Knowledge, attitude and practice of farmers towards goat mastitis through item response theory approaches

Abstract

Goat production, which is largely managed by resource-poor farmers, is hindered by limited infrastructure, education, and low awareness of prevalent diseases, management, and cultural perceptions. Goat mastitis is often overlooked due to poor routine diagnostic culture, low farmer awareness, and management-related risk factors. The objectives of the study were to assess the knowledge, attitudes, and practices (KAP) levels of herders on mastitis in goats. A cross-sectional study with simple random sampling was conducted in selected districts of northeastern Amhara region, Ethiopia, from September 2024 to May 2025. Data was collected from 275 participants using a semi-structured questionnaire. Among the 275 survey participants, 30.7% reported mastitis incidents. The average scores for knowledge, attitudes, and practices related to mastitis were 47.8%, 69.4%, and 46.2%, respectively. The majority of the respondents did not attend school, and their knowledge about the causes of mastitis, zoonotic diseases, and subclinical mastitis was significantly associated with district and age demographics. In conclusion, herders lacked desirable knowledge and practices, and zoonotic risks related to mastitis. Tailored training on the causes, and the zoonotic risk, control and prevention mastitis was suggested for goat farmers. Furthermore, research on bacterial species and resistance factors identification are recommended.

Supplementary Information

The online version contains supplementary material available at 10.1186/s12917-026-05342-w.

Introduction

Goats are raised primarily by poor farmers in Asia and Africa, who have limited access to infrastructure, finance, education, and extension services. Herdsmen are characterized by a lack of understanding of major economically important and zoonotic diseases, as well as they are ineffective in control and prevention practices. Goat milk production in the global dairy market accounts for only 2%, which falls short of its potential [25]. Possible reasons for such a deficit are negative consequences of prevalent diseases, as well as genetic, nutritional, and management factors. Moreover, the purpose of breeding, beliefs and perceptions about the utility and marketing of goat milk contribute to this low production [7, 19].

Mastitis is one of the most important economic diseases of mammals, including goats, and impedes substantial milk production and animal welfare. The disease is better studied in cows and buffaloes than in goats and is mainly bacterial in origin, but it could also be due to fungal, viral, parasitic, or traumatic causes [25]. There are two broad categories of mastitis based on manifestations: clinical mastitis (CM) and subclinical mastitis (SCM). CM is the most severe and primary cause of unwanted culling of highly prolific dairy animals, whereas SCM is the prime reason for a decline in milk production, which is often unnoticed or misestimated. According to [13], approximately 75% of economic impairment is due to SCM.

Mastitis in goats, particularly subclinical mastitis (SCM), is often misidentified, unreported, or underestimated due to several factors, including the low sensitivity and specificity of diagnostic tests, as well as limited farmer awareness and motivation, since clinical signs are frequently not visible and farmers may not recognize the problem [19, 25]. Abrasive bedding, insects (flies), chemicals that disrupt teat skin integrity are among the risk factors of mastitis outbreaks. Additionally, host-related factors such as age, breed, lactation stage, and parity determine the susceptibility of goats to mastitis and influence the severity of the disease [10, 22].

Mastitis poses a threat to public health in rural communities largely reliant on goat milk. There is a lack of evidence on goat herders’ awareness and management practices toward mastitis [5]. Appropriate knowledge, attitude and practice (KAP) levels are essential for herder proficiency in controlling mastitis and other infectious diseases in goats, as well as ensuring milk safety for public consumption [3].

Measuring herders’ proficiency in the control and prevention of mastitis and other infectious diseases requires appropriate analytical approaches. Item Response Theory (IRT) also known as the latent response theory has recently been introduced as a robust method to explain the relationship between latent traits (unobservable characteristic or attribute) and their manifestations (i.e. observed outcomes, responses or performance) [18, 24]. By applying IRT to a KAP survey, researchers can identify which items are most informative, assess the reliability of the scale across different levels of the latent trait, and obtain more valid and nuanced estimates of herders’ knowledge, attitude and practices. Therefore, this study aimed to assess the knowledge, attitudes, and practices (KAP) of goat owners regarding mastitis in their goats.

Materials and methods

Description of the study areas

The study was conducted in the Waghimra areas of the Amhara Regional State, Ethiopia, which is located between 12° 23′N to 13° 16′N latitude and 38° 44′E to 39° 21′E longitude. Farmers in the Waghimra Zone have a long-standing tradition of raising goats, mostly the Abergelle goat breed, which is the primary source of livelihood in the area. According to studies, approximately 80%−98% of households keep goats alone or with other livestock [4, 20]. Milk production is the second most preferred purpose of breeding goats [2]. The study was conducted in three districts: Sekota, Abergelle, and Ziquala. The total livestock population in the study area is shown in Table 1.

