Health beliefs, label orientation and consumer behaviour toward red and processed meats: implications for sustainable diets in South Africa Open Access

Red and processed meat consumption remains a critical topic at the intersection of nutrition, health, and sustainability. While red meat provides important nutrients such as high-quality protein, iron, zinc, and vitamins B12 and B6 (Czerwonka et al., 2017), excessive intake is associated with elevated risks of cardiovascular disease, colorectal cancer, and premature mortality (Bouvard et al., 2015; Wolk, 2017), leading major health authorities, including the International Agency for Research on Cancer, to classify processed meat as carcinogenic and unprocessed red meat as probably carcinogenic (WCRF/AICR, 2018). Alongside these health concerns, meat production is a major contributor to environmental pressures through greenhouse gas emissions, land and water use, and other ecological impacts (Godfray et al., 2018). Global frameworks such as the EAT–Lancet Commission underscore the need for dietary shifts that improve human and planetary health (Willett et al., 2019), but achieving such transitions ultimately hinges on how consumers perceive and act upon health and sustainability information. Understanding these perceptions is therefore essential for informing behaviour-change strategies and supporting progress toward Sustainable Development Goal 12 on responsible consumption and production.

Consumer perceptions of red and processed meat are far from uniform. While the health and environmental consequences of excessive meat intake are widely discussed in high-income countries (HICs), behavioural evidence from low- and middle-income countries, including those in sub-Saharan Africa (SSA), remains limited. In many African societies, meat is not only a nutritional commodity but also a symbol of affluence, celebration, and social identity. As a result, meat consumption decisions are shaped by a blend of cultural, economic, and informational factors. Although awareness of health risks is increasing, consumers in emerging economies often face competing priorities, such as food affordability, accessibility, and trust in labelling systems, that influence their willingness to modify meat consumption (Verbeke et al., 2010; de Araújo et al., 2022). This context underscores the importance of understanding the psychological and motivational drivers underlying meat consumption reduction or substitution in the global South.

The present study adopts the Health Belief Model (HBM) to explain how perceptions and beliefs translate into behavioural intentions regarding red and processed meat consumption. The HBM posits that individuals engage in preventive health behaviours when they perceive themselves to be at risk of a serious health threat (perceived susceptibility and severity), believe that behavioural change will mitigate the risk (perceived benefit), and encounter cues that prompt action (Rosenstock, 1974, Champion and Skinner, 2008). In the context of meat consumption, perceived health risk reflects individuals’ assessment of disease vulnerability due to meat intake, health consciousness captures their motivation to maintain healthy lifestyles, and label orientation functions as an external cue that reinforces awareness and behavioural action (Asioli et al., 2017). Together, these constructs provide a comprehensive framework for predicting how consumers interpret health information and adjust their dietary behaviours.

Several contextual characteristics of SSA may shape HBM constructs differently from Western settings. In many SSA societies, red meat symbolizes hospitality, respect, and social status (Erasmus and Hoffman, 2017), creating a form of cultural attachment that alters how health messages are interpreted. When meat consumption is tied to identity and social belonging, individuals may perceive lower susceptibility and severity because the behaviour is socially valued rather than seen as risky (Graça et al., 2019a, b; Ruby and Heine, 2012). Cultural attachment can also weaken the perceived benefits of reducing meat, as health or environmental gains often appear less compelling than immediate cultural or social expectations, unless framed in culturally meaningful ways (Collier et al., 2021). In parallel, affordability, satiety needs, and access constraints frequently outweigh health considerations for many households (Vorster et al., 2011), heightening perceived barriers to change.

Information environments further complicate motivation. Nutrition and label literacy vary widely across the region (Laar et al., 2022) and fragmented or inconsistently enforced labelling systems limit the effectiveness of labels as cues to action (Bopape et al., 2021). Consequently, many consumers rely on sensory judgement or trust in local vendors rather than formal nutrition information (Nordhagen et al., 2022), meaning that motivation and self-efficacy are shaped more by social cues and local knowledge than by health warnings. The ongoing nutrition transition—marked by rising processed food availability and shifting disease burdens—also influences how risks and benefits are appraised (Popkin et al., 2020). Collectively, these cultural, economic, and informational dynamics demonstrate why HBM pathways established in Western contexts may not fully apply in SSA and underscore the need to examine how perceived risk, health consciousness, and label orientation function within this distinct context.

