Artificial Insemination as Risk-Averse Behavior and Bovine Tuberculosis in Madagascar

The increasing livestock demand in developing countries raises concerns about endemic zoonotic diseases, such as bovine tuberculosis (bTB). Farmer behaviors and social contexts drive transmission, necessitating behavior change interventions. An epidemiological field survey was conducted to assess artificial insemination (AI) use and bTB incidence among dairy farmers in Madagascar, combined with socioeconomic interviews with 114 farmers in Malagasy from 2021 to 2022. A probit regression model analyzed farm-level bTB infection status as the dependent variable. AI was used by 14.1% (15 households), with 93.3% (14 households) employing natural bull mating. The herd-level bTB prevalence was 41.2% (50 cows), and 41.6% (47 households) owned at least one bTB-positive cow. AI use increased the risk of bTB, likely due to inadequate cleaning of AI equipment. Despite AI’s potential to improve productivity, improper implementation may increase the risk of bTB. These findings underscore the need for enhanced farmer education on bTB prevention and proper sanitation of AI equipment.

Introduction

The increasing demand for livestock products in developing countries has led to the growing prevalence of endemic zoonotic diseases (zoonosis) in rural areas. Zoonotic diseases, which are transmissible between animals and humans, have significant economic and public health challenges and are considered externality problems in economic terms.

Cattle are Madagascar’s most significant livestock by population; however, the annual per capita meat and milk consumption remains far below the mainland African average. The country imports substantial quantities of milk, reaching up to 11,000 metric tons annually. In response to domestic demand genetic improvement of dairy cattle using artificial insemination (AI) is of critical importance. The use of AI in developing countries has the potential to maximize the genetic potential of indigenous breeds adapted to local environments. However, improving milk production and reproductive performance through AI depends not only on the selection of high-quality semen combinations, but also on factors such as the state of infrastructure, farm management practices, and the skills of veterinarians.

Although Madagascar is free from economically devastating diseases, such as foot-and-mouth disease, bovine tuberculosis (bTB), a zoonotic disease caused by Mycobacterium bovis, is endemic. Beyond its health risks, bTB reduces cattle weight gain and milk production [1]–[4], leading to economic losses.

A 1969 nationwide survey revealed a 21% prevalence of bTB in cattle [5], which increased to 30% in 2001, as reported by Quirin et al. [6]. Although only 1.3% of human TB cases in Madagascar are caused by M. bovis [7], most bTB research in the country has focused on veterinary epidemiology, leaving gaps in our understanding of its spread from behavioral or social perspectives. It is possible that the risk of zoonotic bTB transmission is underestimated in Madagascar.

Furthermore, social norms in developing countries play a critical role in the spread of infectious diseases in livestock. Studies have revealed that this custom contributes to the spread of infectious diseases in pig farming [8], [9], highlighting how social norms influence the transmission of livestock diseases in developing countries. In Madagascar, the traditional custom of “Fihavanana” plays a key role in societal interactions [10], [11]. However, modifying farmer behavior within social frameworks remains a challenge, despite its potential to mitigate the impact of such diseases. Research in Peru has identified behavioral and structural barriers to implementing rabies prevention strategies [12], whereas studies in the Philippines have shown that behavioral barriers to tuberculosis can be addressed by promoting the perceived benefits of seeking care [13].

Analyzing such social behavioral barriers through the lens of behavioral economics can effectively contribute to the prevention of disease transmission. Behavioral economics, which combines economics and psychology, has recently gained attention for its application in public health [14], [15]. Despite the critical role of human behavior in the spread of infectious diseases in livestock, most studies have focused on veterinary medicine and animal science [14]–[20], with limited research in veterinary economics [21]–[23]. Addressing these research gaps requires a focus on farmer behavior to inform more effective disease control strategies.

This descriptive study aimed to quantitatively analyze the social and behavioral characteristics of surveyed farmers to explore the factors contributing to disease outbreaks and propose strategies to address behavioral barriers. This research was conducted in three phases. First, this study investigated the increasing adoption of AI technology for dairy cattle and examined its role in bTB transmission at the farm level. Second, the incidence of bTB among dairy farmers in Madagascar was determined through an epidemiological field survey conducted in 2022. Third, based on the findings regarding AI utilization and bTB incidence, the behavioral characteristics of farmers were identified, and strategies to overcome social and behavioral barriers to bTB prevention were proposed.

