Cohort profile: The Chikwawa lung health cohort; a population-based observational non-communicable respiratory disease study in Malawi.

Purpose The Chikwawa lung health cohort was established in rural Malawi in 2014 to prospectively determine the prevalence and causes of lung disease amongst the general population of adults living in a low-income rural setting in Sub-Saharan Africa. Participants A total of 1481 participants were randomly identified and recruited in 2014 for the baseline study. We collected data on demographic, socio-economic status, respiratory symptoms and potentially relevant exposures such as smoking, household fuels, environmental exposures, occupational history/exposures, dietary intake, healthcare utilization, cost (medication, outpatient visits and inpatient admissions) and productivity losses. Spirometry was performed to assess lung function. At baseline, 56.9% of the participants were female, a mean age of 43.8 (SD:17.8) and mean body mass index (BMI) of 21.6 Kg/m2 (SD: 3.46) Findings to date Currently, two studies have been published. The first reported the prevalence of chronic respiratory symptoms (13.6%, 95% confidence interval [CI], 11.9 to 15.4), spirometric obstruction (8.7%, 95% CI, 7.0 to 10.7), and spirometric restriction (34.8%, 95% CI, 31.7 to 38.0). The second reported annual decline in forced expiratory volume in one second [FEV1] of 30.9mL/year (95% CI: 21.6 to 40.1) and forced vital capacity [FVC] by 38.3 mL/year (95% CI: 28.5 to 48.1). Future plans The ongoing current phase of follow-up will determine the annual rate of decline in lung function as measured through spirometry, and relate this to morbidity, mortality and economic cost of airflow obstruction and restriction. Population-based mathematical models will be developed driven by the empirical data from the cohort and national population data for Malawi to assess the effects of interventions and programmes to address the lung burden in Malawi. The present follow-up study started in 2019.


Strengths and limitations of this study
• This is an original cohort study comprising adults randomly identified in a low-income Sub-Saharan African Setting.
• The repeated follow up of the cohort has included objective measures of lung function.
• The cohort has had high rates of case ascertainment that include verbal autopsies.
• The study will include an analysis of the health economic consequences of rate of change of lung function and health economic modelling of impact of lung diseases and potential interventions that could be adopted.
• A main limitation of our study is the systematic bias may be introduced through the selfselection of the participants who agreed to take part in the study to date and the migration of individuals from Chikwawa.

