Respiratory symptoms, fractional exhaled nitric oxide & endotoxin exposure among female flower farm workers in Ethiopia

A cross-sectional study was conducted from July to October 2013 in four flower farms in the three main geographical areas for cultivation of roses in Ethiopia (Hollota/Addis-Alem, Debrezeit and Ziway areas). Due to practical reasons such as location and accessibility, we used convenient sampling to select the farms. The workers in each of the farms were divided into two observational groups based on their workstation (inside vs. outside greenhouses) for an interview, FeNO measurement and endotoxin sampling under their normal working conditions. All the four farms were included in the assessment of respiratory symptoms, but due to limited resources only two farms were included in the measurements of endotoxin and FeNO. The farms were visited before the actual data collection and the farm managements were informed about the purpose of the study.

The farms only grow roses, and vary in size from 5–51 hectares, while the total number of workers ranged from 350 to 1,300 (Table 1). The production rate is relatively constant, but at times there is peak production to meet increased demands. All the farms use chemical fertilizers and pesticides to enhance the growth of flowers and to control pests. In addition, Farm I also uses mites as biological control, however, we did not collected information on the types of mites used. The pesticides are manually mixed at a central location inside greenhouses and distributed through pipelines. The sprayers use a nozzle or sprayer gun to spray the pesticide. In all the farms, except Farm IV, spraying takes place in the morning during which time the other greenhouse workers are relocated to other greenhouses until the spraying is completed. In Farm IV spraying is done late in the afternoon after all greenhouse workers have completed their work and left the farm.

The great majority of the workers are females, who either harvest and weed flowers inside greenhouses or trim and pack roses in a pack-house before storage in a cold room until transporting them to Addis Ababa for export. The few men work as pesticide sprayers and also perform other activities including repairing the greenhouses, collecting and disposing of wastes as well as transporting harvested flowers.

Female workers from the selected farms participated in the study. Based on the prevalence of shortness of breath (70%) among Ethiopian flower farm workers [13], a total of 248 participants (divided in two groups, i.e., inside vs. outside greenhouse) were needed to achieve 80% statistical power with significance level set to P < 0.05. The plan was to have equal number of workers from each of the four farms as well as from inside and outside greenhouses, and overall 122 and 126 workers participated from inside and outside greenhouses, respectively. The same participants from Farm I & II (n = 114) were also invited to participate in the FeNO measurements. Participants were randomly selected by the researcher from the personnel registration book in Farm II & III, while in Farm I & IV, the selection was done together with the farm administration. The principal investigator did all the interviews and the FeNO measurements in a separate room inside the farm compound. The FeNO instruction was sometimes done in groups of 2–4 workers.

A part of the standardized questionnaire from the British Medical Research Council (BMRC) on respiratory health was used to assess chronic respiratory symptoms. The questionnaire was adapted to the Ethiopian context, and translated from English to the local language; the symptoms included cough first thing in the morning, day/night time cough, shortness of breath, chest tightness and wheezing as well as previous diseases such as bronchitis, pulmonary TB, asthma and chest injury (yes/no). In addition, questions about upper airway symptoms (runny nose, sneezing and blocked nose) were also included [21]. Moreover, previous flower farm experience (months), socio-demographic characteristics such as age, education (years in school), types of job (cutter & weeder, packing, deleafing, grading, quality control), present and previous smoking habits (yes/no) and the use of Personal Protective Equipment: gloves and respiratory protective device (yes/no) were also asked.

FeNO measurements were taken for each person in a sitting position during the working day between 10:00 and 15:00 hours, using a portable electrochemistry-based sensor (NIOX MINO; Aerocrine AB, Solana, Sweden) in accordance with the American Thoracic Society & European Respiratory Society recommendations for online measurement of FeNO [22,23]. The mean ambient NO, room temperature and relative humidity of the room where FeNO measurement performed were registered daily; below 5 ppb, 22°C (18 – 28°C) and 70%, respectively.