Materials and Methods

The study was approved by the CENA (Approval No. 03-24; Approval Date September 27, 2024) and conducted in Antananarivo, located in the Analamanga region of Madagascar. This region forms a part of the dairy triangle, which produces 80% of the country’s milk. A socioeconomic survey was conducted through interviews with 114 farmers in Malagasy from April to July 2022, followed by additional interviews in the same area between March and October 2023. All participants provided informed consent to participate in the study. The respondents were randomly selected from a list of farmers affiliated with dairy boards. Furthermore, 155 blood samples were collected from dairy cows and tested using an Antigen Rapid Bovine TB Ab Test Kit (GENTAUR, Kampenhout, Belgium) to determine the prevalence of bTB. Our study, based on 114 households, may not fully represent all dairy farmers across Madagascar. However, the study area belongs to the so-called “dairy triangle,” which is the most important milk-producing region in the country, and thus reflects the typical farming practices of Malagasy dairy farmers. The Antigen Rapid Bovine TB Ab Test Kit employs a solid-phase chromatographic immunoassay to qualitatively detect M. bovis antibodies in serum or plasma. We interpret the kit results as serological screening indicators rather than definitive diagnostic confirmation.

Given the small-scale livestock farming system in the study area, with an average of 2.25 cattle per household, blood samples were collected from at least one multiparous cow per household. The number of sampled cows ranged from one to three, depending on the household scale. A single cow was sampled from 81, two cows from 25, and three cows from eight households. This sampling strategy provided insights into bTB prevalence at both the herd and household levels, considering the characteristics and behavior of farmers.

Probit analysis was used to examine the association between bTB infection status at the farm level and farmers’ characteristics. The dependent variable in the probit model represented farm-level bTB infection and was categorized as bTB positive (1) or bTB negative (0).

The probit model can be specified as shown here:

$$\begin{array}{r}{Y}_i=\beta X_i+\epsilon \end{array}$$

with

$$\begin{array}{r}{Y}_i=\{\begin{array}{l}1,ifbTB-positive\\ 0,otherwise\end{array}\end{array}$$

where $Y_i$ is the bTB infection status of the cattle owned by farmer $i$. $X_i$ is the vector of independent variables influencing the infection status, $\beta$ is the vector of parameters to estimate, and $\epsilon$ is the error term [24]. Unlike prior veterinary studies that focused on risk factors at the herd level, this study analyzed bTB incidence at the household level, emphasizing farmer characteristics. The explanatory variables included farmer characteristics and the adoption of livestock production techniques. The questionnaires assessed factors influencing the diffusion of AI, including women’s decision-making in farm management, understanding of bTB, and risk attitudes. Previous studies have highlighted the significant influence of farmers’ risk attitudes on the adoption of new agricultural technologies in developing countries [25]–[29]. This study quantified risk attitudes, categorizing farmers as risk-takers or risk-averse, using Binswanger’s method [30], which involved a questionnaire test that presented alternative gain scenarios for farmers.

Results

Table I presents the characteristics of the surveyed farmers. Each household had an average of 2.25 dairy cattle, indicating small-scale management, with dairy income constituting approximately 50% of total household income. Most dairy farmers (63%) utilized extensive grazing systems on communal land.

Table I. Characteristics of Farmers

Variables	Mean	Std.	Min.	Max.
Farmer age (year)	49.20	12.09	25.00	72.00
Family size	4.43	1.41	1.00	6.00
Sex of the farm head (1: male, 0: female)	0.92	0.27	0.00	1.00
Years of experience in dairy farming	16.82	9.90	0.17	50.00
Number of cattle per farm (head)	2.25	1.14	1.00	5.00
Education (higher than secondary school = 1, other = 0)	0.39	0.49	0.00	1.00
Farming system (extensive = 1, other = 0)	0.63	0.48	0.00	1.00
Artificial insemination (AI) (AI alone, AI and bull = 1, bull = 0)	0.13	0.34	0.00	1.00
Daily milk production (L/day/cow)	8.99	5.08	1.30	27.00
Proportion of income from dairy (%)	50.96	18.20	0.00	100.00
Proportion of income from off-farm activities (%)	10.75	16.28	0.00	60.00
Land area (acre)	95.80	142.85	0.00	900.00

Table II highlights the extent of bTB knowledge among farmers. Only five households (4.4%) indicated “I know a lot,” whereas most households (58.8%) responded with “I know a little.” Regarding “True” or “False” questions on key bTB characteristics, 61 households (59.2%) correctly identified that bTB “Affects humans,” and 62 (60.2%) recognized “Persistent cough” as a symptom. However, misunderstandings were evident in identifying “Fever in cattle” and “Swelling of the udder” as bTB characteristics, with 22 (21.4%) and 42 (40.1%) households, respectively, providing incorrect responses.

