Prevalence of abnormal spirometry and correlates of lung function decline among a population in rural Uganda Marvah Hill Pierre-Louis, Yiqi Qian, Emily N Satinsky, Edwin Nuwagira, Charles Baguma, Patience Ayebare, Patrick Gumisiriza, Megan Lam, Mi-Sun Lee, David C Christiani, Stephen Asiimwe, Bernard Kakuhikire, Alexander C Tsai, Mark J Siedner, Crystal M North Abstract BackgroundChronic lung disease, which is the third leading cause of death globally, places a disproportionate burden of morbidity and mortality on sub-Saharan Africa, where limited infrastructure results in a lack of understanding of how lung function changes over time, whether it varies among susceptible populations, or whether at-risk individuals could be identified before abnormal lung function develops. To address this gap, we aimed to characterise lung function patterns and identify correlates of lung function decline in rural Uganda.MethodsAdults completed health questionnaires and American Thoracic Society/European Respiratory Society-acceptable spirometry up to three times over four years (2015–2019). Our primary outcome was lung function (forced expiratory volume in 1 second (EV1) and FEV1/forced vital capacity (FVC)); our secondary outcome was spirometry patterns. We characterised baseline patterns using Global Lung Initiative (GLI)-Global prediction equations and identified correlates of decline using generalised mixed effects linear regression. We then explored effect modification by sex, smoking, and socioeconomic status using interaction terms, with inverse probability of treatment weights balancing baseline characteristics.ResultsA total of 902 participants completed 1,471 ATS/ERS-acceptable spirometry tests. Their median age was 35 years, 56% were women, 75% were never-smokers, and 31% experienced chronic respiratory symptoms. Abnormal lung function affected 14% of the cohort, more commonly women and those with respiratory symptoms. FEV1 declined by 10mL/year (95% CI = −17, −2), and FEV1/FVC declined by 0.006/year (95% CI = −0.008, −0.005) or 0.53% predicted/year (95% CI = −0.68, −0.37). Among those with normal baseline lung function, FEV1 and FEV1/FVC declined more rapidly (FEV1 −18mL/year; FEV1/FVC −0.66% predicted). There was no effect modification by sex, smoking, or SES.ConclusionsLung function decline may be more rapid in this young, mostly non-smoking population. Our findings extend the understanding of non-obstructive lung disease in sub-Saharan Africa and emphasise that risk persists despite normal spirometry. These data can inform regional public health strategies for risk mitigation, early disease detection, and building healthcare infrastructure for the growing population at risk for or living with chronic lung disease. Chronic lung disease is the third leading cause of death globally, responsible for 3.9 million deaths and 112 million disability-adjusted life years annually []. Chronic obstructive pulmonary disease (COPD) and asthma are the most prevalent chronic lung diseases, affecting 480 million and 262 million people globally, respectively [,]. More recently, additional patterns of abnormal spirometry, referred to as preserved ratio impaired spirometry (PRISm) and restrictive spirometry pattern have been the subject of increased investigations, as these patterns of abnormal spirometry are associated with incident respiratory and cardiovascular diseases, and increased morbidity and mortality [,]. The prevalence, risk factors, and morbidity and mortality of chronic lung disease are known to vary by region and gender. For example, the BOLD study, which is among the first global studies of lung disease burden, estimated an obstructive lung disease prevalence among men ranging from 6% in India to 23% in South Africa and among women from 4% in the Philippines to 21% in Sweden []. A recent systematic review and meta-analysis estimated that the prevalence of COPD in sub-Saharan Africa is 8% [], which is higher than estimates for the USA [] and Europe []. Indeed, by 2050, sub-Saharan Africa is projected to have the highest COPD prevalence in the world []. Similarly, there are 120 million people living with asthma in sub-Saharan Africa [], nearly a third of whom may meet the criteria for severe asthma []. There is also regional variation in the influence of socioeconomic status (SES) on the burden of lung disease []. In high income countries, lower SES was found to be associated with poorer lung health []. Comparatively, data on the relationship between SES and lung health in low-and-middle-income countries (LMICs) are quite limited [], with studies showing both higher [] and lower [] respiratory morbidity among those of lower SES. Smoking is the leading risk factor for chronic lung disease among men and in high income countries, while air pollution is the leading risk factor for lung disease among women and in LMICs []. Tuberculosis and chronic HIV infection, common conditions in many LMICs, were also found to be associated with higher risk of chronic lung diseases and are important drivers of non-smoking chronic lung disease risk [,]. The prevalence, costs, and morbidity and mortality of chronic lung disease have rapidly increased over the last 30 years – disproportionately so in LMICs and particularly among women [,,]. However, diagnosing lung disease requires equipment that is not widely available in many LMICs. Given the