Identifying patient-level risk factors associated with non-β-lactam resistance outcomes in invasive MRSA infections in the United States using chain graphs

Background

MRSA is one of the most common causes of hospital- and community-acquired infections. MRSA is resistant to many antibiotics, including β-lactam antibiotics, fluoroquinolones, lincosamides, macrolides, aminoglycosides, tetracyclines and chloramphenicol.

Objectives

To identify patient-level characteristics that may be associated with phenotype variations and that may help improve prescribing practice and antimicrobial stewardship.

Methods

Chain graphs for resistance phenotypes were learned from invasive MRSA surveillance data collected by the CDC as part of the Emerging Infections Program to identify patient level risk factors for individual resistance outcomes reported as MIC while accounting for the correlations among the resistance traits. These chain graphs are multilevel probabilistic graphical models (PGMs) that can be used to quantify and visualize the complex associations among multiple resistance outcomes and their explanatory variables.

Results

Some phenotypic resistances had low connectivity to other outcomes or predictors (e.g. tetracycline, vancomycin, doxycycline and rifampicin). Only levofloxacin susceptibility was associated with healthcare-associated infections. Blood culture was the most common predictor of MIC. Patients with positive blood culture had significantly increased MIC of chloramphenicol, erythromycin, gentamicin, lincomycin and mupirocin, and decreased daptomycin and rifampicin MICs. Some regional variations were also observed.

Conclusions

The differences in resistance phenotypes between patients with previous healthcare use or positive blood cultures, or from different states, may be useful to inform first-choice antibiotics to treat clinical MRSA cases. Additionally, we demonstrated multilevel PGMs are useful to quantify and visualize interactions among multiple resistance outcomes and their explanatory variables.

Etiology and risk factors for bovine respiratory disease in pre-weaned calves on California dairies and calf ranches

Our study objective was to estimate the magnitude of association of BRD risk factors including failure of passive immunity transfer, sex, age, and the detection of suspected BRD etiological pathogens in pre-weaned dairy calves in California. A conditional logistic regression model and a mixed-effects logistic regression model were used to estimate the association of these potential risk factors with BRD from a matched and nested case-control studies, respectively. For each exposure covariate, the odds ratio (OR) is the ratio of odds of an exposure in a BRD calf (case) to that in a non-BRD calf (control). In the matched case-control study, an interaction term between failure of transfer of passive immunity and sex of calf showed that female calves were more negatively impacted by failure of transfer of passive immunity compared to male calves. The odds ratios comparing failure of transfer of passive immunity in BRD score positive calves versus controls for male calves was 1.34 (95 % CI: 0.87, 2.06) and was 2.47 (95 % CI: 1.54, 3.96) for female calves. The model odds ratios varied from 1.74 (95 % CI: 1.26, 2.42) for Mycoplasma spp. to 9.18 (95 % CI: 2.60, 32.40) for Histophilus somni, with Mannheimia haemolytica and Pasteurella multocida having an OR of 6.64 (95 % CI: 4.39, 10.03) and 6.53 (95 % CI: 4.44, 9.59), respectively. For bovine respiratory syncytial virus positive calves, the OR was 4.60 (95 % CI: 3.04, 6.97). Findings from the nested case-control study showed that based on thoracic ultrasonography findings consistent with BRD, the odds of a calf being 1 day older compared to a day younger were 1.01 (95 % CI: 1.00, 1.02) among BRD cases. For the bacterial and viral pathogens, the OR for Mycoplasma spp. and Pasteurella multocida were 1.85 (95 % CI: 1.24, 2.75) and 1.86 (95 % CI: 1.28, 2.71), respectively. The OR values for these pathogens were similar when both thoracic auscultation and ultrasound findings were used to detect cases of BRD. Based on positive scores for BRD using the California BRD scoring system, the OR for facility type, calf ranch versus dairy farm, was 3.17 (95 % CI: 1.43, 7.01), Mannheimia haemolytica was 3.50 (95 % CI: 2.00, 6.11), Pasteurella multocida was 1.78 (95 % CI: 1.21, 2.60), and bovine coronavirus was 2.61 (95 % CI: 1.85, 3.70). Results from both study designs showed the difference in relative contributions of age, sex, immune status, and pathogens in BRD occurrence between cases and controls in pre-weaned dairy calves.

A novel risk assessment tool for bovine respiratory disease in preweaned dairy calves

Due to the increased morbidity and mortality of bovine respiratory disease (BRD) in dairy calves, as well as an increasing urgency for the judicious use of antimicrobials in farm animals, a comprehensive risk assessment tool for BRD in preweaned dairy calves has been designed based on a longitudinal and a cross-sectional study. As a multifactorial disease complex in which immune function stressors increase susceptibility to respiratory pathology, risk management programs for environmental and husbandry practices may be an effective approach for BRD control. Practices of known or suspected effect on BRD in preweaned calves have been explored in 2 large studies correlating management factors to BRD prevalence (BRD 100 study) and incidence (BRD 10K study) and forming the scores presented here. Priority was given to results from multivariable over univariable model estimates. However, when used, univariable model estimates were adjusted for confounders or stratified by effect modifiers if necessary. Regression coefficients were translated into scores, which are presented in a field-ready tool consisting of (1) a risk assessment questionnaire, which identifies the herd-specific risk factors and the risk scores associated with each; (2) the California BRD scoring system to estimate the BRD prevalence at the time of risk assessment for future comparison with the prevalence after interventions; and (3) the BRD control and prevention herd management plan, which can be used to plan and track the interventions identified. Scores for 100 dairies across California were used to benchmark a dairy's risk on a spectrum. With the help of the risk assessment tool, dairy producers, calf managers, and veterinarians may be able to adjust management factors that affect BRD risk on a farm and objectively monitor BRD prevalence before and after management interventions. As a result, the BRD risk assessment tool described here is the first comprehensive effort for herd-specific BRD control and prevention.