Building on these gaps, the present study examines how perceived health risk, health consciousness, and label orientation shape meat reduction and substitution behaviours among South African consumers. It further investigates whether health consciousness mediates the relationship between label orientation and intentions to reduce red and processed meat consumption, an important mechanism for understanding how health cues translate into behavioural change within an African context.

This cross-sectional, perception-based study assessed adult consumers’ health perceptions, label used and purchasing behaviour for red and processed meats in urban and peri-urban areas of Eastern Cape Province, South Africa. The study population comprised adults (≥18 years) who reported purchasing or consuming red and/or processed meat. Exclusion criteria included respondents who: (1) were vegetarian/vegan, (2) reported never purchasing or consuming red/processed meat, or (3) provided incomplete questionnaires (>30% missing data).

The questionnaire was developed based on prior literature on consumer perceptions of meat quality, health risk awareness, and food labelling behaviour (de Araújo et al., 2022). It included items on; sociodemographic, purchasing behaviour (5-point Likert) (frequency and type of meat products purchased, willingness to pay for quality), health perceptions (perceived risk of red or processed meat for chronic diseases and mortality, consumers’ view or red or processed meat products, purchase of low/reduced sodium red meat products, importance of added hormones to meat animals or ingredients to red meat products, purchase of grass-fed and cage free meat animal products), label use and health consciousness (10 items-measuring attitudes towards healthy purchases practices, packaging trust, preservatives and animal welfare). A pilot test (n = 30) was conducted to assess clarity, internal consistency, and completion time; minor adjustments were made to improve item wording and flow.

Data were collected via interviewer-administered paper questionnaires by trained enumerators following a standardized protocol. Enumerators received training on informed consent, neutral question delivery, and data entry. Informed consent was obtained from all participants. Questionnaires were reviewed daily for completeness; double data entry was employed to minimize transcription errors. To minimize social desirability bias, respondents completed the survey anonymously, were informed that there were no right or wrong answers and were encouraged to provide honest responses. Common method variance was mitigated through (1) psychological separation of constructs in the questionnaire, and (2) varying item formats across sections.

A heterogeneous, location-based sampling approach was used to ensure participation from consumers of diverse demographic backgrounds. Respondents were recruited from supermarkets, open markets, butcheries, and abattoirs across urban and peri-urban areas. Although the initial aim was to use stratified random sampling, a formal sampling frame with explicit strata (e.g. by gender, age, or education) was not available; therefore, the study relied on random selection within recruitment locations to achieve demographic variability. The sample size was determined using Cochran’s formula for categorical data:

Where: z = z-score at 95% confidence (1.96), p = estimated proportion of consumers with health concern (∼0.5), e = margin error (0.05). This yielded a minimum sample size of 384, which was increased to 400 respondents to account for incomplete responses.

All data were coded and analysed using IBM SPSS Statistics version 25.0 (IBM Corp., Armonk, NY, USA). Descriptive and inferential statistical techniques were employed to explore consumer perceptions, purchasing behaviour, and health concerns associated with red and processed meat consumption. Cross-tabulation with Chi-square (χ²) tests was employed to compare proportions of consumers according to purchasing and consumption habits as well associations between socio-demographic variables and categorical outcomes such as frequency of meat consumption, willingness to pay for higher quality meat, and label-reading behaviour.

The internal consistency of multi-item constructs, particularly those assessing consumer health consciousness, purchasing motivations, and perception of meat quality, was evaluated using Cronbach’s alpha. A threshold of ≥0.70 was considered acceptable for scale reliability. The Kaiser–Meyer–Olkin (KMO) measure of sampling adequacy and Bartlett’s test of sphericity were used to assess the appropriateness of the dataset for factor analysis. To reduce dimensionality and identify latent constructs underlying consumer attitudes and perceptions, exploratory factor analysis (EFA) was performed using principal component analysis with Varimax rotation. Factor extraction was based on eigenvalues greater than 1. Items with factor loadings ≥0.40 were retained. The emergent themes guided the construction of composite indices used in subsequent regression models.

