Using registry data to identify individual dairy cows with abnormal patterns in routinely recorded somatic cell counts.

Data from the Danish milk recording system routinely enter the Danish Cattle Database, including somatic cell counts (SCC) for individual animals. Elevated SCC can signal intramammary inflammation, suggesting subclinical mastitis. Detecting mastitis is pivotal to limit severity, prevent pathogen spread, and target treatment or culling. This study aimed to differentiate normal and abnormal SCC patterns using recorded registry data. We used registry data from 2010 to 2020 for dairy cows in herds with 11 annual milk recordings. To create consistency across herds, we used data from 13,996 unique animals and eight different herds, selected based on the amount of data available, only selecting Holstein animals and conventional herds. We fitted log10-transformed SCC to days in milk (DIM) using the Wilmink and Wood's curve functions, originally developed for milk yield over the lactation. We used Nonlinear Least Square and Nonlinear Mixed Effect models to fit the log10-transformed SCC observations to DIM at animal level. Using mean squared residuals (MSR), we found a consistently better fit using a Wood's style function. Detection of MSR outliers in the model fitting process was used to identify animals with log10(SCC) curves deviating from the expected "normal" curve for that same animal. With this study, we propose a method to identify single animals with SCC patterns that indicate abnormalities, such as mastitis, based on registry data. This method could potentially lead to a registry data-based detection of mastitis cases in larger dairy herds.

Lactational treatment of bovine mastitis-Development over time and factors influencing cytological cure.

Lactational treatment of bovine mastitis is a major contributor to antibiotic consumption in dairy cattle and is, therefore, important to address in light of the increasing problem of antibiotic resistance. In this large-scale database-based retrospective observational study, we combined electronic health records and routinely measured somatic cell counts from individual cows to create an overview of lactational mastitis treatment in Danish dairy herds from 2010 to 2019. Furthermore, posttreatment somatic cell count was used to approximate treatment success in terms of cytological cure. A generalized logistic regression with mixed effects was performed to combine knowledge on cow-level factors (treatment-, pathogen-, and cow-related) with the new infection risk at the herd level, and to explore the relative effect on cytological cure. The investigation revealed that the total number of lactational treatments appears to have decreased steadily over the study period, whereas treatment duration increased slightly. The proportion of cases treated with penicillin-based protocols and the proportion of milk samples sent for pathogen analysis also decreased. Meanwhile, results from the statistical analysis confirm the importance of cow-related factors, such as parity and lactation stage, for the probability of cytological cure following lactational treatment of mastitis. However, they also disclose that factors that are easier to adjust, such as optimizing treatment duration, including knowledge on causative pathogens and improving the herd-level new infection risk that can be used to positively influence the outcome. Application of this knowledge could potentially assist in promoting a more prudent use of antibiotics for dairy cattle in the future.

Dynamics of somatic cell count (SCC) and differential SCC during and following intramammary infections.

Somatic cell count is frequently used as an indicator of intramammary infections (IMI) in dairy cattle worldwide. The newly introduced differential SCC (DSCC) can potentially contribute to detection of IMI. The purpose of this study was to investigate the dynamics of SCC and DSCC after IMI. We used a data set with monthly samples from 2 Danish dairy herds through 1 yr, using bacterial culture to identify IMI. The dynamics of SCC and DSCC with regard to IMI were assessed at quarter level following new IMI with each of 3 defined pathogen groups, major, minor, or "other" pathogens, using general additive models. Both SCC and DSCC increased after IMI, with a more pronounced increase if major or other pathogens were detected compared with minor pathogens. We found that DSCC increased after IMI with other pathogens in both herds and, in herd 2, after IMI caused by major and minor pathogens. We also estimated the duration of increased SCC and DSCC when they exceeded a threshold, done separately for each pathogen group. Major pathogens had the longest-lasting effect in both herds for both SCC and DSCC. We conclude that the magnitude and duration of response of SCC and DSCC to IMI differs between herds and causative pathogens.

Economic and epidemiological impact of different intervention strategies for clinical contagious mastitis.

