Prevalence and factors associated with asthma among adolescents and adults in Uganda: a general population based survey

This study was a cross-sectional general population based survey in five districts in Uganda: Kampala (urban) and Iganga, Kiruhura, Maracha and Pader (rural), Fig. 1. The overall calculated sample size was 2936 participants (518 from each of 4 rural districts and 864 from Kampala) based on the assumption of an asthma prevalence of 8%, a precision of 0.03 and a design effect of 1.5 (to account for the cluster design). Clusters (villages) were selected by probability proportionate to size by Uganda Bureau of Statistics using the Uganda National population and housing census of 2014. Households within clusters were selected by simple random sampling from a household list generated by village leaders. All persons aged ≥12 who were members of selected household and provided written informed consent (and assent in case of minors) were surveyed. Exclusion criteria were: residency of congregation settings (schools, prisons, homes) and temporary residents (less than 2 weeks in household of selected villages). According to the Uganda National population and housing census of 2014, the average number of persons 12 years and older in a household was estimated to be 2.5 persons and the average number of households per cluster was 90 households. Based on these estimates we surveyed a total of 1408 households in 60 clusters across the country; 20 clusters in Kampala and 20 households from each of the clusters and in rural districts we surveyed 10 clusters and 25 households from each of the districts.

In this survey three field teams each comprising of one supervisor, two interviewers, one spirometry technician, one district tuberculosis and leprosy supervisor (DTLS), one local council 1 leader (LC1), one driver and community volunteers as needed was used. Each team surveyed one cluster per day (i.e. about 50 participants/day). The implementation of the survey commenced with the training of the survey teams. Thereafter, a pilot was undertaken to test survey human resources, study tools and the designed data system. After the pilot, adjustments to the tools and the data management system were made. The teams were retrained. Halfway into the survey, amid term review was conducted to inform the investigators of any needed adjustments and strategies to enhance the survey quality.

Sampled participants were interviewed by trained research assistants using a standardized questionnaire developed by adapting questions from internationally recognized questionnaires, namely the World Health Organization (WHO) health survey [1, 10], the ISAAC [10] and ECRHS surveys [9]. Participants who reported either wheeze in the last 12 months, history of current use of asthma medications at the time of the survey or history of ever having a physician diagnosis of asthma were considered to be asthmatics.

Anthropometric measurements were measured; height (measured without shoes to the nearest 0.1-cm using a stadiometer [SECA; Hamburg, Germany]) and weight (measured without shoes and in light clothing to the nearest 0.1 kg using a calibrated beam scale). Blood pressure (BP) was measured using an Omron automated sphygmomanometer model HEM-907, which has an adjustable cuff size. Participants assumed a resting seated posture ≥10 min prior to two sequential BP readings taken 10 min apart. We considered the average of the two BP readings as the individual’s BP. Participants with systolic BP > 130 and diastolic BP > 90 were considered to have hypertension for purposes of this analysis.

Participants who fulfilled the criteria for asthma on questionnaire and were ≥ 35 years underwent spirometry testing to assess for presence of fixed airflow obstruction. The 35 year cut off limit was chosen because fixed air flow obstruction increases with age and based on our previous surveys we found many persons with fixed airflow obstruction from age 35 years and older [11]. Participants identified as having asthma were referred to nearest health facilities for further evaluation and management. Spirometry was conducted and interpreted according to American Thoracic Society/European Respiratory Society guidelines using a Pneumotrac® spirometer with Spirotrac® V software (Vitalograph Ltd., Buckingham, United Kingdom) [12]. Spirometry was performed with participant seated and with a nose clip applied. Testing continued until at least three acceptable and reproducible blows with the largest and second-largest values for both forced vital capacity (FVC) and forced expiratory volume in 1 s (FEV ₁₎ within 150 mL or no more than 5% difference; the largest values for FVC and FEV ₁ were considered the best and used for analysis. Spirometers were calibrated every morning with a 3 L syringe. Pre-bronchodilator spirometry was performed. Participants whose FEV₁/FVC ratio was less than 80% underwent post bronchodilator spirometry (i.e. repeat spirometry 15 min after inhalation of 400 micrograms of inhaled salbutamol). On a daily basis, a physician reviewed all spirograms and those that did not meet the quality criteria were repeated the following day. Predicted parameters were based on NHANES III models as in built within the Spirotrac® V spirometers program used [13]. Participants whose post bronchodilator FEV₁/FVC ratio was less than the LLN ie, participants below the fifth percentile of the predicted FEV ₁ /FVC ratio (calculated with GLI2012 Data Conversion software; version 3.3.1) were classified as having fixed airflow obstruction [14, 15]. However these participants were not excluded from asthma participants on this basis.

Ethics approval was obtained from the Mulago Hospital Research and Ethics committee and the Uganda National Council for Science and Technology. Participants provided written informed consent and were free to terminate study participation at any time during the study. For children between the ages of 12–18 years we obtained their written assent and written parental/legal guardian consent.

The planned sample size was 2936 participants, sufficient to provide a precise national, rural vs. urban and male vs. female estimates assuming a national asthma prevalence of 8%. Urban setting was defined as any areas gazette by the government of Uganda as urban during the 2014 national housing and population census [16].

