An exploratory assessment of human and animal health concerns of smallholder farmers in rural communities of Chimborazo, Ecuador

Study area and study population

This study was conducted in the Chimborazo Province of Ecuador, which is predominantly rural. Thirty-eight percent of the province’s population is comprised of indigenous residents making it the province with the highest population of this segment of the population (INEC, 2010). About a third (33.6%) of the residents are employed in agriculture (INEC, 2010), including crop cultivation and animal production. However, the agricultural capacity of communities in Chimborazo varies widely and is heavily influenced by elevation and annual rainfall. Dependence on livestock as a source of income is particularly common in areas >3,500 m above sea level, where few crops can be successfully grown (Córdova, Hogarth & Kanninen, 2018).

Chimborazo is divided into 10 counties (cantones) and further subdivided into parishes (parroquias). Located in the southeast corner of the province is the parish of Cebadas which is a pocket of extreme poverty, with an estimated 97.5% of residents living below the poverty line (Instituto Nacional de Estadistica y Censos, 2014a).

Data collection and management

Sampling was conducted in three communities (San Antonio de Cebadas, Atillo, and Puca Totoras) of Cebadas Parish that have a total of 228 households. Based on provincial data indicating pediatric diarrhea prevalence of 16.1%, monthly pediatric prevalence of 2.0% was estimated (margin of error = .05). Cochran sample size calculations with a finite population correction were performed using Open Epi (Dean, Sullivan & Soe, 0000). A total of 27 respondents were required to achieve a 95% confidence level. Enrollment of additional respondents was desired in order to adequately represent households with children and permit inclusion of multiple predictors in regression analysis.

The three communities included in the study were located at or above 3,500 m above sea level and were accessible by vehicle most or all of the year. Residents of all three communities were predominantly Spanish-speaking or bilingual Spanish- and Quichua-speaking. Participant eligibility criteria included: household resident ≥18 years of age (preferably household head), able to consent to participation, and able to speak Spanish well or very well. Eligibility was assessed for each potential participant before the informed consent process was conducted. The survey was conducted in Spanish, with all participants enrolled and surveyed by a single interviewer in order to ensure consistency in survey conduct and response assignation. The informed consent document was read to participants and a copy provided to each for their records. Participants were encouraged to ask questions and were asked to verbally acknowledge their understanding of the study procedures before signing the informed consent document retained by the investigator.

Study participants were recruited during July/August 2019. Sampling of households was conducted in coordination with the president of each community. Consecutive houses were approached until an eligible individual was identified and agreed to participate. The survey was immediately conducted on-site. The adjacent neighbor of a participating household was skipped and then the process repeated, approaching consecutive households until identification and enrollment of another eligible individual. Enrollment was limited to a single adult resident in each household. Households were not re-approached if adult residents present at approach declined to participate.

Data were collected through structured interviews which gathered information on demographic characteristics of household members, land ownership and characteristics of properties owned, and the role of remittances as a portion of household income. A range of questions were included regarding the number and type of household-owned animals and details of animal interactions engaged in by either the respondents or pediatric household members. Respondents were asked about episodes of diarrheal illness experienced by household members and household-owned animals in the 30 days prior to interview. A diarrheal episode was defined as ≥3 watery stools within a 24-hour period. For participants reporting diarrhea, follow-up questions gathered details regarding how the illness was addressed or treated. Finally, information was elicited on health concerns as well as suggestions regarding potential interventions to address them. Data were collected electronically using Epi Info 7 (Centers for Disease Control and Prevention, Atlanta, GA), which was also used for exploratory review.

Descriptive analysis

Analyses of the survey data were conducted in SAS 9.4 (SAS Institute, Cary, NC, USA). Variables collected were largely dichotomous or categorical in nature. Descriptive analyses included demographic characterization of households and assessment of associations of household characteristics with health concerns, pediatric health status, and health behaviors. The principal dependent/outcome variables assessed were reported pediatric and livestock diarrhea. Percentages of responses in each level of categorical variables and their 95% confidence intervals were also computed.

