Risk factors and transmission pathways associated with infant Campylobacter spp. prevalence and malnutrition: A formative study in rural Ethiopia

Early infection from enteropathogens is recognised as both a cause and effect of infant malnutrition. Specifically, evidence demonstrates associations between growth shortfalls and Campylobacter infection, endemic across low-income settings, with poultry a major source. Whilst improvements in water, sanitation and hygiene (WASH) should reduce pathogen transmission, interventions show inconsistent effects on infant health. This cross-sectional, formative study aimed to understand relationships between infant Campylobacter prevalence, malnutrition and associated risk factors, including domestic animal husbandry practices, in rural Ethiopia. Thirty-five households were visited in Sidama zone, Southern Nations, Nationalities and Peoples’ region. Infant and poultry faeces and domestic floor surfaces (total = 102) were analysed for presumptive Campylobacter spp. using selective culture. Infant anthropometry and diarrhoeal prevalence, WASH facilities and animal husbandry data were collected. Of the infants, 14.3% were wasted, 31.4% stunted and 31.4% had recent diarrhoea. Presumptive Campylobacter spp. was isolated from 48.6% of infant, 68.6% of poultry and 65.6% of floor surface samples. Compared to non-wasted infants, wasted infants had an increased odds ratio (OR) of 1.41 for a Campylobacter-positive stool and 1.81 for diarrhoea. Positive infant stools showed a significant relationship with wasting (p = 0.026) but not stunting. Significant risk factors for a positive stool included keeping animals inside (p = 0.027, OR 3.5), owning cattle (p = 0.018, OR 6.5) and positive poultry faeces (p<0.001, OR 1.34). Positive floor samples showed a significant correlation with positive infant (p = 0.023), and positive poultry (p = 0.013, OR 2.68) stools. Ownership of improved WASH facilities was not correlated with lower odds of positive stools. This formative study shows a high prevalence of infants positive for Campylobacter in households with free-range animals. Findings reaffirm contaminated floors as an important pathway to infant pathogen ingestion and suggest that simply upgrading household WASH facilities will not reduce infection without addressing the burden of contamination from animals, alongside adequate separation in the home.

Oral consent was obtained from adult parents/guardians of child participants. There were no adult participants. Oral consent was obtained because most caregivers were illiterate; this was approved by both IRBs. It was recorded by writing on the survey form. If the data are held or will be held in a public repository, include URLs, accession numbers or DOIs. If this information will only be available after acceptance, indicate this by ticking the box below. For example: All XXX files are available from the XXX database (accession number(s) XXX, XXX.).
• If the data are all contained within the manuscript and/or Supporting Information files, enter the following: All relevant data are within the manuscript and its Supporting Information files.
• If neither of these applies but you are able to provide details of access elsewhere, with or without limitations, please do so. For example: Data cannot be shared publicly because of [XXX]. Data are available from the XXX Institutional Data Access / Ethics Committee (contact via XXX) for researchers who meet the criteria for access to confidential data.
The data underlying the results presented in the study are available from (include the name of the third party • All data created during this research will be openly available from the Cranfield Online Research Data at 10.17862/cranfield.rd.9907385. and contact information or URL). This text is appropriate if the data are owned by a third party and authors do not have permission to share the data.

• * typeset
Additional data availability information: Tick here if the URLs/accession numbers/DOIs will be available only after acceptance of the manuscript for publication so that we can ensure their inclusion before publication.

