Aims and objectives
The aim of this systematic literature review was to identify and evaluate studies providing data that could be used to assess the impact of disease and other causes of morbidity or mortality in backyard chickens in low-income and middle-income countries. The following two objectives were formulated following the Population (P), Exposure (E) and Outcome (O) framework (50) (Table 7):
- to assess the economic impact in monetary terms of disease and other causes of morbidity and/or mortality in chicken production of backyard farms in low-income and middle-income countries.
- to assess the productivity losses due to diseases and other causes of morbidity and/or mortality in chicken production at the backyard farm level in low-income and middle-income countries.
Table 7: Components of the objectives according to the PEO framework.
Population (P)
|
Chickens living in backyard farms
|
Exposure (E)/condition of interest
|
Diseases and other causes of morbidity and/or mortality to chickens
|
Outcomes (O)
|
Economic impact and productivity losses
|
Systematic literature review
This systematic literature review was conducted following the PRISMA guidelines (51) (checklist available in S1 Table of the supplementary material).
Data sources and search strategy
Information sources fall under the category of primary sources (first-hand information) and mainly included journal articles and reports. Main data sources include the following databases: PubMed, Medline in OVID, Scopus, Web of Knowledge, CAB direct, the international information system for agricultural science and technology (AGRIS), research in agricultural and applied economics (AgEconSearch), Agricola, Google Scholar, CyberLeninka, China National Knowledge Infrastructure (CNKI), Literatura Latinoamericana y del Caribe en Ciencias de la Salud (LILACS), Thai-Journal Citation Index (TCI), Scientific Information database (SID) and Civilica. Also, governmental, and international organisation reports and journals such as the Catalogue de l’Ecole Inter-Etats des Sciences et Médecine Vétérinaires (EISMV) of Dakar, the Revue d’élevage et de médecine vétérinaire des pays tropicaux (REMVT) and International Livestock Research Institute (ILRI) were also included. Only published literature was considered.
The following steps were designed to access scientific publications and grey literature.
- A basic search of reports and relevant peer-reviewed publications and reports describing backyard chicken farming to identify relevant keywords in English in the title and abstract/foreword was undertaken.
- The identified keywords/phrases were tested using MEDLINE to ascertain where terms co-occur most, after which they were iteratively refined to improve search terms.
- The refined search terms were used within specific databases to undertake a comprehensive search of the peer-reviewed literature.
- Reference lists from selected studies were checked to find additional papers that were not captured on the data search.
- Records were screened to remove duplicate articles.
- Partners from the GBADs network were contacted to access relevant unpublished studies.
The search was constrained from 1981 to 2021 (40 years) and eligibility criteria included 10 languages: Arabic, English, Farsi, French, Hindi, Portuguese, Russian, Spanish, Standard Chinese and Thai. Search strategy included: (economic OR productivity OR financial OR expenditure OR control) AND (cost* OR loss* OR impact OR benefit*) AND (chick* OR broiler* OR hen* OR poultry OR “gallus gallus”) AND (disease* OR death* OR mortality OR nutrition*) AND (backyard OR “family-based” OR smallholder OR traditional OR “low biosecurity” OR “subsistence farmer”). However, this search was adapted to the different databases and languages. The search strings applied in each language and in each database together with the date of the search and the number of hits obtained can be found in S1 File of the supplementary material.
Selection criteria
This review covered “low-income” and “middle-income” countries as classified by the World Bank(52). This review considered all original studies evaluating the impact of chicken diseases and other causes of morbidity or mortality in backyard farms in selected countries. Studies were accepted or rejected based on three additional criteria: (i) whether the study is about diseases and/or other causes of morbidity or mortality in chicken, (ii) whether the study quantitatively assessed the relationship between the disease and/or other causes of morbidity or mortality and productivity or economic impact, (iii) whether the study was conducted in countries included in the previous categories.
This review focused on backyard farms (including free-roaming chicken farms) and low biosecurity commercial farms following the definition by the FAO (sector 3 and 4)(11). A low biosecurity level and contact with other birds and wildlife were considered as essential selection criteria. Flock size was not considered as a search criterion as this may vary depending on flock dynamics and region. Studies covering all diseases and zoonotic diseases associated with chickens were included. Pathogens identified were cross-checked in ENHanCEd Infectious Diseases (EID2) database (https://eid2.liverpool.ac.uk/). Only studies targeting chicken (Gallus gallus domesticus) were included. The “breed” (indigenous or local, commercial and crossbreed) was not considered as selection criteria because chicken flocks in backyard farms may contain crossbreeds or be referred to using local breed names.
Modelling studies and ex-ante assessment studies were not considered. Targeted study design included experimental studies (clinical field trials) and observational studies (case-controls and longitudinal studies). Studies in which diseases were not diagnosed by an animal health professional (e.g., veterinarian, para-veterinarian) or which have not been confirmed using laboratory diagnosis were excluded. Similarly, poor nutritional status should have been assessed by an animal health professional. Studies that did not report the period covered in the data collection were excluded.
