Theory in the context of human behavior indicates why, when and how a behavior does or does not occur (12). Furthermore, theories identify determinants of influence to be targeted to alter behavior and they reflect integrated knowledge about relevant mechanisms of action and moderators of change of behavior.

Davis et al. (17) identified 82 health psychology theories for describing human behavior. Every theory consists of various key constructs, i.e., specialized terms, to label the theory's elements (18). The ‘Theory of Planned Behavior’ (TPB) (19), an extension of the ‘Theory of Reasoned Action’ (20) is the most commonly used psychosocial theory in human health (21). A central factor in TPB is an individual's intention to perform a given behavior, because it is assumed that intention captures motivational factors that influence a behavior (19). The TPB consists of three conceptually independent determinants of intention. The first is ‘attitude toward the behavior,’ referring to the degree to which a person has a favorable or unfavorable appraisal of the behavior in question (19). Secondly, ‘subjective norms’ refer to perceived social pressures toward a certain behavior, and thirdly, ‘perceived behavioral control’ refers to the perceived ease or difficulty of performing a behavior. The latter includes a reflection on past experiences as well as anticipated impediments and obstacles (19).

According to a systematic review (22), the other most common theories used in human health behavior include the Transtheoretical Model (23), the Social Cognitive theory (24), and the Health Belief Model (25). The Transtheoretical Model conceptualizes behavior change as a process involving a series of six distinct stages: pre-contemplation, contemplation, preparation, action, maintenance, and termination. The Social Cognitive theory (24) states that behavior is influenced directly by goals and self-efficacy expectations and indirectly by self-efficacy, outcome expectations, and sociostructural factors. The Health Belief Model (25) hypothesizes that health-related behavior depends on a combination of perceived susceptibility, perceived severity, perceived benefits, perceived barriers, cues to action, and self-efficacy. These theories include overlapping constructs such as results expectancies, i.e., beliefs about the behavior and expectations for the results of the behavior (21). Additionally, overlap in self-evaluation occurs, described as the individual's subjective evaluation of the amount of control and competency to engage successfully in a behavior. Finally, there is an overlap in social factors, regarded as the influences of other people on behavior change (21).

It is noteworthy that these health psychology theories are mainly driven by an individuals' cognitive processes and actions and individual-level factors (as above) relating to motivation and capability (17, 26). Limited attention has been given to ‘external influences’ and wider interpersonal factors that shape outcomes and behaviors.

In contrast to psychological theories, sociological approaches focus predominantly on the context in which people live and interact. They consider behavior as an outcome of complex inter-relationships and shared social practice. Theories such as social practice theory (27) and the normalization process theory (28) are some dominant approaches in the rural sociology for incorporating impacts of social contexts to understand behavior (29, 30). Also approaches exploring the role of cultural scripts (31), the concept of capital, habitus and field (32, 33) and the identity and lay-expert knowledge reflexivity are often applied for incorporating social contexts to understand behavior (29, 30).

Use of various psychosocial theories can help us to understand key constructs driving behaviors. Although review manuscripts regarding farmers' behavior have been published in the scientific literature (4, 34, 35), there is no comprehensive review assessing what, if any, psychosocial theories have been applied and what constructs or factors were identified.

Given the complexity of behavior and behavior change, there have been attempts to develop comprehensive frameworks that can guide research, intervention design, and assist non-experts such as policymakers to understand human behavior. Two such widely cited frameworks are the social ecological framework or model (SEM) (36) and Behavior Change Wheel (BCW) (26).

The SEM is based on the previously developed ecological systems theory by Bronfenbrenner (36). SEM provides a holistic multilevel framework that considers the complex interplay among personal, interpersonal, and contextual environments to understand behavior (37). The SEM contextualizes behavior of individuals using various dimensions. These dimensions include personal (e.g., knowledge, attitudes, behavior), interpersonal (e.g., social networks, social supports), community (e.g., relationships among organizations/institutions), and public policies (e.g., local, state, and national laws) and how these factors interplay with each other (38). The SEM has been widely used for health promotion since the 1980s (39).