Infant growth, infection and domestic animal exposure 57
Enteropathogen infection and associated diarrhoea in infancy and the relationship with linear 58 growth failure (stunting) is a dynamic area of research in infant malnutrition. Whilst child 59 deaths from diarrhoea dropped by over half in just 15 years between 2000−2015 1 , diarrhoeal 60 episodes have not similarly decreased 2 suggesting a need for better measures to detect and 61 prevent infection. Early diarrhoea and diarrhoea-related sequelae hold both acute and 62 chronic consequences. Whilst good evidence indicates that a heavy early diarrhoeal burden 63 does affect growth and worsen nutritional status [3][4][5] , there is debate about its relative 64 contribution to long-term growth faltering 6,7 . Other direct, biological causes under study 65 include environmental enteric dysfunction (EED): a condition characterised by the 66 disturbance of gut immunity, structure and function, which ultimately impairs nutrient 67 absorption and linear growtheven without diarrhoea [8][9][10] . Nonetheless, the common 68 underlying factor to these different contributors is early exposure to pathogenic bacteria and 69 repeated infection 11,12 . As such it is increasingly evident that stunting will not be resolved by 70 improved nutritional intake or acute rehabilitation alone 13 but with parallel improvements in 71 water quality, sanitation and hygiene (WASH) which act as a primary barrier to infection. 72 73 Recent cluster-randomised control trials have sought to investigate the effect of improved 74 WASH, alone and in combination with nutrition supplementation, on child health. However 75 different study designs and settings have for the most part failed to show consistent 76 evidence for a reduction in diarrhoea or improvements in malnutrition indicators [14][15][16][17][18][19] . One possibility is that despite thorough design, interventions mainly focused on containing human 78 excreta and did not consider (and conventionally have not considered) the role of animal 79 faeces in domestic contamination and illness: 20 surprising given over 60% of infectious 80 diseases in humans are caused by zoonotic pathogens 21 . Transmission pathways are not 81 mutually exclusive, and inadequate separation of animals from the home environment may 82 inevitably result in faecal-oral transmission through direct contact with animal faeces or 83 contaminated soil, or faecal contamination of hands, food, objects or water sources 22-24 . 84 Infants are also vulnerable to transmission routes specific to age-related behaviours, across multiple pathways within the home and ingested through normal infant hand-to-mouth 95 behaviour 25,35 . Among those pathogens of highest concern, Campylobacter consistently 96 emerges as one of the key contributors to diarrhoea and malnutrition 31,32,34 and EED 28

Country context and study sample 146
In Ethiopia, despite substantial recent reductions, linear growth failure affected more than a 147 third of infants in 2016 59 . Ethiopia has one of the highest domestic animal densities per km 2 148 worldwide 60 and poultry are ubiquitous in rural households. Some research in Ethiopia has 149 documented the proximity and exposure of infants to chickens and their faeces in regions 61 150 and the relationship with infant growth 54 , and a few regional studies have associated 151 Campylobacter infection with infant diarrhoea and malnutrition. 62-64 . However further 152 research is required in Ethiopia on the epidemiology of infant Campylobacter prevalence and 153 infant health outcomes and the relationship to poultry ownership and WASH facilities. This 154 small, formative study was conducted in the Southern Nations, Nationalities, and 155 Peoples' region (SNNPR), Sidama zone (regional subdivision), Ethiopia, as the geographical 156 outreach area of the non-governmental organisation People in Need. The study took part in 157 the month of June 2019 -the start of the region's rainy season. Two rural kebeles 158 (neighbourhoods) were chosen from a woreda (zonal subdivision) which remained 159 representative of typical rural livelihoods across Sidama zone. A simple random sampling 160 method was used to identify households fulfilling the eligibility criteria of having an infant 161 aged 10−18 months and owning free-range poultry. The random sample is described as 162 follows. After communication with a government Health Extension Worker (HEW) local to 163 each kebele, the team produced a sampling frame for both kebeles of all infants aged 10−18 164 months from households known by the HEW to own poultry. For both sampling frames, 165 households were sequentially numbered on paper and using a simple lottery method 17 and 166 18 infants were randomly drawn from the two kebele frames respectively for a total sample 167 of 35 infants. Households were visited on a single occasion. 168

