Microbial contamination of coffee during postharvest on farm processing: A concern for the respiratory health of production workers

Abstract

Coffee workers respiratory health problems, found to be associated with dust exposure in the coffee factories, but the content of the dust is not well known. A cross-sectional survey was conducted to assess the microbial contamination of coffee in dry and wet coffee process, from four farms in two regions of Ethiopia. A total of 36 samples of coffee were collected for laboratory investigation. The microbial load in the dry process ranged from 6.9 × 10² to 7.2 × 10⁵ colony forming units (CFU)/mL while the microbial load in wet process ranged from 2.5 × 10² to 4.6 × 10⁵ CFU/mL. The results indicate the presence of gram negative bacteria in dried and stored beans from both the wet and dry process. During further coffee processing possible release of endotoxin from coffee contaminated by gram negative bacteria might affect coffee workers respiratory health. Further studies are required to assess the relation between bacterial contamination of coffee and endotoxin level in coffee factories.

Keywords:

• Coffee
• gram negative bacteria
• postharvest Ethiopia
• endotoxin

Background

Coffee is a major export commodity and a source of foreign currency for Ethiopia, and contributes to about 60% of the total export earnings.¹ It is estimated that 15 million people in Ethiopia depend on coffee production directly or indirectly for their living.²

Studies in primary coffee processing factories in Uganda and Sri Lanka have indicated a higher prevalence of acute respiratory symptoms than among controls,^3,4 while an increased prevalence of chronic respiratory symptoms has been reported among primary coffee factory workers in Papua New Guinea and Tanzania.^5,6 Studies have also observed reduced lung function among coffee workers compared to controls.^5,7 These studies are from Papua New Guinea and Tanzania and indicate that the coffee workers might develop respiratory lung diseases due to dust exposure at work. Similarly, a study recently performed in Ethiopia among workers in primary coffee processing factories shows that coffee workers had a significantly higher prevalence of chronic respiratory symptoms and reduced lung function compared to controls.⁸ This study was performed in the same area as the present study. Primary coffee processing factories refer to mechanical cleaning of debris from parchment coffee and include hulling, grading, hand picking, and packing of green coffee beans. Several studies conducted in coffee roasting factories have also found a higher prevalence of chronic respiratory symptoms and reduced lung function among coffee workers than among controls.^9–13

After harvesting, the coffee cherries pass through different processing stages to remove the layers of the coffee cherries (ie, hull, pulp, mucilage, parchments, and silver skin).¹⁴ In Ethiopia, two types of postharvest on farm processing of coffee are used to remove these layers and they are called wet and dry processes. During dry processing the unpulped coffee cherries are allowed to dry in the sun under natural conditions by spreading on ground, mats, cemented floor, or raised dry bed.^14,15 The spread coffee is heaped in the night and re-spread each morning. This process continues for about two to three weeks depending on the weather condition of the area.¹⁴ During wet processing, the coffee fruits are depulped on the day of harvesting. The pulped coffee beans are allowed to ferment naturally in different tanks for about 16–24 h depending on the weather condition. When fermentation is completed, the coffee beans are washed with running water to remove the remaining mucilage and acids. Finally, the wet coffee beans are spread on a drying table and allowed to dry by sun.² In Ethiopia, about 71% and 29% of coffee is processed by dry and wet process, respectively.¹⁶

Studies have indicated that endotoxin is released to the working environment when agricultural products contaminated with dust containing gram negative bacteria are processed.^17–20 Endotoxin is known to cause respiratory health problems.^21–25 Studies conducted in primary coffee processing factories have shown high levels of total dust exposure.^4,5,26 In our previous study, we indicated that 84% of the dust samples among male coffee workers in primary coffee processing factories in Ethiopia exceeded the occupational exposure limit value,²⁷ but there is little knowledge about the content of the dust. However, Sakwari et al. found high levels of endotoxin in primary coffee processing of Tanzania.^7,26 This finding indicates that coffee can be contaminated with gram negative bacteria like any other agricultural products during postharvest on farm processing of coffee and storage.²⁸ Furthermore, Sakwari et al. found an association between exposure to endotoxin and respiratory symptoms and reduced lung function among the coffee processing factory workers in Tanzania.⁷ The same author suggested that poor storage and drying coffee on ground might have increased bacterial contamination of the beans,⁷ but this suggestion has not been verified.

