Study population: Postpartum anestrus was defined as lack of estrus in a cow that had calved 60 or more days (3). It was evaluated in crossbred (Holstein × Zebu, n = 1286) cows of smallholder farms (n = 273) in the Chittagong, Joypurhat, and Sirajgonj districts of Bangladesh. The farms were registered with the Community-based Dairy Veterinary Foundation (CDVF) of Bangladesh Agricultural University, which provides scheduled preventive and emergency veterinary services to registered farms. Cows were hand milked twice daily; however, during the last part of lactation, many farmers milked their cows only once a day. An intensive dairy production system with tie-stall barns was practiced. Animals were generally fed a mixed ration consisting of agricultural by-products (wheat bran, rice polish, and oil cakes), rice straws, green fodder, and cut-and-carry grass. All cows in the randomly selected farms were taken into consideration to determine anestrus in the study population. Sample size was determined based on an estimated anestrus prevalence of 20% and an allowable error of 5% in the estimated prevalence.
Measurement of cyclicity: Cows not detected in estrus at 60 or more days postpartum were examined by transrectal ultrasonography to identify the underlying disorders. The ultrasonographic examinations were performed using a real-time B-mode ultrasound (PharVision Micro V10, Tequesta, FL, USA). The cows were examined again 10 days later when the result of the first examination was not conclusive. The findings were interpreted as follows: a) cows having 2 small ovaries without a CL at both examinations and without any abnormal finding in the uterus were grouped as true anestrus; b) cows were considered subestrus if a functional CL was found on at least 1 of the ovaries at either of the 2 examinations; c) cows with follicular or luteal structures more than 2.5 cm in diameter on both examinations were diagnosed with cystic ovaries; and d) cows with purulent material in the uterus with a CL on either of the ovaries were diagnosed with pyometra.
Data notation: Data on the following parameters were carefully recorded on a cow basis (body condition, suckling status, calving season, parity, milk production, and milking frequency) and herd basis (frequency of anestrus cows, herdsman’s sex and education, milk market, amount of income from the dairying for livelihood of the farmer, farm size, floor type, bull in farm, supplementation of vitamin/mineral premix in the ration, green grass availability, and time and duration of estrus observation) for every farm. All information was collected by the same person, by interviewing the farmers using a structured questionnaire, retrieving the requested data from the farm record book, or physical measurement throughout the study. Body condition score (BCS) of cows was given using a 1 to 5 scale (1 = emaciated, 5 = obese) with 0.5 increments (10) by the same person throughout the study. The milk market was recorded as traditional, cooperative, and ensured traditional (Pala group) based on the prevailing marketing channels (6). Pala is an informal group of farmers in remote areas collaborating for collective milk marketing (6). In the Pala marketing channel of the area studied, the farmer groups affiliated with CDVF had a higher bargaining power to negotiate a reasonable milk price.
Data management and statistical analysis: After a descriptive analysis, the likelihood of the biologically relevant risk factors was analyzed by using multiple logistic regression (Minitab version 15.1, Minitab Inc., State College, PA, USA). The continuous independent variables were transformed into categorical variables before being used in the statistical models. Cows diagnosed with cystic ovaries (n = 10) and pyometra (n = 6) were not included in the models. Daily milking frequency (once or twice), bull in farm (presence or absence), time of estrus observation (morning or evening; before or after milking), supplementation of vitamin/mineral premix in the ration (used or not), and green grass (available or not) were firstly tested for possible univariate associations with anestrus (P ≥ 0.10) and were excluded. Only risk factors that had univariate associations (P < 0.10) were subsequently evaluated in the multiple explanatory multivariable models of logistic regression analysis to generate odds ratios (ORs) with respective 95% confidence interval (CI). The statistical model (true anestrus vs. subestrus) for evaluating the cow level risk factors consisted of BCS, sucking, calving season, parity, and milk production. In another statistical model (anestrus vs. non-anestrus) the farm-level risk factors for anestrus such as the herdsman, farm size, milk market, estrus observation, sex and education of the herdsman, and dairying in livelihood were evaluated. Differences with P ≤ 0.05 were considered significant in the final model. Goodness-of-fit of the model was assessed by using the comparison of the deviance of the model to a χ2 distribution.