Table II. Bovine Tuberculosis (bTB) Prevalence in the Analamanga Region

Number of tested heads	Herd-level prevalence2)	Number of research farms	Farm-level prevalence3)
155 heads	41.2% (50 heads)	114 farms	41.6% (47 farms)

AI was employed by 15 households (14.1%), with 14 households combining AI with natural bull mating and only one household relying exclusively on AI. Despite the introduction of AI in Madagascar in 1990, the adoption rate has remained limited. The higher cost of AI services, approximately $20 compared to $5 for bull mating, is considered a significant barrier to their widespread use (service fee data were obtained from interviews conducted in March 2023).

Table III summarizes the analysis of 155 blood samples. The herd-level prevalence was 41.2% (50 cattle), with 32.3% of the households (47 households) having bTB-positive cows (one or more) at the farm level.

Table III. Farmers’ Knowledge of bTB¹⁾

Variables	Category of response	Number of farmers (%)
a) Do you know bTB?	I don’t know	11 (9.6)
	I know a little	67 (58.8)
	I know about it	31 (2.7)
	I know a lot	5 (4.4)
b) bTB characteristics2)	Affects human (true)	61 (59.2)
	Fever in cattle (false)	22 (21.4)
	Udder swelling (false)	42 (40.1)
	Normal weight gain (false)	69 (67.0)
	Persistent cough (true)	62 (60.2)

Table IV lists the explanatory variables used for the probit analysis, and Table V presents the results. Three variables significantly influenced the occurrence of bTB at the farm level: the number of veterinarian visits per cow per year (Vet. visits), the farming system, and AI use (Table V, left column).

Table IV. Description of Variables

Variable	Definition	Mean	Std.
bTB	Farm has at least one positive bTB case = 1	0.41	0.49
	Farm has no positive bTB case = 0
Farm years	Years of experience in dairy farming	16.82	9.9
Age	Farmer age (year)	49.20	12.09
Parents	Farm was succeeded by parents = 1	0.68	0.47
	Farm was started by the present head of farmer = 0
Number of cattle	Number of cattle per unit of land (head)	4.43	1.41
Vet. visits	Number of veterinary visits per cattle past year	1.84	1.71
Education	Farmer education higher than secondary school = 1	0.39	0.49
	Farmer education secondary school or lower = 0
Land area	Farm area owned by the farmer (acres)	3.38	1.48
Farming system	Extensive rearing system = 1	0.63	0.48
	Intensive and semi-intensive rearing system = 0
Al	Farmers use AI, AI and bull = 1	0.14	0.34
	Farmer uses bull only = 0
Decision	Female is the main farm decision-maker = 1	0.66	0.48
	Male is the main farm decision-maker = 0
Risk attitude	Risk-taker = 1	0.34	0.48
	Risk-averse = 0

Table V. Probit Regression Analysis of bTB Prevalence and AI Adoption Factors

Dependent variables	bTB			AI
	Coefficient	Std.		Coefficient	Std.
Independent variables
Parents	0.504	0.287		−0.352	0.358
Age	−0.019	0.099		0.088	0.123
Age2	0.001	0.001		−0.001	0.001
Number of cattle	0.788	0.718		−4.190	1.679	**
Vet. visits	−0.224	0.075		0.114	0.095
Education	−0.315	0.265		0.447	0.350
Farming system	−0.988	0.278	***	1.004	0.479	**
Decision	−0.429	0.293		0.820	0.382	**
Risk attitude	−0.315	0.273		−0.698	0.363	*
AI	0.881	0.375	**
Const	0.813	2.475		−3.765	3.189
Log-likelihood	−63.506			−34.902

Additionally, a probit analysis was conducted to identify factors associated with AI adoption, and the results are shown in Table V (right column). The “Number of cattle” variable exhibited a significant negative coefficient, indicating that an increased cattle density per unit area correlated with a lower likelihood of AI adoption. The “Farming system” variable had a positive coefficient, suggesting that farmers utilizing extensive farming systems with local breeds were more inclined to adopt AI.