The overall aim of this study was to compare different intervention strategies for clinical intramammary infections (IMI). We conducted a simulation study to represent a Danish dairy cattle herd with IMI caused mostly by Staphylococcus aureus and 9 different intervention strategies for clinical IMI. A standard intervention of 3 d of treatment consisting of intramammary injections for all clinical cases was used. Two of the strategies reflected the use of more antibiotics and 6 strategies reflected cow-specific treatment or culling decisions. For these strategies, we assessed the cost and effectiveness of culling as an IMI intervention. Our results showed that nearly all strategies could reduce the number of IMI cases [e.g., a median of 37 clinical cases with the extended intramammary treatment over 5 d strategy (Basic5) and 30 clinical cases with the cow culled with recovery probability below 50% (Before50)] compared with the standard intervention (median of 42 clinical cases). This happened alongside either increased antibiotic usage (e.g., from a median of 123 treatment days up to 179 treatment days with strategy Basic5) or an increased number of cows culled in relation to IMI (e.g., from a median of 16 up to 24 cows with strategy Before50). Strategies with more antibiotics or reactive culling had a slightly higher net income (e.g., €190,014 median net income with strategy Basic5 or €196,995 with strategy Before50 compared with €187,666 with the standard strategy). This shows that a cow-specific clinical intervention approach can be cost-effective in reducing IMI incidence.

A strain-, cow-, and herd-specific bio-economic simulation model of intramammary infections in dairy cattle herds.

Intramammary infections (IMI) in dairy cattle lead to economic losses for farmers, both through reduced milk production and disease control measures. We present the first strain-, cow- and herd-specific bio-economic simulation model of intramammary infections in a dairy cattle herd. The model can be used to investigate the cost-effectiveness of different prevention and control strategies against IMI. The objective of this study was to describe a transmission framework, which simulates spread of IMI causing pathogens through different transmission modes. These include the traditional contagious and environmental spread and a new opportunistic transmission mode. In addition, the within-herd transmission dynamics of IMI causing pathogens were studied. Sensitivity analysis was conducted to investigate the influence of input parameters on model predictions. The results show that the model is able to represent various within-herd levels of IMI prevalence, depending on the simulated pathogens and their parameter settings. The parameters can be adjusted to include different combinations of IMI causing pathogens at different prevalence levels, representing herd-specific situations. The model is most sensitive to varying the transmission rate parameters and the strain-specific recovery rates from IMI. It can be used for investigating both short term operational and long term strategic decisions for the prevention and control of IMI in dairy cattle herds.

Determinants of antimicrobial treatment for udder health in Danish dairy cattle herds.

Societal pressure to limit the use of antibiotics in livestock production systems, including dairy cattle systems, is consistently increasing. To motivate farmers to reduce antibiotic usage, it is important to understand the factors that determine whether a cow will be treated with antibiotics or not. If farmers' usual practices regarding antibiotic treatments are taken into account, they may be motivated to adopt control measures that can facilitate prudent use of antibiotics and are at the same time cost-effective. In this study, we analyzed database recordings of milk yield and somatic cell count from the routine milk recording scheme, clinical registrations of mastitis and PCR results, and cow factors such as days in milk and parity in relation to antibiotic treatments for 518 dairy herds in Denmark. Farm-wise logistic regressions were used to predict antimicrobial treatment based on these factors. The resulting regression coefficients of 422 herds were further analyzed by principal component analysis and clustering to determine the driving predictors for treatment in different groups of farms. The results showed that determinants that were most important for predicting antibiotic treatments vary from one farm to another. Health indicators such as PCR or somatic cell count were most indicative for treatment on some farms, whereas other groups seemed to depend more on production factors (milk yield) or later culling of the cows. This shows that farmers behave differently and differences can be identified in register data. This information can be considered when developing cost-effective herd-specific control measures of mastitis to promote prudent use of antibiotics in Danish dairy cattle farms.

Reservoirs of Staphylococcus spp. and Streptococcus spp. Associated with Intramammary Infections of Dairy Cows.

To design cost-effective prevention strategies against mastitis in dairy cow farms, knowledge about infection pathways of causative pathogens is necessary. Therefore, we investigated the reservoirs of bacterial strains causing intramammary infections in one dairy cow herd. Quarter foremilk samples (n = 8056) and milking- and housing-related samples (n = 251; from drinking troughs, bedding material, walking areas, cow brushes, fly traps, milking liners, and milker gloves), were collected and examined using culture-based methods. Species were identified with MALDI-TOF MS, and selected Staphylococcus and Streptococcus spp. typed with randomly amplified polymorphic DNA-PCR. Staphylococci were isolated from all and streptococci from most investigated locations. However, only for Staphylococcus aureus, matching strain types (n = 2) were isolated from milk and milking-related samples (milking liners and milker gloves). Staphylococcus epidermidis and Staphylococcus haemolyticus showed a large genetic diversity without any matches of strain types from milk and other samples. Streptococcus uberis was the only Streptococcus spp. isolated from milk and milking- or housing-related samples. However, no matching strains were found. This study underlines the importance of measures preventing the spread of Staphylococcus aureus between quarters during milking.

