Prevalence and correlates of Campylobacter 

and Salmonella species in small-scale 

broiler poultry, Gauteng, South Africa, 

2020-2021 

 

Shira Rebeka Amar 

 

 

A Research Report submitted to the Faculty of Health Sciences, 

University of the Witwatersrand, in partial fulfilment of the requirements 

for the degree of Master of Science (Epidemiology). 

Johannesburg, June 2022 



ii 

DECLARATION 

I, Shira Rebeka Amar, declare that this research report is my own original work. Where another 

person’s work has been used (either from a printed source, internet or any other source), it has 

been acknowledged and referenced in accordance with the guidelines of the Wits School of 

Public Health, University of the Witwatersrand. I have not used work previously produced by 

another student or any other person to hand in as my own. I have not allowed, and will not 

allow, anyone to copy my work with the intention of passing it off as his or her own work. This 

research has not been submitted previously for any other degree or examination at any other 

institution. 

 

Shira Rebeka Amar 

 

Signed: 

 

Date: 22/06/2022 

 

 

 

 

 

 

 

 

 

  



iii 

DEDICATION 

 

 

In loving memory of my mother, 

Ruth Lapinsky Amar 

(1960-2021) 

 

י“  ”צָמְאָה לְךָ נַפְשִׁ

Psalm 63:2 

 



iv 

ABSTRACT 

Background 

Campylobacter spp. and non-typhoidal Salmonella enterica (NT S. enterica) serovars are 

commonly isolated zoonotic bacteria associated with foodborne disease in humans. 

Colonisation of broilers with these pathogens compromises food safety for consumers of 

poultry products. Prevalence estimates of Campylobacter spp. and NT S. enterica serovars in 

local small-scale broiler production systems are sparse. Furthermore, biosecurity measures 

practiced in these production systems are not documented. This study aimed to estimate the 

prevalence of Campylobacter spp. and NT S. enterica serovars among small-scale broilers and 

examine the correlation of biosecurity practices with Campylobacter spp. positivity in Gauteng 

Province, South Africa. 

Methods 

Secondary cross-sectional data from a survey of 30 small-scale broiler farms was analysed. 

Farmers completed a farm management and biosecurity questionnaire. Cloacal swabs from 900 

broilers were cultured using standard methods for Campylobacter spp. and Salmonella. Isolates 

were identified using a real-time polymerase chain reaction test. Farm and bird-level 

prevalence was calculated. Biosecurity correlates of Campylobacter spp. were identified using 

univariable and multivariable logistic regression while accounting for clustering of broilers by 

farm. Potential confounders such as season, antibiotic use and age at slaughter were adjusted 

for.  

Results 

NT S. enterica serovars were not isolated (0% (0/900), 95% CI (0.0– 4.0%)). In contrast, 

Campylobacter spp. were detected in 49.8% of broilers (448/900, 95% CI (46.5–53.1%)) with 

an overall farm prevalence of 90% (27/30, 95% CI (73.4–97.9%)). Campylobacter coli was 

dominant (65.4%, 316/483) over C. jejuni (34.6%, 167/483). Among sampled broiler farms, 

53.3% (16/30) applied rodent control methods and 56.7% (17/30) used soap and disinfectant 

for poultry house hygiene. Exposures correlated to Campylobacter spp. positivity included 

open-walled naturally ventilated houses compared to free-range housing (OR= 3.82 95% CI 

(2.09 – 6.97)), drinking water treatment (OR=86.81 95% CI (4.56 – 1651.50)) and unsealed 

feed bags or feed spillage (OR= 12.48, 95% CI (2.88 – 54.12)). Conversely, the summer season 

(OR=0.06, 95% CI (0.02–0.20)) was protective.  



v 

Conclusion 

Campylobacter spp. were highly prevalent in sampled broilers while no NT S. enterica serovars 

were isolated. Appropriate drinking water treatment and correct feed management of high-

quality feed and improved hygiene of poultry houses emerged as important biosecurity and 

farm management practices that may reduce the odds of Campylobacter spp. positivity in 

small-scale production systems. These findings may be used to improve the health of broilers 

produced in small-scale systems and promote food safety for consumers in this market sector.  



vi 

ACKNOWLEDGEMENTS 

I am deeply grateful to the following people and organisations. 

My co-supervisors, Dr Sumaya Mall and Dr Liesl De Boni, for their support and mentorship. 

Thank you for meticulously reviewing my work and investing ‘significant’ time and energy 

into this project. 

Dr Zvifadzo Matsena Zingoni and Prof Jonathan Levin for their helpful advice on the statistical 

analysis of the data. 

I would like to thank Gauteng Veterinary Services (GVS) at the Gauteng Department of 

Agriculture and Rural Development for granting me access to their data and for the opportunity 

to assist in the development and upskilling of small-scale poultry farmers.  

I am grateful to Dr Peter Geertsma and all of my colleagues at GVS who welcomed me to the 

team and collected the data. Sincere thanks to Mr Gerbrand van der Zel for his contribution to 

the data collection. 

Thank you to the South African Field Epidemiology Training Programme at the National 

Institute for Communicable Diseases for this opportunity. Special thanks to Dr Alex de Voux 

and Ms Hetani Mdose for their valuable supervision. 

To my siblings, Rachel, and Gilad, my extended family and family-in-law, and friends who are 

like family, thank you for your immeasurable support in trying times. 

To my mother, Ruth, of blessed memory, thank you for providing me with every opportunity 

to excel and for being passionate about my interests always. “May Hashem lift up his face 

toward you and give you peace” (Numbers 6:26). 

Lastly, to my loving husband, Yoni. You have carried me through. When I count my blessings, 

I count you twice.  

 

 

 

 



vii 

CONTENTS 

 
DEDICATION ......................................................................................................................... iii 

ABSTRACT .............................................................................................................................. iv 

ACKNOWLEDGEMENTS ...................................................................................................... vi 

CONTENTS ............................................................................................................................. vii 

LIST OF TABLES ..................................................................................................................... x 

LIST OF SUPPLEMENTARY MATERIALS ......................................................................... xi 

LIST OF APPENDICES .......................................................................................................... xii 

NOMENCLATURE .............................................................................................................. xiii 

CHAPTER 1 - INTRODUCTION ............................................................................................. 1 

1.1 Background ...................................................................................................................... 2 

1.1.1 Foodborne campylobacteriosis and salmonellosis .................................................... 2 

1.1.2 Public health importance of campylobacteriosis and salmonellosis ......................... 6 

1.2 Literature Review ............................................................................................................. 7 

1.2.1 Literature search strategy ........................................................................................... 7 

1.2.2 Prevalence studies of Campylobacter spp. and non-typhoidal Salmonella enterica 

serovars in poultry and poultry products ............................................................................ 7 

1.2.3 Studies of Campylobacter spp. and non-typhoidal Salmonella enterica prevalence in 

poultry and poultry products limited to the African continent ......................................... 13 

1.2.4 Biosecurity and management practices associated with colonisation of poultry by 

Campylobacter spp. .......................................................................................................... 15 

1.2.5 Biosecurity and management practices associated with colonisation of poultry by 

non-typhoidal Salmonella enterica serovars .................................................................... 16 

1.3 Problem Statement ......................................................................................................... 17 

1.4 Justification .................................................................................................................... 18 

1.5 Research Question .......................................................................................................... 18 

1.6 Aim ................................................................................................................................. 18 



viii 

1.7 Objectives ....................................................................................................................... 19 

CHAPTER 2 - METHODS ...................................................................................................... 20 

2.1 The primary data ............................................................................................................ 21 

2.1.1 Data source .............................................................................................................. 21 

2.1.2 Setting ...................................................................................................................... 21 

2.1.3 Study population ...................................................................................................... 21 

2.1.4 Study sample............................................................................................................ 21 

2.1.5 Sampling strategy .................................................................................................... 21 

2.1.6 Data collection process ............................................................................................ 22 

2.2 Present study .................................................................................................................. 22 

2.2.1 Study design ............................................................................................................ 22 

2.2.2 Sampling strategy .................................................................................................... 22 

2.2.3 Precision and power calculation .............................................................................. 22 

2.3 Data management ........................................................................................................... 23 

2.4 Exposure variables ......................................................................................................... 24 

2.5 Outcome variable ........................................................................................................... 25 

2.6 Descriptive analysis........................................................................................................ 25 

2.6.1 Descriptive statistics of study sample ...................................................................... 25 

2.6.2 Prevalence estimates ................................................................................................ 25 

2.7 Inferential analysis ......................................................................................................... 26 

2.7.1 Univariable analysis ................................................................................................ 26 

2.7.2 Multivariable analysis.............................................................................................. 26 

2.8 Ethical considerations .................................................................................................... 26 

CHAPTER 3 – RESULTS ....................................................................................................... 28 

3.1 Characteristics of the study sample ................................................................................ 29 

3.1.2 Biosecurity and management practices of the study sample ................................... 30 

3.2 Campylobacter and non-typhoidal Salmonella enterica prevalence estimates .............. 31 



ix 

3.3 Logistic regression model .............................................................................................. 32 

3.3.1 Univariable correlates of Campylobacter colonisation ........................................... 32 

3.3.2 Multivariable correlates of Campylobacter colonisation ........................................ 34 

CHAPTER 4 – DISCUSSION, RECOMMENDATIONS AND CONCLUSION .................. 35 

4.1 Key study findings ......................................................................................................... 36 

4.2 Contextualisation of key findings .................................................................................. 36 

4.3 Limitations of the study.................................................................................................. 41 

4.4 Recommendations .......................................................................................................... 43 

4.5 Conclusion ...................................................................................................................... 45 

REFERENCES ........................................................................................................................ 46 

SUPPLEMENTARY MATERIAL .......................................................................................... 58 

APPENDICES ......................................................................................................................... 62 

Appendix A: Plagiarism declaration .................................................................................... 62 

Appendix B: Turnitin cover page ......................................................................................... 63 

Appendix C: Ethics waiver .................................................................................................. 64 

Appendix D: Letter of permission to use data...................................................................... 65 

Appendix E: Farmer consent form ....................................................................................... 66 

Appendix F: Poultry biosecurity and management questionnaire ....................................... 67 

  



x 

LIST OF TABLES 

CHAPTER 1 - INTRODUCTION 

Table 1.1: Campylobacter spp. prevalence estimates in poultry and poultry products from 

observational studies .................................................................................................................. 8 

Table 1.2: NT S. enterica prevalence estimates in poultry and poultry products from 

observational studies ................................................................................................................ 10 

 

CHAPTER 3 - RESULTS 

Table 3.1: Description of the study sample, Gauteng Province, 2020-2021…………………. 27 

Table 3.2: Biosecurity and farm management exposure variables and Campylobacter status of 

30 small-scale broiler farms, Gauteng Province, 2020-2021………………………………….28 

Table 3.3: Campylobacter spp. and non-typhoidal S. enterica prevalence estimates, Gauteng 

Province, 2020-2021…………………………………………………………...……….…….30 

Table 3.4: Univariable logistic regression findings of possible exposures correlated to 

Campylobacter spp. positivity, Gauteng Province, 2020 – 2021……………………………...30 

Table 3.5: Variables correlated with Campylobacter spp. positivity in the final multivariable 

model, Gauteng Province, 2020-2021………………………………………………………...32 

 

  



xi 

LIST OF SUPPLEMENTARY MATERIALS 

Supplementary A: Manipulations of exposure and outcome variables .............................. 587 

Supplementary B: Exposure variables of interest in the analysis ........................................ 59 

Supplementary C: Final logistic regression model equation ................................................ 60 

  



xii 

LIST OF APPENDICES 

Appendix A: Plagiarism declaration .................................................................................... 61 

Appendix B: Turnitin cover page ......................................................................................... 62 

Appendix C: Ethics waiver .................................................................................................. 63 

Appendix D: Letter of permission to use data...................................................................... 64 

Appendix E: Farmer consent form ....................................................................................... 65 

Appendix F: Poultry biosecurity and management questionnaire ....................................... 66 

  



xiii 

NOMENCLATURE 

All-in-all-out 

system 

A strategy used in animal rearing to minimise spread of infectious diseases. In 

this system, all animals are emptied from an area/premises and the area is 

cleaned and disinfected without any animals inside, before a new group of 

animals is introduced. Only animals of the same age group are reared at one 

time. 

