Quality of Life After Lower Respiratory Tract Infection

Acute lower respiratory tract infections (LRTIs) are a group of infections affecting the respiratory tract below the level of the larynx, including pneumonia, acute bronchitis, bronchiolitis, acute infective exacerbations of chronic obstructive pulmonary disease (COPD).

In those who survive their initial episode of LRTI, recovery may be slow with adverse outcomes which may last months or even years after the initial episode and have implications for patients, their families and health care resources. A recent prospective cohort study in the U.S. (PNEUMO) reported that at 6 months after hospitalisation with community-acquired pneumonia (CAP) and compared to their pre-hospitalisation status, 12.8% of participants had lost the ability to perform at least one basic activity of daily living (ADL), 22.0% lost the ability to perform at least one instrumental ADL, 41.6% suffered cognitive impairment, 58.7% had loss of employment, and 23.6% had decreased quality of life (Han et al 2025). Additionally pneumonia patients have high rates of readmission to hospital, which in turn is associated with significant inpatient mortality (Lawrence et al 2023), high rates of primary care consultation within 7 days of discharge (Baskaran et al 2021), increased risk of cardiovascular events and heart failure (Eurich et al 2017; Violi et al 2017), recurrent pneumonia (Baskaran et al 2022) and persistent symptoms (Pick et al 2019).

LRTIs therefore heavily impact patients, their families and carers, and the use of healthcare resources. Measuring the quality of life in patients hospitalised with acute LRTIs for up to one year after their discharge is an important component in the broader assessment of the cost of LRTIs in the UK, but long-term quality of life data in these patients are limited.

Every individual's experience of recovery from pneumonia is unique reflecting the complex interplay between their clinical characteristics, existing co-morbidities, frailty, physical and mental status, illness severity, and underlying causative micro-organism. There are few long-term data on outcomes which are important to patients following an episode of LRTI, including patient-perceived quality of life. It is therefore important to study this outcome and to identify those factors which may be associated with quality-of life measures in order to: a) gain a broader picture of the impact of LRTI on patients themselves; b) inform future trials assessing interventions aimed at improving patients' quality of life; c) to feed into analyses of the health economic costs of LRTIs.

The purpose of this study is to gain a better understanding of how health-related quality of life is affected in people who are hospitalised after a LRTI. The primary objective is to document health states and utility indexes for hospitalised LRTI patients and their evolution over the course of one year after the initial episode. The secondary objective is to evaluate associations between patient characteristics and their hospital stay with their health-related quality of life.

Participants will be recruited when they are almost ready to be sent home from hospital. Informed consent will be obtained prior to participation (or for those lacking capacity to provide consent, a consultee acting on their behalf will complete a declaration form). Before discharge participants will be asked to complete the EQ-5D-5L health-related quality of life questionnaire . After discharge from hospital, participants will complete the same questionnaire at one month, three months, six months and one year after discharge. Questionnaires will be administered by telephone, digitally (email form) or by post.

Descriptive statistics will be conducted at each time point (discharge, 1, 3, 6, 12 months):

1. EQ-5D-5L profiles: Number and proportion at each level (1-5) for each dimension.
2. Ceiling/floor effects: Proportion at ceiling (11111) and floor (severe problems on ≥3 dimensions) reported at each timepoint.
3. Utility values: Mean, Standard Deviation (SD), Standard Error (SE), median, Interquartile Range (IQR), min, max, and n.
4. VAS scores: Mean, SD, median, IQR, min, max (descriptive only).
5. Change scores: Mean change from baseline to each follow-up, with 95% Confidence Interval (CI).
6. Population comparison: Utility decrement from age-sex matched norms at each time point.

Quality-Adjusted Life Years (QALYs) will be calculated using area-under-the-curve with trapezoidal rule and linear interpolation:

1. Time zero: Hospital discharge date.
2. Intervals: Discharge→1 month, 1→3 months, 3→6 months, 6→12 months. Actual days used.
3. Formula: QALY per interval = [(Ustart + Uend) / 2] × time in years. Total QALYs =

• (QALY_intervals)

1. Mortality: Deaths contribute QALYs to date of death (utility = 0 thereafter).
2. QALY loss: Calculated as the difference between expected and observed QALYs over 12 months. Expected QALYs derived from age-sex matched English population norms using adjusted limited dependent variable mixture models (ALDVMM) with EQ-5D-5L responses mapped to 3L using Hernández Alava algorithm.

Health state utility values reported by severity tier and time point will serve as inputs for cost-effectiveness models, where survival is incorporated through model structure using transition probabilities derived from the linked CPRD-HES cohort.

All cohorts will be stratified by age, sex, severity tier, diagnostic category, index of Multiple Deprivation quintile, number of co-morbidities, ECOG Performance Status, CURB-65 score.

To model utility trajectories over time and identify predictors of recovery, the investigators will use mixed-effects linear regression with random intercepts. Fixed effects: time, age, sex, severity tier, performance status, comorbidity count. Time × severity interaction explored to assess differential recovery rates. Predicted 12-month utilities derived from model.

Because EQ-5D utility values have an upper limit and many patients may report full health at follow-up, the investigators will check whether the model fits the data and produces realistic predicted values. If predictions fall outside the valid range or the model performs poorly, alternative approaches (such as Tobit regression, generalised linear models with clustered responses, or two-part models) will be explored in sensitivity analyses.

Missing data: Completion rates will be reported by timepoint and baseline characteristics compared between completers and non-completers. The following sensitivity analyses for missing data will be performed (a) multiple imputation by chained equations (MICE) conditional on baseline characteristics and severity tier, with last observation carried forward (LOCF) as conservative comparator; (b) Complete case: restricted to participants with EQ-5D-5L data at all five timepoints.