Chi-squared analyses were used to assess the relationship between the dependent variables (pediatric and livestock diarrhea) and categorical independent variables. Independent variables assessed included respondent factors (educational attainment), household factors (total household residents, running water in home, indoor toilet, household receipt of remittances), smallholding factors (property adjacency, size, running water availability, running water type, animal enclosures, animal waste management system), and pediatric factors (consumption of unpasteurized milk, consumption of untreated water, exposure to recreational water sources, participation in animal butchering, participation in milking dairy cattle, and interaction with animal waste). Pediatric diarrhea was also assessed for association with livestock diarrhea.

Univariable and multivariable models

Potential predictors of pediatric and livestock diarrhea were first identified by fitting a series of univariable logistic regression models. Independent variables with univariable p-values ≤.20 were considered for assessment in the multivariable models. Subsequent to the univariable analyses, two multivariable logistic regression models (pediatric and livestock diarrhea) were fitted using a backwards stepwise approach with a critical p-value of .05. Since pediatric diarrhea was a relatively rare outcome, the Firth logistic model was used to investigate its predictors in order to correct for small-sample bias associated with maximum likelihood estimation of the ordinary logistic regression model. However, an ordinary logistic model was used for livestock diarrhea. Confounders were assessed by running the models with and without the suspected confounder and evaluating the changes in coefficients of the variables remaining in the models. Changes of 20% or greater indicated confounding and hence the variable was retained in the model, regardless of significance level. Odds ratios (ORs) and their 95% confidence intervals (95 CIs) were calculated for all variables included in the final models. Goodness-of-fit of the logistic regression models was assessed using the Hosmer-Lemeshow test.

Ethics approval

Approvals for study implementation were granted by both the University of Tennessee Institutional Review Board (IRB Number 19-05199-XM) and the Pontifical Catholic University of Ecuador Committee for Ethics in Human Subjects Research (Approval Number 2019-95-EO).

Demographic and household characteristics

A total of 58 respondents participated in the study. Self-identified race of the respondents was principally Indigenous (65.5%, 38/58) (Table 1). Many respondents (56.9%, 33/58) reported their primary language to be Spanish, but more than a quarter (27.6%, 16/58) spoke both Spanish and Quichua. Most (82.8%, 48/58) of the respondents owned their land, while the rest either farmed on property belonging to a family member (13.8%, 8/58) or rented the smallholding (1.7%, 1/58). A total of 70.7% (41/58) of the respondents reported having completed primary school education while 3.4% (2/58) reported having completed secondary school or higher level of education.

The median age of respondents was 39.5 years (interquartile range: 32, 50). Although total household size ranged from 1 to 11 members, the median household size was 4 (interquartile range: 3, 6). Most households comprised a bi-generational family unit including either middle-aged adults with children or seniors with unmarried adult children (Table 1). Multi-generational households were also noted, although extended families sharing resources and domestic responsibilities more commonly inhabited adjacent houses, rather than a single-family home. Sources of income outside the household were reported by 3.4% (2/58) of the respondents (Table 1). Both households reporting such remittances received less than 10% of total household income from the non-resident contributors.

A high proportion (80.3%, 122/152) of adult household residents were reported to be employed in agriculture, representing 94.8% of all households (Table 1). Regardless of employment type or status, all respondents reported either cultivating crops or raising animals on one or more plots of land. While these varied widely in size, from a micro-parcel of 200 square-meters (0.02 hectares) to a property of 20 hectares, a large proportion (75.9%, 44/58) of respondents reported that their principal smallholding was adjacent to their home (Table 1). The median size of adjacent smallholdings was 2.0 hectares (interquartile range: 0.64, 3.0). A smaller proportion of respondents had non-adjacent smallholdings, with a median size of 2.5 hectares (interquartile range: 1.0, 4.0). For those respondents with a primary smallholding site non-adjacent to their home, the median travel distance was 1 Km (interquartile range: 0.5, 2).

Animal ownership and husbandry

The respondents reported a number of animal species reared on the smallholdings (Table 2). Nearly all respondents kept cattle (98.3%) and pigs (97.8%) (Table 2). Respondents’ principal revenue source was dairy or mixed dairy/beef production. Eighty-one percent of the respondents reported keeping cattle on the holding contiguous to the home while 17.3% kept cattle on holdings not contiguous to the land where the home was built. Only large animals (cattle, donkeys, and horses) were kept on non-adjacent smallholdings. Numbers of cattle owned varied widely among respondents, from single cow/calf pairs to multiple head of cattle in beef operations. The median number of cattle owned was 8 (interquartile range: 6, 10) (Table 2). Nearly three-quarters (72.4%, 42/58) of the respondents reported that their smallholdings did not have animal enclosures. All respondents who reported having enclosures used barbed-wire (93.8%, 15/16) or electric (6.2%, 1/16) fencing, while those without fencing reported tethering their livestock to pasture or allowing them to roam free. None of the respondents reported having a barn or similar structures for livestock.