169
As this study was designed to provide formative evidence, a sample size calculation was not 170 performed. Formative research is often conducted as part of the process of a larger study 171 design and provides data for research teams to plan interventions or further data collection. 172 Formative research is early phase data and is not powered to detect differences between 173 groups. As such, this study results must be interpreted in this context, where it provided 174 indicative data towards the hypothesis but was not sufficiently powered for conclusive 175 techniques were followed and samples weighed using sterile disposable weighing boats to 1 220 ± 0.05 g wet weight. Samples were then aliquoted into sterile plastic centrifuge tubes 221 containing 9 mL of prepared sterile peptone water and vortexed well. For poultry faecal 222 samples only, 100 µL of sample was pipetted into sterile tubes containing 900 µL of peptone 223 water to prepare a 10-fold serial dilution up to 10 5 dilution. 100 µL of floor surface and infant 224 faecal samples and poultry faecal sample dilutions of orders 10 1 , 10 3 and 10 5 were drop 225 plated on pre-labelled plates and spread using disposable L-shaped spreaders. At the start of each household visit, the study was introduced by the field team and HEW and 246 informed consent was described to the caregiver in their first language of Amharic or 247 Sidamo. Fieldworkers tested the caregivers' understanding of consent by asking them 248 questions regarding the study and the consent process, and explained all data was 249 anonymised. As most adult caregivers were illiterate, oral consent and assent for their infant 250 was recorded. The survey was written in English, translated to Amharic by the field team and 251  were classified as wasted (WLZ <-2 SD), eleven (31.4%) as stunted (LAZ <-2 SD) and four 289 infants both wasted and stunted (11.4%, WLZ and LAZ <-2 SD). Of those infants classified 290 as wasted (n=5), all had experienced diarrhoea within the past seven days (p<0.001; OR 291 1.83, 95% CI 1.07−3.14). Diarrhoeal prevalence was not significantly related to stunting 292

Campylobacter prevalence and correlation with infant health measures 302
The following sections describe the relationships between survey variables, prevalence of 303 The associations between risk factors and transmission pathways in relation to infant health 333 outcomes are detailed in Table 2. Fig. 1

350
Results from this small cross-sectional study suggest that in these rural Sidamo households 351 raising free-range domestic poultry, the prevalence of infants testing positive for infection on growth is related to agehighlighting an increased level of risk as infants start 400 to crawl 31 . Whilst this study did not capture hand-to-mouth contact events, previous research 401 by this team in the same geographical area recorded infants mouthed their own hands or 402 those of their caregiver a mean 31 and 21 times respectively over one hour, which were 403 often visibly dirty (90.0% and 86.0% respectively) 50  were not correlated with stool samples negative for Campylobacter, perhaps suggesting that 420 simply providing WASH facilities will not prevent transmission and infection. However it is 421 possible facilities are also not used, particularly by children, which remains a limitation. In 422 rural communities it can be difficult to assess and accurately report the use of latrines and 423 soap for handwashing. Whilst in this study the visual inspection of latrines suggested they 424 were all used, soap ownership would often be reported but not seen. Regardless, it seems 425 logical that when sharing living spaces so closely, domestic animals contribute to infection 426 from zoonoses and widespread contamination of multiple pathways . There are intrinsic and 427 inseparable connections between these various transmission pathways. This is illustrated in 428

malnutrition. Specifically, evidence demonstrates associations between growth shortfalls and 27
Campylobacter infection, endemic across low-income settings, with poultry a major source. to humans mainly via contaminated food and water. In low-income countries, Campylobacter 60 is a common cause of infant diarrhoea and is linked to malnutritionboth wasting (low 61 weight-for-height) and stunting (low height-for-age). However despite evidence for this, we 62 know little about how, why and when infants become infected. This small study tested the 63 theory that infant health is negatively affected by Campylobacter infection due to domestic 64 animal exposure, and to assess risk factors associated with infection. We found that in 65 households with free-roaming poultry, almost half of infants had Campylobacter infection. does affect growth and worsen nutritional status [3][4][5] , there is debate about its relative 83 contribution to long-term growth faltering. 6,7 . Other direct, biological causes under study 84 include environmental enteric dysfunction (EED): a condition characterised by the 85 disturbance of gut immunity, structure and function, which ultimately impairs nutrient 86 absorption and linear growtheven without diarrhoea 8-10 . Nonetheless, the common 87 underlying factor to these different contributors is early exposure to pathogenic bacteria and 88 repeated infection. 11,12 . As such it is increasingly evident that stunting will not be resolved by 89 improved nutritional intake or acute rehabilitation alone 13 but with parallel improvements in 90 water quality, sanitation and hygiene (WASH) which act as a primary barrier to infection. facilities. This small, formative study was conducted in Sidama zone, the Southern Nations, 176 Nationalities, and Peoples' region (SNNPR), Sidama zone (regional subdivision), Ethiopia,177 as the geographical outreach area of the non-governmental organisation People in Need. 178 The study took part in the month of June 2019 -the start of the region's rainy season. Two 179 rural kebeles (neighbourhoods) were chosen from a woreda (zonal subdivision) which 180 remained representative of typical rural livelihoods across Sidama the zZone. A simple 181 random sampling method was used to identify households fulfilling the eligibility criteria with 182 of having an infant aged 10−18 months and owning free-range poultry. The random sample 183 is described as follows. After communication with a government Health Extension Worker 184 (HEW) local to each kebele, the team produced a sampling frame for both kebeles of all 185 infants aged 10−18 months from households known by the HEW to own poultry. For both 186 sampling frames, households were sequentially numbered on paper and using a simple lottery method 17 and 18 infants were randomly selected drawn from the two kebele frames 188 respectively for a total sample of 35 infants. Households were visited on a single occasion. 189