Antimicrobial 

resistance 

The ability of a microbial pathogen to survive exposure to an antimicrobial 

agent that previously was able to destroy and control the microbe. 

Antimicrobial resistance arises through genetic mutations and transfer of 

genes. 

Biosecurity A set of management and physical measures that aim to reduce the risk of 

introduction, establishment and spread of animal diseases, infections or 

infestations, to, from and within an animal population.  

Broiler Chicken raised for the purpose of meat production. 

Caecum / caeca A pouch within the digestive tract: the blind-ending tube connected to large 

intestine in digestive tract of poultry. 

Carcass rinse 

fluids 

Fluids sampled after rinsing an animal carcass with water. 

Cloaca  A common cavity at the end of the digestive tract for the release of excretory 

and reproductive products in birds and reptiles. 

Colonise/ 

colonisation 

The presence of a microorganism on or in a host, with growth and 

multiplication of the organism, but without interaction between host and 

organism. 

Commensal/ 

commensalism 

An organism (such as a bacteria) in a biological relationship that benefits from 

the interaction while the other organism is not harmed nor receives any benefit. 

Contaminate/ 

contamination 

The presence of a microorganism on a body surface or an inanimate object. 

Flock A group of animals of one kind (referring to birds or sheep). 

Litter A mixture of material used for bedding in a poultry house (e.g., sunflower 

husks, wood shavings, spilled feed, and feathers/excretions from the poultry 

that were housed.) 

https://www.oie.int/fileadmin/Home/eng/Health_standards/tahc/2018/en_glossaire.htm#terme_risque


xiv 

Partial 

depopulation 

Early removal of a portion of birds from a poultry house while leaving some 

birds remaining in the house. 

Poultry All domesticated birds, including backyard poultry, used for the production of 

meat or eggs for consumption, for the production of other commercial 

products, for restocking supplies of game, or for breeding these categories of 

birds, as well as fighting cocks used for any purpose [1]. 

Prevalence The proportion of a population with a specific characteristic in a given time 

period. 

Salmonella 

enterica, non-

typhoidal (NT S. 

enterica) 

In this research report, this refers to serovars in the genus Salmonella, species 

enterica and subspecies enterica (I) that are associated with foodborne disease. 

Examples of these serovars are S. Enteritidis and S. Typhimurium. Non-

typhoidal refers to all serovars excluding Typhi, Paratyphi A, Paratyphi B, 

Paratyphi C, or Sendai [2]. 

Serovar A grouping of micro-organisms based on their cell-surface antigens. 

Small-scale 

farm 

A farm that produces livestock or agriculture in a small area of land without 

advanced equipment or technology. Small-scale farms generally have a lower 

output compared to the commercial sector.  

Zoonosis A disease of animals that may be transmitted to humans. 

 



1 

CHAPTER 1 - INTRODUCTION 

This chapter presents an overview of foodborne disease related to Campylobacter spp. and non-

typhoidal Salmonella enterica (NT S. enterica) serovars. The chapter focuses on the disease 

epidemiology in developing countries. The public health impact of foodborne disease is 

discussed. A critical review of the literature, synthesising studies of the prevalence of 

Campylobacter spp. and NT S. enterica serovars in poultry and poultry products worldwide; 

and associated farm biosecurity and management factors is then presented. This is followed by 

the aim and objectives of the research study. 

  

 

The chapter headings are:  

1.1 Background  

1.1.1 Foodborne campylobacteriosis and salmonellosis 

1.1.2 Public health importance of campylobacteriosis and salmonellosis 

1.2 Literature Review  

1.2.1 Literature search strategy 

1.2.2 Prevalence studies of Campylobacter spp. and non-typhoidal Salmonella 

enterica serovars in poultry and poultry products 

1.2.3 Studies of Campylobacter spp. and non-typhoidal Salmonella enterica 

prevalence in poultry and poultry products limited to the African continent 

1.2.4 Biosecurity and management practices associated with colonisation of poultry 

by Campylobacter spp. 

1.2.5 Biosecurity and management practices associated with colonisation of poultry 

by non-typhoidal Salmonella enterica serovars 

1.3 Problem Statement 

1.4 Justification 

1.5 Research Question 

1.6 Aim 

1.7 Objectives  

 
 
 



2 

1.1 Background  

1.1.1 Foodborne campylobacteriosis and salmonellosis  

 

Foodborne diseases including campylobacteriosis and salmonellosis compromise food safety 

and contribute to morbidity and mortality globally and in Africa [3–7]. The World Health 

Organisation (WHO) estimated that 582 million cases of foodborne illness occurred worldwide 

in 2010, resulting in 25.2 million disability adjusted life years (DALYs) lost [3]. Despite a high 

burden of disease, a relatively low estimate of worldwide mortality due to foodborne disease 

was reported by the WHO, at 5 deaths per 100 000 population [3]. The public health threat 

posed by foodborne disease is not uniform in all nations as regional mortality rates vary widely 

[3]. Africa had a disproportionately high death rate due to foodborne disease (14 deaths per 

100 000 population), compared to a low of 0.5 deaths per 100 000 population in Europe [3].  

Second only to viral causes, bacterial pathogens predominate in foodborne disease cases 

worldwide [5,8]. Campylobacter spp. and non-typhoidal Salmonella enterica (NT S. enterica) 

serovars are among the most commonly isolated zoonotic bacteria associated with foodborne 

disease [3,5,6]. In 2010, 95.6 million campylobacteriosis cases and 78.4 million non-typhoidal 

salmonellosis cases occurred globally [3]. These ubiquitous pathogens are associated with 

substantial morbidity [3,5]. It is estimated that foodborne Campylobacter spp. and NT S. 

enterica serovars resulted in a combined 4.3 million DALYs lost globally in 2010 [3]. 

Foodborne diseases may also cause society to  incur considerable economic losses [9,10]. The 

average cost per foodborne disease case in the United States was estimated at $2 422, while 

total annual costs specific to Campylobacter spp. and non-typhoidal Salmonella amounted to 

$7.9 million and $6.5 million respectively [9].  

Aetiology and epidemiology  

Campylobacter spp. are gram-negative bacteria in the genus Campylobacter and family 

Campylobacteraceae. Several Campylobacter spp. may cause disease in humans, animals or 

both. Campylobacter jejuni and Campylobacter coli are commonly implicated species in 

foodborne disease cases [11–13] and C. jejuni is the predominant species in most geographic 

areas [12]. In 2017, 84.4% of Campylobacter cases with species information in the European 

Union (EU) were identified as C. jejuni and 9.2% were reported as C. coli cases [4]. Although 

the clinical signs caused by either species are often indistinguishable, C. coli may cause up to 



3 

25% of all campylobacteriosis infections [12]. Transmission of Campylobacter spp. from 

environmental sources or through horizontal transmission results in colonisation of the caecum, 

cloaca, large intestine and jejunum [14]. Once colonised, broilers act as a bacterial reservoir, 

contaminate the rearing environment and transmit Campylobacter spp. horizontally throughout 

the flock. Poultry carcasses may become contaminated with intestinal microflora during the 

slaughter process [14]. Although Campylobacter spp. will not proliferate outside of the 

digestive tract of warm-blooded animals, these microbes survive for several weeks in food 

products, especially if stored at low temperatures [11,12]. In contrast to NT S. enterica 

serovars, vertical transmission through eggs is not relevant to the epidemiology of the disease 

[15]. 

Salmonella enterica subspecies enterica is an intracellular pathogen of warm-blooded 

mammals and comprises more than 2000 serovars [16]. These gram-negative bacilli of the 

genus Salmonella in the family Enterobacteriaceae are important contributors to the 

worldwide foodborne disease burden [17]. S. enterica serovars may be classified as typhoidal 

or non-typhoidal. Typhoidal serovars are specialised pathogens that are restricted to human 

hosts and have no animal reservoir. Human infection results in systemic disease known as 

enteric fever or typhoid, mainly in developing countries [16]. Non-typhoidal serovars occur 

worldwide and are generalised pathogens with a broader host range of animals and humans 

[16]. Of the NT S. enterica serovars, S. Enteritidis (SE) and S. Typhimurium are the most 

frequently reported serotypes associated with foodborne disease [18]. Foodborne disease 

caused by NT S. enterica serovars results from consumption of contaminated animal-derived 

products, most notably eggs, fresh produce or poultry products. Infection may also occur from 

direct contact with an infected animal or more rarely from person-to-person transmission 

[17,19]. Salmonella serovars that are specific to poultry (S. Gallinarum and S. Pullorum) are 

not included in the scope of this research. 

Clinical symptoms  

Individuals affected by foodborne campylobacteriosis or non-typhoidal salmonellosis present 

with gastroenteritis of varying severity [11,25,26]. Campylobacteriosis symptoms include 

fever, abdominal cramps, vomiting, dehydration and diarrhoea which may be bloody. Usually 

the disease will resolve after five to seven days but there is risk of severe complications 

including septicaemia, peripheral neuropathies, arthritis and cardiovascular abnormalities 

[11,12,27]. Guillain-Barré syndrome is associated with C. jejuni infection specifically [12,13]. 



4 

Foodborne salmonellosis presents clinically as diarrhoea, abdominal cramps and fever. In some 

cases it may progress to bacterial meningitis, reactive arthritis or endocarditis [11,21]. More 

severe or fatal cases of campylobacteriosis and salmonellosis may occur in children under five 

years of age and in immune-compromised or elderly individuals [21,26,27]. 

Prevalence, distribution and risk factors 

Prevalence data for human campylobacteriosis in developing countries is limited, in part by the 

lack of dedicated surveillance systems [13,28], under-reporting and non-specific disease 

symptoms [29]. A number of longitudinal studies have been conducted. A Malawian study 

conducted over a ten-year period detected Campylobacter spp. in 21% of children hospitalised 

with diarrhoea [30]. Similarly, in Kenya, a cohort study isolated Campylobacter spp. from 5% 

of children who died in hospital due to diarrhoea. Notably, NT S. enterica was isolated from 

20% of the mortalities [31]. Prevalence of Campylobacter species in children aged less than 

five years in Ethiopia was estimated at 10% using meta-analysis techniques [32]. These African 

studies suggest that Campylobacter is commonly detected in children suffering from diarrhoea 

in multiple African states [30–32]. 