Table 1.

Distribution of livestock species by district in Waghimra zone

Livestock species	Sekota	Abergelle	Ziquala	Sub total
Goat	98,788	118,797	117,516	335,101
Sheep	42,141	55,413	37,809	135,363
Cattle	79,501	50,992	57,101	187,594
Equine	21,847	10,616	11,912	44,315
Poultry	69,041	52,785	23,318	166,991
Total	292,313	288,603	247,650	828,566

(Source: Waghimra Livestock and Fishery Office; 2023/24)

Study design and sampling approaches

A cross-sectional study was conducted from September 2024 to May 2025. The study population comprised households that were engaged in goat rearing during the study period. A multistage top-down sampling approach was used, with simple random sampling at the household level. Three districts (Sekota, Abergelle, and Ziquala) were chosen purposively on the basis of logistics, accessibility, and goat-keeping traditions. Three villages from each district and households within these villages were randomly selected. Local residents aged 18 years and older with goat-breeding experience were considered eligible. Selected households were visited by a team of researchers from the Sekota Dryland Agricultural Research Center for face-to-face interviews. Participants were briefed on the study objectives, the estimated duration of the interview, their voluntary participation, and their right to withdraw at any time. As all participants were adults (≥ 18 years) and participated voluntarily, written ethical approval was not required. Oral informed consent was obtained from all participants, and their anonymity was fully ensured.

Sample size determination

The minimum sample size (N) for the household questionnaire survey was determined using the formula [8]:

The sample size was calculated as 278 households, assuming a standard error (Se) of 3% for item response scores of any binary outcome. The number of households involved in goat production was obtained from the 2023 baseline data provided by the Livestock and Fishery Office of each district. The final sample size was distributed across the districts using a proportionate sampling method, based on the minimum required sample size, the number of households practicing goat production in each district, and the total number of households across all districts. Accordingly, 115 households from Sekota, 88 from Abergelle, and 72 from Ziquala were included in the study.

Questionary preparation and data collection

The investigators collaborated to develop relevant items for each domain, refining them to measure farmers’ KAP level, taking into account the local context of extensive systems of goat farming. The semi-structured questionnaires translated into local language (Amharic) were pilot-tested with 14 households, and necessary modifications were made; data from the pilot phase were excluded from the final analysis. Subsequently, the finalized questionnaire was administered to the participants, with the interviewer recording their responses.

The KAP assessment constructs consisted of 21 items: nine for knowledge, eight for practice, and four for attitudes. The knowledge items aimed to assess participants’ understanding of goat mastitis, covering case observation, clinical presentation, animal health impacts, causes, seasonal variation, public health risks, and recognition of subclinical mastitis. The practice items were designed to evaluate participants’ behaviors related to mastitis prevention and control, including separating sick and healthy animals during feeding, isolating (quarantining) affected animals, maintaining barn and milking hygiene, consuming raw milk, milking animals with mastitis, and seeking veterinary care. The attitude construct included items assessing participants’ beliefs regarding raw milk consumption, the impact of mastitis on milk yield, the effectiveness of antimicrobials, and the risks associated with poor bedding.

Data management and statistical analysis

All the data collected were curated and cleaned in Microsoft Excel (MicrosoftCorp., Redmond, WA, USA). The responses of knowledge items were initially multiple choice (5 items) or “yes vs. no” (4) and were reclassified as “correct” vs. “incorrect”. Similarly, the yes/no and multiple-choice response practice items were reclassified as “desirable” vs. “undesirable” and the responses for the attitude items were “yes vs. no” and were reclassified as “desirable” vs. “undesirable” [3, 4]. Multiple-choice item responses were dichotomized to simplify the analysis, ensure compatibility with IRT models, and reduce the complexity introduced by distractors that do not provide meaningful information about respondents’ ability. The participants’ average knowledge, practice, and attitude levels were computed and graded as good or poor, with a 50% cutoff. STATA Version 17 (StataCorp, Texas, USA) software was used for the statistical analysis.

Descriptive statistical analysis summarized the data as frequencies, means, standard deviations, and percentages. Cronbach’s alpha (α > 0.6) were used to assess internal consistency among items and the Shapiro–Wilk test was used to examine the normality of the latent trait (theta) before performing inferential statistical analyses. Univariable logistic regression models were fitted for each independent variable (demographics such as age, sex, district, and education level) against the dependent variable (correct/incorrect response). Associations were reported as crude odds ratios, with statistical significance considered at P < 0.05.

The knowledge and practice items were analyzed via a two-parameter logistic regression (2PL) item response theory (IRT) model. The IRT model was constructed through a systematic, stepwise backward elimination approach. Unidimensionality was assessed using exploratory factor analysis, with eigenvalues and factor loadings examined to evaluate the dominance of a single underlying latent trait. Items with factor loadings ≥ 0.40 were retained, and the unidimensionality assumption was satisfied, as the first factor having an eigenvalue greater than 1 and accounting for the majority of the total variance.

