This study was conducted in and around Gondar town in Central Gondar zone, in Amhara National Regional State. Gondar town is located between 12°36' North latitude and 27°2' East longitude. The annual mean minimum and maximum temperature vary between 12.3–17.7°C and 22–30°C, respectively, with an annual average temperature of 19.7°C (1).

All cows used in the study were Holstein Friesian × Zebu crossbreed cows. The majority of herds were kept indoors and were hand-fed. Hay, wheat bran, crop residue, and brewery byproduct were used to feed the cows. Cows were milked by hand two times a day early in the morning and late in the afternoon by stripping (milking by thumb and index finger), squeezing (milking by five fingers), or both, depending on the habit of the farmer and the size of the teats.

This study consists of two separate studies: prevalence and risk factors study and milk production losses and economics study. In the prevalence and risk factors study, data were collected from 81 dairy farms to gain insight into the prevalence and risk factors associated with SCH. The 81 dairy farms were obtained by random selection from the extensive list of dairy herds in Gondar town. In the milk production losses and economics study, data were collected on 12 farms to evaluate the effect of SCH on milk production losses. For this later study, dairy farms that had relatively “large herds,” dairy farms with a herd size > 6 cows, with “the same” breed and management were selected based on their willingness to participate in the study.

The data collected from 81 farms that were used to test for potential herd-level and cow-level risk factors were analyzed as follows. The univariable screening was done using mixed logistic regression models with a random herd effect. Correlations between pairs of independent variables were evaluated using the Spearman rank correlations. If two variables had a correlation coefficient of ≥ |0.7|, only one of the variables was included in the further multivariable analysis. Both herd-level and cow-level risk factors that were statistically significant at P < 0.25 in the univariate analyses were tested starting from the most significant variable by removing one variable at a time in the same multivariate multilevel logistic regression models using backward reduction. Variables in multivariable models with P < 0.1 from the Wald test were retained. All the two-way interactions between variables in the final multivariable models were tested, but no significant interactions were found. Confounding was checked during the model building process by evaluating the change in the beta estimate of other variables when a variable was removed from the models. If this change in beta estimate was > 30%, the variable was considered a confounder. Dystocia and retained placenta were strongly correlated with metritis (r = 0.78); therefore, these two variables were excluded from the analyses. The data analysis was performed using Stata release14 (StataCorp LLC, USA).

Economic losses of SCH were estimated after estimating the average daily milk production losses of SCH by subtracting the average daily milk yield in cows with SCH from the average daily milk yield of cows without SCH over the 8 days period. Costs of SCH were calculated based on the average daily milk production loss due to SCH per cow in combination with information obtained from the questionnaire. As farmers did not have the information on the average daily milk production loss, we used the half-a-day milk yield measurement over the 8 days to estimate the average daily milk production losses associated with SCH. Economic losses were determined as costs per cow per lactation. To estimate these costs, we made the following assumptions:

Cows with blood total calcium threshold <2 mmol/l were considered subclinical hypocalcemic (9).

Where:

ADML_CSCH = Average daily milk production losses per cow associated with SCH based on the half a day milk yield measurement over 8 days

DMY_CȼSCH = Total half a day milk yield in liter in cows without SCH over an eight-day period

N_CȼSCH = Number of cows without SCH

DMY_CSCH = Total half a day milk yield in liter in cows with SCH over an eight-day period

N_CSCH = Number of cows with SCH

Economic losses as a result of the reduction in milk yield associated with SCH per cow per day was estimated as

Where:

ELCD_SCH = Economic losses associated with SCH per cow per day

ADML_CSCH = Average daily milk production losses per cow associated with SCH based on the half a day milk yield measurement over 8 days

P_M = The average price of milk per liter in Ethiopian Birr (ETB) during the study period

Economic losses per cow per lactation as a result of the reduction in milk yield associated with SCH was estimated as

Where:

ELCL_SuM = Economic losses per cow per lactation associated with SCH

ELCD_SCH = Economic losses associated with SCH per cow per day

L_L = Lactation length

Milk prices used in the calculations were derived from a questionnaire survey that was conducted among the owners of the participating farms. Data used to calculate costs of SCH are available in Supplementary Material 2.

The prevalence of SCH was 20.3% (51/221, 90% CI = 18.6–28.1%). There was a great variation in the prevalence of SCH among the different groups of cows. The highest prevalence of SCH was estimated in cows with poor body condition (40%) while the lowest was in cows with metritis (4.17%). Based on the univariate logistic regression analysis, six of the variables were significantly associated with SCH (p < 0.25) (Table 3).

Four variables were excluded from the final multivariate model. Grazing on pasture, membership to dairy cooperative, feeding meeting and above the cows' requirement (based on farmers' opinion), and previous experience of metritis reduced the odds of SCH (Table 4).

Cows with and without SCH had an average daily milk yield of 9.45 and 9.90 L, respectively, as estimated by half-a-day milk yield measurement of 8 days. The minimum, average, and maximum values of the most important input items and economic loss associated with SCH are shown in Table 5. Daily milk yield of cows and milk price varied largely between farms.

The objectives of this study were to estimate prevalence, identify factors associated with SCH, and estimate average economic losses associated with reduced milk production as a result of SCH in urban and periurban dairy farms keeping Holstein Friesian × Zebu crossbreed lactating dairy cows in Northwestern Ethiopia.