The global and regional burden of non-typhoidal salmonellosis has been investigated by 

multiple researchers using various methods [33–36]. A recent systematic review synthesised 

data from 35 studies and estimated the global incidence of non-typhoidal salmonellosis in 2017 

at 7.5 cases per 100 000 person-years. Sub-Saharan Africa had the highest incidence of all 

regions at 34.5 cases per 100 000 person-years [35]. It is estimated that 93.8 million 

gastrointestinal illnesses (of which 80.3 million are foodborne) caused by NT S. enterica 

serovars occur annually [34]. African incidence estimates ranged from 1.4 to 2520 cases per 

100 000 population per year, with higher incidence rates occurring in children and in 

inhabitants of rural areas [36]. 

Risk factors for campylobacteriosis and salmonellosis may vary in high versus low-to-middle 

income countries [37–42]. In high-income countries, important risk factors for salmonellosis 

include eating raw or undercooked poultry meat and eggs [41]. In contrast, in developing 

countries, using pit latrines or rivers and canals as toilets and drinking water from a non-piped 

source are associated with disease [39]. Case-control studies in the Netherlands and Tanzania 

demonstrated that foodborne transmission involving consumption of poultry and undercooked 

meat were the greatest risk factors for campylobacteriosis [38,40,42]. In a systematic review 

and meta-analysis carried out using eight observational studies conducted in Ethiopia, 



5 

consumption of animal products, illiterate mothers and the status of mothers’ personal hygiene 

were significantly associated with the prevalence of Campylobacter spp. in children aged under 

five [32]. Contact with farm animals [38] and domestic animals [32] may also increase the risk 

of campylobacteriosis. Risk factors for foodborne disease are diverse and highlight that animals 

and animal products are important sources of disease, regardless of socioeconomic conditions. 

Bacterial colonisation of poultry 

Campylobacter spp. are commensal organisms in poultry. Therefore, birds that are colonised 

with the bacteria are clinically healthy and are asymptomatic [15,20].  Isolated studies have 

suggested that young poultry birds with intestinal bacterial colonisation may experience 

diarrhoea and weight loss with mild gastrointestinal pathological lesions [15]. Similarly, 

poultry that are colonised with generalist NT S. enterica serovars (this excludes the poultry-

specific agents S. Gallinarum and S. Pullorum) are sub-clinically infected since these microbes 

are commensals in poultry [21]. There is no evidence that production at the farm-level is 

reduced by colonisation of poultry with generalist NT S. enterica serovars or Campylobacter 

spp. The poultry farm presents a critical control point for the reduction of human foodborne 

campylobacteriosis and salmonellosis. Poultry farmers may be able to reduce the incidence of 

human disease by limiting colonisation of poultry at the farm stage of the poultry production 

chain [22].  

Disease surveillance in small-scale broilers 

Many small-scale broiler farmers do not produce for export purposes, for which regular 

Salmonella monitoring is a requirement [23]. These producers may also bypass the use of 

registered abattoirs where meat inspection is conducted [24]. Therefore, there is limited 

justification to maintain broilers free of NT S. enterica and Campylobacter spp. in small-scale 

production systems that cater to informal local markets. It is challenging to incentivize broiler 

farmers to control NT S. enterica serovars and Campylobacter spp. as bacterial colonisation 

does not directly reduce production [15] or profit. To promote food safety in small-scale 

production systems, the production of broilers free of NT S. enterica serovars and 

Campylobacter spp. should enable improved access to high-quality markets and an associated 

increased profitability.  

 

  



6 

1.1.2 Public health importance of campylobacteriosis and salmonellosis 

 

Campylobacter spp. and NT S. enterica serovars are established causes of zoonotic foodborne 

disease and pose a risk to human health worldwide [3,4,11,25]. The WHO’s 2010 data 

synthesis on the global and regional disease burden of foodborne diseases established that the 

rate of DALYs for foodborne diseases was highest in the African region [3]. Of the 22 

foodborne diseases considered, those caused by invasive and non-invasive NT S. enterica 

serovars resulted in the greatest (4.07 million) foodborne DALYs in 2010. Campylobacter 

pathogens caused 2.1 million foodborne DALYs in 2010 and 1390 illnesses per 100 000 person 

years globally [3]. The impact of these pathogens was reiterated in a similar study focused on 

the high-income region of the United States. Both NT S. enterica serovars and Campylobacter 

spp. resulted in considerable DALYs lost annually (32900 and 22500 DALYs respectively) 

[5]. Moreover, vulnerable populations such as children aged five and under are more likely to 

experience severe disease caused by foodborne pathogens and are disproportionately affected 

by foodborne disease, with 38% of an estimated 582 million foodborne illnesses occurring in 

children under five years [3]. Economic fallout resulting from campylobacteriosis and 

salmonellosis may be substantial. It is estimated that the food-pathogen pairs that incurred the 

highest costs in the United States were Campylobacter spp. in chicken ($6.9 billion) and 

Salmonella serovars in poultry ($2.8 billion) [9]. Morbidity and mortality due to 

campylobacteriosis and salmonellosis is associated with social and economic losses and may 

burden the healthcare sector [3,11]. 

Further negative implications of foodborne disease relate to antimicrobial drug resistance 

(AMR) [43,44]. The routine use of fluoroquinolones and other antibiotics in poultry farms has 

been associated with increased AMR of Campylobacter spp. and NT S. enterica serovars in 

both humans and animals [13,21,45–47]. Campylobacter spp. are documented to continuously 

develop new mechanisms for AMR [48,49]. Some of these mechanisms lead to resistance to 

single drugs but multi-drug resistance has also developed and cases of drug-resistant human 

campylobacteriosis are an emerging food safety risk [48,50]. Effective treatment of 

salmonellosis in humans is limited by resistance of the pathogen to certain antimicrobials 

[51,52]. In a study in the United States, 2.8% of NT S. enterica isolates were resistant to 

clinically important fluoroquinolones and 2.5% of isolates were resistant to a third-generation 

cephalosporin [51]. Overall, in the United States and Europe there is a low to moderate level 



7 

of antimicrobial resistance in NT S. enterica isolates however certain serovars have 

demonstrated an increasing trend of single or multi-drug resistance [51,52]. 

1.2 Literature Review  

1.2.1 Literature search strategy 

PubMed and Google Scholar databases were searched on a regular basis between July 2020 

and October 2021. Material discussing foodborne disease caused by Campylobacter species 

and NT S. enterica serovars, prevalence studies of these bacteria in poultry or animal-derived 

food products were retained. Publications that studied risk factors or associations with these 

pathogens in food-producing animals or on farms were also extracted. Material was restricted 

to English publications. Search terms that were used included: (‘Campylobacter’ OR 

‘Campylobacteriosis’ OR ‘Campylobacter jejuni’ OR ‘Campylobacter coli’) OR (‘Salmonella’ 

OR ‘Salmonellosis’ OR ‘S. enterica’ OR ‘Salmonella Enteritidis’ OR ‘non-typhoidal 

Salmonella) AND (‘prevalence’ OR ‘burden’ OR ‘presence’ OR ‘risk factor’ OR ‘factor’ OR 

‘correlates’ OR ‘incidence’ OR ‘association’) AND/OR (‘poultry’ OR ‘broiler’ OR ‘chickens’ 

OR ‘poultry products’ OR ‘chicken meat’ OR ‘food-producing’ OR ‘farm’). Different 

permutations of these terms were used. The abstracts were used to select literature that would 

be relevant to the current study. The full text papers were downloaded and managed using 

Mendeley reference software. Studies that were included in the literature review were in most 

cases observational studies, meta-analyses or systematic reviews about the prevalence of 

Campylobacter spp. or NT S. enterica serovars in poultry or poultry products and factors 

associated with the presence or absence of disease.  

1.2.2 Prevalence studies of Campylobacter spp. and non-typhoidal Salmonella 

enterica serovars in poultry and poultry products 

A number of studies estimating the prevalence of both Campylobacter spp. and NT S. enterica 

serovars in food producing animals and animal products have been conducted in countries 

across all income levels [4,47,53–72]. The studies were predominantly conducted in the large-

scale, commercial poultry sector with knowledge gaps concerning the small-scale poultry 

sector [56–60,71,72]. The prevalence studies were conducted between 2001 and 2020 with 

small (n=7) and larger (n=2835) sample sizes [61,62]. Poultry at different stages of the 

production chain including at the farm-level, abattoir, processing plant and in the retail, 

environment were sampled. Since prevalence estimates have been found to vary at these 



8 

different stages [73,74], it is difficult to compare the prevalence findings for Campylobacter 

and NT S. enterica serovars. Additionally, these prevalence studies utilised a variety of sample 

types including cloacal and caecal contents, carcass swabs and washes, duodenal samples, 

faeces or litter and environmental samples. Isolation of Campylobacter spp. was not equally 

successful for all types of samples. For example, Campylobacter spp. was detected from a 

greater proportion of broilers using laboratory testing of cloacal swabs in charcoal Amies 

medium compared to testing of caecal contents collected during necropsy [75]. Importantly, 

multiple studies have demonstrated that PCR methods to screen and confirm Campylobacter 

spp. from a variety of samples are superior to culture techniques [76–78]. Therefore, prevalence 

studies that relied only on traditional culture methods may present under-estimates. Due to the 

varying sample types and diagnostic tests used, it has been challenging to estimate a global 

prevalence statistic of either pathogen in broiler populations. 

Broadly, Campylobacter spp. are highly prevalent in broiler populations in developing and 

developed nations (see Table 1.1, tabulated in order of descending prevalence). A combination 

of bacterial culture and polymerase chain reaction testing of gastrointestinal samples including 

caecal contents and cloacal swabs detected a prevalence range of 5.9%-100% [55,56,58–

60,64,65,71,76,79]. Prevalence findings from faecal samples had a more narrow range of 7%-

78% [57,61,62,64,72,80]. Despite these findings, it is not known if small-scale production 

systems have comparable Campylobacter spp. burdens. 