Subsequently, the latent trait (θ) for each respondent was estimated using a Bayesian latent trait model, which combines prior information with observed responses to produce probabilistic estimates of individuals’ positions on the underlying trait continuum. To examine the associations between the estimated latent trait (θ) and demographic factors, the Mann–Whitney U test was used for comparisons of θ between two groups (e.g., sex), while the Kruskal–Wallis test was applied for comparisons of θ between more than two groups (e.g., district, education, age).

Results

Sociodemographic characteristics

The KAP assessment surveyed 287 households involved in goat production, with data from 275 households found eligible for analysis while 12 were excluded because of missing data. Table 2 summarizes the participants’ sociodemographic data. The majority of the participants were male, and 73.1% had never attended school. The median respondent age was 40 years, with a majority of respondents being between 21 and 40 years old.

Table 2.

Sociodemographic characteristics indicated by four categorical variables in the KAP survey on 275 participants in three districts of Amhara regional State, Ethiopia

Characteristics	Category	Frequency	Percentages
Sex	Female	72	26.2
	Male	203	73.8
Age	< 20	9	3.3
	21–40	148	53.8
	41–60	106	38.6
	> 60	12	4.4
District	Sekota	115	41.8
	Abergelle	88	32
	Ziquala	72	26.2
Education	No formal education	201	73.1
	Primary	55	20.0
	secondary and above	19	6.9

Goat management practices

Table 3 illustrates farmers’ livestock ownership and management practices. Majority (90.6%) of the respondents kept goats alongside other livestock, 7.6% raised goats exclusively, and 1.8% of participants sold or lost their goats in recent months. The grazing system was communal, and during the drought season, most households fed their goats crop residues, hay, and homemade alcohol byproducts. Goats were allowed to graze on croplands after harvest, and during droughts, flocks moved to highland areas until the rains returned.

Table 3.

Management practices of goat-owning households (N = 275) surveyed in three districts of Amhara regional State, Ethiopia

Management practice	Category	N	Percentages
Livestock species holdings	Goat alone	21	7.6
	Goat with other livestock	249	90.6
	Sold or lost in recent month	5	1.8
Keeping goat at daytime	Separately	125	45.5
	With other livestock	150	54.6
Keeping goat at night	Goats alone	112	40.7
	With sheep	156	56.9
	More than three species	6	2.2
Housing type	Staked	10	3.7
	Enclosed with roof	199	72.6
	Enclosed without roof	45	16.4
	Cave	10	3.7
	Open area	10	3.7
Feeding system	Communal	275	100
	Others	–	–
Supplementation	No	48	17.6
	Yes	227	82.6

Respondents’ knowledge, attitudes, and practices toward goat mastitis

Respondents’ knowledge

Around 47.8% of the respondents had good knowledge of mastitis, and the percentage of correct responses per knowledge item ranged from 12.3% to 96.6%. Majority of participants reported the occurrence of mastitis, with 30.7% recalling cases in their flocks during the study year. The vast majority of participants correctly identified the clinical signs and consequences of mastitis. They classified mastitis as non-seasonal (42.8%), seasonal (57.2%) and 19.9% occurs during the dry season and 37.3% during the rainy season. Other majority (80%) of the respondents was not knowledgeable enough to mention the causes of mastitis, and 87% had never heard of SCM. Most of the respondents were also confused about zoonotic diseases. Table 4 summarized the respondents’ knowledge level.

Table 4.

Assessment of households’ knowledge on goat mastitis (N = 275) surveyed in three districts of Amhara regional State, Ethiopia

Knowledge Item	Level	Response	Frequency	Percentages (SD)
K_1_Do you know about mastitis?	Correct	Yes	262	95.3 (0.01)
	Incorrect	No	13	4.7 (0.06)
K_2_Do you experience mastitis (Yefiyel git Beshita in Amharic language) in 12 months?	correct	Yes	81	30.7 (0.05)
	Incorrect	No	183	69.3 (0.03)
K_3_What clinical signs did you observe?	Correct	Udder and milk abnormalities	265	96.7 (0.01)
	Incorrect	None, incorrectly	10	3.6 (0.06)
K_4_What are the consequences?	Correct	Mention animal and public risks	258	93.8 (0.01)
	Incorrect	None/incorrectly	17	6.2 (0.06)
K_5_Do you know the causes of mastitis?	Correct	Yes	111	40.4 (0.05)
	Incorrect	No	164	59.7 (0.04)
K_6_Could you mention causes of mastitis	Correct	Bacterial, viral, fungal, physical	52	20.0 (0.06)
	Incorrect	None/incorrectly	208	80.0 (0.03)
K_7_In which season is mastitis prevalent?	Incorrect	Not specified	101	42.8 (0.03)
	Correct	Dry season	88	37.3 (0.05)
		Wet season	47	19.9 (0.06)
k_8_Could you mention some zoonotic diseases due to mastitis	Correct	Bacterial, viral, fungal,	62	23.6 (0.05)
	Incorrect	None/incorrectly	201	76.4 (0.06)
K_9_Have you heard of SCM (Yemaytay fiyel git Beshita)	Correct	Yes	32	12.3 (0.06)
	Incorrect	No	228	87.7 (0.02)
Average Knowledge	Correct			47.8 (0.01)
	Incorrect			52.2 (0.01)