Whilst the purpose of the SEM is having broad inclusion of constructs that impact behavior, more recently developed frameworks tend to not only focus on identifying influences on behavior change, but also on understanding mechanisms of change and implementation of interventions. The BCW, which is linked to the Capability-Opportunity-Motivation (COM-B) model of behavior and is derived from human health research, is an example of such a framework (Figure 1). COM-B, designed by Michie et al. (26), consolidates the overlapping constructs of various theories. It provides a systematic way of characterizing interventions that enable their outcomes to be linked to mechanisms of action (12). The BCW is a synthesis of 19 health psychology theories and frameworks and includes the assumption that behavior is not only driven by beliefs and perception, but also by unconscious biases, mental shortcuts, and physical and contextual environments (12). The BCW consists of three layers: the hub of the wheel [the Capability-Opportunity-Motivation Behavior (COM-B) model], an intervention layer with nine functions, and a policy layer with seven categories (Figure 1). The hub of the wheel identifies the three constructs of behavior systems that could prove to be targets for intervention (12). “Capability” is subdivided into psychological capability and physical capability. “Opportunity” can be subdivided into social and physical opportunity, and “motivation” into automatic and reflective motivation. All three constructs (capability, opportunity, and motivation) are conceptualized as being essential for behavior (12). The BCW was successfully used in human health interventions to change human behavior [e.g., (40, 41)]. To successfully apply this method, it is crucial to be specific about the target group and its behavior.

The COM-B model and SEM enable exploring constructs that are not only related to the individual context but are also inclusive of broader contextual and social constructs. Therefore, by targeting interventions on various levels, we assume these frameworks to be more suitable for sustainable behavior change than frameworks that do not enable contextual and social constructs.

Understanding broader and complex behavioral structures is necessary before behavioral changes can be made, but also before key constructs of behavior can be incorporated into any animal disease prevention and control models. There is very limited work in the veterinary domain to design and test theory-informed behavior interventions (42).

In addition to choosing the appropriate theory or framework to successfully understand behavior, it is crucial to use robust design and data collection methodologies, including validity of questions or scales to measure constructs and/or behavior.

PubMed and Web of Science were screened on 26 October 2020 for potentially relevant articles. We developed a search strategy consisting of relevant keywords describing the following themes: farmer, veterinarian, behavior/behaviour, perception, attitude, beliefs, and disease control. The broad themes were combined into a single query. A search containing the following words was conducted:

(farmer^*or producer^*or veterinarian^* or veterinary or vet or vets) AND (cattle or beef or dairy or cow^*or calf or calves or heifer^*) AND (belie^*or behaviour^*or behavior^*or attitude^*or perception^* or driver^*or barrier^*or enabler^*or motivat^*) AND (manag^*or control^*or implement^*or uptake or prevent^* or use^*or usage^*).

As an additional analysis, influences of veterinarians on farmers were explored, as veterinarians are deemed to be very important in farmers' decision making. Therefore, studies focusing on veterinarians' behavior were also assessed for eligibility. Although no limits on publication date nor language were applied during our initial screening, only English and Dutch articles were considered for full-text reviewing. Queries were adapted to database-specific terms, as deemed necessary. Using these search criteria, 6,667 manuscripts were identified in both databases, and after duplicate removal, abstracts of 5,134 manuscripts were screened. For inclusion in the final review, we formulated additional criteria: (1) the subject of the study had to be related to prevention or control of cattle disease; and (2) the study should have a focus on behavioral key constructs of farmers. Finally, reference lists from all articles included in this review were reviewed for potential inclusion (Table 1).

Eight additional manuscripts, identified by screening reference lists of included manuscripts, were eligible according to the inclusion criteria and were included in the full text screening. After applying all criteria, 166 manuscripts remained for full text screening (Figure 2). The web-based software Covidence^® (61) was used for screening and data extraction. Screening and data extraction was conducted by the first author and discussed in detail with author JK after each phase. The 166 identified manuscripts are included in the Supplementary Material.