Sanitation and waste management characteristics

Fewer than half (41.4%, 24/58) of the respondents reported having indoor running water. The largest proportion of respondents had only outdoor toilets (36.2%, 21/58), while 22.4% (13/58) reported having only indoor toilets and 6.9% (4/58) had both indoor toilets and pit latrines (Table 3). About a quarter (27.6%, 16/58) of the households had only pit latrines while 6.9% (4/58) had neither toilets nor latrines.

Most (75.9%, 44/58) of the smallholdings were unimproved, lacking waste management systems for both human and animal waste. Just under half (48.2%) of the smallholdings had on-site irrigation systems while 72.4% (42/58) had running water via either piped municipal water (17.2%) or surface water (55.2%) sources (Table 3). However, commensurate with the low proportion of piped water on smallholdings (17.2%, 10/58), very few respondents reported having a toilet (3.6%, 2/55) or latrine (5.4%, 3/55) onsite at the smallholding. In the absence of sanitary sewer systems, most smallholdings relied on improvised human waste management techniques, including burning/burying (29.3%, 17/58), covering with grass (22.4%, 13/58), and using rivers or streams for defecation (1.7%, 1/58) (Table 3). Similarly, respondents reported very limited use of animal waste management systems, with just two respondents (3.5%) reporting the use of a wastewater holding pond. Consistent with the proportion of respondents reporting no waste management plan (84.5%, 49/58), most smallholdings (74.1%, 43/58) relied on regular spreading of animal manure on pasture (63.8%, 37/58) or crops (10.3%, 6/58) to manage animal waste.

Prevalence of diarrhea

In the preceding 30 days, 12.1% of the respondents had experienced diarrhea, yet very few (28.6%, 2/7) of these had sought care for their illness (Table 4). In 15.2% (7/46) of households in which children lived, one or more children had diarrhea in the previous month. Children with diarrhea ranged in age from 1 to 13 years of age with a median age of 5 years (interquartile range: 1, 11). Post hoc power of 0.936 was calculated for detection of the observed pediatric diarrhea prevalence. There was no significant (p = .48) difference in the median number of days of diarrhea experienced by children and adult respondents. A much higher proportion (71.4%, 5/7) of respondents sought care for the child or children that experienced diarrhea than for their own diarrhea. At least one animal in 55.2% of the households was reported to have experienced diarrhea in the previous 30 days. Despite the high frequency of diarrheal illness among livestock, fewer than half of these respondents (40.6%, 13/32) reported having sought veterinary care for their animals because of the diarrhea. Respondents indicated that, when sought, animal care was usually obtained from more skilled community members or para-veterinary professionals.

Potential sources of pathogen exposure

Exposures to potential sources of gastrointestinal pathogens were widespread. The percentage (93.5%) of children that was exposed to food- and water-borne potential sources of pathogens (drinking unpasteurized milk, drinking untreated water, and bathing or playing in surface water) was not significantly (p = .61) different from that of adults (91.4%). However, the percentage of children (80.4%) exposed to blood-borne and fecal transmission exposures (butchering animals, milking dairy cattle, and handling animal waste) was significantly (p < .0001) lower than that of adults (100.0%) (Table 4). The most common exposure categories among adults were (in rank order): milking dairy cattle (98.3%, 57/58), handling animal waste (87.9%, 51/58), and drinking untreated water (74.2%, 43/58). The categories of exposure most frequently reported for children were the same, but varied in rank: drinking untreated water (80.4%, 37/46), milking dairy cattle (76.1%, 35/46), and handling animal waste (63.0%, 29/46) (Table 4).