190
As this study was designed to provide formative evidence, a sample size calculation was not 191 performed. Formative research is often conducted as part of the process of a larger study 192 design and provides data for research teams to plan interventions or further data collection. 193 Formative research is early phase data and is not powered to detect differences between At the start of each household visit, the study was introduced by the field team and HEW and 275 informed consent was described to the caregiver in their first language of Amharic or 276 Sidamo. Fieldworkers tested the caregivers' understanding of consent by asking them 277 questions regarding the study andor the consent process, and explained all data was 278 anonymised. As most adult caregivers were illiterate, oral consent and assent for their infant 279 was recorded. The survey was written in English, translated to Amharic by the field team and 280 were calculated for length-for-age and weight-for-length (LAZ and WLZ respectively) using 295 the WHO 2006 Child Growth Standards 72 . Z-scores were categorised into stunting and 296 wasting using the standard cut-off value less than −2 standard deviations of the reference 72 . 297 analysis and further analysed using SPSS (version 22.0, IBM, New York). Simple frequency 299 distribution tests described survey response data, anthropometric data and Campylobacter 300 prevalence. Fisher's exact test for independence tested associations between variables for 301 the small sample size (5% significance). Results with significant p-values from the Fisher's 302 exact test reported odds ratio (OR) risk estimates with corresponding 95% confidence 303 intervals (CI). 304 305

Campylobacter infection prevalence and correlation with infant health measures 328
The following sections describe the relationships between survey variables, prevalence of 329

380
Results from this small cross-sectional study suggest that in these rural Sidamo households 381 raising free-range domestic poultry, the prevalence of infants testing positive for infant 382 Lastly, the use of culture-based method alone (and without sample pre-enrichment) holds 504 limitations: firstly due to changes in Campylobacter cell physiology and loss of viability 505 between sample deposition, collection, transport and plating (whereby cells enter the viable 506 but non-cultivable [VBNC] state). This may have underestimated the true prevalence. and by 507 On the other hand, culture holds limited sensitivity and high rates of false detection; 86 . whilst 508 there is evidence for good specificity of the agar in comparison and evaluation studies, there 509 is no certainty of the rate of false positives in this study. Secondly,Lastly, whilst the culture 510 media shows high specificity, it was not possible to differentiate between or quantify different 511 Campylobacter species. The parallel use of qPCR alone or PCR with ELISA methods would 512 enhance culture-based findings 31,87 . 513