Full-shift personal endotoxin air samples were collected in the two selected farms (inside vs. outside greenhouses). The number of endotoxin samples was based on the Rappaport and Kupper’s recommendations for exposure studies; 10–20 measurements per observational group, two measurements of 5–10 randomly selected individuals per group [24]. Totally 46 female workers were invited for sampling; 25 from Farm I, and 21 from Farm II. Among these, 31 workers had repeated samples.

Sampling was performed from the workers’ breathing zone during the summer months, which is a rainy season in Ethiopia. Side Kick Casella Pump at a flow rate of 2 L/minute connected to 25 mm closed-face conductive Millipore cassettes containing a glass fiber filter with 0.2 μm pore size was mounted onto each worker for sampling of the “total dust” fraction. The mean sampling times were 350 (275 – 425) minutes and 358 (240 – 405) minutes for Farm I and II, respectively. The pumps were calibrated using a Rotameter every day before starting the measurement, and the air flow rate was also measured at the end of the measurements. Two samples were excluded due to a reduction in air flow of more than 10% during the sampling day. Filters were kept cold at about 4°C inside a box filled with ice bags until the samples were transported to Norway and then to Sweden for analysis at Lund University Medical Laboratory. The samples were transported as hand luggage on the flights, and were outside the cold box for a total of about 15 hours. The glass fiber filters were immersed into 0.05% Tween-20 pyrogenic free water and shaken on a rotary shaker for 1 h. Endotoxin extracts were analyzed using kinetic chromogenic Limulus amebocyte Lysate (LAL) Assay. The results were expressed as Endotoxin Unit (EU) per filter. A total of 27 samples had endotoxin values below the Limit of Quantification (LOQ), which was set to 10 EU per filter by the laboratory. In calculations of exposure levels (EU/m³), the amount of endotoxins on the respective filters were divided by the air volume that had passed through that filter. The measurements below LOQ were set as LOQ/2^1/2 in further data handling. This cut-off level has been suggested when less than 50% of the data are nondetectable, and the geometric standard deviation is below 3.0 [25]. The resulting limit of detection (LOD) for the endotoxin exposure corresponding to the 27 samples < LOQ varied from 8.7 to 14.7 EU/m³ due to different air volumes.

The research proposal was submitted and approved by the Ethical committees both in Norway and Oromia Regional State Health Bureau, Ethiopia. All the participants were informed about the purpose of the study and written consent was obtained.

The data was plotted into SPSS version 21 and analyzed using descriptive statistics. The distribution of endotoxin and FeNO data were skewed, and were therefore log-transformed in order to compare the levels between the groups. Chi-square test for categorical variables, and independent t-tests for continuous variables were used to test differences between the groups. The significance level was set to 0.05. A linear mixed-effects model was used to analyze the determinants of endotoxin exposure among the female workers, since we had repeated measurements of endotoxins on several of the workers. The individual female participants were included in the model as random effects. The farms (Farm I vs. Farm II) and workstation (inside vs. outside greenhouses) were set as fixed effects. Logistic regression was used to test differences in symptoms between groups by adjusting for education. Education was the only potential confounder among those tested that was significantly different between workers inside vs. outside greenhouses.

The workers were categorized by workstation based on whether they work inside the greenhouse or not. The response rate for the interview was 100%. The mean age of all female workers was 24 years (range 14 – 60 years). There were no differences in age, previous respiratory diseases and duration of work experience between inside and outside greenhouse workers (Table 2). Workers outside the greenhouse had significantly higher education than those working inside (Table 2). None of the female workers were current smokers or had any previous history of smoking. The majority of the workers (97%) used biofuel (wood, cow dung & charcoal) as energy source for cooking and other domestic purposes. A quarter of the participants had domestic animals, and 7% shared their living space with the animals.