Table 1.1: Campylobacter spp. prevalence estimates in poultry and poultry products from observational 

studies 

Country Year Sample Sample size Prevalence Reference 

Brazil 2010 Farm litter 40  100%, (40/40) [71] 

Pooled caecal 

content 

40 100%, (40/40) 

Carcasses 40 with faecal 

contamination 

17.6%, (7/40) (C. 

jejuni) 

40 without faecal 

contamination 

58.8%, (23.5/40) (C. 

jejuni) 

France 2008 Pooled caeca and 

carcasses 

425 carcasses 87.5%, (372/425) [68] 

425 pooled caecal 

samples 

77.2%, (328/425) 

South 

Africa  

2016 Faecal samples 100 78%, (78/100) [80] 

Iran  2009 Caecal swabs 100 76%, (76/100) [60] 

Belgium/  

Denmark 

2002-

2004 

Gastrointestinal 

tract 

56 broiler flocks 73%, (41/56)  [69] 

Morocco 2017 Cloacal swabs 105 71.4%, (75/105) [79] 

South 

Africa 

2008-

2009 

Caecal contents 56 rural chickens 68%, (38/56) [55] 

140 commercial 

free-range chickens 

47.1%, (66/140) 



9 

Country Year Sample Sample size Prevalence Reference 

133 industrial 

broilers 

47.4%, (63/133) 

Lebanon 2016 Caecal contents 227 67%, (152/227) [65] 

Neck skin 227 17.2%, (39/227) 

Kenya 

(Kibera) 

 

2018 Chicken meat 

(retail) 

28 64%, (18/28) [61] 

Faeces/litter 7 43%, (3/7) 

Kenya 

(Dagoretti) 

2018 Chicken meat 

(retail) 

25 60%, (15/25) [61] 

Faeces/litter 18 33%, (6/18) 

Thailand 2012-

2014 

Caecal samples 442 57%, (252/442) [56] 

New 

Zealand 

2007 Whole carcasses 

(skin) 

163 carcasses 44.8%, (73/163) [70] 

Bangladesh 2019 Cloacal swabs 84 pooled cloacal 

swabs from 84 

broiler farms  

40.5%, (34/84) [59] 

Spain 2010-

2012 

Cloacal swabs  2221 38.1%, (846/2221) [58] 

South 

Africa 

2003 Fresh and frozen 

chicken carcasses  

99 carcasses 32.3%, (32/99) [54] 

United 

States of 

America 

2010-

2011 

Faecal samples 400 29.5%, (118/200) [72] 

Australia 2016 Faecal and 

environmental 

samples 

1856 28.3%, (526/1856) [57] 

Denmark 2009-

2010 

Faecal samples 2835 flocks 14%, (388/2835) [62] 

Iceland  2016-

2018 

Faecal and caecal 

samples 

857 flocks 7%, (41/857) [64] 

South 

Africa 

Unknown Cloacal swabs 408  5.9%, (24/408) [76] 

 

Campylobacter spp. have been isolated from numerous food-producing animals and animal-

derived food products [63,67]. A recent meta-analysis of observational studies conducted 

between 1980 and 2019 in predominantly high-income countries in Europe and North America 

analysed the global pooled prevalence of Campylobacter spp. in food products of animal origin 

[67]. Poultry meat was found to be the food source primarily responsible for human exposure 

to the pathogen. The highest prevalence of Campylobacter spp. in sampled livestock (including 

cattle, pigs, broilers, hens, goats and sheep) was detected in broilers. Broiler carcasses and 

livers had a prevalence of 52.3% and 65.5% respectively. The meta-analysis suggested that the 

highest prevalence (14.1%) of C. coli was identified in broilers and that prevalence of C. jejuni 

in broiler meat was 33.7% (95% CI 30.7%-36.8%) [67]. Importantly, studies that were included 

utilised varying diagnostic methods (morphological identification, biochemical or PCR testing) 

which may differ in sensitivity and bias the results. 



10 

Observational studies that estimated the prevalence of NT S. enterica serovars in broiler poultry 

have generally isolated NT S. enterica at a lower level than Campylobacter spp. (see Table 1.1 

and Table 1.2, tabulated in order of descending prevalence). Developing countries had a high 

burden of NT S. enterica-colonised or contaminated broilers, (38.6% prevalence in Senegal 

[81] and 42% prevalence in Bangladesh [82]) but, there is evidence that some high-income 

countries have comparably high prevalence rates (22% prevalence in Canada [83] and 13% 

prevalence in a combined survey of Belgium and Denmark [69]). NT S. enterica prevalence 

estimates ranged between 0%-42% and were determined from a range of samples (including 

cloacal swabs, carcasses and litter) (see Table 1.2) using diverse diagnostic methods [82,84]. 

Gaps regarding NT S. enterica prevalence at the on-farm rearing stage and specifically in small-

scale production systems remain. 

Table 1.2: NT S. enterica prevalence estimates in poultry and poultry products from observational studies  

Country Year Sample Sample size Prevalence Reference 

Bangladesh 2016 Cloacal swabs 60 42%, (25/60) [82] 

Senegal Unknown Pooled faecal samples 57 35.1%, (20/57) [81] 

Chicken carcass skin 57 38.6%, 22/56 

Chicken carcass muscle 57 29.8%, 17/56 

Chicken meat (restaurant)  42 14.3%, (6/56) 

Canada 2012-

2013 

Chicken carcasses and 

breast/thigh pieces 

2732 22%, 

(597/2732) 

[83] 

South Africa Unknown  Fresh and frozen chicken 

carcasses  

99  19.2 %, 

(19/99) 

[54] 

Nigeria 2019 Faecal and environmental 

samples 

558 15.9%, 

(89/558) 

[85] 

Trinidad Unknown Carcass swabs and carcass 

rinse fluids 

450 14.2%, 

(64/450) 

[86] 

Belgium/ 

Denmark 

2002-

2004 

Gastrointestinal tract 56 broiler 

flocks 

13%, (7/56)  [69] 

Iran 2018 Faecal samples 110 11,8%, 

(13/110) 

[87] 

Zimbabwe 2003-

2005 

Pooled cloacal swabs 2833 10%, 

(283/2833) 

[88] 

Nepal 2019 Faecal and environmental 

samples 

288 9%, (26/288) [89] 

United States of 

America 

2010-

2011 

Faecal samples 400 8.8%, (35/400) [72] 

France 2008 Pooled caeca and carcasses 425 batches 7.5%, (32/425) [68] 

Brazil 2010 Farm litter 40  5%, (2/40) [71] 

Pooled caecal content 40 0%, (0/40) 

Carcasses 120 0%, (0/120) 

Ethiopia 2013-

2014 

Pooled fresh faecal droppings 549 4.7%, (26/549) [90] 

South Korea 2013 Raw poultry meat 80 3.7%, (3/80) [91] 

Ethiopia 2017-

2018 

Cloacal swabs, faecal 

droppings, feed samples and 

floor swabs 

836 2.9%, (24/836)  [84] 

New Zealand 2007 Whole carcasses 163 0%, (0/163) [70] 

 



11 

European prevalence estimates for Campylobacter spp. and NT S. enterica serovars in animals 

and animal products have been reported by multiple researchers [4,68,69,92]. The European 

Food Safety Authority (EFSA) reviewed data from European Union-coordinated zoonoses 

monitoring systems to determine a 37.4% Campylobacter prevalence in fresh broiler meat and 

12.3% prevalence in live broilers in 2017. At flock level, the European Union had a 3.31% 

Salmonella prevalence [4]. The EFSA findings present an overall picture of the burden of 

disease on the continent. Cross-sectional studies limited to a single European state offer higher 

prevalence estimates for both pathogens. French and Danish research produced comparable 

prevalence results for Campylobacter spp. of 77.2%-87.5% prevalence depending on the type 

of sample in France [68] and 73% flock-level prevalence in Denmark [69]. NT S. enterica 

prevalence findings were 7.5% and 13% in France and Denmark respectively [68,69]. The 

Danish study employed multi-sampling techniques (duodenum, crop and caecum) in contrast 

to other prevalence studies [69]. The researchers emphasised that this multi-sampling technique 

was necessary to detect NT S. enterica and may be related to the slightly higher NT S. enterica 

prevalence estimate compared to others in Europe [4,69]. Sampling strategies such as using 

slaughter batches as the sampling unit and pooling samples for analysis [68] may have led to 

an over-estimation of the true bacterial prevalence. The available research indicates that 

Campylobacter spp. are present at a greater prevalence than NT S. enterica serovars in 

European broiler populations. 

The selection of sample type or the stage of broiler production at which sampling occurs may 

affect prevalence findings in observational studies. Higher Campylobacter spp. prevalence 

estimates were found based on whole carcass sampling (87.5%) versus caecal sampling 

(prevalence of 77.2%) [68]. This may illustrate that contamination of the skin of the carcass 

occurs in the slaughterhouse. Evisceration of carcasses has been shown to contaminate 

neighbouring carcasses and can contaminate the following slaughter batches which may have 

entered the slaughterhouse disease-free [46,68]. The slaughter process itself, rather than true 

colonisation of the digestive tract of poultry may cause an apparent increased prevalence of 

Campylobacter in samples tested at the point of slaughter. Conflicting results concerning the 

effect of faecal contamination on prevalence estimates were reported from Brazil [71]. Caecal 

contents had a prevalence of 100% and carcasses without obvious faecal contamination had a 

higher Campylobacter jejuni prevalence (58.8%) than those with obvious faecal contamination 

(17.6%). In contrast, prevalence estimates for Campylobacter coli were lower and suggested 

that faecal contamination increased the bacterial burden: 70.6% based on caecal samples, 



12 

11.6% in carcasses with macroscopic faecal contamination and 9.8% in carcasses without 

visible faecal matter [71]. These findings demonstrate the importance of the slaughter process 

in determining the risk of exposure to the consumer. These existing data also point to the gaps 

in knowledge of pre-slaughter bacterial prevalence and on-farm bacterial loads. 

Investigation of the prevalence of Campylobacter spp. and NT S. enterica serovars in chicken 

products in the retail or restaurant chain has established that chicken products pose a risk to 

food safety. Heavy burdens of Campylobacter spp. were detected in Kenyan (prevalence of 60-

64%) [61], Thai (prevalence of 56.8%) [93] and Australian (prevalence of 84%) [94] chicken 

meat and offal products. A high proportion (70.7% and 44.8% respectively) of raw chicken 

samples in the United States [95] and New Zealand [70] were also contaminated with 

Campylobacter spp. These prevalence findings do not necessarily measure colonisation with 

Campylobacter spp. Rather, the results are probably an indication of contamination of the 

carcass which may occur at slaughter, during processing or in a retail setting. Similar to 

prevalence studies conducted at the abattoir or processing plant, NT S. enterica serovars were 

detected at a lower prevalence than Campylobacter spp. in retail or restaurant chicken products. 

Chicken served at street restaurants in Senegal had a NT S. enterica prevalence of 14.3% [81] 

and NT S. enterica was not detected in 163 retail chickens in a New Zealand based study [70]. 

Research focused on the retail stage of production fails to account for possible on-farm sources 

of bacteria. 

The bacterial burden of broiler poultry may fluctuate depending on the stage of production. A 

recent systematic review and random-effects meta-analysis investigated the prevalence of 

Campylobacter and NT S. enterica in the farm environment, processing and retail stages in the 

United States [74]. Campylobacter prevalence was estimated at 59.2% at the retail level. The 

highest prevalence of Campylobacter was found during processing (60.9%-97.9%) and 

environmental samples at farm level had the lowest prevalence at 15.8%. NT S. enterica 

prevalence in these three stages (farm environment, processing and retail) followed a different 

pattern with the farm environment at a higher prevalence (22.9%) than the retail stage (19%). 

Processing-level prevalence estimates ranged from 14.3%-68.6% [74]. The meta-analysis was 

limited since some study groups included a low number (n=2) of studies in the analysis. 

Although there was high heterogeneity between the study groups, the meta-analysis model was 

fitted with multiple moderating variables to account for this concern. 



13 

1.2.3 Studies of Campylobacter spp. and non-typhoidal Salmonella enterica 

prevalence in poultry and poultry products limited to the African continent 

Animal reservoirs of Campylobacter spp. or NT S. enterica serovars are recognized to be 

highly prevalent in the African region. A random effects meta-analysis published in 2019 

addressed the prevalence of both Campylobacter spp. and NT S. enterica serovars in food-

producing animals and food products in 27 African countries [66]. The research indicates that 

Campylobacter could be isolated from 37.7% (95% CI 31.6–44.3) of all poultry samples. 