Conclusion 515
This formative study adds further preliminary evidence to the body of research documenting 516 infant Campylobacter carriage and infection in households rearing free-range poultry-rearing 517 households. In these households, increased wasting and diarrhoea was seen in in infected 518 infants positive for presumptive Campylobacter highlights the potential impact of acute 519 infection. Repeated symptomatic infection and low weight weight loss may mean infants risk 520 entering a spiral of lower weight gain weight loss and subsequent growth deficits. 521 Alternatively, frequent carriage, or asymptomatic infection, and a high prevalence of stunting 522   Table 1, I would suggest change the header "% of total" to "Average or Percent", and move the values of "average age" and "average duration of diarrhea" from the "n" column to the "Average or Percent" column. Please also give the counts of diarrhea during last 7 days.
We thank the editor for their personal feedback on the manuscript and believe the comments have helped improve the overall draft. We appreciate the opportunity to respond to these comments.
This change has been made in Table 1. I believe the count of diarrhoea during the last 7 days is already noted (n=11)apologies if we have misinterpreted your suggestion. Page 13, lines 316-318. This sentence is very unclear to me. Are you comparing prevalence of infection between children with diarrhea during past 7 days and those without? If so, there should be an odds ratio. How are "symptomatic" vs "asymptomatic" carriers defined? Infections with diarrhea and infections without diarrhea in past 7 days? Thank you. Yes, symptomatic and asymptomatic are defined as you mention (with and without diarrhoea respectively, where diarrhoea was current [the stool sample was diarrhoeal]). This is clarified in lines 291-293 and again lines 335-336. Line 321, how was the OR of 1.41 calculated? If wasted kids were all infected, the odds of infection among wasted kids is then infinity, and the OR would be infinity too, as long as the infection rate among non-wasted children is not 100%. Given 95% CI for these ratios as well.
Thank you. This has been worded wrongly as all infants were compared. This is amended. This has also been amended in lines 338-343.
I would suggest making a table for all important OR estimates. This is included, thank you. Figure 1, again, not sure what is the difference between "symptomatic (diarrhea)" and "Diarrhea within past 7 days". One is 58.8% and the other is 31.4%, so they have to be different. Please give clear definition.
Thank you. We classified 'symptomatic' as infants who tested positive for Campylobacter who also had a diarrhoeal stool. Diarrhoea during the last 7 days was not tested for an association with positive infant faeces (although one might argue that it should be). This has been clarified again in lines 291-293 and lines 335-336. We have also added some detail in the figure legend (Fig 1, lines 372-373) if this helps to clarify, but can be removed if more appropriate.
Were poultry feces collected indoor or outdoor? Were poultry feces common indoor?
This is clarified lines 221-222. Thank you for making this point also, it is valuable: we did note the samples that came from outdoors versus indoors and I have made this point in the discussion (lines 425-427). If you developed a questionnaire as part of this study and it is not under a copyright more restrictive than CC-BY, please include a This is included, thank you. copy, in both the original language and English, as Supporting Information.
Funding information should not appear in the Acknowledgments section or other areas of your manuscript. This is removed.

1
Without confirmation of the taxonomic nature of the isolates by molecular and sequencing approach, I am highly suspecting that the Campylobacter prevalence reported in this study are highly overestimated. Further recovering Campylobacter from field samples can be very challenging in low income countries (I speak from experience); therefore, the use of a direct plating approach might not be the best strategy. This approach must be combined with more sensitive and specific methods. We respectfully disagree with the reviewer that culture overestimates Campylobacter infection and may actually underestimate due to VBNC (viable but non-culturable) bacteria. This is already noted in the limitations in lines 495-504. The authors are aware that there will be a rate of false positives that we cannot determine, and we have added this to the limitations section (line 501). Whilst PCR demonstrates high sensitivity, there are also pitfalls to using this detection method. Indeed it has also been noted that PCR will pick up Campylobacter of low burden, potentially over-estimating infection prevalence. Further, qPCR can also overestimate viable cells due to the presence of extracellular DNA which contributes to the signal. Culture alone is not a perfect method but in this study, was used as a simple method to provide formative data towards a hypothesis. We hope this response addresses the reviewer's important concern. Abstract: Missing word in the sentence line 25-26 Thank you. We cannot identify a missing word here and think the sentence reads clearly. All numbers starting at the beginning of a sentence must be fully written (not numeric) Thank you, this has been changed throughout.
Full name of SNNPR This has been changed. Please provide the r2 for the correlation data between the campylobacter prevalence and other variables studied (line 42) The authors respectfully comment that it is possible the reviewer has misinterpreted the analysis performed in this paper. As stated in the statistical analysis section (lines 301-304), we used Fisher's exact test to test associations between variables which predicted positive infant faeces. We used Fisher's exact test as a more appropriate measure for correlation due to the small sample size, which also gives an odds ratio and corresponding p-value, as stated in the text. We did not perform a regression between variables as it was felt by the team that the sample size was not sufficiently large to perform such an analysis and we felt we would be 'over-analysing' the data. As such, a limitation is that the associations here are not part of a model where other factors would be held constant/controlled for, but present associations between exposure variables and the outcome. Only the use of microbiology approach (direct plating on selective medium) to identify Campylobacter is not good sufficient enough. Additional molecular approach (PCR) must be performed to confirm the isolates belong to the Campylobacter genus.
We thank you reviewer for their comment and insight here. We have addressed this particular comment above.