Non-parametric correlations (Spearman’s rho) examined associations between health perceptions, label orientation, health consciousness, and consumption/purchasing behaviour. Correlation strength was interpreted as: ρ = 0.10–0.29 (small), 0.30–0.49 (moderate), ≥0.50 (strong). Multiple regression tested predictors of “reduction in red meat and processed meat consumption” with independent variables including, age, gender, household size, monthly income, label-score, health consciousness score, and perceived health risk that high consumption of red/processed meat increases chronic disease and preterm mortality. Missing data was handled by listwise deletion. Multicollinearity was assessed via tolerance/VIF; all VIFs were <2, indicating no collinearity concerns. Model fit and significance were evaluated with R²/Adjusted R², standard error (SE) of estimate, and the F-test from ANOVA.

To test the proposed mediation hypothesis, the PROCESS macro version 4.0 for SPSS (Hayes, 2017) was employed, specifying Model 4 to examine whether health consciousness mediated the relationship between label orientation (independent variable) and red meat reduction (dependent variable). The analysis used 5,000 bootstrap samples and a 95% bias-corrected confidence interval (CI) to estimate indirect effects. Mediation was considered significant when the 95% CI for the indirect effect did not include zero. All variables were mean-centred prior to analysis to reduce potential multicollinearity.

Chi-square analyses (Table 1) revealed several significant associations between sociodemographic factors and consumer behaviour. Age was significantly associated with preferred place of meat purchase (p < 0.001), type of meat consumed (p = 0.001), and purchase of low-fat and low-sodium products (p < 0.05). The reported significant association between age groups and the purchase of low-fat and low-sodium meat products could be linked to the health risks associated with high sodium and fat intake from meat products, such as hypertension and cardiovascular diseases (Kotopoulou et al., 2023). These health concerns might have driven certain age groups, particularly older adults, to opt for healthier meat options (Mendes et al., 2024). Gender was associated with frequency of meat consumption (p = 0.050) and portion size (p = 0.016). The association of gender with frequency and portion size of meat consumption is well documented across various studies. While meat consumption is often associated with masculinity in men and thus influences dietary choices in terms of consumption of larger portions of meat, women are more likely to reduce meat consumption for health or ethical reasons (Hayley et al., 2015; Rosenfeld and Tomiyama, 2021).

Household size was strongly associated (p < 0.05) with preferred outlet, product type, knowledge of meat labelling, media influence, and health-related product choices such as low-sodium and hormone-free meats (p < 0.05). Larger households may rely more on informal outlets and bulk purchases, with health considerations often secondary to affordability and quantity. Prior studies confirm that household composition strongly influences dietary choices, with larger families prioritizing affordability over individual health preferences (Revoredo-Giha et al., 2019).

Income and source of income were linked to purchase of low-fat/low-sodium products, label knowledge, and perceived health risks of red/processed meat (p < 0.05). Higher-income respondents were more likely to consider clean label and ethical attributes such as grass-fed or hormone-free meat, consistent with previous research showing that affordability remains a barrier to healthier or ethically produced meat consumption (Bryant, 2020; Collier et al., 2021). Educational attainment showed strong associations with nearly all behavioural outcomes, including frequency of meat consumption, label knowledge, and purchase of “clean label” products (p < 0.01). Highly educated respondents were more likely to interpret labels, seek healthier meat options, and reduce red meat. This supports evidence that nutritional literacy enhances risk perception and adoption of healthier eating behaviours (Asioli et al., 2017).

Figures 1-3 presents consumers’ ethical and production-related concerns regarding free-range and grass-fed meat products. A large majority of respondents (79.8%) considered grass-fed products as “somewhat to extremely important”. Purchasing frequency varied: for free-range products, 2.7% of respondents purchased them always, 20.3% most of the time, 61.8% sometimes, and 15% rarely. Similarly, for grass-fed products, 2.3% purchased always, 24.3% most of the time, 60.5% sometimes, and 12.3% rarely.