Isolation of Campylobacter was highest (73.8%) from external poultry samples (defined as 

skin or feathers) and lowest based on poultry meat or organs (21.3%). Gastrointestinal samples 

yielded a prevalence of 40.2%. Regionally, Central Africa had the highest prevalence (91.2%) 

while Northern Africa had a prevalence of only 24.2%. Contamination of the external carcass 

by Campylobacter spp. from other animals or faecal matter during slaughter or possibly pre-

slaughter can be reinforced by the meta-analysis’ findings of higher prevalence from external 

samples as opposed to all other sampling areas. NT S. enterica serovars were detected in 13.9% 

(95% CI 11.7–16.4) of poultry samples. Similarly, external poultry samples had the highest 

prevalence of NT S. enterica (28.5%) and poultry meat or organs had the lowest prevalence 

(13.2%). The prevalence in gastrointestinal samples was similar at 13.4%. NT S. enterica 

prevalence was higher in Southern Africa than in Northern Africa with prevalence values of 

28.2% and 10.5% respectively [66]. The meta-analysis included 247 studies, 25 of which were 

carried out before 2000 and four of these were published nearly 70 years ago. However, the 

remaining 222 studies took place between 2000 and 2016. The studies that were analysed 

employed various laboratory techniques and some diagnostic procedures were not fully 

explained. Therefore, comparability of study findings is limited. While Campylobacter spp. 

and NT S. enterica serovars were detected throughout the continent, this study emphasises that 

these pathogens are not equally prevalent in all African countries. 

A comprehensive review of studies estimating the prevalence of Campylobacter spp. in food-

producing animals and in humans by African region and country also illustrated the variability 

of Campylobacter prevalence by geographic location [63]. This review did not include NT S. 

enterica prevalence estimates nor did the authors undertake meta-analysis. The studies 

reviewed collected various types of poultry samples including fresh and frozen chicken from 

the retail environment and carcasses at point of slaughter [63]. African estimates of 

Campylobacter prevalence in poultry products range from 9.6% (n=680) in Egypt to 90% 



14 

(n=150) in Cameroon [63]. Though all of the studies included in the review used culture 

techniques for isolation of Campylobacter spp., prevalence estimates varied widely [63]. 

South African studies that focus explicitly on Campylobacter spp. or NT S. enterica bacterial 

isolation from poultry or poultry products are limited. A local study to examine the incidence 

of NT S. enterica serovars in food-producing animals, animal feed and environmental sources 

isolated NT S. enterica serovars in 5% (9031/180298) of samples [53]. SE was the most 

commonly isolated serovar – with 21.5% of the isolated NT S. enterica serovars detected from 

poultry on farms, poultry meat and poultry houses [53]. In an observational study limited to 

poultry carcasses in Gauteng Province, Campylobacter was isolated from carcass rinse fluids 

in 32.3% of sampled fresh and frozen retail chickens and NT S. enterica was detected in 19.2% 

of samples. Importantly, this prevalence study may have been limited by the small sample size 

(99 samples) included in the analysis [54]. This study was conducted at the retail level and did 

not address the prevalence of these pathogens at farm-level or at the abattoir specifically.  

Data concerning bacterial prevalence of Campylobacter spp. in South Africa’s commercial 

poultry sector is sparse but more available than that of the small-scale poultry sector. A cross-

sectional study conducted in KwaZulu-Natal Province, South Africa compared the prevalence 

of Campylobacter spp. among chickens in different production systems (rural production, 

commercial free-range systems and industrial broiler production) [55]. The findings indicated 

that chickens reared in a rural setup had the highest prevalence of Campylobacter (68%) while 

commercial free-range chickens and industrial broilers had lower and equal prevalence 

estimates of 47% [55]. Though this study was limited by a small sample size (56 rurally-

produced chickens, 140 samples from the commercial free range farm and 133 samples from 

industrial broilers) it offers a unique perspective on the effect of differing production systems 

and related farm management practices on the prevalence of the pathogen in broilers [55]. No 

similar study that contrasts production systems could be identified for comparison. Discordant 

prevalence estimates of Campylobacter spp. in commercial chickens have been estimated in 

KwaZulu-Natal Province (78% based on 100 faecal swabs) [96] and North West Province 

(5.9% from 408 cloacal swabs) [76]. The available research indicates that South African 

poultry and poultry products are a reservoir of Campylobacter spp. and NT S. enterica serovars 

and act as a human disease exposure risk [53–55,76,96]. Prevalence data specific to the small-

scale poultry industry in South Africa is lacking.  



15 

1.2.4 Biosecurity and management practices associated with colonisation of 

poultry by Campylobacter spp. 

In principle, the risk of bacterial contamination of poultry carcasses during later stages of 

processing may be reduced by implementing strategies at the primary stages of production 

which take place at the farm level [97]. Specifically, faecal contamination of carcasses at the 

abattoir may be prevented if the presence of Campylobacter bacteria in poultry faeces is 

addressed at the source [98]. It is estimated that the prevention of high rates of Campylobacter 

spp. colonization in poultry has the potential to reduce human cases related to poultry meat 

consumption by 7-10% [99]. 

Numerous epidemiological studies have examined the effect of biosecurity on colonisation of 

poultry with Campylobacter spp. [45,50,58–62,97,99–102]. However, the majority of these 

studies were conducted in high-income countries in Europe and North America. The literature 

review could not identify a South African study. The literature that examines the relationship 

between biosecurity practices and Campylobacter prevalence is largely focused on the 

commercial poultry sector [58,60,62,99–101] as opposed to small-scale or backyard 

operations. A single study of small-scale chicken farms in a low to middle income peri-urban 

area was identified [61].  

Various biosecurity and management practices associated with an increased risk of colonisation 

with Campylobacter spp. are presented in the literature. These include: increasing age of birds 

in the flock [45,58,59,102], age at slaughter greater than 45 days [60] and 35 days [62], 

increasing number of poultry houses on the farm [45,50,100,102], increasing age of the poultry 

house building [59,62,101], the practice of partial depopulation [45,58,102], increased number 

of farm workers in the poultry houses [62,102] and increased number of personnel visits to the 

houses per day [102]. Poor cleaning and disinfection of poultry houses and the surrounds 

[50,59,102] as well as a non-cement floor material that is not easy to clean [61] were also 

associated with an increased risk of Campylobacter colonisation. Less than 10 years of farming 

experience and the use of rice husk as a litter material were significantly associated with 

Campylobacter prevalence in a 2019 Bangladeshi cross-sectional study, although causality 

cannot be determined using this study design [59]. It was argued that the presence of other 

livestock on the farm or nearby [45,58,100] and rodents or insects in the poultry house [45,58] 

are factors associated with an increased risk of bacterial colonisation.  



16 

A number of exposures have had contradictory effects on Campylobacter spp. colonisation of 

poultry in reported literature. Examples include the presence of other livestock on the farm 

[50,102], the presence of wildlife [50,102], pets (dogs and cats) [50,58,102], rodents and rodent 

control, insects [50,102] and length of time the poultry houses are left empty between flocks 

(downtime) [50,100–102]. Additional management practices with inconclusive predictive or 

protective effects include: water source and water disinfection [45,50,58,100,102], flock size 

and stocking density [45,50,59,102]. The heterogeneity of these findings may be related to 

unmeasured confounders (possibly broiler breed and breed-related disease susceptibility, or 

effect of in-feed antibiotics), varying laboratory diagnostic methods and different study 

environments (namely commercial farming versus informal production). Aspects of 

methodological design such as sample type and stage of production would also possibly 

contribute to these conflicting findings. 

The literature highlighted that scientific knowledge of biosecurity interventions that could 

contain the spread of Campylobacter was inconsistent. Practices that could reduce the risk of 

Campylobacter colonisation are mostly related to the use of a hygiene barrier or entrance room. 

This is a station at which hand-washing, house-specific boots and clothes, overshoes and foot 

baths are used [45,50,58,101,102]. Conversely, foot baths may also increase the risk of 

colonisation if not properly used, replenished and maintained [45]. The all-in-all-out system in 

place of partial depopulation [45,50,102], use of dedicated equipment for each poultry house 

[50], staff training and limited access to key personnel only [50] are also important biosecurity 

measures to reduce colonisation in broilers [50]. A review of biosecurity-related literature 

suggested that vehicle disinfection and gates for each poultry house may be important in 

reducing the prevalence of Campylobacter on the farm [50]. An Iranian cross-sectional study 

conducted on commercial broiler farms suggested that the use of antibiotic drugs in the first 

two weeks of production and an increasing level of education of the farmer were protective 

factors [60]. Though many possible sources of disease on the farm were identified, evidence 

for control was conflicting. No clear target for improved disease control at the farm-level was 

identified [45,50,102].  

1.2.5 Biosecurity and management practices associated with colonisation of 

poultry by non-typhoidal Salmonella enterica serovars 

Given that the epidemiology of NT S. enterica serovars and Campylobacter spp. in broiler 

production systems differ, biosecurity and management practices of importance to these 

diseases may also diverge. Limited research is available concerning NT S. enterica serovars 



17 

and associated biosecurity and management practices. The available literature primarily 

discusses commercial poultry farming which generally operates with greater resources than the 

small-scale poultry industry. Most studies discuss vaccination as a tool for Salmonella 

prevention without addressing the role of basic biosecurity in the spread of Salmonella serovars 

[103–105]. Totton et al.’s [103] random effects meta-analysis of the effect of biosecurity and 

vaccination on NT S. enterica serovars in broiler poultry was limited by the minimal number 

of relevant studies available for analysis. Of the 22 studies addressing vaccination against NT 

S. enterica serovars, the majority were conducted in high income countries in Europe and North 

America. Vaccination with a modified-live S. Typhimurium vaccine (investigated in nine 

studies) was found to reduce the risk of colonisation in broilers [104]. A single observational 

study in Canada that investigated risk factors for colonisation with NT S. enterica serovars in 

broilers, found that the odds of bacterial colonisation for chicken houses that were not 

permanently locked was 2.6 times the odds of colonisation for chicken houses kept 

permanently locked [106]. A similar study in the Netherlands identified that NT S. enterica 

prevalence increased with increasing flock size [73]. The feed supplier also had a significant 

effect on prevalence, suggesting that contaminated feed material is an important risk factor 

[73]. Additional factors associated with colonisation and contamination with NT S. enterica 

serovars include: a downtime less than one month, having less than five fans in the poultry 

house, farms with less than four houses [87] and on-farm disposal of waste [85]. In common 

with findings from studies dealing with Campylobacter colonisation [45,58,100], the presence 

of other livestock on the farm increased the risk of NT S. enterica in poultry [85]. 

Epidemiological research concerning biosecurity and management practices associated with 

NT S. enterica colonisation in broilers is not as comprehensive as that related to Campylobacter 

colonisation. Based on the available literature, farm-level biosecurity and management 

practices offer important control points which may be used to limit the entry of NT S. enterica 

serovars into the human food-chain. However, specific biosecurity practices relevant to small-

scale producers have not been conclusively identified. 

1.3 Problem Statement 

Colonisation of broiler poultry with Campylobacter spp. and NT S. enterica serovars 

compromises food safety and facilitates the transmission of foodborne zoonotic disease to 

consumers of poultry products [11,21,25–27,46,107]. A limited number of studies have 

estimated the prevalence of Campylobacter spp. or NT S. enterica serovars in the South African 



18 

commercial poultry population [53–55] and even less research has been conducted in the small-

scale broiler sector [55]. Although it is accepted that effective biosecurity practices can reduce 

harmful bacteria at the farm-level, few data concerning management and biosecurity correlates 

of disease exist for African-based small-scale poultry production systems. 