Reservoirs of Corynebacterium spp. in the Environment of Dairy Cows.

Although Corynebacterium spp. can be regularly associated with subclinical and clinical mastitis cases in dairy cows, knowledge on their reservoirs in dairy farms is sparse. Therefore, samples were collected at 10 visits with 14 day intervals from bedding material (n = 50), drinking troughs (n = 20), different walking areas (n = 60), cow brushes (n = 8), fly traps (n = 4), the passage to pasture (n = 9) as well as milking liners (n = 80) and milker gloves (n = 20) in one dairy cow farm. Additionally, quarter foremilk samples from all lactating cows (approximately 200) were collected at each visit. All samples underwent microbiological examination and cultured isolates were identified using MALDI-TOF MS. Most Corynebacterium spp. that were cultivated from milk were also isolated from the housing environment and milking-related niches (C. amycolatum, C. confusum, C. stationis, C. variabile, C. xerosis) or from milking-related niches only (C. frankenforstense, C. pilosum, C. suicordis). C. bovis was not cultivated from any environmental niche, while being the dominant species in milk samples. This study demonstrates that many Corynebacterium spp. present in milk samples can also be isolated from the cows' environment. For C. bovis, the most relevant Corynebacterium species with regard to intramammary infections, it indicates that environmental reservoirs are of little relevance.

A stochastic modelling approach to determine the effect of diverse Staphylococcus aureus strains on the economic and epidemiological outcomes of mastitis intervention strategies in dairy cattle.

Staphylococcus aureus (S. aureus) strains with considerable genetic and phenotypic differences have previously been identified. The economic and epidemiologic impact of S. aureus mastitis has been investigated, but none of these studies took differences between strains into account. Here we aimed to investigate how differences between S. aureus strains affect the economic and epidemiologic outcome of various intervention strategies against clinical and subclinical intramammary infections. Five S. aureus strains were modelled using a stochastic bio-economic model simulating a dairy herd of 200 cows using single-day time steps. The strain characteristics of the five simulated S. aureus strains (general, contagious, spill-over, clinical and persistent) were based on divergent phenotypes as described in literature. Outcomes of the model included incidence (both clinical and subclinical), number of antibiotic treatment days, number of culled cows, and net income. Intervention strategies against clinical and subclinical intramammary infections were based on (variations of) intramammary antibiotic treatment, testing, and culling. Both single and multiple pathogen (intramammary infection caused by S. aureus, Escherichia coli, and non-aureus staphylococci) scenarios were simulated to determine the effect of the five S. aureus strains on the impact of 19 different intervention strategies. The results showed that the incidence (both clinical and subclinical), number of treatment days, number of culled cows, and net income varied considerably for the different S. aureus strains. Comparison of the model outcomes within and between strains showed that for most intervention strategies the relative impact differed per strain. However, the intervention strategy with the best outcome for most variables and strains was the culling of cows with a recovery probability lower than 50%. This shows that the relative economic and epidemiologic impact of most of the modelled intervention strategies were strain-dependent, while some intervention strategies were not strain-dependent. From this, we conclude that, depending on the intervention strategy applied on a farm, it could be advantageous to type S. aureus to determine whether it would be economically and epidemiologically beneficial for the existing intervention strategy to be changed.

A simulation study to investigate the added value in using differential somatic cell count as an additional indicator for udder health management in dairy herds.

Mastitis is one of the most costly diseases in dairy herds worldwide. Somatic cell count (SCC) is widely used as an indicator for subclinical intramammary infections (IMI) that may eventually cause mastitis in dairy herds. Differential somatic cell count (DSCC) has recently been introduced as an additional indicator for IMI. The objective of this study was to investigate the value of using DSCC as an additional indicator to select cows for testing and subsequent intervention for subclinical mastitis during the lactation. We parameterized an existing bio-economic simulation model for dairy herds to include DSCC. Then, we simulated three Danish dairy cattle herd situations with different pathogen distributions where the main pathogens were 1) Staphylococcus aureus, 2) Streptococcus agalactiae, and 3) Streptococcus uberis. In these herds, we simulated two different selection strategies for testing (bacterial culture) for subclinical IMI and various intervention strategies for test positive cases. The first selection strategy considered only SCC; cows were selected for testing if they had a low SCC measurement followed by two high SCC measurements. In the second selection strategy, cows additionally had to have a high DSCC measurement. Results showed that both selection strategies led to a similar net income and to a similar number of clinical and subclinical cases for all investigated intervention strategies. However, when using DSCC in the selection of animals, the number of treatment days and the number of cows culled in relation to IMI was reduced: The median annual number of treatment days was reduced by 25-38 days in herd 1, by 25-42 days in herd 2, and by 30-48 days in herd 3, depending on the intervention strategy. The median annual number of cows culled in relation to IMI was reduced by up to 8 cows (10 cows in herd 3) for one of the intervention strategies. Subject to limitations associated with model assumptions, these results suggest that considering DSCC when selecting cows for testing can reduce IMI related culling and the use of antibiotics without changing in-herd prevalence nor resulting in economic loss.