Health perceptions emerged as the primary reason for purchasing both product types, with 62.5% choosing free-range and 65.4% choosing grass-fed products because they were viewed as healthier. Animal welfare also played an important role, with 22.3% favouring free-range products due to more humane treatment of animals, while 21.6% associated grass-fed production with sustainability. A smaller proportion of consumers (15.3% for free-range and 13% for grass-fed) highlighted reduced processing and better taste, respectively, as key motivators.

These findings confirm that health considerations are the dominant driver of consumer preference for alternative production systems, consistent with previous studies linking such systems with superior nutritional quality and safety compared to conventional meat (Verbeke et al., 2010). Ethical concerns, particularly around animal welfare, were also significant, echoing earlier research showing that humane treatment strongly influences purchasing behaviour, even when price premiums are involved (Clark et al., 2016). Furthermore, the association of grass-fed products with sustainability reflects increasing consumer awareness of environmental impacts in livestock production (McNeill and Van Elswyk, 2012).

The rotated component matrix for health-conscious purchase behaviour items, derived from principal component analysis with Varimax rotation, is presented in Table 2. The KMO measure verified sampling adequacy (KMO = 0.688) and Bartlett’s Test of Sphericity was significant (χ² (45) = 454.22, p < 0.001), supporting factorability. Two conceptually distinct factors emerged: (1) label and outlet health/quality concern and (2) product safety and welfare concern. Factor 1 comprises items related to trust in nutritional labels, purchasing based on nutritional value, preference for credited outlets, and buying neatly packed products, with loadings ranging from 0.52 to 0.79. Factor 2 includes items reflecting concern about preservatives, animal welfare and hormone use, and cleanliness of the sale environment, with loadings ranging from 0.75 to 0.81. Together, Factors 1 and 2 account for 39.2% of the rotated variance (20.9 and 18.3%, respectively). Although the initial extraction indicated three components (Eigenvalues >1), the rotated solution revealed that the third component lacked a coherent pattern of strong loadings and did not represent a meaningful construct. Consistent with recommended EFA guidelines (Tabachnick and Fidell, 2019), only the two interpretable factors were retained for subsequent analyses.

The emergence of a label/outlet dimension and a distinct safety/welfare dimension is consistent with recent evidence that both information cues (labels, outlet credibility) and food safety/ethical production independently shape consumer decisions around meat (including substitution and reduced red meat choices). However, the modest communality for the actual reduction behaviour indicates that reducing red meat is influenced by additional drivers (cost, taste, habit, cultural norms), and thus predictors beyond health-concern constructs should be included in predictive models (Font-i-Furnols, 2023; Taillie et al., 2021, 2023).

Table 3 is the result of the factor analysis of meat quality perception before purchase. The factor analysis revealed a three-factor structure explaining 62.14% of the total variance (KMO = 0.657; Bartlett’s Test: χ² (45) = 472.96, p < 0.001), indicating the adequacy of the data for dimensional reduction. Consistent with established consumer behaviour theory, the extracted factors aligned with intrinsic and extrinsic quality cues, but with an important distinction: intrinsic attributes emerged as two separate subdimensions, highlighting nuanced differences in how consumers process meat quality information before purchase.

The intrinsic sensory quality (Factor 1) encompassed colour of meat (0.832), freshness (0.827), and smell (0.511). These attributes represent immediate sensory impressions that consumers can assess visually and olfactorily at the point of purchase, functioning as rapid heuristics for safety, freshness, and palatability. The extrinsic information and assurance (Factor 2) included price (0.528), quality of packaging materials (0.818), labelling (0.663), and place of slaughter/purchase (0.833). These cues are external to the product itself but signal quality, authenticity, and safety. They provide assurance when sensory cues are ambiguous and influence trust, value perception, and risk reduction. Factor 3 (intrinsic physical composition) consisted of fat content or lumps (0.889) and extent of marbling (0.440). These structural attributes are part of the product’s physical make-up and relate to expectations of tenderness, juiciness, and flavour, but also to health and nutritional concerns for certain consumer segments.