Introduction
Globally, non-communicable respiratory diseases (NCRD) are the third leading cause of noncommunicable disease (NCD) mortality, causing an estimated 4 million deaths each year (1).
Amongst the NCRD, asthma and chronic obstructive pulmonary disease (COPD) are the most prevalent, affecting approximately 358 million and 174 million people respectively (2). Annually, COPD causes 3 million deaths accounting for 6% of all deaths worldwide (2)(3)(4). Furthermore, the deaths from these diseases are rising globally (5) in part due to increased longevity and changes in population structure (6).
The majority of the burden of NCRD mortality and morbidity is in low and middle-income countries (LMIC) (1,7), which now account for 90% of COPD deaths (8). Several community based studies in LMIC have documented a high prevalence of abnormal lung function, both obstructive and restrictive (low lung volumes) (9-15), whilst several couple have documented low prevalence of COPD (16,17) but high prevalence of respiratory symptoms (17). On the other hand, very few observational cohort studies have reported and described the health and economic burden of NCRD (18,19), especially in LMIC settings. Their prevalence means that there is a pressing need to better document the life course epidemiology and the related health and economic burden of abnormal (obstructive and restrictive) lung function in LMIC (10,11).
Malawi remains one of the poorest countries in the world (20) with 83% of its 18 million inhabitants living in rural areas (21). With a GDP per capita of $300, over half the households live below the poverty line (using the international poverty line of US$ 1.90 per person per day) (22), and about 50% of the national health expenditure is funded from external donors (23,24). Like many sub-Saharan African (SSA) countries, Malawi is at the intersection of high rates of . CC-BY 4.0 International license It is made available under a author/funder, who has granted medRxiv a license to display the preprint in perpetuity.
is the (which was not peer-reviewed) The copyright holder for this preprint . https://doi.org/10.1101/2020.04. 16.20068643 doi: medRxiv preprint communicable respiratory diseases (Tuberculosis (TB), pneumonia), and increasing NCRD (25-27). Recent studies have reported substantial levels of abnormal lung function with spirometric evidence of low lung volumes and obstructive deficits present in 34.8% (95% CI: 31.7%, 38.0%) and 8.7% (7.0%, 10.7%) of rural adults and 38.6% (34.4%, 42.8%) and 4.2% (2.0%, 6.4%) of urban adults respectively (10,11). Spirometric deficits were defined according to the NHANES III Caucasian references (28). What is not known is whether and how these spirometric deficits impact on the everyday lives of the country's people and health system. Potentially, as in other low-income situations, the economic burden of NCRD may have serious adverse outcomes for households including unpredictable household expenditures due to complications and catastrophic health expenditure (29).
To examine the health and economic burden of NCRD, including abnormal lung function in Malawi, our prospective study aims to follow up a population-based cohort of participants in the rural district of Chikwawa, in southern Malawi, who were recruited to a longitudinal follow-up spirometry study conducted between August 2014 and July 2015 (the Chikwawa lung health cohort) (11,15). The primary objectives of the current study are to; (i) estimate the annualised rate of change in lung function by age and sex as determined by repeating spirometry; (ii) to develop a mathematical population model based on the cohort findings that estimates the lifetime health impact of airflow obstruction in Malawian adults in disability-adjusted life years (DALYS); (iii) estimate the health resource use and lifetime costs in the cohort of Malawian adults with airflow obstruction in international dollars (Int$); (iv) produce model estimates of the lifetime cost effectiveness (Int$/DALY) of selected key intervention compared with current practice to define optimum packages of interventions; and (v), recreate these analyses for . CC-BY 4.0 International license It is made available under a author/funder, who has granted medRxiv a license to display the preprint in perpetuity.
is the (which was not peer-reviewed) The copyright holder for this preprint . https://doi.org/10.1101/2020.04. 16.20068643 doi: medRxiv preprint Malawian adults with low lung volumes. The economic cost will be from a societal perspective and will include health sector costs, patient/family and carer costs and productivity losses (30).
Presently, the Malawian health system recommends the use of salbutamol and beclomethasone inhalers and prednisolone as interventions for chronic asthma management and salbutamol inhalers, prednisolone and hydrocortisone injections as interventions for acute asthma (31) but these interventions are only available in 8% of urban health facilities and 2% of rural health facilities in Malawi (32).
. CC-BY 4.0 International license It is made available under a author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

Setting
The study is currently conducted in Chikwawa district, located in Southern region of Malawi (see figure 1) Figure 1: Districts in Malawi. Inset map highlights Chikwawa district, the study area.
. CC-BY 4.0 International license It is made available under a author/funder, who has granted medRxiv a license to display the preprint in perpetuity.
is the (which was not peer-reviewed) The copyright holder for this preprint . https://doi.org/10.1101/2020.04. 16.20068643 doi: medRxiv preprint Who is in the study?
The Chikwawa lung health cohort was initiated alongside the Cooking and Pneumonia study (CAPS) (11,33) (Trial registered with ISRCTN, number ISRCTN59448623). CAPS was a cluster randomized trial that investigated the health effects of a cleaner-biomass fuel cookstove intervention (33). The aim of setting up the Chikwawa lung cohort was to determine the prevalence and determinates of lung disease amongst adults in Chikwawa, rural Malawi (11). In addition, two rounds of follow-up studies have been done with the Chikwawa lung cohort aiming to assess the determinants of lung function trajectories as affected by personal air pollutant exposures, including the CAPS cookstove intervention (15). The current study will provide longitudinal data by following up participants from the Chikwawa lung health cohort who still reside in Chikwawa and were recruited to the baseline study in 2014 -2015 (11) and associated risk factors, health utilisation use and economic burden.