1.4 Justification  

Currently, the prevalence of Campylobacter spp. and NT S. enterica serovars in small-scale 

broiler units in South Africa is not known. The local small-scale broiler sector probably 

presents substantial food safety and public health challenges related to Campylobacter spp. and 

NT S. enterica serovars. Baseline prevalence estimates are important for monitoring disease 

trends and evaluating the effect of disease control interventions in poultry populations. In 

addition, information on the prevalence of these organisms may be used to guide policies and 

control measures within this production sector. This would assist to improve the health of 

small-scale poultry populations and possibly expand poultry product exports. Numerous 

studies [45,50,97,99,102] in high-income countries and commercial poultry enterprises 

examined the benefits of biosecurity practices on these diseases in poultry. However, small-

scale farms in developing countries have not been the subject of such investigations. The 

findings of this study could provide evidence for effective and practical biosecurity measures 

to target important foodborne bacterial pathogens in small-scale broiler production systems. 

Practical methods to reduce colonisation of birds with pathogens at the farm may be useful in 

mitigating the risk of foodborne disease in humans [99,107].  

1.5 Research Question 

Which biosecurity or farm management practices are associated with the prevalence of 

Campylobacter spp. and NT S. enterica serovars in small-scale broiler production systems in 

Gauteng Province, South Africa? 

1.6 Aim 

This study aimed to estimate the prevalence of Campylobacter spp. and NT S. enterica serovars 

in small-scale broiler production systems in Gauteng Province and determine the correlation of 



19 

farm biosecurity and management practices with isolation of Campylobacter spp. and NT S. 

enterica serovars.  

1.7 Objectives 

1. Estimate the prevalence of Campylobacter spp. and NT S. enterica serovars in the small-

scale broiler sector in Gauteng Province. 

2. Describe the farm-level biosecurity and management practices implemented on small-

scale broiler farms in Gauteng Province. 

3. Examine the association of implemented biosecurity and management practices with 

isolation of Campylobacter spp. and NT S. enterica serovars in small-scale broilers in 

Gauteng Province. 

 

 

  



20 

CHAPTER 2 - METHODS 

This chapter provides information about the primary data source and data collection processes. 

For the present study, the methods used for data management and descriptive analysis of the 

small-scale farms and biosecurity practices are then presented. Finally, the methods applied for 

inferential analysis to investigate farm management and biosecurity correlates of 

Campylobacter spp. positivity using logistic regression are indicated.  

 

The chapter headings are:  

2.1 The primary data 

2.1.1 Data source 

2.1.2 Setting 

2.1.3 Study population 

2.1.4 Study sample 

2.1.5 Sampling strategy 

2.1.6 Data collection process 

2.2 Present study 

2.2.1 Study design 

2.2.2 Sampling strategy 

2.2.3 Precision and power calculation 

2.3 Data management 

2.4 Exposure variables 

2.5 Outcome variable 

2.6 Descriptive analysis 

2.6.1 Study participants 

2.6.2 Prevalence estimates 

2.7 Inferential analysis 

2.7.1 Univariable analysis 

2.7.2 Multivariable analysis 

2.8 Ethical considerations 

  



21 

2.1 The primary data 

The present study is a secondary analysis of the primary data described below. 

2.1.1 Data source 

The data source was operational data collected by Gauteng Veterinary Services (GVS), 

Gauteng Department of Agriculture and Rural Development (GDARD). The primary data were 

not collected for the purposes of a primary scientific study. The data were collected as part of 

routine GVS disease surveillance activities for Campylobacter spp. and Salmonella serovars. 

Routine biosecurity audits were conducted simultaneously at the sampled farms by GVS 

officials to assess and improve the farming practices on emerging small-scale broiler farms.  

2.1.2 Setting 

Data were collected from November 2020 to May 2021 across three areas of Gauteng Province 

(Pretoria, Randfontein and Germiston). 

2.1.3 Study population 

The study population was all small-scale broiler farms in Gauteng Province that had a flock 

size of between 100 and 120 000 broilers. 

2.1.4 Study sample 

GDARD’s study sample was 30 consenting small-scale broiler farmers (10 farms each in 

Pretoria, Randfontein and Germiston). At each farm, 30 broiler birds were sampled, therefore, 

there were 900 poultry samples in total. 

2.1.5 Sampling strategy 

No complete register of small-scale broiler farms in Gauteng Province exists. Therefore, the 

size of the small-scale broiler industry is not known and can only be estimated. A representative 

sample of small-scale broiler farms were selected from a list of 45 farms known to GDARD 

using stratified random sampling. Small-scale broiler farms were divided into three strata based 

on area and 10 farms were randomly selected per area using simple random sampling. At each 

farm, 30 broiler birds were sampled using simple random sampling.  



22 

2.1.6 Data collection process 

After the farmer provided written informed consent [Appendix E], a trained GVS veterinary 

official conducted an inspection of the farm and completed a questionnaire with the farmer 

using the EpiCollect5 mobile application [Appendix F]. The questionnaire collected 

demographic information about the farmer and data about farm management and biosecurity 

practices. A unique reference number was generated for each farm and used to label the 

laboratory submission form. The veterinary official then collected cloacal samples from 30 

broiler birds using swabs with Amies charcoal-based transport medium. The cloacal swabs 

were kept at a cool temperature and transported to the Agricultural Research Council – 

Onderstepoort Veterinary Research (ARC-OVR) laboratory on the same day as sample 

collection. Standard methods for bacterial culture of Salmonella (ISO 6579 of 2002) and 

Campylobacter spp. (ISO 10272-1 of 2006) were applied. ISO 6579 of 2002 is used to detect 

any Salmonella serovar and is not limited to NT S. enterica serovars. Speciation of positive 

Campylobacter culture results was performed using a real-time polymerase chain reaction 

(PCR) test. Since these bacteria are not normally present, growth of any Salmonella serovar or 

Campylobacter spp. by bacterial culture qualified as a positive result, and no cut-off value was 

required for interpretation. 

2.2 Present study 

2.2.1 Study design 

A cross-sectional study design was applied for this secondary data analysis. 

2.2.2 Sampling strategy 

Data collected from all 30 farms (900 birds) were included in the analysis. 

2.2.3 Precision and power calculation 

The sample size was not calculated since it was predetermined by the primary data. A precision 

calculation was manually performed using the below formula where p was the expected 

prevalence and n was the sample size. For this calculation, a prevalence of 75% was assumed. 

A precision estimate, or measure of uncertainty, of 2.8% was obtained. This suggests that the 

true prevalence may lie between 2.8% lower or higher than the calculated value. 



23 

𝑃𝑟𝑒𝑐𝑖𝑠𝑖𝑜𝑛 =  ±1.96 √
𝑝 (1−𝑝)

𝑛
× 100 =  ±1.96 √

(0.25)(0.75)

900
× 100 = 2.8%  

Power of the prevalence analysis was calculated with Stata version 15 (StataCorp, College 

Station, TX, USA) using the power calculator [108]. A moderate design effect of 1.5 was used 

to account for clustering of birds within farms, therefore, the effective sample size was 300. At 

a significance level of 5%, the sample had over 90% power (93%) to detect as significant a 

difference in prevalence of 10% versus 25% between two equally sized groups. When a more 

conservative design effect of two was used (effective sample size = 228), a power of at least 

80% (84%) was still demonstrated to detect as significant a difference between a prevalence of 

10% and one of 25%. Therefore, the study is adequately powered to detect any large effects on 

prevalence. 

2.3 Data management 

The questionnaire data and the laboratory results data were received in two separate Microsoft 

Excel (Office 365, 2016) files. Using Stata version 15 (StataCorp, College Station, TX, USA), 

the files were merged using the unique reference number that was captured in both documents. 

The data were cleaned using Microsoft Excel and Stata version 15. 

A variable to denote Campylobacter status of a farm was created using the frequency of 

Campylobacter cases per farm, where a farm with one or more cases was positive and a farm 

with zero cases was negative. In order to describe the study sample of farms by Campylobacter 

status, sociodemographic variables and basic farm characteristics (type of housing, size of the 

flock and type of management system) were included. 

Farm-level prevalence estimates were estimated using the binary outcome variables, 

Campylobacter status and NT S. enterica status that had a positive or negative value. Bird-

level prevalence was calculated using numerical outcome variables, total Campylobacter 

positive birds and total NT S. enterica positive birds, which took on values from 0 to 30. 

The investigation of possible correlates between bacterial prevalence and farm management 

and biosecurity practices required the Campylobacter binary outcome variable and 31 exposure 

variables of interest from the biosecurity questionnaire. Some exposure variables were created 

from a combination of two or more variables in the original data. The details of these 

adjustments are presented in Supplementary A.  



24 

2.4 Exposure variables 

The exposure variables included in the analysis are presented in four categories. These are 

socio-demographic, farm management, biosecurity and other. Supplementary B contains 

additional information about these variables. 

Socio-demographic variables:  

Sex of owner 

 Education level of owner 

 Area  

 State support 

 

Farm management variables: 

 

 Target market 

 All-in-all-out system 

 Flock size 

 House structure 

 Number of poultry houses 

 Stocking density ≤ 15 birds/m2 

 Days rested between cycles 

 Age at slaughter or selling in weeks 

 Drinking water treatment  

 Open water body less than 1 km away 

 Water source 

 Storeroom 

 Number of farm workers 

 Feed bags open/spilled  

Dogs/cats around houses 

 Antibiotic use 

 Salmonella vaccination 

 

Biosecurity variables: 

 

 Overalls provided 

 Gumboots provided 

 Access control 

 Foot bath used and replenished at least daily 

 Fenced and gated chicken houses 

 Bird-proofed houses 

 Rodent control 

 Hand hygiene 

 Clean house with water/soap and disinfectant 

 Disinfection of vehicles entering farm 

 

Other variables:  

 

Season 

 



25 

2.5 Outcome variable 

For the inferential analysis, a binary outcome variable based on the laboratory results for 

Campylobacter spp. (positive or negative) and NT S. enterica serovars (positive or negative) 

was used.  

2.6 Descriptive analysis 

2.6.1 Descriptive statistics of study sample 

Characteristics of the small-scale broiler farms were described by Campylobacter spp. status 

(positive or negative). Categorical variables were described using frequency and percent of 

farms. The median and interquartile range for the non-normally distributed continuous 

variables were presented. 

2.6.2 Prevalence estimates 

Prevalence estimates and 95% confidence intervals were calculated using the proportion 

command in Stata version 15. If one or more broilers from a farm was positive on culture for 

Campylobacter spp. or NT S. enterica serovars, the farm was considered as a positive farm. 

Farm-level prevalence was estimated using the total number of Campylobacter-positive farms 

as the numerator and the total number of farms sampled for Campylobacter as the denominator. 

Bird-level prevalence of Campylobacter was calculated using the total number of 

Campylobacter-positive bird samples as the numerator and the total number of birds sampled 

for bacterial culture as the denominator. The proportion of C. coli and C. jejuni species was 

also determined. Since there were no indeterminate results, all laboratory results were used to 

calculate the prevalence of Campylobacter. The same methods were used to calculate the farm-

level and bird-level prevalence of NT S. enterica serovars. The bird-level prevalence of 

Campylobacter spp. in three areas of Gauteng Province was estimated using the number of 

Campylobacter-positive birds in Germiston, Tshwane and Randfontein respectively as the 

numerator, and the total number of birds sampled in each area as the denominator.  