In line with established frameworks, meat quality perception is shaped by intrinsic cues, which are inherent to the product, and extrinsic cues, which are external but related to how the product is presented and marketed (Rajic et al., 2021, 2022). However, this study extends the traditional two-category model by revealing that intrinsic cues are not always perceived as a single dimension. The extrinsic dimension remained consistent with prior research, combining pricing, packaging, labelling, and origin cues into a unified construct that provides reassurance about product quality and authenticity (Machiels and Orth, 2017; Javeed et al., 2022). These findings imply that intrinsic quality perception may be multi-dimensional in certain consumer contexts, with freshness-related cues and composition-related cues playing distinct roles in decision-making.

A simple linear regression examined the effect of label orientation on consumers’ red meat reduction behaviour (Table 4). Results indicated that label orientation significantly predicted meat reduction, F (1, 299) = 7.54, p = 0.006, explaining 2.5% of the variance (β = 0.157, p = 0.006).

Although label orientation explained only a small proportion of the variance in red meat reduction (R² = 0.025), this finding is consistent with behavioural nutrition research, where individual psychological cues typically account for modest shares of variance in complex dietary behaviours. Food choice is inherently multidimensional and influenced by factors such as habit, income, taste preferences, and cultural norms; thus, low explanatory power is expected when a single predictor is modelled in isolation. In contrast, label orientation did not significantly predict substitution behaviour, (F (1, 299) = 0.33, p = 0.569, R² = 0.001), indicating that consumers who pay attention to meat labels are not necessarily more likely to purchase low-fat, grass-fed, or reduced-sodium meat products. This outcome aligns with previous studies showing that food labels often raise awareness without necessarily translating into behavioural substitution, particularly in contexts where consumers face price constraints, low label literacy, or limited product differentiation (Campos et al., 2011; Hoek et al., 2011; Aschemann-Witzel and Grunert, 2015). In low- and middle-income settings, label information may increase perceived health risks but fail to overcome economic or habitual barriers to product switching (Carlsson et al., 2022).

The result of mediation analysis examining whether health consciousness mediates the relationship between label orientation and intention to reduce red and processed meat consumption is shown in Table 5. Label orientation score served as the independent variable, health consciousness score as the mediator, and red and processed meat reduction as the dependent variable. The results indicated that label orientation had a significant positive effect on health consciousness (path a: $B=0.5989$⁠, SE = 0.0466, t = 12.865, p < 0.001), accounting for approximately 35.6% of the variance in health consciousness (⁠$R2=0.3563$⁠). Health consciousness, in turn, had a significant positive effect on meat reduction (path b: $B=1.1183$⁠, SE = 0.2078, t = 5.381, p < 0.001). The total effect of label orientation on meat reduction (path c) was significant and positive (⁠$B=0.4805$⁠, SE = 0.1749, t = 2.747, p = 0.0064). However, when health consciousness was included in the model, the direct effect (path c′) became negative and statistically non-significant (⁠$B=−0.1893$⁠, SE = 0.2085, t = −0.908, p = 0.3647), indicating full mediation.

The significant a path indicates that individuals who pay greater attention to food labels tend to exhibit higher health consciousness. This is consistent with earlier research showing that label use enhances awareness of diet–health relationships and fosters self-regulation in food choices (Campos et al., 2011; Koch et al., 2022). The significant b path demonstrates that heightened health consciousness leads to stronger intentions to reduce consumption of red and processed meats, a relationship supported by prior evidence linking health motivations to meat avoidance (Hoek et al., 2011).

The non-significant direct effect (c′) and significant indirect effect suggest that the impact of label orientation on meat-reduction behaviour is completely channelled through health consciousness. In other words, label orientation increases consumers’ awareness of the health implications of their diet, which subsequently encourages intentions to reduce red and processed meat intake. This form of “full mediation” indicates that without the cognitive activation of health concerns, label exposure alone may not directly translate into behavioural intentions. The present findings align with recent evidence that labelling and warning messages influence consumers’ meat-related choices primarily by enhancing perceived health risks rather than by direct persuasion. For instance, Taillie et al. (2023) found that health warning labels on red meat products significantly reduced the likelihood of their selection, largely by increasing perceived unhealthiness. Similarly, Koch et al. (2022) observed that pictorial health warnings on red meat dishes lowered purchase intentions, especially among consumers with higher baseline health awareness.

