Milk consumption was found to be both significantly correlated with other markers of a diverse and high quality diet (i.e. vegetable, egg and meat consumption) and elevated among households owning dairy animal(s) (OR: 2.11), even after controlling for multiple confounders. The association between dairy animal ownership and milk consumption was not statistically significant, which might be explained by the relatively small sample size (n = 195). Of note, we included all households in the four villages that will be affected by the project. It is conceivable that families owning dairy animals would consume milk when the dairy animal is producing milk. However, families owning dairy animals were, in general, different from those not owning dairy animals. For example, they often owned greater assets (land, access to irrigation and motorised vehicle(s)) than those that did not. This was also consistent with the finding that woman-headed and SC households owned dairy animals far less frequently. This suggests ownership of dairy animals was associated with an overall higher socioeconomic status in the study area. The investment of purchasing and managing a high-yielding dairy animal may be prohibitive to those without assets [7, 12]. However, even after matching for these socioeconomic covariates through PSM, a positive association was found between dairy cow ownership and milk consumption, suggesting that this finding is not entirely dependent on socioeconomic status.
The advantage of owning land and irrigation access for meeting fodder and water needs of dairy animals was reported by other studies [7, 12]. Indeed some local farmers from nearby villages revealed cultivating only fodder crops in their irrigated fields, focusing solely on dairy for livelihood [13]. This indicates that while dairy animals have the potential to contribute dietary quality and diversity, the impact may be disproportionately higher for richer households, as has been shown in an earlier study [12].
Interpretation from a household nutrition perspective
The main finding of association between dairy animal ownership and household milk consumption was corroborated by a large study from India [11], and also smaller studies from Ethiopia (23% increased frequency) [8], Uganda [12, 22] and Kenya [23]. This was found to be especially important in areas without access to markets [8], which was not the case in our study area where dairies have been established.
While milk was significantly correlated with other markers of a diverse and high quality diet, it was not the only source of protein and micronutrients in the study area, as has also been reported in literature [9]. There is also consumption of finger millet, pulses (a regular feature in meals), eggs and meat, the latter two being more frequent among households owning dairy animals. These findings are in contrast to what was observed in some villages in northern India where consumption of milk and milk products were found to be more critical to dietary quality [7]. The importance of understanding local context in the contribution towards household nutrition is emphasised [7].
The proportion of households that reported having experienced food insecurity during the last two years was similar for households with or without dairy animal(s) (17.9% vs 22.5%). These percentages do not indicate the frequency and severity of the experienced food insecurity. In addition, it is difficult to draw causal inferences in the context of dairy animal ownership as this is a cross-sectional study.
Food consumption at household level cannot be extrapolated to nutritional status of individuals within the household, as shown before [6, 11]. A study from Ethiopia indicated positive impact on reducing stunting [8]. Studies from Uganda found significant positive impact [12], no impact [22], or even negative impact of dairy animal ownership on child nutrition, and hence, there must be other contextual factors, such as availability and use of sanitation and intra-household competition for resources. Small ruminants (e.g. goats and sheep) were found to contribute to better nutrition outcomes in Uganda [21] and the poorest households in Kenya [24]. Several other factors complicating this relationship have been elucidated in the literature, including wealth, resource constraints and experience of financial shocks [8].
Interpretation in the light of WSD projects
WSD projects locally have helped overcome the obstacle of high initial investment by providing grants and loans to procure livestock, preferentially to poor woman-headed households through SHGs [13]. Currently SHG membership was somewhat lower among households without dairy animals (33.3% against 41.7%), and this can be expected to improve through the planned WSD project [15]. Beneficiaries in earlier local WSD projects perceived financial and nutritional benefits following the adoption of dairy animal(s) [13]. On similar lines, an intervention study in Rwanda on donation of livestock to households was able to demonstrate impact on child nutrition [25]. However, keeping in mind that managing dairy animals also takes a lot of work and harbours various costs [26, 27] – including accessing water and feed [7], all households may not be able to adopt it.
Interventions encouraging dairy animal ownership as part of the WSD project should take into account whether it is feasible for low-income households to maintain a dairy animal long-term. Additionally, challenges of water and feed are worsened during droughts [7], which occur regularly in Kolar district. Local anecdotal evidence (assimilated during a recent study) [13] reported that several households sold their dairy animals a few years ago following a period of intense drought. Financial returns from dairy animals were also reportedly lower in areas with high groundwater exploitation [27], such as in the study area. In addition, promotion of dairy animals comes with health and ethical challenges especially for high-yielding varieties [28]. Therefore this strategy could be reviewed accordingly.
Interpreting effects of household size and wage labour
We found a strong association of milk consumption and household size (OR: 1.88, 95% CI: 1.34, 2.77), which is in contrast to findings from a large Indian dataset [11]. Two factors might explain this observation. First, wealthy households in the study area lived as joint families, as they have the financial and human resources to buy and manage dairy animals. Second, the poorest households were those of elderly women living alone. The strong association between wage labour and milk consumption (OR: 2.89, 95% CI 1.04, 9.03) may also be related to few households consisting only of elderly poor women living alone unable to engage in wage labour. Reportedly, regular wage labour in construction industry and domestic work in nearby cities was providing adequate returns to young people from this area [13].
Limitations of the study
It is not possible to draw conclusions on causal relationships from cross-sectional data. Reverse causality between household milk consumption and dairy animal ownership is plausible if milk consumption be considered a proxy for wealth/income. However, keeping the literature and context in mind, this is unlikely. However, as ownership of cattle was strongly associated with wealth indicators, the association with household milk consumption should be interpreted with caution. Another limitation of the analysis was the lack of data on other milk products. In our preceding work in the same region, we found that part of the milk was consumed in fermented form (curd) [13]. As this curd was made from fresh milk within the household, we assumed it was represented within the data on milk consumption. Finally, the findings of the present study mainly apply to the study area, but may also provide insights on what can be expected in the drought-prone rural regions in southern India.
Scope for future research and practice
The data analysed here stem from a baseline survey for a proposed WSD project. We will get further insights on how things changed due to the interventions in 2024–2025. Further research could adopt a prospective mixed-method design, focus on differential benefits experienced by various types of adopting households and also study challenges being faced by each in taking up and managing dairy animals. Adding health outcome measures as part of the survey (e.g. nutritional status among children within the household) and also micronutrients among adults (e.g. haemoglobin levels) would be good to estimate the size and distribution of direct health impacts of these interventions. This kind of evidence is currently lacking [5].
As the data used in this study came from a baseline survey for a HIA, it shows the benefits gained by conducting HIAs, as it fosters empirical research in settings that are usually neglected [29, 30]. Indeed, baseline survey data can be leveraged to better understand about agriculture and nutrition linkages, besides potentially other locally relevant health outcomes.