Baseline participant recruitment
The participants were originally recruited in 2014 -2015. The participants were selected through random sampling of a list of adults living in each of the 50 villages participating in CAPS (11). The participants included those who took part in the CAPS intervention and those who did not but resided in villages where the CAPS intervention was being implemented. The list of adults was obtained from local community liaison personnel from each village following a series of community engagement events with the village leaders such as chiefs and other community representatives (11). The random selection was conducted by an independent statistician at the Burden of Obstructive Lung Disease (BOLD) centre in London in accordance with the BOLD protocol (34). The identified individuals comprised a population-representative, age and gender . CC-BY 4.0 International license It is made available under a author/funder, who has granted medRxiv a license to display the preprint in perpetuity.
is the (which was not peer-reviewed) The copyright holder for this preprint . https://doi.org/10.1101/2020.04. 16.20068643 doi: medRxiv preprint stratified, sample of adults who were then invited to participate in the 2014 -2015 baseline study. Participants had to provide written informed consent or an independently witnessed thumbprint to be included in the study (11). Those who were acutely unwell or pregnant women or were non-permanent residents of Chikwawa were excluded from the baseline study (11).
A total of 3000 adults were invited to participate in the baseline study of which 1481 (49.3%) agreed to participate (11). Participants were stratified into two age groups: 18 -39 years and 40 years and above. In order to provide an estimate of chronic airflow limitation prevalence in the stratum with a precision (95% CI) of +3.3% to 5.0% and assuming a prevalence of 10% to 25%, a total sample of 1200 participants was estimated allowing for unequal age and gender distribution, refusals and inability to provide spirometry measurements of acceptable quality (11). Table 1 below summarises the age and sex characteristics of those who agreed to participate in the study compared to those who did not. . CC-BY 4.0 International license It is made available under a author/funder, who has granted medRxiv a license to display the preprint in perpetuity.
is the (which was not peer-reviewed) The copyright holder for this preprint . https://doi.org/10.1101/2020.04. 16.20068643 doi: medRxiv preprint How often has the cohort been followed up? Study participants have been followed up twice prior to the current study. The baseline study was done between August 2014 -July 2015 (11)  . CC-BY 4.0 International license It is made available under a author/funder, who has granted medRxiv a license to display the preprint in perpetuity.
is the (which was not peer-reviewed) The copyright holder for this preprint . https://doi.org/10.1101/2020.04. 16.20068643 doi: medRxiv preprint In the current phase of follow-up, verbal autopsies were conducted for the 2014 -2015 baseline participants who have died, and a questionnaire was administered to the next of kin for those . CC-BY 4.0 International license It is made available under a author/funder, who has granted medRxiv a license to display the preprint in perpetuity.
is the (which was not peer-reviewed) The copyright holder for this preprint . https://doi.org/10.1101/2020.04. 16.20068643 doi: medRxiv preprint who were unobtainable due to being no longer resident in Chikwawa. The data and variables collected in the Chikwawa lung health cohort are described in table 2.
. CC-BY 4.0 International license It is made available under a author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

Participant and public involvement
Participants were not involved in in setting research question or the outcome measures but have been instrumental in implementation of the study.
Participants and the public were involved in the dissemination of baseline information nationally through the Ministry of health, and in the Chikwawa community from which the data was collected through the Chikwakwa Health Research Committee and the Chiefs and community leaders from the villages from where we collected our data. These activities have encouraged community buy-in and involvement in the subsequent rounds of follow-up within the study.
. CC-BY 4.0 International license It is made available under a author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