Respondents’ attitudes

Approximately 69.4% of the respondents had a desirable attitude toward mastitis, and the item-level correct attitude ranged from 25.6% to 94.1%. The vast majority of respondents (90.6%) believed that mastitic milk caused human illness, 67.3% believed that it reduced milk yield, and 74.4% believed that antimicrobials were ineffective against mastitis. Furthermore, 94.1% believed that poor housing hygiene (pen/barn) increased the risk of mastitis (Table 5). The majority of the respondents in this study attributed mastitis to bad luck, evil eyes, or other beliefs.

Table 5.

Assessment of households’ attitudes toward goat mastitis (N = 275) surveyed in three districts of Amhara regional State, Ethiopia

Attitude item	Level	Response	Frequency	Percentages (SD)
A_1_ Do you think mastitis can cause human illness?	Desirable	Yes	230	90.6(0.02)
	Undesirable	No	24	9.5 (0.06)
A_2_ Do you think mastitis can cause a decrease in milk yield?	Desirable	Yes	171	67.3 (0.04)
	Undesirable	No	83	32.7 (0.05)
A_3_Do you think mastitis can be effectively treated via antimicrobials?	Desirable	Yes	65	25.6 (0.05)
	Undesirable	No	189	74.4 (0.03)
A_4 Do you think poor goats’ housing increases the risk of mastitis?	Desirable	Yes	239	94.1 (0.02)
	Undesirable	No	15	5.9 (0.06)
Average	Desirable	Yes	705	69.4 (0.02)
	Undesirable	No	311	30.6 (0.03)

Respondents’ practice

Table 6 summarizes the respondents’ practices for the control and prevention of mastitis. The study found that behavioral practices among herders were inadequate, with only 36.1% implementing appropriate measures. Specifically, only 34.3% practiced separate feeding for sick and healthy animals, 9.8% provided separate pens for diseased and healthy animals, and 85.3% never washed their hands before or after milking. Moreover, the majority practiced traditional mastitis treatments, such as applying plant juices directly to udders as topical lotion or drenching, applying salt to the wound, milking with hot metal or stone, and steaming the affected udder. Additionally, 77.8% consumed raw milk, and 14.9% consumed milk from mastitic animals.

Table 6.

Assessment of households’ practices on goat mastitis (N = 275) surveyed in three districts of Amhara regional State, Ethiopia

Practice items	Level	Response	Frequency	Percentages (SD)
P_1_ Do you separately feeding and watering for D+ and D–	Desirable	Yes	91	34.3 (0.06)
	Undesirable	No	174	65.7 (0.04)
P_2_ Do you provide isolation pen for D+ from D–	Desirable	Yes	26	9.8 (0.01)
	Undesirable	No	239	90.2 (0.08)
P_6_ Have you brought sick animals to vet clinics?	Desirable	Yes	55	94.7 (0.05)
	Undesirable	No	210	5.3 (0.05)
P_3_ Do you practice traditional healing?	Desirable	Yes	200	75.5 (0.06)
	Undesirable	No	14	24.5 (0.05)
P_4_ Do you consume raw milk?	Undesirable	Yes	214	77.8 (0.02)
	Desirable	No	61	22.2 (0.04)
P_5_ Do you consume unprocessed milk from infected does?	Undesirable	Yes	40	14.9 (0.05)
	Desirable	No	225	85.1 (0.04)
P_7_ How frequently do you clean the goat’s barn?	Desirable	Daily/weekly	107	40.4 (0.04)
	Undesirable	Monthly/none	158	59.6 (0.05)
P_8_ Do you wash your hands before and after milking the goats?	Desirable	Yes	39	14.7 (0.02)
	Undesirable	No	226	85.3 (0.01)
Average practice	Desirable			36.1 (0.01)
	Undesirable			63.9 (0.01)

D⁺ = diseased, D^- = healthy

Associations between respondents’ demographics and their KAP level of mastitis

A statistically significant association was found between mean knowledge and district, but no association was found between mean practice and any of the demographic risk factors. As a result, only item-level data for each domain, with demographics as independent variables, were analyzed.