Moreover, several studies have identified health consciousness as a key psychological construct linking information exposure and dietary behaviour (Szakály et al., 2012). Health-conscious consumers actively use food labels to make dietary decisions that align with their well-being goals (Annunziata and Vecchio, 2016). Consequently, the present study extends this literature by demonstrating that labels not only inform but are associated with latent health concerns that translate into behavioural change intentions.

Furthermore, since the direct effect of labels was negative and non-significant, this may reflect countervailing processes, such as scepticism toward labelling or cognitive dissonance among habitual meat consumers. These findings are consistent with the notion of inconsistent mediation (Hayes, 2017), where the mediator accounts for the main effect and the residual direct path operates in the opposite direction.

Table 6 presents Spearman’s rank-order correlations between key perception, attitude, and behaviour variable. Health consciousness scores correlated moderately and positively with self-reported reduction of red meat (ρ = 0.375, p < 0.001), indicating that more health-oriented consumers are more likely to cut back. While this effect size is moderate by behavioural standards, it is both statistically and practically meaningful, reinforcing health consciousness as a key determinant of dietary change. Perceived health risk of red/processed meat consumption correlated negatively with views of red meat (ρ = −0.178, p = 0.002) and, unexpectedly, negatively, with reported low-fat meat purchasing (ρ = −0.171, p = 0.003). Although both correlations are small in magnitude, they signal consistent patterns: higher perceived risk was associated with less positive attitudes toward red meat and a lower likelihood of substituting with low-fat variants.

By contrast, label score showed no significant association with low-fat purchasing meat purchasing (ρ = −0.021, p = 0.722) and only a weak non-significant tendency to align with perceived health risk of red/processed meat consumption (ρ = 0.104, p = 0.071). Health consciousness score was unrelated to low-fat meat/processed meat purchasing (ρ = −0.005, p = 0.934).

The negative relationship between perceived health risk of red/processed meat consumption and low-fat meat purchasing could imply that consumers with heightened risk perceptions may view low-fat products as insufficient to offset perceived harm and instead reduce or eliminate red meat altogether, consistent with “risk-avoidance” substitution patterns documented in health behaviour models (Hartmann and Siegrist, 2017). In our sample, health-concerned consumers appear to favour risk-avoidance (reduction/elimination) over risk-minimization (switching to low-fat variants). The absence of correlation between health-consciousness and low-fat meat purchasing reinforces a broader shift in health-oriented purchasing priorities: recent studies suggest consumers increasingly value processing, additives, and animal welfare claims over macronutrient-focused claims such as fat content (Asioli et al., 2017). Thus, low-fat positioning in red meat may not resonate as strongly with today’s health-conscious buyers compared to broader clean label or ethically produced messages.

The negative correlation between red meat views and perceived health risk of red/processed meat consumption mirrors experimental evidence demonstrating that risk-framing and warning labels can alter perceptions of red meat’s healthfulness and reduce purchase intentions. This aligns with Graça et al. (2019), who argue that risk perceptions foster attitudinal dissonance, prompting consumers to reconcile perceived harm with behaviour, often by reducing or rejecting meat consumption. This association underscores the perceptual link between risk awareness and attitudinal shift away from red meat, which is central to public health communication strategies.

The absence of an association between label orientation and low-fat meat and processed meat purchasing suggests that strong risk perception consumers who habitually consult labels may not necessarily translate this into within-category substitution (e.g. choosing low-fat over regular red meat). Instead, label-oriented, health-motivated consumers may exit the category entirely, reducing overall red and processed meat consumption rather than switching to “healthier” variants. This interpretation aligns with evidence from randomized choice experiments where front-of-pack health warnings led to category-level declines in red meat selection rather than preferential selection of leaner products (Sanchez-Sabate and Sabaté et al., 2019).