The search criteria led to the inclusion of 166 manuscripts, of which 133 studied farmers' behavior, 13 a combination of farmers' and veterinarians' behavior, and 20 studied only veterinarians' behavior. Results of the 20 studies on veterinarians' behavior will be presented independent from all other results. In the 146 remaining studies, 24 themes of disease prevention and control were identified, with the most intensively studied theme being the application of biosecurity measures (n = 22; 15%) and bovine paratuberculosis (Johne's disease) (n = 22; 15%), followed by behavior related to use of antimicrobials and prescriptions for antimicrobials by farmers and veterinarians (n = 20; 14%), and mastitis (n = 19; 13%) (Table 2). The relatively large number of studies on antimicrobial use and antimicrobial resistance was attributed to emergence of antimicrobial resistance in human and animal health (62), and regulations to decrease antimicrobial use (63).

Manuscripts that studied farmers' behavior within cattle disease prevention and control were identified from 27 countries worldwide (Figure 3). Researchers from the United Kingdom published the most manuscripts (n = 35), followed by researchers from the Netherlands (n = 15), Canada (n = 13), and the USA (n = 13). Researchers from six African countries published a total of 11 manuscripts; researchers from five Asian countries published nine manuscripts, and South American researchers published one manuscript originating from Brazil. In Europe, research in this area was mainly done by researchers from north-western European countries (Figure 3). Specifically, European countries were Belgium (n = 6), Denmark (n = 5), France (n = 4), Germany (n = 3), Ireland (n = 6), Spain (n = 3), Sweden (n = 9), Switzerland (n = 4) and a combination of the Nordic countries (n = 1). Some publications were a result of a collaboration of researchers and data from more than one country.

The first manuscript identified was published in 1995 by Australian authors (64). The increased appreciation of social science studies in the veterinary field is substantiated by the total number of manuscripts published between 1995–2009 (n = 15), 2010–2014 (n = 34) and 2015–2019 (n = 97) (Figure 4; IRR = 2.6, P < 0.001).

Although application of social sciences in farmers' behavior studies has increased, between 2008 and 2020, the average percentage of manuscripts that applied a theory varied from 17 to 89% per year (Figure 4). Overall, 58% of the published studies lacked any theoretical underpinning. Therefore, there is a lack of theory-informed tools to research in farmers' behavior related to cattle disease prevention and control. The percentage of studies based on theory increased from 2012 until 2014, relative to the period until 2011 (OR = 3.5, P = 0.04). However, in the years thereafter (2015–2018), this percentage tended to decrease compared to 2012–2014 (OR = 0.42, P = 0.09). The number of manuscripts based on theory in 2019–2020 did not differ from 2015 to 2018 (P = 0.90) (Figure 4). Not all manuscripts of 2020 could be included in our review (search date October 26th), but it seems unlikely that the current percentage of 29% will increase substantially when the search term is used again. Reasons for the declining percentage of theoretical underpinned veterinary research remains unclear. It might be related to challenges of interdisciplinary research, and that most research on theories is published in sociological journals, which likely have limited veterinary readership. Presumably dissemination of developments in social science research to veterinary research is limited or at least suboptimal. Further, use of inductive approaches are a common feature of qualitative studies and half (51%) of the qualitative studies in the review were identified as exploratory. However, in studies using quantitative or mixed methods, 61 and 62%, respectively, lacked use of a theory.