Associations between udder health and culling in dairy cows.

Culling is an important management tool in dairy herds, as it affects herd economics and animal welfare. In relation to health, culling is usually studied as a consequence of health disorders, but it can also be regarded as a tool to manage health in the herd by making strategic culling decisions. In this study, we used data from the Danish Cattle Database in herd-wise survival analyses to investigate factors associated with culling, in relation to udder health, in 1,452 dairy herds. The data included milk yield, somatic cell counts (SCC), parity, and different disease related factors with a special focus on udder health. In each herd, observations and survival analyses were divided into five groups: mid lactation heifers, late lactation heifers, early lactation cows, mid lactation cows, and late lactation cows. The results showed that a high average milk yield reduced the culling hazard, and a number of risk factors (e.g., parity, a high SCC or treatment of mastitis) were associated with an increased hazard for culling. Importantly, the strength and direction of many of these associations was dependent on the lactation stage. The resulting coefficients were further analysed by principal component analysis and clustering to explore variations in culling risk factors amongst herds. In some herds, parity was an important factor for culling, while in other herds, average milk yield, SCC, or udder health were more important factors. However, clusters were substantially overlapping, indicating that the decision making process underlying culling is complex and multifactorial.

Economic and epidemiological impact of different intervention strategies for subclinical and clinical mastitis.

The objective of this study was to evaluate and compare different combinations of intervention strategies for contagious or opportunistic subclinical and clinical intramammary infections (IMI). We simulated two different Danish dairy cattle herds with ten different intervention strategies focusing on cow-specific treatment or culling, including three baseline strategies without subclinical interventions. In one herd, the main causative pathogen of IMI was Staphylococcus (S.) aureus. In the other herd, Streptococcus (St.) agalactiae was the main causative agent. For both herds, we investigated costs and effectiveness of all ten intervention strategies. Intervention strategies consisted of measures against clinical and subclinical IMI, with baselines given by purely clinical intervention strategies. Our results showed that strategies including subclinical interventions were more cost-effective than the respective baseline strategies. Increase in income and reduction of IMI cases came at the cost of increased antibiotic usage and an increased culling rate in relation to IMI. However, there were differences between the herds. In the St. agalactiae herd, the clinical intervention strategy did not seem to have a big impact on income and number of cases. However, intervention strategies which included cow-specific clinical interventions led to a higher income and lower number of cases in the S. aureus herd. The results show that intervention strategies including interventions against contagious or opportunistic clinical and subclinical IMI can be highly cost-effective, but should be herd-specific.

Author Correction: Methods for estimating disease transmission rates: Evaluating the precision of Poisson regression and two novel methods.

A correction to this article has been published and is linked from the HTML and PDF versions of this paper. The error has been fixed in the paper.

Methods for estimating disease transmission rates: Evaluating the precision of Poisson regression and two novel methods.

Precise estimates of disease transmission rates are critical for epidemiological simulation models. Most often these rates must be estimated from longitudinal field data, which are costly and time-consuming to conduct. Consequently, measures to reduce cost like increased sampling intervals or subsampling of the population are implemented. To assess the impact of such measures we implement two different SIS models to simulate disease transmission: A simple closed population model and a realistic dairy herd including population dynamics. We analyze the accuracy of different methods for estimating the transmission rate. We use data from the two simulation models and vary the sampling intervals and the size of the population sampled. We devise two new methods to determine transmission rate, and compare these to the frequently used Poisson regression method in both epidemic and endemic situations. For most tested scenarios these new methods perform similar or better than Poisson regression, especially in the case of long sampling intervals. We conclude that transmission rate estimates are easily biased, which is important to take into account when using these rates in simulation models.