Associations between respondents’ demographics and knowledge of mastitis

The univariable logistic regression result revealed statistically significant associations between demographics (age and district) with some of the knowledge items. Respondents aged 21–40 and 41–60 had significantly higher odds of responding yes to “Do you know mastitis?” compared to the reference group under 20 (p < 0.00). Besides, older respondents had higher odds of identifying clinical signs of mastitis (OR = 24.75 for 21–40; OR = 17.03 for 41–60; and OR = 35.48 for over 60) and the consequences of mastitis (OR = 10.85 for 21–40 and OR = 7.92 for 41–60) compared to the reference group, under 20 (P < 0.000). Recalling mastitis incidence during the study year was also strongly associated with participants’ age. Moreover, older participants also had lower odds of understanding mastitis causes (OR = 0.3 for 21–40 years; OR = 0.24 for 41–60 years; OR = 0.13 for over 60; P < 0.000) and zoonotic diseases (OR = 0.19 for 21–40 years; OR = 0.23 for 41–60; P < 0.000). Compared to the reference Sekota district, Ziquala (OR = 0.22; p < 0.009) for the item “experienced mastitis during the study year” and Abergelle (OR = 2.75; p < 0.001) for the item “mention clinical sign of mastitis” showed a statistically significant association (Table 7).

Table 7.

Associations between demographic variables and households’ knowledge outcomes on goat mastitis (N = 275) surveyed in three districts of Amhara regional State, Ethiopia

Variable	Category	Knowledge outcome	OR	P-value
	21–40	Heard of about mastitis	4.73	< 0.001
	41–60		4.27	< 0.001
	21–40	Clinical signs presentations	24.75	< 0.001
	41–60		17.03	< 0.001
	> 60		35.48	< 0.001
Age (Reference ~ 20)	21–40	Consequences	10.85	< 0.001
	41–60		7.92	< 0.001
	21–40	Recall of cases	0.31	< 0.001
	41–60		0.35	< 0.001
	> 60		0.19	< 0.021
	21–40	Understand causes	0.30	< 0.001
	41–60		0.24	< 0.001
	> 60		0.13	< 0.001
	21–40	Understand zoonotic risk	0.19	< 0.001
	41–60		0.23	< 0.001
District (Sekota)	Ziquala	Recall of cases	0.22	< 0.009
	Abergelle	Clinical signs presentations	2.75	< 0.001

Associations between respondents’ demographics and practices related to mastitis

In terms of practice, the age and sex of participants were found to be statistically associated with good or poor practice. Older respondents were more likely to prepare a separate pen for mastitic animals than younger ones (OR = 10.85: 21 to 40 and OR = 7.92: 41 to 60 years old). Females had lower odds of traditional mastitis healing practices (OR = 0.34, P < 0.001) and raw milk consumption (OR = 0.36; P < 0.003) than males. Besides, adults had higher odds of traditional healing (OR = 6.67 for 21–40; OR = 4.37 for over 60, P < 0.001) and raw milk consumption (OR = 7.06 for 21–40; OR = 3.59 for 41–60, P < 0.001) than the younger ones. Conversely, adults had lower odds of avoiding milking mastitic goats (OR = 0.13–0.22, P < 0.001) and hand washing before and after milking (OR = 0.12–0.24, P < 0.001) than younger participants (Table 8).

Table 8.

Associations between demographic variables and households’ practice outcomes on goat mastitis (N = 275) surveyed in three districts of Amhara regional State, Ethiopia

Variable	Category	Practice outcome	OR	P-value
Age (reference < 20)	21–40	Separate pen for mastitic animals	10.85	< 0.001
	41–60		7.92	< 0.001
	21–40	Traditional healing use	6.67	< 0.001
	> 60		4.37	< 0.001
	21–40	Raw milk consumption	7.06	< 0.001
	41–60		3.59	< 0.001
	21–40	Avoid milking mastitic goats	0.22	< 0.001
	41–60		0.13	< 0.001
	21–40	Hand washing before/after milking	0.24	< 0.001
	41–60		0.12	< 0.001
Sex (male)	Female	Traditional healing use	0.34	< 0.001
	Female	Raw milk consumption	0.36	< 0.003

Item response theory

The Cronbach’s alpha values for the knowledge, practice, and attitude questionnaire-constructed domains were 0.67, 0.63, and 0.32, respectively (Supplementary file 1). The first two had acceptable item internal consistency in measuring knowledge and practice, so inferential analysis was continued; for attitudes, only simple descriptive statistics were computed. Factor analysis revealed five factors for practice items with eigenvalues of 2.36, 0.74, 0.20, 0.11, and − 0.02 (Supplementary file 2) and four factors for knowledge items with eigenvalues of 2.44, 0.49, 0.1, and 0.00; both domains were deemed unidimensional (Supplementary file 3). Overall, five knowledge items and four practice items with factor loadings > 0.4 were used for the ICC analysis and TIF.