The prevalence of SCH was 20.3%. We collected samples regardless of the stages of lactation although the prevalence of SCH is expected to be high at the early stages of lactation (22). Therefore, the prevalence of SCH in the current would be much higher if the sampling were done at the early stage of calving, particularly in the first 48 h (9). To our knowledge, there are no studies on SCH in Ethiopia and other developing countries even though SCH is more common in dairy farms (23, 24) and is responsible for large economic losses than the clinical form (24). Higher milk yield in high-producing dairy cows leads to increased metabolic demand on the cow (4), resulting in an increased incidence of metabolic disorders (5). Therefore, it was not surprising to find the prevalence of 20.3% of SCH in the study farms which keep improved, high producing animals. Researchers from many parts of the world described that hypocalcemia (both clinical and subclinical) occur most commonly at the early stage of lactation (22, 25). For instance, Reinhardt et al. (9) from the United States of America reported a prevalence of SCH varying from 25 to 54%. However, the report was based on data at early lactation and there is a difference in production system and breed of cows that made comparison impossible.

Grazing on pasture, level of feeding, metritis, and membership to dairy farmers' cooperative were associated with SCH. It is not surprising to find lower odds of SCH in herds grazing on pasture compared to herds that did not graze on pasture as a low calcium diet promotes efficient resorption of calcium from bone and absorption from the small intestine. Cows that were, according to the farmers, fed above their requirement had lower odds of SCH. This might be due to the fact that dairy farmers may feed dietary calcium that was enough to meet their cows' requirements. Feeding low calcium diets in the late dry period activate calcium homeostatic mechanisms, and the cow is thus capable of absorbing calcium more efficiently from the intestine and drawing calcium from the bone around the time of calving (20). However, there is no availability of formulated rations sufficiently low in calcium in the Northwestern Ethiopian market to implement the low-calcium principle. Alternatively, these farmers can prevent SCH by feeding their cows calcium-rich rations 3–4 days before parturition, vitamin D supplementation, reducing the dietary cation-anion difference, and magnesium supplementation in the late gestation period (4).

In this study, metritis was found to be associated with reduced odds of SCH. Hypocalcemia impairs immune function and muscle contraction (18). In such a situation, metritis is more prone to occur (26). Martinez et al. demonstrated that cows with low calcium at least once between 0 and 3 days in milk had 4.5-fold increased odds of metritis (26). In a recent study in multiparous cows, Rodríguez et al. (27) also found greater odds of SCH in cows with metritis. Contradictory to many reports, in this study, cows with metritis have lower odds of SCH. This could be due to unobserved confounding in our analysis.

Membership to dairy cooperatives reduced the odds of SCH. Farms registered in dairy cooperatives had a better opportunity of getting training about different health problems of dairy cows. It can also be explained in relation to the experience in dairy farming. Farmers register in dairy cooperatives when their experience in dairy management is enough to decide to continue in dairy farming. In our study, we used serum electrolyte analysis which is not a perfect test to identify the true status of SCH, with both the imperfect sensitivity and specificity. Therefore, the strength of the associations between independent and dependent variables has probably been underestimated suggesting that some risk factors could not be identified.

People first have to consider they have a problem before they will be motivated to solve that problem (28). Accordingly, one of the first prerequisites for dairy farmers to implement measures to control SCH is creating awareness that SCH is an economic problem. It is, however, less likely that Northwestern Ethiopian dairy farmers are aware of SCH as costly, and even may not know SCH as a problem at all. To our knowledge, there has never been a study that estimated the economic losses associated with SCH and milk fever in Ethiopia. By estimating the economic impact of hypocalcemia, it may be possible to motivate the prevention of hypocalcemia in Ethiopia.

The average economic losses associated with SCH were 11.25 ETB (1 ETB = 0.023 US Dollar) and 3,026.25 ETB per cow per day and per lactation, respectively. Although we tried to base the estimate of the costs of SCH as much as possible on actual data, it was not possible to obtain all the needed data from this study. Therefore, we had to make assumptions, for instance, it was assumed that milk production level was considered constant throughout lactation. Because of this assumption, there could be potentially under or overestimation of the economic impacts of SCH. Moreover, converting the 8 days measurement of SCH toward a whole year would impact the precision of the estimate of the economic losses associated with SCH. As there is no previous study that estimated economic losses associated with SCH in Ethiopia, it was not possible to compare the estimate in this study with the economic losses associated with SCH in other previous studies. However, as most of the input values in this study were based on data collected directly from cows and dairy farmers, the study gives valuable insight into the economic impacts associated with SCH.

It is less likely that Ethiopian dairy farmers are aware of the existence of SCH because there is no routine measurement of calcium level or any other parameters to monitor SCH. In a situation where farmers are neither aware of the prevalence nor of the economic consequences of SCH, it is unlikely that farmers will make efforts to prevent milk SCH. Therefore, counseling dairy farmers about the impact of SCH and the potential profit of milk fever control measures is important to motivate and change in behavior of dairy farmers to improve health of dairy cows. Moreover, dairy farmers should be informed about the importance of feeding cows with diets low in calcium in the late dry period and/or calcium-rich rations 3–4 days before parturition, vitamin D supplementation, and magnesium supplementation in the late gestation period in preventing SCH. Even though it was not possible to compare with previous results, the prevalence of SCH was high in Northwestern Ethiopian dairy farms. Few risk factors were associated with SCH indicating the importance of a better understanding of factors that predispose to deficiency of calcium in Holstein Friesian × Zebu crossbreed cows in North-West Ethiopia. Milk production losses estimated in this study were moderate. However, the economic estimate had shown that Northwestern Ethiopian dairy farmers are losing larger income due to SCH. The results in this study may be useful to aware other smallholder dairy farmers with low milk yields in Ethiopia and other countries with similar circumstances.