The “Decision” variable had a positive coefficient, indicating that women’s involvement in farm management decision-making increased the likelihood of AI use. Conversely, the “Risk attitude” variable had a negative coefficient, indicating that risk-taker farmers were more likely to adopt AI, whereas risk-averse farmers were less inclined to use this technology.

Discussion

The “Vet. visits” had a negative coefficient, indicating that an increase in veterinarian visits per cow per year was associated with a reduced likelihood of bTB occurrence. Similarly, the “Farming system” variable also demonstrated a negative coefficient, suggesting that farmers practicing extensive cattle farming, characterized by outdoor grazing, had a lower likelihood of bTB occurrence. Regular veterinary visits in the study area enhanced cattle observation and management, thereby reducing the likelihood of bTB infection. Conversely, the likelihood of bTB infection was higher in intensive farming systems where cattle were kept indoors. AI use had a significantly positive coefficient, indicating a higher likelihood of bTB occurrence on farms using AI.

The prevalence rate observed in this study (41.2% at the herd level and 32.3% at the farm level) was relatively higher than the herd-level prevalence rates of 20 and 30% reported in other studies [20], [6].

The higher bTB occurrence among farmers using AI may be attributed to the reuse of needles and inadequate disinfection of AI equipment. Interviews conducted in March 2023 with local technicians offering AI services confirmed that while AI equipment was properly disinfected before use, the reuse of injection needles was a common practice. This finding aligns with the result of this study and previous research [31]–[33], which suggested that bTB infections can be linked to the quality of semen used for AI and the inability of veterinary assistants to maintain sanitary conditions during the procedures, thereby facilitating disease transmission. Therefore, attention should be directed towards disinfecting AI equipment and improving semen quality. This issue warrants further investigation in future studies.

Moreover, the finding that a higher number of cattle per unit area was associated with a lower likelihood of AI adoption suggests that the high cost of AI services per insemination, in comparison to bull mating, may hinder AI adoption. According to a local research institute, extensive farms are more likely to maintain local breeds and farmers aiming to improve productivity and secure superior genetics for their cattle are more likely to adopt AI. This information, gathered from a meeting with a local research institute in March 2023, highlights the economic considerations that influence AI adoption in the region.

AI allows for the insemination of numerous cows through semen dilution, which increases the number of high-milk-producing cows, particularly in dairy farming. A survey in Madagascar showed that farmers using AI had an average milk production of 14.1 L/day/cattle, compared to 8.3 L/day/cattle for those not using AI. These differences were statistically significant.

Interviews conducted in March 2023 with female farmers who did not own bulls revealed that one reason for adopting AI was concern about introducing diseases from outside the farm through bull mating. Our model incorporated two variables directly derived from behavioral theory: risk attitude and women’s decision-making in farm management. Both the “Decision” and “Risk attitude” variables showed similar trends, with farmers seeking to minimize the risks of livestock diseases more likely to adopt AI. The findings highlight the necessity of socio-cultural dimension in shaping livestock management practices.

This adoption contributed to higher milk production in households using AI. However, the analysis also revealed a paradox, indicating a higher likelihood of bTB infection in households that have adopted AI.

As this study relies on cross-sectional data analysis, the results primarily demonstrate an association rather than a definitive causal link between AI use and bTB infection. However, qualitative interviews with AI technicians revealed practices such as needle reuse and insufficient disinfection, which suggest a plausible causal mechanism.

Conclusion

AI, a new technology in Madagascar, is likely to be adopted by risk-averse farmers in the context of behavioral economics. Paradoxically, the adoption of AI behavior increases the risk of bTB infection. In Madagascar, milk and dairy consumption are essential for improving the livelihoods of farmers and enhancing the nutritional status of rural communities by boosting cattle productivity. Therefore, managing the consumption and quality of milk and dairy products is crucial.

These results also highlight the importance of properly disinfecting AI equipment and improving farmers’ understanding of bTB.

Given that bTB is a zoonotic disease, infection can spread to other cattle and affect human health. As defined by economics, this presents an externality problem that necessitates effective public health measures. Therefore, At the same time, as the demand for livestock products continues to expand in low- and middle-income countries, research on bTB remains limited [34], [35]. bTB control has therefore been recognized as a shared challenge across developing countries. Communicating these findings to local agencies is crucial for implementing corrective action.

Acknowledgment

We acknowledge the Malagasy Dairy Board (MDB) is acknowledged for providing the list and contact information of dairy farms for the field survey.

Conflict of Interest

The authors declare that they do not have any conflict of interest.