26 

2.7 Inferential analysis 

Inferential analysis was conducted at the bird-level using logistic regression. The data were 

aggregated by farm; therefore, frequency weights were applied to represent the frequency of 

Campylobacter positive and negative birds per farm. Observations in the dataset were also 

clustered by farm. This was accounted for using the clustered sandwich estimator (vce (cluster 

clustervar) option [109] in Stata version 15. Using this method, the standard errors allow for 

intragroup correlation instead of requiring that observations are independent within groups. 

2.7.1 Univariable analysis 

Univariable analysis to determine the association of the individual exposure variables with the 

binary outcome variable was conducted using simple logistic regression. All exposure factors 

with a p-value less than or equal to 0.2 were considered as potentially significant and were 

eligible for inclusion in the multivariable model.  

2.7.2 Multivariable analysis 

Using the potentially significant variables identified from the univariable analysis, backward 

stepwise elimination with a p-value cut-off of 0.1 was used to select the final multivariable 

model. Variables that were expected to behave as confounders as indicated by literature or 

through logical reasoning (routine antibiotic use, season [92,110–112] and age at slaughter 

[92,110–112]) were retained in the model regardless of the p-value in order to account for 

potential bias. In the final model, those variables with a p-value less than 0.05 were considered 

significant. Variables with a p-value between 0.05 and 0.1 were considered as marginally 

significant. A post-regression link test for model specification was applied [113]. The 

unadjusted (univariable analysis) and adjusted (multivariable analysis) odds ratios together 

with 95% confidence intervals and p-values for each included predictor variable were reported.  

2.8 Ethical considerations 

An ethics waiver was obtained from the Animal Research Ethics Committee of the University 

of the Witwatersrand in December 2020, certificate reference ‘Waiver 04-12-2020-O’ 

[Appendix C]. Permission was granted by GDARD to access to the primary data used in the 



27 

present study [Appendix D]. Personal identifiers were removed prior to accessing the data. 

Participants of the primary surveillance activities gave written informed consent [Appendix E]. 

 

  



28 

CHAPTER 3 – RESULTS 

In this chapter, the characteristics of the small-scale broiler farms sampled, including farm 

biosecurity and management practices are described. The prevalence estimates for 

Campylobacter spp. and NT S. enterica serovars at farm and bird level are then presented. 

Finally, the exposure variables that correlated with Campylobacter spp. colonisation in the 

univariable and multivariable analyses are reported.  

 

The chapter headings are:  

3.1 Characteristics of the study sample 

3.1.2 Biosecurity and management practices of the study sample  

3.2 Campylobacter spp. and non-typhoidal Salmonella enterica serovars prevalence estimates 

3.3 Logistic regression model 

3.3.1 Univariable correlates of Campylobacter spp. colonisation 

3.3.2 Multivariable correlates of Campylobacter spp. colonisation 

 



29 

3.1 Characteristics of the study sample 

Descriptive analysis was conducted at farm-level (Table 3.1). Among sampled small-scale farms 

practicing informal poultry production, local trade in live birds was predominant (24/27, 88.9% 

of Campylobacter positive farms and 2/3, 66.7% of non-colonised farms). In the 

Campylobacter positive group, the majority of farm owners were male (16/27, 59.3%) and had 

attained either further or higher education levels (11/27, 40.7% each). Farms located in 

Tshwane were positive for Campylobacter (10/10, 100%). Farms from which Campylobacter 

spp. were isolated mainly practiced all-in-all-out management (17/27, 63%) and had a smaller 

median flock size of 900 birds (IQR 235 – 2500) compared to a median of 1000 birds (IQR 

1000 – 1200) in Campylobacter negative farms. Open-walled and naturally ventilated poultry 

houses were predominant in both Campylobacter positive (26/27, 96.3%) and Campylobacter 

negative farms (3/3, 100%). Most farms were sampled during summer (19/30, 63.3%).  

Table 3.1: Description of the study sample, Gauteng Province, 2020-2021 

Characteristic 

Campylobacter 

negative farms 

(N=3) 

Campylobacter 

positive farms 

(N=27) 

Total sampled 

farms (N=30) 

n % n % n % 

Sex of farm 

owner 

Male 0 0 16 59.3 16 53.3 

Female 3 100 11 40.7 14 46.7 

Education level 

of owner 

Declined/unknown 0 0 2 7.4 2 6.7 

Intermediate (grade 4-6) 0 0 1 3.7 1 3.3 

Senior (grade 7-9) 0 0 2 7.4 2 6.7 

Further education (grade 10-12) 1 33.3 11 40.7 12 40 

Higher education 

(college/tertiary) 

2 66.7 11 40.7 13 43.3 

Area Germiston 2 66.7 8 29.6 10 33.3 

Tshwane 0 0 10 37 10 33.3 

Randfontein 1 33.3 9 33.3 10 33.3 

State 

supported farm 

No 3 100 21 77.8 24 80 

Yes 0 0 6 22.2 6 20 

Target market Abattoir  1 33.3 2 7.4 3 10 

Live birds 2 66.7 24 88.9 26 86.7 

Informal slaughter 0 0 1 3.70 1 3.3 

All-in-all-out 

management 

No 1 33.3 10 37 11 36.7 

Yes 2 66.7 17 63 19 63.3 

Flock size Median [IQR] 1000 [1000-1200] 900 [235-2500] 1000 [300-2400] 

Structure of 

poultry house 

Free range permanently/part of 

day 

0 0 1 3.7 1 3.3 

Open walls, natural ventilation 3 100 26 96.3 29 96.7 

Season at 

sample 

collection 

Autumn 0 0 11 40.7 11 36.7 

Summer 3 100 16 59.3 19 63.3 



30 

3.1.2 Biosecurity and management practices of the study sample  

Biosecurity measures to prevent the introduction and spread of pathogens were inconsistently 

implemented in the small-scale broiler farms (Table 3.2). Treatment of drinking water (such as 

chlorination) was not practiced in the majority of farms (25/27, 93% of Campylobacter positive 

farms). Just over half of all farms utilised borehole water. In most Campylobacter positive 

farms, feed, cleaning equipment and disinfectant were stored separately. Spillage of feed bags 

occurred in a third of colonised poultry farms and the presence of dogs or cats near to the 

poultry houses was common. Antibiotic use was common across the study sample. The 

provision of gumboots and overalls was inconsistent in both Campylobacter positive and 

Campylobacter free farms. Use of footbaths and the regular change of disinfectant used in the 

footbaths were poorly implemented. Rodent and bird control methods were not used in the 

majority of Campylobacter positive farms. Few farms practiced hand hygiene and disinfection 

of vehicles entering the farm. It was also evident that 48% (13/27) of farms had poor poultry 

house cleaning and disinfection practices. 

Table 3.2: Biosecurity and farm management exposure variables and Campylobacter status of 30 small-

scale broiler farms, Gauteng Province, 2020-2021 

Exposure variable 

Campylobacter 

negative farm (N=3) 

Campylobacter 

positive farm (N=27) 

Total sampled 

farms (N=30) 

n % n % n % 

Number of 

poultry houses 

Median [IQR] 1 [1-1] 1 [1-2] 1 [1-2] 

Stocking density 

≤ 15 birds/m2 

No 0  0 5  18.5 5 16.7 

Yes 3  100 22  81.5 25 83.3 

Poultry house 

rest days 

Median [IQR] 8 [7-21] 9 [7-14] 8.4 [7-14] 

Age at 

slaughter/selling 

(weeks) 

4  1  33.3 3  11.1 4 13.3 

5  0  0 8  29.6 8 26.7 

6  2  66.7 13  48.1 15 50 

7  0  0 3  11.1 3 10 

Drinking water 

treatment 

No 3  100 25  93 28 93.3 

Yes 0  0 2  7 2 6.7 

Open water body 

≤ 1km from farm 

No 3  100 20  74 23 76.7 

Yes 0  0 7  26 7 23.3 

Water source Borehole 1 33.3 15 55.6 16 53.3 

Municipal 2 66.7 12 44.4 14 46.7 

Storeroom  None 0 0 9 33.3 9 30 

Separate for feed, 

cleaning equipment 

and disinfectant 

1 33.3 11 40.7 12 40 

Single storeroom 2 66.7 7 25.9 9 30 

Number of farm 

workers 

Median [IQR] 3 [2-3] 2 [1-4] 2[2-4] 

Feed bags 

open/spilled 

No 3  100 19  70.4 22 73.3 

Yes 0  0 8  29.6 8 26.7 



31 

Exposure variable 

Campylobacter 

negative farm (N=3) 

Campylobacter 

positive farm (N=27) 

Total sampled 

farms (N=30) 

n % n % n % 

Dogs/cats 

around poultry 

house 

No 0  0 11  40.8 11 36.7 

Yes 3  100 16  59.2 19 63.3 

Antibiotic use No 0  0 6  22.2 6 20 

Yes 3 100 21 77.7 24 80 

Salmonella 

vaccination 

No 3  100 26  96.3 29 96.7 

Yes 0 0 1  3.7 1 3.3 

Overalls 

provided 

No 1  33.3 16  59.2 17 56.7 

Yes 2  66.7 11  40.8 13 43.3 

Gumboots 

provided 

No 2  66.7 10  37 12 40 

Yes 1  33.3 17  63 18 60 

Access control No 1  33.3 14  51.9 15 50 

Yes 2  66.7 13  48.1 15 50 

Footbath used 

and disinfectant 

changed daily 

No 0  0 14  51.9 14 46.7 

Yes 3 100 13  48.1 16 53.3 

Fenced farm and 

gated poultry 

house 

No 0  0 12  44.4 12 40 

Yes 3  100 15  55.5 18 60 

Bird-proofed 

poultry houses 

No 1  33.3 16  59.2 17 56.7 

Yes 2  66.7 11  40.8 13 43.3 

Rodent control No 1  33.3 13  48.1 14 46.7 

Yes 2  66.7 14  51.9 16 53.3 

Hand hygiene No 2  66.7 18  66.7 20 66.7 

Yes 1  33.3 9  33.3 10 33.3 

Clean house with 

water/soap and 

disinfectant 

No 2  66.7 11  40.8 13 43.3 

Yes 1  33.3 16  59.2 17 56.7 

Disinfection of 

vehicles entering 

farm 

No 2  66.7 25  93 27 90 

Yes 1  33.3 2  7 3 10 

3.2 Campylobacter and non-typhoidal Salmonella enterica prevalence 

estimates  

The prevalence of Campylobacter spp. at farm-level was high (27/30, 90%), and half of the 

broilers sampled were positive for Campylobacter on culture (448/900, 49.8%). Prevalence 

ranged widely between three areas of Gauteng Province (59/300, 19.7% to 214/300, 71.3%). 

The dominant Campylobacter species isolated was C. coli (316/448, 70.5%). No NT S. enterica 

serovars or any other Salmonella serovar were cultured from the broiler samples (Table 3.3). 

The 95% confidence interval suggests that a prevalence of up to 17.2% at farm-level and 4.0% 

at bird-level was possible.  