Findings to date
The Chikwawa lung health cohort has provided data characterising the burden of chronic  Table 3. At baseline, a total of 1481 participants were recruited of which 637 (43.0%) were male and 844 (57.0%) were female(11). The mean age was 43.9 years (SD: 17.8), mean body mass index (BMI) was 21.6 Kg/m 2 (SD: 3.46). Cigarette smoking rates were 22.1% (n=327) were current or ever smokers of which the majority were men (n = 255, 78.0%).
There was no difference in ages between the men and women (see table 3).
The frequency of chronic respiratory symptoms and abnormal spirometry. Among the participants with interpretable and reliable spirometry (34) (n = 886), spirometric obstruction ( defined as FEV1/FVC < 0.70) and spirometric restriction (defined as FEV1/FVC > 0.70 and post-bronchodilator FVC < 80% predicted) (28) were present in 8.7% (7.0%, 10.7%) and 34.8% (95% CI: 31.7%, 38.0%) of the participants respectively according to the NHANES III Caucasian references (11). 13.7% reported either having a 'cough without having a cold', 'bringing up phlegm from your chest', 'wheezing in your chest', 'shortness of breath when hurrying on the level or walking up a slight hill', or 'breathing problems interfering with your daily activity' while 11.3% reported a 'cough on most days of the month for at least three months per year'. 3.4% were diagnosed with asthma while 4.0% were diagnosed with either asthma, emphysema, chronic bronchitis, or COPD (see table 3). Presently, we are able to trace over 85% of the participants in the Chikwawa lung cohort and have invited them to participate in this current phase of follow-up.
. CC-BY 4.0 International license It is made available under a author/funder, who has granted medRxiv a license to display the preprint in perpetuity.
is the (which was not peer-reviewed) The copyright holder for this preprint . https://doi.org/10.1101/2020.04. 16.20068643 doi: medRxiv preprint  . CC-BY 4.0 International license It is made available under a author/funder, who has granted medRxiv a license to display the preprint in perpetuity.

Strengths and limitations
The Chikwawa lung health cohort appears to be the only one of its kind in a low-income country setting aiming to investigate the economic costs over the life course of non-communicable respiratory disease. This cohort represents an opportunity to develop and model cost-effective interventions and programmes for this setting. The baseline cohort was conducted alongside a rigorously conducted cluster randomised control trial. Despite local complexities, we presently have identified over 85% of the baseline cohort to be included in the current phase of follow-up.
Systematic bias may be introduced through the self-selection of the participants who agreed to take part in the study to date and the migration of individuals from Chikwawa. Although we have been able to track over 85% of the original Chikwawa lung health cohort and have invited them to participate in the current phase of follow-up, the participants who can be traced and from whom data are collected may differ from those who cannot be traced or do not attend followup. Similarly, at baseline, the participants who agreed to be consented were slightly older and mainly women. The process of verbal autopsies for those who have died (46), and collection of data from the next of kin of those who have moved away, may shed some light on the status of those who have moved away from Chikwawa and deaths from respiratory causes will be of particular interest in the current follow-up. The other limitation identified in this study is recall bias. This is due to most of the data being collected through administering questionnaires in a structured interview format, one can expect recall bias over the follow-up period. We are using tested and validated tools in addition to well-trained experienced interviewers to minimize this bias.
. CC-BY 4.0 International license It is made available under a author/funder, who has granted medRxiv a license to display the preprint in perpetuity.
is the (which was not peer-reviewed) The copyright holder for this preprint . https://doi.org/10.1101/2020.04. 16.20068643 doi: medRxiv preprint The main strength of the cohort is the collection of initial objective measures of lung function using spirometry conducted to internationally agreed standards (34,42) and on two further occasions over a 3-year period. This will provide valuable insights into the health relevance and

Competing interests
There are no competing interests.

Data sharing statement
Further information about the data can be obtained from the corresponding author (martin.njoroge@lstmed.ac.uk). All the from the Chikwawa lung health cohort presented in this article are stored by the research group on safe servers at the Malawi Liverpool Wellcome Trust . CC-BY 4.0 International license It is made available under a author/funder, who has granted medRxiv a license to display the preprint in perpetuity.
is the (which was not peer-reviewed) The copyright holder for this preprint . https://doi.org/10.1101/2020.04. 16.20068643 doi: medRxiv preprint programme (MLW), Malawi and the BOLD centre at Imperial College London, UK and handled confidentially.
. CC-BY 4.0 International license It is made available under a author/funder, who has granted medRxiv a license to display the preprint in perpetuity.
is the (which was not peer-reviewed) The copyright holder for this preprint . https://doi.org/10.1101/2020.04. 16  . CC-BY 4.0 International license It is made available under a author/funder, who has granted medRxiv a license to display the preprint in perpetuity.
is the (which was not peer-reviewed) The copyright holder for this preprint . https://doi.org/10.1101/2020.04. 16  author/funder, who has granted medRxiv a license to display the preprint in perpetuity.