Prevalence of asthma was calculated as the proportion of participants with asthma in the survey population and presented with 95% confidence intervals (95% CI). Weighting to account for clustering due to the cluster design of the survey was performed. A weight, which is the reciprocal of the overall selection probability (p) was generated as 1/p where p = p₁*p₂*p₃ with p_1, p₂ and p₃ being the probabilities of selecting a district, a cluster within a district, and a household within a cluster, respectively. Later, “svy:” command in Stata was used to apply the weights when estimating the prevalence and other statistics. Because weighted and unweighted prevalence estimates differed, we present the weighted prevalence estimates in this manuscript. Descriptive statistics was used to summarize participants’ characteristics.

To obtain factors independently associated with asthma, a random-effects model was fitted to the data [17]. All factors that were individually associated with asthma with p-value< 0.20 and demographic factors were subjected to multivariable analysis using a random-effects model. To arrive at a better fit, backward model building was conducted using likelihood ratio test (LRT), the multicollinearity was checked using the variance inflation factor (VIF). The results from a better fit and free from multicollinearity (VIF < 10) are presented as adjusted estimates. Data was analyzed using STATA (StataCorp. 2011. Stata Statistical Software: Release 12. College Station, TX: StataCorp LP).

From September 15th to October 10th, 2016, 4310 participants were invited and 3416 participated (participation rate of 79.3%). Of 3416 participants, 61.2% (2088) were female, 22.78% (778) were of urban residence and the median age was 30 years (IQR 20–45). Further details of participants’ characteristics are shown in Table 1.

Overall 323 participants were found to have asthma. Three hundred and eighteen of 323 asthmatic participants (9.3%), 58/323 (1.7%), and 25/323 (0.7%) reported to have had wheezing the past 12 months, had ever had physician’s diagnosis of asthma, and were currently using asthma medications at the time of the survey, respectively. A Venn diagram showing overlaps between these three measures of asthma is presented in Fig. 2. The weighted prevalence of asthma was 11.02% (95% CI: 8.87–13.17), males 10.27% (95% CI: 7.88–12.65), females 11.40% (95% CI: 8.71–14.09), urban 12.99% (95% CI: 9.03–16.95), rural 8.86% (95% CI: 7.74–9.98), Table 2. Among both males and females, the asthma prevalence increased with increasing age, Fig. 3.

More asthmatics than non-asthmatics reported tobacco smoke exposure 14.2% vs. 6.3%, p < 0.001, biomass smoke exposure 28.0% vs. 19.7%, p < 0.001, family history of asthma 26.9% vs. 9.4%, p, <0.001, history of tuberculosis (TB) 3.1% vs. 1.3%, p = 0.010, and hypertension 17.9% vs. 12.0%, p = 0. 003, Additional file 1: Table S1.

The proportions of participants with allergy and respiratory symptoms by asthma status are presented in Additional file 1: Table S2A &2B. Nasal congestion in the past 12 months was reported by 40.3% of asthmatics vs. 13.2% non-asthmatics, p < 0.001). Itchy watery eyes were reported by 40.6% of asthmatics vs. 20.6% non-asthmatics, p < 0.001) while skin rash was reported by 20.7% of asthmatics vs. 11.0% non- asthmatics, p < 0.001. The proportions of the different respiratory symptoms by asthma vs. non-asthma status respectively were: cough (51.7% vs. 17.6%, p = < 0.001), shortness of breath (40.3% vs.5.8%, p < 0.001), chest pain (56.7% vs. 22.3%, p < 0.001) and sputum production (28.5% vs. 5.3%, p < 0.00).

The factors independently associated with asthma in this survey as obtained from an adjusted random-effects model were: smoking, adjusted odds ratio (AOR) 3.26 (95% CI:1.96–5.41, p < 0.001), family history of asthma, AOR 2.90 (95% CI: 1.98–4.22 p- <0.001), nasal congestion in the past 12 months, AOR 3.56 (95% CI: 2.51–5.06, p < 0.001), biomass smoke exposure, AOR 2.04 (95% CI: 1.29–3.21, p = 0.02) and urban residence, AOR 2.01(95% CI: 1.23–3.27, p = 0.05), Table 3. All respiratory symptoms were associated with asthma, AORs (95% CIs) of: cough 2.41 (1.66–3.50, p < 0.001), shortness of breath 6.84 (4.57–10.23, p < 0.001), chest pain 3.00 (2.15–4.19, p < 0.001) and sputum production 1.81 (1.16–2.88, p = 0.009), Table 3. The factors associated with asthma in a model that considers only factors associated with asthma with a p-value less 0.05 at a bivariate stage are shown in Additional file 1: Table S3.

Of the 323 participants who were classified as having asthma on the questionnaire, 138 (42.72%) were 35 years and older and therefore eligible for spirometry. Of these, 120 (86.96%) underwent spirometry and 18(13.04%) did not. We obtained interpretable spirometry in 106 of the 120 (88.33%). After post bronchodilator testing, 16 of the 106 participants who underwent spirometry were confirmed to have fixed airflow obstruction (15.09%), 13(12.26%) had significantly reversible airflow obstruction (i.e. FEV₁ reversibility of > 12% or > 200mls) and 9 (8.49%) had a restriction.