Discrimination powers of knowledge and practice items among participants

The item-level discrimination and difficulty powers for the knowledge items ranged from 0.2 to 5.08 and from − 6.05 to 9.29, respectively. The item “Do you know the causes of mastitis?” had the highest discrimination power (a = 5.08), whereas the items “Have you heard of SCM?” and “Do you experience/identify mastitis in 12 months?” had the poorest discrimination power (a < 0.25). Overall, two knowledge items had poor discrimination power (< 0.25), five had moderate discrimination power (0.25–3.25), and two had excellent discrimination power (> 3). The study revealed that some knowledge items were simple, whereas others were difficult, implying that the knowledge section effectively distinguishes between people who understand mastitis to varying degrees. Seven knowledge items had moderate to excellent discrimination power and are therefore suitable for assessing herders’ knowledge of mastitis. The most important knowledge items, however, proved to be the most difficult, necessitating revisions.

Among the eight practice items, 4 were found to be constrained in fitting to the 2PL model for discrimination and difficulty estimation, and only the other 4 items were used for analysis. The item “Separate feeding and watering for D⁺ and D^–” in the practice item had better discrimination power (a = 2.12) and all included items had moderate discrimination power (> 0.25) (Table 9). In contrast to knowledge, practice items had less variability in discrimination and difficulty parameter power; despite some practice items being easier, they may not effectively discriminate between respondents of varying abilities. Therefore, revising these practice items may be beneficial. The ICC indicated that the likelihood of answering knowledge items correctly increases as the underlying trait (ability) increases.

Table 9.

Item discrimination and difficulty parameters for knowledge (K) and practice (P) items on goat mastitis, estimated using an item response theory model (N = 275)

Item	Discrimination (se)	P > z	Difficulty (se)	P > z
K_1	3.46 (0.19)	0.014	–2.56 (1.02)	0.12
K_2	0.2 (0.16)	0.197	3.96 (3.02)	0.194
K_3	0.9 (0.31)	0.004	–3.03 (0.88)	0.001
K_4	0.36 (0.26)	0.166	–6.05 (4.22)	0.152
K_5	4.17 (0.77)	0.056	0.39(0.06)	0.065
K_6	5.08 (1.05)	0.000	0.89 (0.08)	0.00
K_7	0.75 (0.18)	0.000	–0.14 (0.18)	0.449
K_8	0.51 (0.21)	0.014	3.33 (1.26)	0.008
K_9	0.21 (0.21)	0.318	9.29 (9.14)	0.31
P_1	2.12 (1.19)	0.074	1.00 (0.25)	0.000
P_2	1.64 (0.71)	0.022	–2.61 (0.91)	0.004
P_6	1.07 (0.54)	0.046	–3.17 (1.2)	0.008
P_7	0.76 (0.31)	0.014	0.07 (0.09)	0.481

Mean comparison of latent traits (abilities)

The latent trait (ability) for each respondent was estimated via the Bayesian empirical latent trait model for the practice domain and the knowledge domain. The relationships between the latent traits and demographic factors were subsequently assessed. The findings revealed a statistically significant relationship between the district and the latent trait of knowledge. Compared with the Sekota and Abergelle districts, the Ziquala district demonstrated significantly greater ability, with a mean score of 159.69 (p < 0.025) (Table 10).

Table 10.

Comparison of estimated latent trait scores for knowledge and practice on goat mastitis across demographic factors (N = 275)

Domain	Variable	Category	N	Mean	Z	P value
Knowledge	Education	No formal education	201	137.19	–0.28	0.933
		Primary	55	141.55	0.37
		Secondary and above	19	136.34	–0.09
	District	Sekota	115	131.98	–1.06	0.025
		Abergelle	88	128.12	–1.41
		Ziquala	72	159.69	2.69
	Age	< 20	9	163.94	1.0
		21–40	148	137.36	–0.14	0.299
		41–60	106	140.84	0.47
		> 60	12	101.33	–1.63
	Sexa	Male	203	136.29	–0.60	0.549
		Female	72	142.82	0.60
Practice	Sexa	Male	203	138.4	0.14	0.889
		Female	72	136.88	–0.14
	Age	< 20	9	134.17	–0.15
		21–40	148	136.81	–0.27
		41–60	106	138.08	0.01
		> 60	12	154.83	0.75
	District	Sekota	115	147.6	1.70	0.061
		Abergelle	88	121.88	–2.31
		Ziquala	72	142.38	0.54
	Education	No formal education	201	134.38	–1.25	0.250
		Primary	55	141.93	0.4094
		Secondary and above	19	164.97	1.532

Figure 1 illustrates the relationship between the estimated probabilities of correct response and latent trait theta (θ) for practice items (A) and knowledge items (C). The practice domain achieved the maximum number of correct responses at θ + 4, whereas the number of correct responses in the knowledge domain was + 4. Fig. 1’s TIF curve indicates that knowledge items provide better information at a θ value of 1 (C) and a practice of 0 (B)

Fig. 1.