The personal endotoxin exposure ranged from < LOD to 180 EU/m³ with a geometric mean of 22.8 EU/m³ (Table 3). Only one of the samples exceeded the Dutch recommended health-based exposure limit value for endotoxins in inhalable dust of 90 EU/m³ [26]. The endotoxin exposure inside the greenhouses was significantly higher than outside the greenhouses (p < 0.05) (Table 3). Workers in Farm I had significantly higher endotoxin exposure than workers in Farm II (GM = 29.2 Vs. 17.3 EU/m³; P < 0.01). The greenhouse workers in Farm I were the subgroup with the highest endotoxin exposure (GM = 37.8 EU/m³) (Table 3).

The day-to-day endotoxin exposure variability (within workers) was higher than the between worker variability (Table 4). In a mixed-effects model analysis, the fixed factors workstation and farm, explained 17% of the total variance in endotoxin exposure (Table 4). The fixed factors explained parts of the between-worker variance only. Work in Farm I was associated with 1.7 times higher personal exposure to endotoxin compared to Farm II, while those working inside the greenhouse had 1.4 times higher exposure than those working outside.

None of the female workers in any of the four flower farms used any type of Respiratory Protective Devices (RPD). Almost all (99.6%) of the workers indicated the reason for not using RPD was because it is unavailable or not provided at the work place. Gloves at work were used by 64% of all female workers (78% and 50% among inside and outside greenhouse workers respectively) with significant differences between the farms (p < 0.01). The reported reason for not using gloves was mainly because gloves were not available or provided.

Among all invited workers in the four farms shortness of breath (53%), chest tightness (27%) and morning cough (25%) were the most frequently reported symptoms (Table 5). Morning cough, runny nose and blocked nose were significantly higher among greenhouse workers compared to those working outside (Table 5). When adjusting for education by logistic regression analyses, only the prevalence of blocked nose remained significantly higher among workers inside greenhouses. The prevalence of morning cough and day/night time cough with sputum were significantly different among the four flower farms (p = 0.009 & 0.028 respectively), while the other symptoms did not differ among these groups. Farm I had the highest prevalence for most of the symptoms (results not shown). When comparing results of Farm I & II, where we also have measurements of personal endotoxin exposure and Fractional exhaled Nitric Oxide (FeNO), Farm I has relatively higher prevalence for most of the symptoms. The prevalence inside the greenhouse was higher than outside for most of the symptoms. Due to small numbers, the groups inside and outside the greenhouse in these two farms were not statistically compared after stratification by farm.

A total of 114 female flower workers from Farm I & II participated in measurements of Fractional exhaled Nitric Oxide (FeNO), and valid results were obtained from 108 of these.

Among all participants FeNO ranged from 5 to166 ppb with a GM of 14 ppb (Table 6). Only two workers (1.8%) had FeNO concentration greater than 50 ppb, a level often used to indicate the presence of asthma [27]. The mean FeNO of those working inside and outside the greenhouses did not differ significantly, either among workers in Farm I and Farm II together or within each of the farms (Table 6). Workers in Farm I had significantly higher mean FeNO than in Farm II (GM = 15.8 ppb and 11.8 ppb; P = 0.009) respectively (Table 6). When excluding the two values above 50 ppb, the statistical differences between the groups did not change.

Strengths of the present study are that objective measurements were done both for endotoxin exposure and FeNO. Moreover, the subgroups (inside vs. outside greenhouses) were comparable for most variables except education, which we controlled for in the analysis.

However, it is a weakness that we only measured endotoxin exposure during one summer (rainy and wet). This may not represent the winter (sunny and dry) exposure. A high fraction of the endotoxin samples had values below the limit of quantification at the laboratory. The high fraction of these low levels has presumably influenced the estimates of mean exposures and the fixed factors in the mixed-effects model, but presumably they did not affect the main findings. Pesticides and fertilizers of different types were widely used in the flower farms [13], but we were unable to measure these exposures in this study. There might be a healthy worker effect as most workers had short work experience due to high turnover. In addition, the administration may have picked the healthiest workers for examination. However, the administrative personnel was not likely to know all the details of the workers’ health.