Prevalence of Respiratory Symptoms and Chronic Obstructive Pulmonary Disease, and Reference Values for Lung Function Testing in Kinondoni District, Dar Es Salaam, Tanzania

In the wake of socioeconomic development, changing demographics towards ageing populations and the expatiation of non-communicable diseases put additional strains and new challenges unto health services and disease management in low-income countries. Chronic Obstructive Pulmonary Disease (COPD) is by the WHO Global Burden of Disease Study projected to rank as third leading cause of death worldwide in 2020, though studies concerning prevalence of COPD and presence of risk factors in low-income settings are yet rare. The expected frequency and burden of COPD is however anticipated to be of same magnitude in these countries, where exposures to occupational and domestic hazards are considerable, as for infections and morbidities, influences known to be detrimental to respiratory health. Assessing prevalence and diagnosis is an important step towards clinical management and awareness regarding implementation of prevention initiatives. A necessity in diagnosis of COPD and measuring disease severity is the use of reference values for normal lung function. Adhering to American Thoracic Society (ATS) recommendations, such prediction equations should be based on healthy people with the same anthropometric characteristics and ethnic origin as the subjects being tested. This study aims at deriving reference values of a healthy adult suburban population in Dar Es Salaam, Tanzania, and to estimate prevalence of COPD from a random sample of the population. The study is based on a descriptive cross-sectional design. The source population comprised all adults above the age of 15 in Kinondoni district. Estimating prevalence to 15 % and accounting for reduced precision due to cluster sampling procedure, the sample size required was set to 300 participants. Approval was obtained from the Ethical Committee of Western Norway (REK-Vest) and the Medical Research Coordinating Committee of the National Institute for Medical Research (MRCC). Subjects within the age-span examined and who were willing to participate, were eligible for participation. A total of 365 subjects were enrolled in the study. Sub-selection to the reference sample and for generation of prediction equations were based on ATS recommendations, where subjects with negative responses to core questions from ATS-DLD regarding respiratory symptoms and doctor diagnosed heart/chest illnesses were selected, providing their spirometric data met ATS criteria. Lung volumes was tested using the ndd EasyOne spirometer, and the following parameters were measured; peak expiratory flow (PEF), forced expiratory volume in one second (FEV1), forced expiratory volume in six seconds (FEV6) and the ratio of these two measurements 7 (FEV1/FEV6). All spirograms were reviewed by an experienced chest physician, adhering to ATS acceptability and reproducibility criteria for selection of best pulmonary function curve. Anthropometric data, including weight, standing height and sitting height were collected, and age was recorded as birth-date or as mid year in the year of birth. Respiratory symptoms were recorded using the ATS-DLD validated questionnaire, and questions regarding socioeconomic conditions were derived from a culture specific questionnaire used in the Tanzania Demographic Health Survey from 1999 and 2002. Prevalence of COPD was estimated according to the Global initiative for chronic Obstructive Lung Disease (GOLD) and ATS criteria, and severity was determined by GOLD disease stages. Based on the inclusion criteria, a total of 150 subjects, 52 men (32,7 %) and 98 women (47,6 %) were selected to the reference value group. Median age for men and women in the reference sample was 34 years. The spirometric parameters FEV1, FEV6, FEV1/FEV6 and PEF were regressed against sample mean age, height and weight, and the models were assessed in terms of whether a linear or curvilinear prediction produced best fit for the present data. In the male strata, an exponential model was selected as regression equation for all of the spirometric variables, with the exception of prediction of PEF when sitting height was independent variable, where a linear model was applied. For women, a linear model was chosen in further analyses, except for the prediction equations for PEF, where an exponential equation best fitted the data. For women, weight was retained in the final model when standing height was part of the independent variables and regressed against PEF. For the other spirometric variables, weight was non-significant, and was removed from the equations in further analyses. Height, and height square in the exponential equations for men and women, entered all the regression models with the exception of the model predicting FEV1/FEV6, as it did not provide a significant contribution to the variance of the dependent variables. All the spirometric parameters were negatively related to age, and all increased with height, with exception of FEV1/FEV6. The lower limit of normal (LLN) is presented as – 1,645 x SEE which is the age and height specific, estimated 5th percentile for the reference sample. The reference equations derived from our study do not allow direct comparison with previously published predictions due to differences in source populations and for the effects of altitude, however, on average, our equations generated lower reference values. When comparing our predictions to those published by Mustafa in 1977, no secular changes in FEV1 values could be traced. . 8 Both presence of respiratory symptoms and illnesses were more frequently reported among the female participants, though considerable in both sexes. Regarding questions on smoking exposure, a proportion of 14,0 % of the study population responded that they were current smokers, and 4,7 % was ex-smokers. Patterns of cigarette smoking differed substantially between the sexes, where the proportion of current smokers was considerably higher for men (30,2 %) than for women (1,5 %). Prevalence of COPD when applying a fixed cut off ratio of FEV1/FEV6 < 0,73, was 12,6 %, and it was equally distributed among men (13,9 %) and women (11,5 %). Prevalence of COPD when classified as below the lower limit of normal was 7,9 %, also equally distributed with 6,9 % and 8,7 % of men and women respectively. In our study, prevalence of stage I and II COPD was 6,1 % and 5,8 % respectively. Direct logistic regression was conducted to assess the impact of certain determinants on COPD. Separate analyses were carried out for GOLD and ATS/ERS defined COPD. The multivariate models contained the following independent variables; sex, age, domestic exposure, occupational exposure, smoking status and socioeconomic position. When GOLD defined COPD was dependent variable, the following covariates made a statistically significant contribution to the model; being ex-smoker, (p = 0,01), and age (p < 0,0005). The strongest predictor was ex-smoker, with an odds ratio (OR) of 5,37. In the model using the ATS defined COPD prevalence, the independent variable, ex-smoker was still significant (p = 0,004) with an OR of 7,92. In addition, domestic exposure made a statistically significant contribution to the variance of the model (p = 0,013). However, the OR was less than 1 (0,19), indicating a negative association.Master of philosophy in international healthMAMD-INTHINTH39