Item characteristic curves and test information functions from Item Response Theory (IRT) analysis. A Item Characteristics Curve (ICCs) for practice items showing the probability of a correct response across the latent trait (θ), highlighting differences in item difficulty and discrimination; B corresponding test information function and standard error across θ for the practice domain; C Item Characteristics Curves for five knowledge items and (D) their test information function with standard error, indicating the θ range of maximum information and minimum measurement error.

Discussion

To the best of the authors’ knowledge, the existing literature has largely overlooked comprehensive assessments of herders’ knowledge, attitudes, and practices regarding mastitis in goats, with only a few studies addressing isolated practice-related aspects [21]. This study is the first that systematically analyzed KAP measuring data using IRT approaches and validated the suitability of the items for accurately measuring KAP levels.

The study indicated that while most respondents were aware of mastitis, had experienced it during the past 12 months, and recognized clinical signs, their overall knowledge level was unsatisfactory. The herders were largely unaware of SCM, which is not even comparable to a study conducted in Serbia by [21], where 56% of participants reported awareness of the condition. Additionally, their understanding of mastitis as source of zoonosis was poor, which is in line with a study by Alemayehu et al. (2021) from Ethiopia, where the average knowledge score was 29.92%. Consequently, farmers might not apply proficient knowledge, despite its importance in goats’ health management and public health protection. The majority of the respondents in this study attributed mastitis to bad luck, evil eyes, or other beliefs, which was consistent with the findings of [5]. A lack of understanding of the disease hampers animal health care and productivity [7]. The respondents incorrectly identified malaria and tapeworms as public health risks associated with contaminated milk. However, these diseases are transmitted via vectors and undercooked red meat, not through milk. The vast majority of the respondents thought that drinking raw milk was safe. This result was consistent with findings in Borena, Ethiopia, where lowland pastoralists routinely consume raw milk and believe that boiled milk is equivalent to dead milk, that essential nutrients are lost and that it has an unpleasant taste [5]. The consumption of raw milk from mastitis-infected goats can endanger public health, especially in areas where SCM is not noticed [11, 15].

Substantial knowledge gaps among livestock keepers contribute to high-risk food-handling behaviors and increase the risk of zoonotic disease transmission (Alemayehu et al., 2021). Foodborne diseases resulting from poor food hygiene are responsible for approximately 600 million illnesses and 420,000 deaths worldwide each year, including 125,000 deaths among children under the age of five [6].

There were statistically significant associations between the demographics of herders (age and district) and their knowledge of specific items. Herders in the Abergelle district had lower relative knowledge scores than Sekota districts. Older people are more familiar with mastitis, its clinical signs, and its impact. Their experience in keeping goats and exposure to more cases may explain this association. Elders, on the other hand, are less knowledgeable about the causes of mastitis and the risk of zoonosis. This could be related to illiteracy and poor knowledge absorption, as many Ethiopians saw modern education, media, and extension coverage in the 1980s. However, some KAP‑studies in Ethiopia dealing with livestock diseases or dairy practices show variation in knowledge or practice with demographic variables [9, 12].

The mean practice score was lower than the mean score of knowledge; apparently, farmers have knowledge, which is quite an intangible measure, but the practice decreases when it becomes implemented. The most difficult practice item was to provide separate pens for diseased and healthy animals, which is a significant gap in controlling transmission of contagious infectious disease, such as contagious intra mammary infections. This result was consistent with the report of [21], who reported that 38.33% of farmers in Serbia did not isolate goats with severe systemic mastitis. However, this conclusion contradicts with the finding of [4], who reported that 50% of lowland small ruminant producers used diseased animal isolation pens. On the basis of observations, farmers rarely segregated sick animals into separate constructed enclosures; instead, they frequently shared grazing and watering.

Adults often milk goats immediately after other tasks and handle all lactating goats without washing their hands. This result was contradicted with the findings of [16], who reported that 95% of farmers washed their hands after and before milking goats. This malpractice behavior could explain why mastitis was reported by most of the respondents. Manual hand milking was one of the risk factors associated with increased in bacteriological mastitis in Italy [17]. Mastitis prevention measures such as pen cleaning, isolation, separate feeding, and regular screening [1, 14, 22] were neglected or unknown there.

