratory infections and aetiologies, 1990-2021: a sys-tematic analysis from the global burden of disease study 2021. Lancet Infect Dis2024; 24:974–1002. Vaughn VM, Dickson RP, Horowitz JK, Flanders SA. Community-acquired pneumonia: a review. JAMA 2024; 332:1282–95. Cohen S, Doyle WJ, Alper CM, Janicki-Deverts D, Turner RB. Sleep habits and susceptibility to the common cold. Arch Intern Med 2009; 169:62–7. Prather AA, Janicki-Deverts D, Hall MH, Cohen S. Behaviorally assessed sleep and susceptibility to the common cold. Sleep 2015; 38:1353–9. Robinson CH, Albury C, McCartney D , et al. The relationship between duration and quality of sleep and upper respiratory tract infections: a systematic review. Fam Pract 2021; 38:802–10. Spiegel K, Rey AE, Cheylus A, et al. A meta-analysis of the associations between insufficient sleep duration and antibody response to vaccination. Curr Biol 2023; 33:998–1005.e2. Besedovsky L, Lange T, Haack M. The sleep-immune crosstalk in health and dis-ease. Physiol Rev 2019; 99:1325–80. Irwin MR, Opp MR. Sleep health: reciprocal regulation of sleep and innate immu-nity. Neuropsychopharmacology 2017; 42:129–55. Prather AA, Leung CW. Association of insufficient sleep with respiratory infec-tion among adults in the United States, JAMA. Intern. Med 2016; 176:850–2. Lin C-L, Liu T-C, Chung C-H, Chien W-C. Risk of pneumonia in patients with insomnia: a nationwide population-based retrospective cohort study. J Infect Public Health 2018; 11:270–4. Su VY-F, Liu C-J, Wang H-K, et al. Sleep apnea and risk of pneumonia: a nation-wide population-based study. CMAJ 2014; 186:415–21. St-Onge M-P, Aggarwal B, Fernandez-Mendoza J, et al. Multidimensional sleep health: definitions and implications for cardiometabolic health: a scientific state-ment from the American Heart Association. Circ Cardiovasc Qual Outcomes2025; 18:e000139. Sudlow C, Gallacher J, Allen N, et al. UK biobank: an open access resource for identifying the causes of a wide range of complex diseases of middle and old age. PLoS Med 2015; 12:e1001779. Fan M, Sun D, Zhou T, et al. Sleep patterns, genetic susceptibility, and incident cardiovascular disease: a prospective study of 385,292 UK biobank participants. Eur Heart J 2020; 41:1182–9. Li X, Zhou T, Ma H, et al. Healthy sleep patterns and risk of incident arrhythmias. J Am Coll Cardiol 2021; 78:1197–207. Kaplan EL, Meier P. Nonparametric estimation from incomplete observations. J Am Stat Assoc 1958; 53:457–81. Meng X-L, Rubin DB. Performing likelihood ratio tests with multiply-imputed data sets. Biometrika 1992; 79:103–11. VanderWeele TJ, Ding P. Sensitivity analysis in observational research: introduc-ing the e-value. Ann Intern Med 2017; 167:268–74. Rubin DB. Multiple imputation for nonresponse in surveys. New York: John Wiley and Sons, 1987. Barnard J, Rubin DB. Small-sample degrees of freedom with multiple imputation. Biometrika 1999; 86:948–55. Patel SR, Malhotra A, Gao X, Hu FB, Neuman MI, Fawzi WW. A prospective study of sleep duration and pneumonia risk in women. Sleep 2012; 35:97–101. Boira I, Chiner E. Sleep and respiratory infections. Semin Respir Crit Care Med2025; 46:170–8. Dolsen EA, Crosswell AD, Prather AA. Links between stress, sleep, and inflamma-tion: are there sex differences? Curr Psychiatry Rep 2019; 21:8. Irwin MR, Olmstead R, Carroll JE. Sleep disturbance, sleep duration, and inflam-mation: a systematic review and meta-analysis of cohort studies and experimental sleep deprivation. Biol Psychiatry 2016; 80:40–52. Bakour C, Schwartz S, O'Rourke K, et al. Sleep duration trajectories and systemic inflammation in young adults: results from the national longitudinal study of ad-olescent to adult health (add health). Sleep 2017; 40:zsx156. Quinton LJ, Walkey AJ, Mizgerd JP. Integrative physiology of pneumonia. Physiol Rev 2018; 98:1417–64. Korkmaz FT, Traber KE. Innate immune responses in pneumonia. Pneumonia (Nathan) 2023; 15:4. Shirato K, Sato S. Macrophage meets the circadian clock: implication of the cir-cadian clock in the role of macrophages in acute lower respiratory tract infection. Front Cell Infect Microbiol 2022; 12:826738. Roenneberg T, Pilz LK, Zerbini G, Winnebeck EC. Chronotype and social jetlag: a (self-) critical review. Biology (Basel) 2019; 8:54. Taillard J, Sagaspe P, Philip P, Bioulac S. Sleep timing, chronotype and social jet-lag: impact on cognitive abilities and psychiatric disorders. Biochem Pharmacol2021; 191:114438. Hall MH, Smagula SF, Boudreau RM, et al. Association between sleep duration and mortality is mediated by markers of inflammation and health in older adults: the health, aging and body composition study. Sleep 2015; 38:189–95. Furukawa Y, Sakata M, Yamamoto R, et al. Components and delivery formats of cognitive behavioral therapy for chronic insomnia in adults: a systematic review and component network meta-analysis. JAMA Psychiatry 2024; 81:357–68. Edinger JD, Arnedt JT, Bertisch SM, et al. Behavioral and psychological treat-ments for chronic insomnia disorder in adults: an American Academy of Sleep Medicine clinical practice guideline. J Clin Sleep Med 2021; 17:255–62. Fry A, Littlejohns TJ, Sudlow C, et al. Comparison of sociodemographic and health-related characteristics of UK biobank participants with those of the general population. Am J Epidemiol 2017; 186:1026–34.