Data collection and data extraction
As part of the selection process, titles were first examined to indicate whether each study contained information on productivity and/or economic impact due to diseases and/or other causes of morbidity or mortality in backyard chicken from low-income and middle-income countries. The full abstract was then assessed independently or in combination with the title in those databases that allowed it and then, the full text of selected articles was assessed. The methods and risk of bias were assessed using an appraisal tool in the form of a checklist for inclusion in the literature review. The appraisal tool is available in S2 Table of the supplementary material and it was based on Sargeant et al. (2005) (53). The purpose of this quality assessment tool was to exclude those studies with a poor-quality profile. When ambiguities arose during the screening and assessment process, consensus was reached between at least two of the co-authors (VMG, PR, AS, PT). Data retrieval was conducted by more than one person and guidelines including a data management plan were provided to ensure consistency across the team in this process. When selected studies included several countries with different income classifications, only data from low-income and middle-income countries were extracted. Similarly, when selected studies included several animal species or zoonotic diseases, only disaggregated data corresponding to chicken species were extracted
Outcomes of interest
Primary outcomes include: the effect of disease or other causes of morbidity or mortality on productivity (i.e., production losses) and their economic impact in monetary terms at backyard farm level. Secondary outcomes include: gross margin and enterprise income. Definition of primary and secondary outcomes can be found in Table 8.
Table 8- Definitions of outcomes of interest for the systematic literature review.
Primary outcomes
|
Production parameters:
Definition: All eggs produced during the period of reference independently of their utilization (laying, hatching, consumption). This can be expressed as: total weight of eggs/hen, total number of eggs/hens (54).
Definition: Meat produced from the birds on the farm, including meat produced for home consumption, meat sold, meat used to pay for labour, meat given as payment in-kind, meat given as a gift. This could be expressed as: number of birds slaughtered (heads), average live weight/chicken or total kilograms of live weight (54).
Definition: Number of chickens that died on the farm (excluding culled animals) divided by the total number of chickens present on the farm during the same period and multiplied by 100.
This can be expressed as: daily mortality rate or cumulative mortality rate. Mean and standard deviation of the variable were collected, if available. (55)
Definition: Removal of sick/unproductive birds from the flock. Mean and standard deviation of the variable were collected, if available. (15)
|
Economic impact:
Definition: It refers to the farm level financial impact of a disease or other cause of mortality or morbidity. This includes effects on productivity parameters expressed in monetary terms, expenses associated with disease control and prevention by farmers (e.g. veterinary costs, feed costs, etc).
|
Secondary outcomes
|
Definition: Computed as the enterprise output less the variable costs that are attributed to it over one year. To calculate the total farm gross margin, the gross margins of each enterprise should be summed. The focus was on the chicken gross margin and it was assumed that the one year enterprise output calculation includes changes in the value of the livestock over that period(56).
Definition: Computed as the enterprise gross margin less the fixed costs attributed to that enterprise. However, it is acknowledged that there might be challenges to attribute fixed costs to individual enterprises (56).
|
Meta-analysis of mortality data
Meta-analyses were conducted separately by group of mortality cause, for all groups for which at least two mortality estimates were found. Mortality estimates were classified by groups depending on the aetiology/condition on mortality. These groups included “viral diseases”, “fungal diseases”, “bacterial diseases”, “parasitic diseases”, “bacterial and parasitic diseases”, “bacterial and viral diseases”, “predation”, “cachexia”, “weather” and “injuries”. In those groups that are formed by two different etiological groups (e.g. bacterial and parasitic diseases), the same chickens were diagnosed with diseases from both groups.
In addition to this, mortality estimates were grouped as “epidemic” and “endemic” (57) depending on the epidemiological situation described in the study. To conduct the meta-analysis, the number of dead chickens and the number of chickens in the sample size from which the dead chickens arise in each selected study, were adjusted for 12 weeks considering the period of the data collection reported in each study. The 12 week- period was considered as an approximation of the average length of a production cycle in backyard chickens. This approach allowed us to merge mortality estimates for a production cycle from studies reporting two different epidemiological situations, “epidemic” and “endemic”.
Mortality was interpreted as the incidence of death in a chicken population during a production cycle of 12 weeks. This approach is comparable to incidence risk (IR= A/B), where “A” are the newly affected chickens in a defined period and “B” the total number of chickens at risk of the condition in that population during that period.