This study indicated that herders often travel 5–10 km to seek veterinary care and, when services are inaccessible, they resort to traditional approaches for managing mastitis. This finding aligns with [4], who reported that access to veterinary services for small-ruminant producers in Ethiopia is limited. The preference for traditional remedies may also stem from limited accessibility or inadequate understanding of veterinary services, including doubts about the effectiveness of antimicrobials and concerns about the cost of drugs. Males and older farmers were more engaged in traditional healing practices than females and younger people. Farmers use plant parts (leaves, roots, stems, and bark) to treat mastitis in different ways, including applying plant juices directly to udders as topical lotion and drenching. Traditional therapy methods also include applying salt to the wound, milking with hot metal or stone, and steaming the affected udder. These practices were consistent with the findings of [5]. Moreover, farmers face confusion regarding vaccination and the use of antimicrobial medications. According to [23], approximately, 9.5% of farmers lacked proper knowledge and 84.3% had negative attitudes about antimicrobial use. Goat production in Ethiopia is the most traditional extensive system; the vast majority of goat farmers are unaware of dry period therapy or vaccination against common, economically significant mastitis pathogens. There are no commercially available selective treatments, such as teat sealants or cloxacillin injections.

The study revealed that some knowledge items were simple, whereas others were difficult, implying that the knowledge section effectively distinguishes between people who understand mastitis to varying degrees. Seven knowledge items had moderate to excellent discrimination power and are therefore suitable for assessing herders’ knowledge of mastitis. The most important knowledge items, however, proved to be the most difficult, necessitating revisions. In contrast, practice items had less variability in discrimination and difficulty parameter power; despite some practice items being easier, they may not effectively discriminate between respondents of varying abilities. Therefore, revising these practice items may be beneficial. The ICC indicated that the likelihood of answering knowledge items correctly increases as the underlying trait (ability) increases.

There were no statistically significant associations between mean ability and demographic variables for knowledge (except district). The respondents in the Ziquala district had better mean knowledge proficiency than that of the respondents from the Abergelle and Sekota districts, where there were no cases of CM mastitis and the SCM prevalence was lower.

Limitation of the study

This research is the first to apply the item response theory approach to investigate herders’ knowledge, attitudes, and practices related to mastitis. However, the psychometric measurement items for each domain may not comprehensively cover all relevant aspects, and the limited inclusion of mastitis-specific items could affect the results. Readers should take this limitation into consideration.

Conclusion and recommendations

This study revealed that mastitis is an important production constraint for goats in the study area. Many respondents lacked basic goat management skills related to mastitis and were unaware of the zoonotic risks associated with unprocessed or raw milk consumption habits. Older age groups and districts showed statistically significant associations with specific mastitis knowledge items, with older respondents being more likely to recognize the disease and its signs but less likely to understand its causes and zoonotic risks. Age and sex significantly influenced mastitis-related practices, with older participants being more likely to isolate mastitic animals but engaging more in risky behaviors, whereas females were less involved in traditional healing and raw milk consumption than males. This underscore, the need of tailored training on the causes of mastitis and zoonotic risk, control and prevention for goat farmers. Further research on bacterial species and resistance factor identification is recommended. Knowledge items demonstrated validity in distinguishing herdsmen with varying levels of knowledge, whereas the practice items lacked such robustness; therefore, the knowledge items from this study can be used to assess mastitis knowledge in similar settings, while the practice items need to be revised.

Abbreviations

CM

Clinical Mastitis

SCM

Subclinical Mastitis

IRT

Item Response Theory

KAP

Knowledge, Attitude and Practice

ICC

Item Characteristics Curve

TIF

Test Information Function

OR

Odds Ratio

D+

Diseased

D-

Non diseased

2PL

Two Parametric Logistic Regressions

Authors’ contributions

AT: conception, methodology, investigation, data collection, formal analysis, draft writing, and reviewing; YT: conceptualization, supervision, draft writing, and reviewing; HT: conceptualization, supervision, draft writing, and reviewing; AB: conceptualization, supervision, draft writing, and reviewing; SAM: conceptualization, supervision, draft writing, and reviewing.

Funding

This work was supported by the Amhara Agricultural Research Institute [S/LRD/AH/12/2023], Ethiopia.

Data availability

Data will be available upon responsible request to the corresponding author.

Declarations

Ethics approval and consent to participate

The study was approved by the Sekota Dryland Agricultural Research Center Animal Health Research Review Committee (SDARC-AHRRC/23/05/2023), and the Declaration of Helsinki was followed for human data collections. Informed consent from the participants was obtained via oral understanding prior to the actual face‒to‒face interview.

Consent for publication

Verbal informed consent was obtained from participants that the information will be used anonymously for publication.

Competing interests

The authors declare no competing interests.