The most frequent theoretical underpinning was based on the TPB (14%). The TPB “surge” was also noticed in the human health studies but occurred a decade before the TPB “surge” in farmers' behavior studies, with the main increase between 1997 and 2012 (21). A critical limitation of the TPB is that it is unable to explain sufficient variability in the data (65). For example, in human health studies, the TPB explained 39% of the variance in intentions but only accounted for 27% of human behavior variance (66). Another concern about the TPB is its utility, as it fails in the primary function of a theory; namely, it does not accurately communicate accumulated empirical evidence and therefore does not enable practitioners to develop helpful interventions (Sniethotta et al., 2014). The TPB is more useful for predicting self-assessed behavior than objective measurement of behavior and predicts behavior better in the short vs. long term (21). In human health studies, there is a shift to frameworks that are composite of theory domain frameworks and are better for understanding and changing behavior, such as the COM-B model (26).

The second most used method in these studies was the grounded theory (8%). The overall aim in grounded theory is to generate theories inductively by collecting data about a phenomenon, identifying key elements, and categorizing the relationship of those elements to each other (67). However, only 4 of the 11 manuscripts based on grounded theory ultimately mentioned the development of own theory, indicating failure to ensure that the analysis moved beyond narrative description to generate theoretical concepts (67). This finding illustrates that the theoretical and philosophical underpinning of grounded theory was often not completely understood.

Health psychology theories and constructs are validated in human health research, but there is limited evidence of measurement scale validation and very few have been statistically validated in the context of animal health (68–70). In the reviewed work, there was a lack of confirmatory factor analysis testing theories and constructs. Human health and animal health environments differ considerably. Although farmers make decisions concerning themselves, similar to human health, they also make decisions that affect their animals and business. Therefore, it might be that these theoretical constructs may have even less explanatory value in the context of management of animal health. This hypothesis needs to be explored further.

Data were collected quantitatively in 51% of the studies, while in 29% of the studies, data were collected qualitatively. A mixed approach was used for the remaining 20% (Table 2). Questions to determine farmers' perceptions, beliefs, and attitudes were often measured with 5 or 6-point Likert scales. For countries that published ≥5 manuscripts, quantitative or mixed approaches were more often used, except for India and Canada, where data were mostly collected qualitatively (Table 3). Researchers from Scandinavian countries and the United Kingdom used qualitative and quantitative methods with approximately equal frequency.

To illustrate the complexity of factors impacting farmers' behavior, and the consequences of the lack of theoretical underpinning, key behavioral constructs of farmers' behavior related to cattle disease prevention and control were determined per study and summarized according to the SEM and the BCW of the COM-B model.

In 73 of the farmers' behavioral studies, originating from 18 countries, components of the farmer-veterinarian relationship were explored. The likely large influence of veterinarians on farmers' behavior was also supported by the reviewed literature; veterinarians are often regarded as the primary source of information on animal health and disease control (68, 72–76). It is, therefore, important to consider influences of veterinarians on farmers' decision-making processes. To achieve a better understanding of veterinarians' behavior, a subset of 20 veterinarians' behavior studies was explored. Veterinarians' behavior was studied in 12 countries, including the United Kingdom (N = 4), the USA, Canada, the Netherlands, Sweden, Switzerland (N = 2 per country), Belgium, Ireland, Australia, Argentina, Italy, and Peru (N = 1 per country). The first manuscript was published in 2009 by US authors (77). Themes studied were mainly related to antimicrobial use or attitude toward antimicrobial resistance (39%) and perception of on-farm biosecurity (28%).

More often, constructs mentioned were related to the personal environment (N total mentioned = 103), than to the interpersonal or contextual environment (N total mentioned = 27 and 27, respectively). There was a limited list of contextual factors influencing veterinarians' behavior, perhaps due to harmonization of practice standards or accreditation requirements under which veterinarians operated (Table 6). Personal determinants listed include veterinarians' own beliefs and perceptions as well as how they perceived farmers. Based on the evidence, we inferred that how veterinarians frame barriers to disease control differed from how farmers frame these barriers; unlike farmers, veterinarians put more emphasis of individual and interpersonal factors, as was apparent in the review (8). To adapt to changing policy and practice environments, veterinary practitioners must more effectively blend scientific and evidence-based approaches to veterinary care with the pragmatic concerns facing farmers by fully understanding farmers' determinants of disease control (8, 78).