Reported health concerns

Respondents reported specific concerns regarding their own health more frequently than any other personal health concern, with 55.2% (32/58) naming a particular disease or condition which they had experienced or about which they worried (Table 5). Specific concerns regarding family health were reported by 58.6% (34/58) of the respondents, but relatively few (29.4%, 10/34) of the family concerns specified pediatric health issues. One-fifth of respondents expressed concern regarding healthcare access for themselves (20.7%, 95% CI [12.3–32.8]) but only half as many reported such concern for family members (10.3%, 95% CI [4.8–20.8]). However, among households with cases of pediatric diarrhea, the proportion of respondents expressing concern about access to healthcare rose to 42.9% (3/7, 95% CI [15.8–75.0]). Concern regarding stress or mental health of either self or family members was reported by just 10.3% (6/58) of the respondents. In comparison, concern for physiological stress of animals was reported by 60.3% (35/58, 95% CI [47.5–71.9]) of respondents, ranking as the most frequently cited animal health concern (Table 5). Respondents listed a range of potential causes for illness in their livestock, including both general categories of disease and specific pathogens, as well as climatic and environmental influences (e.g., “altitude sickness”). The respondents reported high levels of concern regarding the impact of health concerns, particularly stress, on their animals’ productivity. However, households reporting animal diarrhea did not demonstrate a greater level of concern regarding healthcare access than all households (Table 5).

Predictors of pediatric and animal diarrhea

Results of both the univariable and final models investigating the associations between potential predictors of pediatric and animal diarrhea are shown in Table 6. Variables that had potential univariable associations (critical p ≤ .20) with pediatric diarrhea were: household with indoor running water (OR = 4.0; 95% CI [0.7–23.3]; p = .12), diarrhea in any household-owned animal (OR = 5.7; CI [0.6–51.9]; p = .12), any child in household consumes raw milk (OR = 3.4; CI [0.6–18.9]; p = .16), and cattle density per hectare at the smallholding (OR = 1.9; CI [0.9–4.0]; p = .11). None of the variables had a significant association with pediatric diarrhea at a critical p ≤ .05 and, therefore, none were retained in the final model.

Variables that had potential univariable associations (critical p ≤.20) with animal diarrhea were: animal enclosure on smallholding (OR = 3.3; CI [0.9–11.9]; p = .07) and animal waste management plan for smallholding (OR = 8.7; CI [1.0–75.0]; p = .049). The latter was the only significant predictor in the final animal diarrhea model (Table 6).

Strengths and limitations

Studies of health concerns of rural smallholding families in Latin American countries and other LMICs are scarce and therefore the present work provides a unique insight into the issues in these communities. The structured interview format elicited numerous topics of health concern that can be addressed in future community interventions. Additionally, this study considers both demographic factors and animal management practices associated with diarrhea prevalence, which have been poorly described in the literature to date. Use of the Firth logistic model is an analytical strength of the study as it corrects for small sample bias associated with ordinary logistic models. However, the study is not without drawbacks. The findings are based on respondent reports and therefore subject to recall biases, particularly related to reports of pediatric diarrhea and medical care sought. This could have been addressed by asking additional questions of the main caregivers of children. However, to minimize respondent burden (which would in turn impact response rate), the interview time had to be limited and questions prioritized. Study findings should be interpreted with caution, given the relatively low number of respondents involved in the study. Future studies should be conducted across a broader geographic area, implementing a multistage randomized sampling scheme to improve representativeness and generalizability. Confirmation of diarrheal illness through medical record evaluation or specimen collection should also be sought. These limitations notwithstanding, the findings of the current study contribute to the otherwise sparse body of literature on the health and livelihoods of rural smallholder Ecuadorian farmers and provide useful information to guide future larger studies of these populations in Ecuador and other LMICs.

Conclusions

The levels of diarrheal illness among the children of smallholder families in this study is similar to national and international estimates. Ongoing surveillance is necessary to permit accurate identification of diarrheal illnesses among both household members and livestock and develop more detailed incidence estimates. A natural next step is to evaluate specific gastrointestinal pathogen prevalence and assess the role of livestock in specific pathogen disease burden among smallholder farmers and their children. This would help guide the development of economically feasible interventions that are technologically appropriate for the region. Mitigation of waste exposure through improvement of sanitation and waste management infrastructure is recommended to reduce the burden of diarrhea both in humans and livestock. Future studies must also involve community members in the design and implementation of educational interventions aimed at decreasing human and animal diarrheal diseases.