limited infrastructure available for tracking population lung health over time in these regions, few cohorts have attempted to characterise longitudinal lung health, leaving us with limited knowledge regarding how lung function changes over time, whether it varies among particularly susceptible populations, or whether at-risk populations could be identified prior to the development of abnormal lung function. To address these knowledge gaps, we measured serial lung function over four years within a well-characterised cohort of 902 adults living in rural Uganda to characterise lung function patterns and identify correlates of lung function decline. METHODS Adults who reported stable primary residence in Nyakabare Parish in southwestern Uganda were enumerated in a population census from 2014 to 2015 [], with parish residents who were at least 18 years of age invited to attend free health screening events that occurred in 2015, 2017, and 2019. Each health screening event lasted five days and was geographically distributed across three locations in the parish to ensure easy accessibility for all residents (Figure S1 in the ) []. At each health screening event, parish residents could consent to have their health information collected for research purposes. We included those who were at least 18 years old, attended at least one health screening event, and consented to have their health information collected for research purposes in this analytical cohort. We otherwise set no other exclusion criteria. At each health screening event, participants completed health history questionnaires that included demographics, SES, medical history, smoking history, and respiratory symptoms. Study staff measured participants’ height using a stadiometer and weight using electronic scales, and body mass index (BMI) was calculated as weight in kilograms divided by the square of height in meters. We measured SES through a principal components analysis based upon a household asset index, in which ownership of more household assets defines greater wealth and thus higher SES []. Then, we created quintiles of asset ownership (and thus SES) from the asset ownership index, in which quintile 1 represented those with the most assets (i.e. the ‘richest’ quintile) and quintile 5 represented those with the least assets (i.e. the ‘poorest’ quintile). Chronic cough was defined as the presence of a cough on most days; for those diagnosed with chronic cough, we ascertained the frequency of sputum production. Chronic dyspnoea was defined as having difficulty breathing on most days. Participants completed spirometry with the guidance of study staff using EasyOne® Plus handheld spirometers (nDD Medical Technologies, Andover, Massachusetts, USA) in accordance with American Thoracic Society/European Respiratory Society (ATS/ERS) standards. Spirometry was considered ATS/ERS acceptable if the forced expiratory volume in 1 second (FEV1) and forced vital capacity (FVC) from the two best manoeuvres were within 200 mL of one another. Per automated spirometer algorithms, participants with FEV1/FVC<0.7 completed repeat spirometry 10 minutes after receiving four puffs of albuterol (Ventolin, GlaxoSmithKline, Philadelphia, Pennsylvania, USA), the results of which we interpreted using Global Lung Initiative (GLI)-Global prediction equations and ATS/ERS interpretation guidelines [,] and only ATS/ERS-acceptable tests were included in the analytic cohort. For those with ATS/ERS acceptable post-bronchodilator spirometry, COPD was defined as a post-bronchodilator FEV1/FVC < lower limit of normal (LLN) [,]. Statistical analysis The primary outcome of interest was change in lung function (pre-bronchodilator FEV1 and FEV1/FVC), while the secondary outcome of interest was prevalence of abnormal spirometry, which included obstruction, spirometric restriction, and PRISm. We chose not to focus on change in FVC over time because FVC is more effort- and time-dependent than other spirometry metrics in field studies and therefore less reproducible over time [,]. To evaluate for selection bias, we compared baseline demographics between participants who had completed ATS/ERS-acceptable vs. unacceptable spirometry, and between participants in our final analytic cohort vs. participants excluded from our final analytic cohort. We tested all subgroup differences using Wilcoxon rank sum, chi-squared, or Fisher’s exact tests, as appropriate. Similar to our prior work [], we used inverse probability of treatment (IPT) weights to estimate population-representative estimates by reducing selection bias. We first estimated the propensity for parish adults to attend at least one health screening event using a logistic regression model consisting of census variables anticipated to correlate with event attendance. Variables included age, sex, education, village, self-reported overall health, and SES. Using this model, we calculated stabilised IPT weights [] and incorporated these weights in our regression models. We then summarised baseline demographics for the whole cohort and by sex, and the proportion of study participants with chronic respiratory symptoms and abnormal lung function, both with and without the weights. We considered each spirometry metric as an absolute value and as a percent predicted value based on GLI-Global prediction equations []. We categorised lung function as obstruction (FEV1/FVC