The aim of the present study was to assess microbial contamination of coffee in different stages of both wet and dry postharvest on farm coffee processing. Thus, this study will identify points of coffee contamination, and could provide new information on how to prevent and control bacterial contamination of coffee beans.

Materials and methods

Study area

The study was conducted in Yebu woreda of Jimma Zone in Oromia Regional State and Shebedino woreda, Sidama Zone of Southern Nations Nationalities Peoples’ Region (SNNPR) of Ethiopia. The two wordeas were selected because they are the largest coffee growing areas in their respective regions.

Sampling procedure

This study assessed two wet and two dry coffee processes on four different farms. In wet processing the beans pass through six processing steps, where the dry process comprises three different processing steps (Figure 1). Coffee samples were collected two times on different days from different batches of coffee from each stage of both the wet and the dry process. Thus 24 samples from the two wet processes and 12 samples from the two dry processes were collected for laboratory investigation. Sampling was conducted in the dry season of the year (from October 2016 to January 2017).

Figure 1. Stages of wet and dry processes and sampling of coffee beans were performed in different stages of dry and wet processing, described inside boxes.

Sample collection

The sample collection form contained date and time, types of sample, location, name of stages and sites. All equipment and materials were sterilized prior to use. Coffee cherries and beans, each weighing about 25 g, were sampled from each stage of the wet and dry processes. The principle investigator together with one senior microbiologist collected the samples from each stage in sterile plastic bag, labeled with types of sample, location, name of stages, and sites and transferred to ice boxes (triple package). The samples were transported within 6 h of collection to Addis Ababa City Administration Health Bureau Laboratory (Public Health Microbiology Laboratory) and analyzed as soon as possible after receipt in the laboratory.

Sample processing and analyses

Three experienced microbiologists at Addis Ababa City Administration Health Bureau Laboratory analyzed the samples. The samples were analyzed for heterotrophic plate count. The coffee cherries were transferred to a flask containing 225 mL of sterile buffered peptone water (1% peptone, 5% weight per volume NaCl) and swirled gently for 20 min using orbit shaker (lab-line instruments.inc, Model 3521, USA).²⁸ The orbit shaker was used to detach microorganism from coffee bean surfaces. Peptone water was used to make a serial dilution (1:10) for each sample. Then serial dilution was made to 10⁻³ for each samples to get appropriate number of colony which ranges from 30 to 300. From each dilutions, 1 mL of sample was pour plated on plate count agar (PCA) (Park Scientific (USA) as described in the Food and Drug Administration Bacteriological Analytic Manual.²⁹ Then the plates were incubated at 37 °C for 72 h (296, South Africa). Fresh media was utilized and its sterility was checked by overnight incubation. Quality control was used in each batch of the samples.

Colony count was made by YLN-30 lab colony counter magnifying digital display apparatus (UK) and undistinguished colonies were ignored. All distinguishable colonies were counted. The best two consecutive dilutions were used, as n₁ and n₂ to calculate the results. Total bacteria colony count was presented as organisms per milliliter of Coffee colony forming unit (CFU/mL). The average plate count was calculated using this formula: $$N =C/V (n_{1 }+0.1n_2)d$$ where C is the sum of colonies on all plates counted; V is the volume applied to each plate; n₁ is the number of plates counted at first dilution; n₂ is the number of plates counted at second dilution; and d is the dilution from which first count was obtained.

Bacteria were identified as gram negative and gram positive bacteria based on standard Gram-stain technique and microscopic observation.³⁰ Yeast and Filamentous fungi were identified based on their morphology from gram stain via microscopic observation (3H30RF200, Germany). All the stained slides were cross checked by another microbiologist from the Ethiopian Public Health Institute. When gram negative bacteria identified in the coffee process the result was presented as “Yes” and when it was absent it was reported as “No.”