  



32 

Table 3.3: Campylobacter spp. and non-typhoidal S. enterica prevalence estimates, Gauteng Province, 2020-

2021 

Campylobacter spp. n/N 
Prevalence 

estimate (%) 

95% confidence 

interval 

Farm-level 27/30 90 73.4 – 97.9 

Bird-

level 

Any Campylobacter spp. 448/900 49.8 46.5 – 53.1  

C. coli 316/448* 70.5 66.1 – 74.7 

C. jejuni 167/448* 37.3 32.8 – 41.9 

Germiston area 59/300 19.7 15.3 – 24.6 

Tshwane area 175/300 58.3 52.5 – 64.0 

Randfontein area 214/300 71.3 65.9 – 76.4 

Non-typhoidal S. enterica serovars n/N 
Prevalence 

estimate (%) 

95% confidence 

interval 

Farm-level 0/30 0 0.0 – 17.2 

Bird-level 0/900 0 0.0 – 4.0 
 

*Prevalence estimates for C. coli and C. jejuni include 35 samples that were co-infected with both Campylobacter 

species.  

3.3 Logistic regression model 

3.3.1 Univariable correlates of Campylobacter colonisation 

Inferential analysis was conducted at bird-level, using 900 poultry observations. Findings from 

the univariable analyses of all exposure variables are tabulated (Table 3.4). The demographic 

variable for education level of the owner was eligible for inclusion in the multivariable model. 

Variables relating to farm management that were potentially significant were state support of 

the farm and poultry house structure. Biosecurity related exposures that were considered for 

inclusion in the multivariable model were: the presence of an open water body less than one 

kilometre from the farm, treatment of drinking water, feed spillage or open feed bags, the 

provision of gumboots and the application of rodent control measures. The summer season was 

also an exposure of potential significance. 

Table 3.4: Univariable logistic regression findings of possible exposures correlated to Campylobacter spp. 

positivity, Gauteng Province, 2020 - 2021 

Variable % Campylobacter 

positive, n/N 

Unadjusted OR (95% 

confidence interval) 

p value 

Education level of 

owner 

 

Declined/unknown 18.3%, 11/60 Reference 

Intermediate (grade 4-6) 90%, 27/30 40.10 (13.37 – 120.25) <0.001 

Senior (grade 7-9) 46.7%, 28/60 3.90 (0.60 – 25.23) 0.153 

Further education (grade 10-12) 58.9%, 212/360 6.38 (1.43 – 28.55) 0.015 

Higher education 

(college/tertiary) 

43.6%, 170/390 3.44 (0.91 – 13.02) 0.069 

State support Yes (vs. no) 17.8%, 32/180 3.52 (0.81 – 15.34) 0.094 

All-in-all-out Yes (vs. no) 48.8%, 278/570 0.90 (0.29 – 2.76) 0.848 



33 

Variable % Campylobacter 

positive (n/N) 

Unadjusted OR (95% 

confidence interval) 

p value 

Poultry house 

structure 

Open walls, natural ventilation 

(vs. free range permanently/ part 

of the day) 

51.3%, 446/870 14.73 (8.41 – 25.78) <0.001 

Flock size 1.00 (1.00 – 1.00) 0.810 

Poultry house rest days 1.00 (0.92 – 1.08) 0.941 

Open water body 

≤ 1km from farm 

Yes (vs. no) 76.7%, 161/210 4.61 (1.14 – 18.75) 0.033 

Water source Borehole (vs. municipal) 57.3%, 275/480 1.92 (0.63 – 5.82) 0.252 

Storeroom  None 61.9%, 167/270 Reference  

Separate for feed, cleaning 

equipment and disinfectant 

48.9%, 176/360 0.59 (0.16 – 2.18) 0.428 

Single storeroom 38.9%, 105/270 0.39 (0.09 – 1.75) 0.221 

Stocking density ≤ 

15 birds/m2 

Yes (vs. no) 47.1%, 353/750 0.51 (0.12 – 2.27) 0.380 

Access control Yes (vs. no) 46%, 207/450 0.74 (0.24 – 2.24) 0.592 

Number of poultry houses 1.09 (0.79 – 1.51) 0.589 

Age at slaughter/selling (weeks) 1.01 (0.52 – 1.93) 0.983 

Drinking water 

treatment 

Yes (vs. no) 96.7%, 58/60 33.46 (7.01– 159.71) <0.001 

Footbath used and 

disinfectant 

changed daily 

Yes (vs. no) 49%, 235/480 0.93 (0.31 – 2.81) 0.901 

Dogs/cats around 

poultry house 

Yes (vs. no) 53%, 302/570 1.42 (0.45 – 4.48) 0.549 

Fenced farm and 

gated poultry 

house 

Yes (vs. no) 43.3%, 234/540 0.52 (0.17 – 1.59) 0.253 

Hand hygiene Yes (vs. no) 42.3%, 127/300 0.64 (0.19 – 2.16) 0.470 

Number of farm workers 1.15 (0.89 – 1.47) 0.287 

Bird-proofed 

poultry houses 

Yes (vs. no) 47.2%, 184/390 0.83 (0.27 – 2.54) 0.747 

Antibiotic use Yes (vs. no) 51.4%, 370720 1.38 (0.45 – 4.25) 0.572 

Feed bags 

open/spilled  

Yes (vs. no) 68.8%, 165/240 2.93 (0.82 – 10.49) 0.098 

Gumboots 

provided 

Yes (vs. no) 58.1%, 314/540 2.34 (0.74 – 7.47) 0.150 

Overalls provided Yes (vs. no) 56.3%, 219/390 1.57 (0.50 – 4.92) 0.437 

Rodent control Yes (vs. no) 39.2%, 188/480 0.40 (0.13 – 1.20) 0.101 

Season at sample 

collection 

Summer (vs autumn) 34.4%, 196/570 0.16 (0.05 – 0.49) 0.001 

Clean house with 

water/soap and 

disinfectant 

Yes (vs. no) 52.7%, 269/510 1.32 (0.44 – 3.98) 0.627 

Disinfection of 

vehicles entering 

farm 

Yes (vs. no) 36.7%, 33/90 0.55 (0.10 – 2.91) 0.483 

  



34 

3.3.2 Multivariable correlates of Campylobacter colonisation 

The unadjusted and adjusted odd ratios of variables correlated with Campylobacter positivity 

in the final multivariable model are presented (Table 3.5). The odds of Campylobacter 

colonisation among broilers that received treated drinking water was 86.81 times the odds of 

colonisation for broilers drinking untreated water. Campylobacter spp. positivity among 

broilers reared in an open-walled poultry house structure was 3.82 times as likely compared to 

broilers reared in free-range production systems. The odds of Campylobacter positivity where 

feed bag spillage occurred was 12.48 times the odds of positivity where feed bags were sealed. 

The summer season had a protective effect, with 94% reduced odds of colonisation compared 

to the autumn season. The following two variables were not statistically significant since the p 

values were greater than 0.05 and the 95% confidence intervals included the value of 1. 

Antibiotic use appeared to decrease the odds of Campylobacter positivity by 60%, but this 

exposure was only marginally significant (p value = 0.088). Lastly, for every one-week 

increase in the age of broilers at slaughter compared to four weeks of age, the odds of isolating 

Campylobacter were reduced by 37% (p value = 0.148). The equation for the final logistic 

regression model is presented in Supplementary C. 

Table 3.5: Variables correlated with Campylobacter spp. positivity in the final multivariable model, 

Gauteng Province, 2020-2021 

Variable Unadjusted OR (95% 

confidence interval) 

p value Adjusted OR (95% 

confidence interval) 

p value 

Poultry house 

structure 

Open walls, natural 

ventilation (vs. free 

range permanently/ 

part of the day) 

14.73 (8.41 – 25.78) <0.001 3.82 (2.09 – 6.97) <0.001 

Age at slaughter/selling (weeks) 1.01 (0.52 – 1.93) 0.983 0.63 (0.34 – 1.18) 0.148 

Drinking water 

treatment 

Yes (vs. no) 33.46 (7.01 – 159.71) <0.001 86.81 (4.56 – 1651.50) 0.003 

Antibiotic use Yes (vs. no) 1.38 (0.45 – 4.25) 0.572 0.40 (0.14 – 1.15) 0.088 

Feed bags 

open/spilled 

Yes (vs. no) 2.93 (0.82 –10.49) 0.098 12.48 (2.88 – 54.12)  0.001 

Season at sample 

collection 

Summer (vs autumn) 0.16 (0.05 – 0.49) 0.001 0.06 (0.02 – 0.20) <0.001 

 

The linktest to investigate misspecification errors showed that the t-test for hatsq was not 

significant (p value = 0.869). The linktest provided some evidence that the final model was 

correctly specified, accounted for all relevant variables and did not include irrelevant variables.  

  



35 

CHAPTER 4 – DISCUSSION, RECOMMENDATIONS AND 

CONCLUSION 

In this chapter, the major prevalence findings for Campylobacter spp. and non-typhoidal S. 

enterica serovars and significant biosecurity or farm management correlates are discussed in 

the context of the available previous literature. This discussion is then followed by the 

limitations of the study and recommendations for small-scale broiler farmers and consumers, 

as well as for future research in this field. Finally, the key contributions of this research are 

presented in the conclusion.  

 

The chapter headings are:  

4.1 Key study findings 

4.2 Contextualisation of key findings  

4.3 Limitations of the study 

4.4 Recommendations 

4.5 Conclusion 

  



36 

4.1 Key study findings 

This cross-sectional study aimed to estimate the prevalence of Campylobacter spp. and NT S. 

enterica serovars in small-scale broiler production systems in Gauteng Province. Farm 

management and biosecurity correlates of these pathogens were also investigated.  

A high prevalence of Campylobacter spp. in small-scale broiler production systems, dominated 

by C. coli, was determined. Conversely, NT S. enterica serovars were not isolated from 

sampled broilers. Important biosecurity measures such as hand hygiene, water treatment and 

disinfection of vehicles entering the farm were not practiced by many sampled farms. The 

multivariable logistic regression model that accounted for clustering of observations by farm, 

identified three farm management and biosecurity correlates of Campylobacter spp. positivity. 

Open-walled and naturally ventilated poultry house structures compared to free-range 

production systems was positively correlated with isolation of Campylobacter spp. Predictive 

practices included feed spillage or unsealed feed bags and drinking water treatment. The 

summer season had a protective effect. Factors that remained in the final model but were not 

statistically significant included age in weeks at slaughter or selling and the use of antibiotics 

in the poultry cycle.  

4.2 Contextualisation of key findings  

The high prevalence of Campylobacter spp. in this study sample concurs with a previously 

estimated 47.1% prevalence in commercial free range chickens in South Africa [55]. However, 

the same study reported a Campylobacter spp. prevalence that was nearly 20% higher (68%) 

in chickens produced in a rural setting [55]. The authors did not define ‘rural farming systems’ 

but these may be comparable to the small-scale production systems sampled in the present 

study. Therefore, present study results indicate a lower Campylobacter spp. prevalence in 

small-scale broilers than previously demonstrated in South Africa. A declining local prevalence 

was also supported by an observational study conducted in 2016 in Kwa-Zulu Natal Province 

that described a 78% Campylobacter spp. prevalence in commercial broilers [96]. Conversely, 

lower Campylobacter spp. prevalence estimates (40.1%) were suggested by a recent meta-

analysis of African-based prevalence studies [66]. This difference in prevalence may be due to 

different sample types (caecal contents and faecal swabs versus cloacal swabs) [75], diagnostic