Study design and setting
This cross-sectional study was conducted among female street traders in Warwick Junction between the months of March and May 2015.
Traders were sampled from the various markets within the trading hub; namely mealie (corn) cookers, bovine head market, traditional herb market, clothing market, fresh produce market and roadside vendors. The registry of traders and registered stall holders were obtained from an informal census conducted by a local non-governmental organisation, Asiye eTafuleni (AeT) in 2013. (figure A).
Only female traders above 18 years of age who were currently working in WWJ for any given time period were included in the study. Participants were included into the study following informed consent.
Population and sample selection
The total number of female traders sampled in the Warwick Junction trading hub was 305, with 25% in traditional herb market and 59% in cooking activities (bovine head market and mealie cookers).
The traders were divided into exposed (n=168 (55%)) and non-exposed (n=137 (45%)) groups, based on their occupational exposures. The exposed group consisted of traders from the bovine head market, mealie cookers and traders in the traditional medicine market, all exposed to either biomass fuels used in cooking, or dusts, while the non-exposed were from the clothing market, fresh produce market and roadside vendors.
Due to the limited number of traders working solely in the cooking processes, the full complement of mealie cookers (n=50) were recruited and all the bovine market traders available during the sampling period were recruited (n=43 (72% of the bovine traders)).
The remaining traders were selected randomly from registry of traders present at the market. The lists of traders from the various markets were provided to the research team during the sampling period. From these lists an estimated 10-15% of the traders were randomly selected to participate in the study.
Participant Interviews
Demographic data, relevant medical history and data on current and previous employment were recorded via a questionnaire.
Reproductive health was assessed via previously validated questionnaire used in the National Birth Defects Prevention Study conducted in the United States[17]. Participants were asked to report on five previous pregnancies (where applicable) and their most recent pregnancy. Detailed questions were asked on characteristics of birth based on recall of the participants (foetal weight, sex, year, outcome), doctor diagnosed conditions during the respective pregnancies (gestational diabetes, eclampsia, hyper-emesis gravidarum), use of over the counter medication and traditional medicines and history of smoking, alcohol and drug use during pregnancy. Infertility was assessed via questions on difficulties experienced during conceiving, being diagnosed as infertile by a healthcare practitioner, treatment taken for infertility and current and past history of relationships with other partners (duration of relationship, duration of un-protected sex, if partner had conceived previously).
Statistical analyses
Data was coded and captured using double entry into Microsoft Excel. Statistical analysis was conducted via Stata IC version 13.1 software for Windows.
Dependent variables were reproductive outcomes: (1) number of pregnancies, (2) spontaneous abortions, (3) low birth weight and (4) doctor diagnosed infertility.
Low birth weight was defined as weight at birth of less than 2 500 grams[1]. Spontaneous abortions/ miscarriages was defined as the premature loss of a foetus up to 23 weeks of gestation[18]. Infertility was defined as the inability of a sexually active couple to achieve pregnancy in one year[19].
Independent variables were defined as exposure to trading in WWJ (calculated in years) and trading area.
Univariate analysis, frequency tables and descriptive statistics was used to describe means, range and standard deviations. Means and standard deviations were compared for numerical variables using student’s t test. Categorical variables were analysed using Pearson’s chi square test to determine measures of association between respiratory symptoms and exposure.
Multivariate regression modelling (linear and logistic) was used to determine the association between adverse reproductive outcomes (dependent variables) and independent exposure variables. The multivariate analysis adjusted for covariates, including age, diagnosed chronic conditions, biomass fuel use at home and type of trading activity. Models were designed for reproductive outcomes among traders who were pregnant while working at WWJ. These models were adjusted for age, doctor diagnosed chronic condition, years working in WWJ and biomass fuel use at home. The statistical level of significance was maintained as p < 0.05 with 95% confidence intervals.