Ethical approval for the study was provided by the Institutional Review Board of the School of Medicine, Makerere University College of Heath Sciences and permission confirmed through the Uganda National Council for Science and Technology (UNCST).
The study conformed to the principles outlined in the declaration of Helsinki. Each study participant aged 18 years and above gave a written informed consent. Parents/next of kin gave a written informed consent on behalf of the minors/children participants. Study participants or their next of kin were asked to fill a questionnaire of socioeconomic and environmental factorsAll data were kept in confidence with each participant assigned a unique identifier code.
This was an unmatched case control study in which 243 cases and 243 controls were recruited using systematic random sampling in Mulago National Referral Hospital.
A random sample of cases and controls from a variety of geographical and social origins was identified at Mulago National Referral Hospital. Controls recruited for the study were selected using systemic random sampling from among people referred for cardiac evaluation for suspected heart disease but found to have normal echocardiograms.
Eligibility and Exclusion Criteria
Cases were subjects aged 5 to 60 years, diagnosed with RHD based on history of acute rheumatic fever, clinical examination finding of a murmur and standard echocardiographic criteria. Only cases living at the current location for the last five years were considered. Cases with possible and probable RHD were excluded from the study. Comparable eligibility criteria for controls included subjects aged 5 to 60 years with no previous history of RHD and had normal echocardiograms and living at the current location for the last five years were considered. Controls with raised antitreptolysin O titres on laboratory evaluation were excluded.
Echocardiography was done according to standard criteria 
Briefly, Standard transthoracic echocardiography (GE, Vivid 8, Chicago, USA) was performed where 2D, M-mode, colour and doppler modalities were done with images taken from parasternal long axis, parasternal short axis, apical four chamber, apical five chamber and subcostal views. Mitral stenosis was diagnosed by two-dimensional echocardiographic and hemodynamic evidence, as well as characteristic mitral valve morphology, including thickened mitral leaflets, cusps, commissures, anterior mitral leaflet doming in the long-axis view, anterior motion of the posterior mitral leaflet on M-mode echocardiography. Mitral valve area was graded based on planimetry and pressure half time methods. Mitral stenosis was graded as mild for valve area between 1.6–2.0 cm2, moderate for valve area 1.1–1.5 cm2, and severe for valve area <1.0 cm2. Aortic stenosis was diagnosed based on the presence of commissural fusion of the aortic leaflets, possible increased echogenicity along the leaflet edges, and systolic doming of the aortic leaflets. Valve area was assessed using planimetry and classified as mild (vale area >1.5 cm2), moderate (valve area 1–1.5 cm2) and severe if valve area was <1 cm2. Mitral and aortic regurgitation were assessed using quantitative methods.
Socioeconomic and Environmental Status
The following parameters were documented; age, gender, level of education, monthly earning (in Uganda Shillings and later converted to United States dollars), number of people per house, distance from the nearest health center (kilometers) and space area per person (square feet).
Level of Income
For standardization purposes, we used the modified Prasad classification of social status which divides individuals into 4 social strata including lower class (below a dollar a day) middle class, middle-high and highest class. 
The Prasad classification was adapted to local conditions by converting the Indian Rupees to Uganda Shillings and to United States dollars. For study participants below 18 years, we used their mothers’ income as an estimate.
Number of People Per House and Space Area
The number of people per house, and space area were determined to define overcrowding.
A household with over 8 people was classified as overcrowded. Similarly, basing on the American Public Health Association recommendation that in a shared household, each individual should have at least 90 square feet of space 
, we surveyed the homes of at least 100 cases and 100 controls (50%) of each study group, and determined the average house area per study group (120 square feet for controls versus 80 square feet for cases). We then divided the average house area of the control group, which was higher, by the number of people per house in the controls and cases study groups to determine space area per person, which was compared between the two groups.
Distance from the Nearest Health Centre
Distance from the nearest health centre was determined to assess access to health care. Previous studies in developing countries have showed that distance of 5 kilometers or less to a health unit predicted better access to health care and therefore uptake of prevention programs such as vaccination. 
Data was double-entered and stored in EPI data version 3.0. (EpiData Association, Odense M, Denmark) Analysis was done using STATA 10.0 statistical package. (Stata Corporation, College Station, TX, USA).
With rheumatic heart disease status as an outcome (dependent) variable, the following were analyzed as independent variables; age (years), income level (Uganda Shillings/United States dollars), education status, number of people per house, space area per person (square feet), distance from the nearest health unit (kilometers) and employment status (yes/no).
At univariate analysis, the Chi square test was used for categorical variables and Student t and was used for continuous variables, respectively for statistical significance.
The following covariates that were significant at the univariate level (at p-values ≤0.2) were entered into the multivariate analysis; Age (years), education level, employment status (yes/no), number of people per house, space area per person (square feet) and distance from home to the nearest health centre (kilometer). Odds ratios were computed with 95% CI.
Logistic regression was used at multivariate analysis where covariates listed above were considered and tested for confounding and interaction. The backward likelihood ratio method was used to get the space area (sq feet) as most significant variable in the multivariate model (main predictor). To test for interaction, product terms were formed between space area (main predictor) and other basic variables in the model. Using the chunk test, where a fit of a model with all interaction terms included is compared with a fit of a model with none of the interaction terms 
. The negative two log likelihood (−2LL) of the full model with interaction terms were included and the reduced model containing only basic variables were compared. Interaction was considered present when the difference between −2LL was significant (at p≤0.05) with a Chi-square test. Confounding would be considered present if the difference between crude and adjusted odds ratios would greater than or equal to 10%.