Data analysis

The data were presented using descriptive statistics. Independent t-test was used to compare mean microbial load between the two types of on farm postharvest processing of coffee and between the two study areas. The analysis was performed using SPSS version 22.³¹ Statistical significance level was set to a p-value less than 0.05.

Ethical considerations

The Institutional Review Board of the College of Health Sciences at Addis Ababa University and the National Research Ethical Review Committee of the Ethiopian Ministry of Science and Technology approved the study. Permission to conduct the study was obtained from each on farm postharvest coffee processing owners.

Results

When merging the two regions the mean microbial load in the dry process (2.0 × 10⁵ CFU/mL; range 6.9 × 10² to 7.2 × 10⁵ CFU/mL) did not differ significantly (p = 0.22) from the microbial load in the wet process (9.0 × 10⁴ CFU/ml; range 2.5 × 10² to 4.6 × 10⁵ CFU/mL). Furthermore, there was no significance difference in microbial load between the two regions (1.7 × 10⁵ CFU/mL versus 8.4 × 10⁴ CFU/mL; p = 0.18) or between the wet and dry processes within the SNNPR (p = 0.74) or within the Oromia region (p = 0.16) (Table 1).

Table 1. Microbial load in different stages of wet and dry coffee processing in Ethiopia.

Regions	Types of postharvest on farm processing of coffee	n	Coffee samples from stages of coffee processing	Aerobic plate count (CFU/mL)	p-value comparing dry and wet types of postharvest on farm processing of coffee
SNNPR	Dry process	2	Coffee cherries	1.8 × 10^5	0.74
		2	Dried beans	1.7 × 10^4
		2	Stored beans	1.9 × 10^4
		6	Mean for the 3 dry process	7.3 × 10^4
	Wet process	2	Coffee cherries	1.0 × 10^5
		2	Pulped beans	8.6 × 10^4
		2	Fermented beans	2.7 × 10^4
		2	Washed beans	1.2 × 10^2
		2	Dried beans	1.4 × 10^5
		2	Stored beans	1.9 × 10^5
		12	Mean for the 6 wet process	9.0 × 10^4
Oromia	Dry process	2	Coffee cherries	1.8 × 10^4	0.16
		2	Dried beans	3.6 × 10^5
		2	Stored beans	6.2 × 10^5
		6	Mean for the 3 dry process	3.3 × 10^5
	Wet process	2	Coffee cherries	8.4 × 10^4
		2	Pulped beans	2.1 × 10^4
		2	Fermented beans	3.7 × 10^4
		2	Washed beans	2.6 × 10^5
		2	Dried beans	1.1 × 10^4
		2	Stored beans	1.4 × 10^5
		12	Mean for the 6 wet process	9.1 × 10^4
Both regions	Dry process	4	Coffee cherries	1.0 × 10^5	0.22
		4	Dried beans	1.9 × 10^5
		4	Stored beans	3.2 × 10^5
		12	All in dry process	2.0 × 105
	Wet process	4	Coffee cherries	9.5 × 10^4
		4	Pulped beans	5.4 × 10^4
		4	Fermented beans	3.2 × 10^4
		4	Washed beans	1.3 × 10^5
		4	Dried beans	7.4 × 10^4
		4	Stored beans	1.6 × 10^5
		24	All in wet process	9.0 × 104

Independent t-test were used to compare mean difference between types of postharvest processing of coffee and between regions.

Microbials identified in different stages of postharvest on farm coffee processing

Bacteria, yeast, and fungal filaments were identified in both wet and dry process in the two study regions. In the wet process, yeast cells were dominant among pulped, fermented, washed, and dried beans (Table 2).

Table 2. Microbial identified in different stages of postharvest on farm coffee processing.