Introduction:
The introduction is too long We thank the reviewer for their comment on the introduction. We politely disagree that the introduction is longer than is usually submitted to PLOS One. We endeavoured to frame the study within current evidence and to address the research need. We will be happy to cut the introduction if further reviews suggest to.

Materials and Methods
Remove text from line 161 to 167.
We thank the reviewer for this suggestion. We respectfully choose to keep the sentence to frame the study within context. The second reviewer also suggests to include more contextspecific examples of research within the field. Line 173: Replace Zone by zone. Further, what do you mean by zone (=Woreda or kebele)?
Thank you, this is clarified in line 176.
Line 173-175, More details are required concerning the simple random sampling method. Did the number of free-range chickens per household was recorded? Did the authors saw a correlation between the campylobacter prevalence or other recorded data and the number of free-range chickens.
Thank you. The random sample is already described in the text following (lines 180-188). A line has been added to clarify that the description follows.
Line 178, "simple lottery method" Can you clarify? Thank you. The simple lottery method is clarified within the description.
Line 180-182: I strongly disagree with the authors. The authors used these data to estimate the prevalence of Campylobacter in this specific area in Ethiopia. Therefore, the sampling method and the sample size are crucial in order to properly estimate the prevalence of Campylobacter. Please remove this sentence. Further, the authors must acknowledge that the population size low and restricted to only two kebeles. Please can you provide information concerning the expected statistical power of your analysis based population size. These information must be acknowledge in the manuscript as limitations of your study We thank the reviewer for their feedback here. We respect the view, however we politely disagree here. Formative research is not intended to test a hypothesis, but is intended as an initial part of the process of a larger study design, and can use both qualitative and quantitative methods to provide data for research teams to plan interventions or further data collection. Formative data is early phase data not powered to detect differences between groups (Gittelsohn et al., 2006). We have added this description to the methods (lines 190-196) to clarify the positioning of this study. We hope this addresses the reviewer's concern.
Line 194-195, please can you explain how the comparison was performed. Did you have reference chart for the consistency (i.e. solid, pasty, liquid)?
Thank you for your comment. The reporting of caregiver-reported diarrhoea is a standard measurement using WHO guidelines, as is referenced in lines 203-206. We have also described in lines 206-208 how we matched with visible diarrhoea. We did not use a chart but diarrhoea was clear from samples as is stated (6 or 7 on the Bristol stool chart). Line 202: what is the estimated time between the sampling and the incubation of the inoculated selective plates in microaerophilic conditions?
Thank you, we assume this refers to line 227 and have added a sentence to clarify timing.
Line 220: did the authors used a positive control in parallel during the processing the field samples (i.e. pure culture of C. jejuni or C. coli for example)?
Thank you. We did not use a positive control within the field. As described in lines 263-266, the media was tested prior to fieldwork using a pure culture and blank samples were used in the field to confirm absence of contamination.
Line 233: For how long the plates drying? This step is not necessary and might cause the death of Campylobacter.
Thank you for your insight. We left the plates to dry as per manufacturer instructions. Line 233-237: I know by experience that this protocol is not sufficient to isolate Campylobacter from feces in Ethiopia due to the development of dense background (green colonies for example) and/or the presence of red campylobacter-like colonies that are not Campylobacter after confirmation via genus specific PCR (16S primer).
We thank the reviewer for their insight. We are unsure how this sentence aligns with lines 233-237 in the manuscript. We note in lines 261-263 the prior testing of this Chromagar which shows high sensitivity and specificity. We did see a dense background in some plates when incubated: However, we discussed this with the manufacturer. It was stated that Chromagar media is designed for the isolation of thermotolerant Campylobacter (jejuni, lari, coli) and is very good at suppressing growth of background flora (the enterics such as E.coli and Enterococcus); thus there is confidence that any red colonies on the plate are Campylobacter spp.
Without confirming these results with PCR or other tests, there is little way we can respond: however, we trust in the validity of findings which have shown Chromagar selects for Campylobacter, and have indicated that Chromagar was used for the isolation and presumptive identification of thermotolerant Campylobacter (line 240).
To avoid over-claiming on our findings, we use the term 'presumptive' where possible in the manuscript.
We would also like to respectfully express we think it is important that these sorts of studies are replicated across different settings and contexts. Where the media may not have been sufficient in one context, that may not be the case across all studies. As with any scientific area, it is important to have more than one set of findings to make a judgement: particularly with new methods (Chromagar) where other authors may replicate protocols and add further evidence.
Line 241-243: Unfortunately, this statement is not true. Our studies (unpublished data) showed that other genus can grow in this medium, which give campylobacter-like colonies. Due date, the only way we could differentiate them was via aerobic growth test or 16S sequencing.
Thank you for your comment and sharing your experience in the field. We believe we have addressed this comment above. We would like to stress the same point made above about the importance of sharing findings from similar methodologies in order to strengthen the area of research. Line 249-252: I am scared that without PCR confirmation, you cannot affirm that your isolates are from the Campylobacter genus. This limitation must be acknowledged in your manuscript. I am suspecting that you are highly overestimating the Campylobacter prevalence based on your direct plating data. Our team faced the exact same issues in a previous project. Out of 1200 suspected Campylobacter isolates collected in Ethiopia using CHROMagar, less them 5% of the isolates were positive for Campylobacter via PCR. I would recommend buying external generators, which allow to continue your PCR cycles for 3-6 hrs.
We thank the reviewer for this insight and for sharing their experience. We believe we have addressed this comment with our first comment in response to the same comment from them. We agree an external generator would have been greatindeed back-up generators also failed on this occasion.
Line 273: you do not know whether Campylobacter or another microorganism of the gut microbiota is the causal agent of the diarrhea symptoms observed in the infant; thereby, you cannot use the term infected. It needs to be changed throughout the manuscript.
Apologies we cannot see how this comment refers to line 273. If the reviewer refers to lines 388 onwards, we have described that this is an association and the relationship with the excretion period of Campylobacter within the stool. We add a line that of course other microorganisms may be present which caused diarrhoeal symptoms (lines 397-398).
Respectfully, we would like to add the point that if we see suspected growth in a stool sample, it is highly likely the infant is infected -Campylobacter is not normally isolated in a stool, and only very small amount of subtypes are part of 'commensal' gut floramost of them are disease-causing.