The multiple linear regression predicting reduction in red meat consumption is shown in Table 7. The full model was statistically significant, F (7, 293) = 11.047, p < 0.001, explaining 20.9% of the variance in reduction of red meat consumption (Adj. R² = 0.190). The model’s explained variance (Adj. R² ≈ 0.19) is typical for complex dietary behaviours determined by multiple, partly unmeasured factors—e.g. taste and habit, cultural norms/identity, availability, price, and social influences (Piazza et al., 2015; Sanchez-Sabate and Sabaté, 2019). Although label orientation showed a significant positive association with reduction intention in the bivariate model (Table 4), it was not a significant predictor in the multivariable model (β = 0.228, p = 0.166). Two predictors were independently associated with greater reduction: health consciousness score (B = 0.899, β = 0.294, p < 0.001) and perceived health risk from red/processed meat consumption (B = 0.328, β = 0.250, p < 0.001). Age showed a positive trend that did not reach conventional significance (p = 0.066). Gender, household size, monthly income, and label score were not statistically significant (p > 0.05) in the multivariable model. Multicollinearity was minimal (all VIFs <2).

The multivariable model shows that health consciousness and perceived health risk are the strongest independent correlates of intentions to reduce red and processed meat consumption. These effects mirror contemporary evidence that health orientation and risk salience are central drivers of meat-reduction behaviour, with studies consistently demonstrating that heightened disease-risk perceptions and stronger health motivation decrease willingness to consume red or processed meat (Bianchi et al., 2022; Graça et al., 2019a, b; Clarke et al., 2021a, b).

Once these psychological predictors are considered, the direct effect of label orientation dissipates, indicating that label attention alone is insufficient to prompt behaviour change. Instead, label use appears to influence reduction indirectly by strengthening broader health motivation, a pattern aligned with prior research showing that label effects are often mediated by health consciousness rather than exerting independent behavioural force. This helps explain why label orientation was significant in bivariate models but non-significant when adjusted for other predictors. Consistent with previous work, labels may shape perceptions and narrow choice sets, but their behavioural impact is modest compared with underlying attitudes and risk beliefs (Fetscherin et al., 2024), and they are more likely to encourage category reduction than substitution (Carrieri and Principe, 2022).

The tendency of South African consumers to reduce rather than substitute red meat further extends the HBM by highlighting how cultural and contextual factors shape the evaluation of benefits and barriers. Although HBM assumes individuals adopt the least costly behaviour that mitigates health risk, our findings show that when cultural attachment to red meat is high and acceptable substitutes are limited, consumers may view reduction, not substitution, as the more feasible and culturally coherent response. Substitution carries social, sensory, and economic barriers that make it less viable than simply consuming less. This suggests that in contexts where cultural meaning, product availability, and affordability strongly influence health decisions, behavioural responses may involve strategic category exit rather than within-category adjustment. Such patterns underscore the need to expand HBM applications to account for cultural norms and market realities that shape how benefits and barriers are weighed.

These findings, interpreted as exploratory, provide first-hand evidence from an African consumer setting where little empirical work has been conducted. Unlike studies in HICs that often report substitution behaviours (e.g. shifting to low-fat variants), our results suggest that South African consumers primarily respond to health concerns by reducing overall intake. This study also demonstrates that health consciousness and perceived risks play a central role in shaping consumer decisions regarding red and processed meat in South Africa.

A key limitation is the reliance on self-reported intentions, which may be susceptible to social desirability bias and may not always translate into actual behaviour. Although we partially addressed this by conducting a robustness check against reported consumption frequency, future studies should incorporate objective measures or longitudinal follow-up to validate whether reported intentions predict actual reductions. Also, future research should build on these exploratory insights by adopting longitudinal and experimental designs, comparing rural and urban populations, and incorporating qualitative approaches to capture deeper cultural influences. Such work will be essential for developing context-appropriate dietary guidelines and consumer education programs that address the unique realities of African populations undergoing rapid nutrition transition.