Types of on farm postharvest processing of coffee	Stages of coffee processing	Shebedino woreda SNNPR				Yebu woreda, Oromia region
		Gram positive	Gram negative	Yeast	Fungi	Gram positive	Gram negative	Yeast	Fungi
Dry process	Coffee fruits	×	×	×		×	×
	Dried beans	×	×	×		×	×		×
	Stored beans	×	×				×	×
Wet process	Coffee fruits	×	×		×	×			×
	Pulped beans	x	×	×	×	×		×
	Fermented beans	×		×		×		×
	Washed beans		×	×				×
	Dried beans		×	×			×	×
	Stored beans	×	×			×	×

“×” refers the presence of microorganisms indicated in the column head and the blank cells indicate the absence of microorganisms.

The results indicate the presence of gram negative bacteria in almost all stages of dry process in both regions. In the wet process, gram negative bacteria were identified in coffee cherries and stored beans in both sampling periods and regions. Gram negative bacteria were identified in dried beans in one of the samples in both regions (Table 3).

Table 3. Presence of gram negative bacteria in different stages of postharvest on farm coffee processing.

Types of postharvest on farm processing of coffee	Stages of coffee processing	Presence of gram negative bacteria in the sample from Shebedino woreda SNNPR		Presence of gram negative bacteria in the sample from Yebu woreda Oromia
		1st round	2nd round	1st round	2nd round
Dry process	Coffee fruits		×	×	×
	Dried beans	×	×	×	×
	Stored beans	×	×	×	×
Wet process	Coffee fruits	×	×	×	×
	Pulped beans			×	×
	Fermented beans
	Washed beans	×
	Dried beans		×	×
	Stored beans	×	×	×	×

“×” refers the presence of gram negative and the blank cells indicate the absence of gram negative.

Discussion

This study identified bacteria, yeast, and fungal filaments at different stages of on farm postharvest coffee processing. In the last stage, the stored beans, gram negative bacteria were present in both the wet and the dry process. This is in line with the high endotoxin levels found in primary coffee production in Tanzania, as the gram negative bacteria may produce endotoxin.

The range of colony count in our study (2.5 × 10² to 7.2 × 10⁵ CFU/mL) was narrower than in previous studies conducted in coffee cherries of Coffea arabica in Brazil (3.3 × 10⁴ to 2.2 × 10⁹ CFU/cherry).²⁸ This might be due to differences in the study period in which both rainy and dry seasons were included in the Brazilian study while only the dry season was considered in the present study. If the present study was conducted in the rainy season of the year the microbial load reported here could also have been higher. Silva et al. found a higher level of microbial load in wet season than the dry season.²⁸ Rain may create favorable condition for proliferation of bacteria.²⁸ In addition the difference could be due to variations in harvesting, transporting, drying, and storage condition.

However, the range of microbial load in the present study was broader than previous studies in Yemeni and Saudi Arabia green coffee (0.6 × 104 to 5.5 × 104 CFU) and (19 × 10 to 18 × 104 CFU), respectively.^32,33 The reasons for the difference in counts could be due to only dried beans were involved in previous studies while in this study in addition to dried beans, coffee fruits, fermented beans, and washed beans were included.

The presence of different microorganisms at different stages of wet and dry processing in our study is in line with previous studies.^34–36 Bruyn et al. indicated that even newly harvested cherries are not free from microorganisms as they contain gram negative bacteria, fungi, and soil microorganisms.³⁷ The type of microorganism present at different stage of coffee processing depends on several factors such as sugar concentration, water activity, availability of oxygen, temperature, acidity, and time.³⁸ The presence of these microorganisms in coffee processing are associated with diverse functionalities such as the degradation of pulp pectin and the depletion of mucilage carbohydrates.^37–39 In the wet method of coffee production, pulped, washed, and fermented beans were dominated by yeast and gram positive bacteria. This finding is in harmony with several previous studies.^15,35,36,40 This could be due to the anaerobic or low oxygen conditions created in wet fermentation that could facilitate the development of lactic acid bacteria which in turn cause the pH to drop, preventing the proliferation of other bacteria and favoring the growth of yeast.⁴¹