Results:
Line 311-323: You cannot use the term infected. It needs to be changed throughout the manuscript. All correlation data presented in the manuscript should be associated with a r2 value to fully appreciated the power of the correlation.
Thank you for your comment. As described in the point above, we have described the association. We acknowledge in our paper that Campylobacter is likely not the only cause of infection. We have also altered how we determine 'infection' in this paper, as infant stools 'positive or negative' for presumptive Campylobacter (lines 288-293). We do not wish to appear we are over-claiming on the data and as such, we have edited both the long and short titles. We hope these are appropriate and reflect the limitations of the small sample size.
We thank the reviewer for their suggestion on the r2 value; we have responded to this in a comment above.
Line 311: Did you observed difference in Campylobacter prevalence between the two kebeles. Why the authors did not present the CFU data from the direct plating?
Thank you for your comment and questions. We did not separate between the two kebeles. Given the small sample size, we did not believe this would give a meaningful finding. We did not present CFU as this was not the primary aim of the study, we intended to gauge prevalence only.
Line 341-342: This sentence is for the discussion, not the results Apologies but we cannot see how lines 341-342 is intended for discussion.
Line 343-34: This sentence is for the fig legend, not the results Again apologies but we cannot see how these lines should be removed. We have removed two lines, 360 and 364 if these are those that the reviewer referred to.