This analysis included mortality data reported in the selected studies with and without a comparator. A comparator group was considered a reference group without the condition/treatment of interest. All analyses were conducted in R software (version 4.1.2, 2022-10-31 ucrt). All meta-analyses were undertaken following a random-effect model in logit using the meta package (58). The obtained pooled mortality was then transformed from logit to proportions. When a group of mortality causes had mortality estimates from “epidemic” and “endemic” epidemiological scenarios, the meta-analysis was conducted considering subgroups. Results of the meta-analyses were visualized using forest plots. The asymmetry of the results was assessed using funnel plots and formal statistical testing. Funnel plots were constructed with outcome (mortality, x axis) against the sample size as a measure of variability (y-axis) as recommended for meta-analysis of proportion studies in Hunter et al, 2014 (59). A Peter regression test, which is based on study size, was conducted to test for asymmetry (59) when there were at least ten estimates for the same group of mortality cause. Forest plots and funnel plots were conducted using the meta and metafor package, respectively (60).
Sensitivity analysis
The assumption behind the adjustment of mortality estimates for a production cycle from epidemic and endemic situations is that epidemics occur in every production cycle. We conducted a sensitivity analysis for infectious causes by doing two meta-analyses following the methodology as described in meta-analysis in mortality data, under two different scenarios. In the first scenario, we assumed epidemics occurring once per year. For this, the number of chickens in each sample (denominator) from studies categorized as “epidemic” was multiplied by 4.3 (52/12=4.3). For the second scenario, we assumed epidemics occurring every two years and for this, the number of chickens in each sample (denominator) from studies categorized as “epidemic” was multiplied by 8.6 (104/12=8.6). For both scenarios, the number of dead chickens (numerator) was taken as reported in each study, regardless of the epidemiological situation. For studies categorized as “endemic”, the number of chickens in the sample was taken as reported in each study.
Estimation of economic losses
This was undertaken in four stages.
Literature search of prices
A literature search was conducted to find prices of a broiler chicken and a layer hen at the end of their productive life, and day-old-chicks (DOC) prices for broilers and layers from countries covered in selected studies. Whenever possible, farm gate prices were gathered. The year of the price was differentiated from the year of the published reference when this information was available.
Price conversion in each country
Prices captured in the literature search were converted to the same year and currency. For this, the inflation rate for each country was estimated using the consumer price index (CPI). Index CPI data were obtained from the World Bank (WB) (61). Prices captured in the literature search were adjusted to 2020 levels, used as a common base and called the ‘final year’, using the formula:
The year 2020 was identified as the most convenient to adjust because CPI data were fairly recent, and it was available for most countries. Once the ‘final year’ price was estimated in the local currency unit, it was converted to US dollars using the currency’s exchange rate from 2020 as given by the WB (62). When the local price had been reported in a foreign currency (e.g., 1.5 Euros for a spent hen in Bangladesh in 2014), the price was converted into the local currency for that year using the WB exchange rate dataset as before (62). All prices were assumed to apply to the beginning of the year. When prices were captured from 2021, prices were converted to 2020 values using formula (1) and when prices were captured from 2022, they were assumed to be from 2021 (i.e., no inflation). When prices were captured from 2020, the exchange rate to US dollars was directly applied. Further information can be found in S2 File of the supplementary material.
Estimation of broiler and layer prices in each country
For each country covered in selected studies, the average price of a broiler and a layer hen were computed using the prices that were previously converted. The average price of a broiler chicken was estimated as the mean between the DOC price of a broiler and the price of a broiler chicken at the end of its production life. Similarly, the average price of a layer hen was calculated as the average between the DOC price of a layer hen and the price of a layer hen at the end of its production life. Where possible, the average price of a broiler and a layer were estimated using the DOC price for a chick of the same production purpose. See further details in the supplementary material.
Once the average prices of a broiler and a layer hen were computed for each country, the prices were multiplied by the purchasing power parity (PPP) conversion factor to estimate international prices. The PPP conversion factor of 2020 was extracted from WB(63).
The production purpose (meat/eggs/dual) of the chickens from each study was gathered. For those studies in which the production purpose was not stated a “NA” was assigned. For the studies in which the production purpose was “meat” and “eggs”, the average price of a broiler and a layer hen were assigned. For the studies in which the production purpose was “dual”, the average price of chickens was estimated as the average between the price of a broiler and the price of a layer hen in that country. In the case of studies where the production purpose was captured as “NA”, the average chicken price weighted according to national chicken production. This means that if the national chicken population consisted of 70% broilers and 30% layers, the average chicken price for that country was computed as an appropriately weighted average of the broiler and layer prices.
Calculation of economic losses
Meta-analyses were conducted for each group of mortality cause to estimate the pooled value lost as a proportion of the monetary value of dead chickens over the monetary value of the sample from which dead chickens arise in each study. The meta-analyses were carried out following the same methodology as described in the meta-analysis of mortality data.
The estimated number of dead chickens was computed by multiplying the pooled mortality rate obtained in the meta-analysis on mortality by the total population of chickens in each group of mortality cause.