In this study, gram negative bacteria were identified in the final stages of dried and stored beans in both the wet and the dry method. This finding is consistent with Bruyn et al. who found relative abundances of gram negative bacteria over the course of drying.³⁷ However, in principle gram negative bacteria are not expected in the final stages of dried and stored beans due to the expected low moisture level at this last stage that do not favor microbial growth, particularly gram negative bacteria as they are less resistant to low moisture content.³⁶ This can be explained by the possibility of higher moisture content of the beans than recommended. The recommended moisture range of coffee beans is 11–12%. According to a study conducted in the southern part of Ethiopia more than 94.4% of the coffee farmers do not determine the exact moisture content of coffee beans for storage.⁴² Thus, a higher moisture content than recommended might be the reason for the observed gram negative bacteria development in the dried and stored beans in the present study.

A second reason could be related to environmental contamination. Generally, coffee beans should be stored in clean areas to prevent contamination of coffee beans.⁴³ However, a previous study conducted in Ethiopia indicated that 54% of coffee farmers did not have proper storage facilities.⁴² This might increase the contamination of coffee beans by gram negative bacteria. This assumption was supported by Balows et al. who found that 38.6% of the bacteria were actual or opportunistic pathogens from coffee beans.⁴⁴ Belay et al suggested that coffee beans should be stored in a place free from potential contaminants, such as cow dung, soils, and chickens.¹⁶

The presence of gram negative bacteria in dried beans and stored beans might not pose a risk to public health because of the high temperatures used during roasting. However, the presence of gram negative bacteria in dried and stored beans might affect the coffee workers respiratory health due to endotoxin exposure when the coffee beans are further processed. In our previous study in primary coffee processing factories located in the regions where we conducted this study indicated a higher prevalence respiratory symptoms and lower lung function among coffee workers than among controls.⁸ These findings might be associated with release of endotoxin from the gram negative bacteria on the coffee beans that have been contaminated during on farm coffee processing.

Thus, endotoxin exposure in coffee processing factories can be prevented by reducing contamination of coffee beans during drying and storage.

This study assessed the contamination of coffee at different stages of on farm postharvest processing of coffee, but it was not designed to identify potential sources of coffee contamination such as quality of harvesting materials, transporting mechanism, type and quality of drying surfaces, season of the year and quality of water used to wash the coffee. Therefore, future studies should consider all these factors to identify the possible source of microbial contamination of coffee, which is important to prevent and control contamination of coffee.

The present study considered only bacterial contamination on farm. However, coffee can also be contaminated during transport of coffee from the farm to the processing factories, or even in the coffee factories coffee can be contaminated during storage, handling, and processing. Therefore, future studies should consider all chains of coffee processing to get a even clearer picture of coffee contamination. Furthermore, this study did not assess respiratory health of coffee workers at the farm. Future studies should include these health aspects. The coffee samples were not blinded for the microbiologist analyzing them, he knows where they were coming from. However, in this situation, there was no specific hypothesis on which stage or type of coffee processing that was contaminated, and a blinding was not considered necessary.

Other limitations of this study were the sampling numbers which were low, with few sampling days. Repeated sampling was conducted only on two different days from different batches of coffee. However, this study collected samples from the two main coffee growing regions of Ethiopia, and the results are likely to be representative to Ethiopian on farm postharvest processing of coffee.

Conclusion

This is the first study to study microbial contamination of coffee during coffee production. This study found gram negative bacteria in the final stage of coffee bean production, in both a dry process and a wet process. As these bacteria might release endotoxin which may cause respiratory health problems among coffee production workers, an effort should be made to reduce the presence of these bacteria in the coffee production.

Competing interest

The authors declare that they have no competing interest.

Acknowledgments

We would like to thank the coffee preprocessing facilities management teams and Addis Ababa City Administration Health Bureau Laboratory for permission to conduct the study and allow us to use their laboratory.

Additional information

Funding

This research was funded by the Norwegian Agency for Development Cooperation (Norad) through the Norwegian Program for Capacity Building in Higher Education and Research for Development (NORHED) through a research project “Reduction of the burden of injuries and occupational exposures through capacity building in low income countries” (project number: 1300646-12).