Management factors associated with bovine respiratory disease in preweaned calves on California dairies: The BRD 100 study

The objective of this cross-sectional study was to determine how management practices on California dairies may be associated with bovine respiratory disease (BRD) in preweaned calves. A convenience sample of 100 dairies throughout California, providing a study population of 4,636 calves, were visited between May 2014 and April 2016. During each farm visit, in-person interviews with the herd manager or calf caretaker were conducted to collect information about herd demographics, maternity pen, colostrum and calf management, herd vaccinations, and dust abatement. A random sample of preweaned calves was identified and evaluated for the presence of BRD using a standardized tool. A survey-adjusted generalized linear mixed model with a logit link function was fitted with calf as the unit of analysis and dairy as the random effect. Mean study herd size (±SE) was 1,718 (±189.9) cows. Survey-adjusted estimates of breed types in the sample were 81.6% (±0.6) Holstein, 13.1% (±0.4) Jersey, and 5.3% (±0.5) crossbred or other purebred breeds, and calf sex proportions were 73.8% (±1.0) female and 26.2% (±1.0) male. Overall survey-adjusted BRD prevalence in the study herds was 6.91% (±0.69). Housing factors positively associated with BRD were metal hutches compared with wood hutches [odds ratio (OR) = 11.19; 95% confidence interval (CI) = 2.80–44.78], calf-to-calf contact in calves >75 d of age (OR = 9.95, 95% CI = 1.50–65.86), feeding Holstein calves <2.84 L of milk or replacer per day (OR = 7.16, 95% CI = 1.23–41.68), and lagoon water used for flushing manure under hutches compared with no flush (OR = 12.06, 95% CI = 1.93–75.47). Providing extra shade over hutches (OR = 0.08; 95% CI = 0.02–0.37), feeding calves at least 90% saleable milk (OR = 0.27, 95% CI = 0.13–0.54) or pasteurized milk (OR = 0.10; 95% CI = 0.03–0.36), and feeding >5.68 L of milk or replacer per day to Jersey calves (OR = 0.04; 95% CI = 0.01–0.28) were negatively associated with BRD. Our study identified management practices on California dairies with variability and that may contribute to differences in BRD prevalence, which will be incorporated into a risk-assessment tool to control and prevent BRD in preweaned dairy calves.

Shared Multidrug Resistance Patterns in Chicken-Associated Escherichia coli Identified by Association Rule Mining

Using multiple antimicrobials in food animals may incubate genetically-linked multidrug-resistance (MDR) in enteric bacteria, which can contaminate meat at slaughter. The U.S. National Antimicrobial Resistance Monitoring System tested 14,418 chicken-associated Escherichia coli between 2004 and 2012 for resistance to 15 antimicrobials, resulting in >32,000 possible MDR patterns. We analyzed MDR patterns in this dataset with association rule mining, also called market-basket analysis. The association rules were pruned with four quality measures resulting in a <1% false-discovery rate. MDR rules were more stable across consecutive years than between slaughter and retail. Rules were decomposed into networks with antimicrobials as nodes and rules as edges. A strong subnetwork of beta-lactam resistance existed in each year and the beta-lactam resistances also had strong associations with sulfisoxazole, gentamicin, streptomycin and tetracycline resistances. The association rules concur with previously identified E. coli resistance patterns but provide significant flexibility for studying MDR in large datasets.

Regional management practices and prevalence of bovine respiratory disease in California's preweaned dairy calves

The objective of this cross-sectional study was to estimate the prevalence of bovine respiratory disease (BRD) in California preweaned dairy calves and identify management practices that are associated with BRD. A convenience sample of 100 dairies in the 3 distinct dairy regions of California was surveyed. Regions evaluated were Northern California (NCA), northern San Joaquin Valley (NSJV), and greater Southern California (GSCA). A questionnaire on calf management practices and demographic information was administered via in-person interviews at each dairy and a random sample of preweaned calves was evaluated using the California BRD scoring system on the same day. Prevalence of BRD varied between the 3 dairy regions: 9.30% in NCA, 4.51% in NSJV, and 7.35% in GSCA. Breed was not associated with BRD prevalence at the statewide level, but differences in prevalence were observed between breeds across the regions with a higher prevalence in NCA for Jerseys and in GSCA for Holsteins, compared with NSJV. Prevalence of BRD was not different between organic and conventional dairies. Colostrum management practices, including heat treatment and feeding colostrum from multiparous cows, varied by region and were associated with lower BRD prevalence. Calves housed in group pens, a practice observed primarily in NCA, had a higher BRD prevalence than those in individual housing. Feeding salable milk was also more common in NCA and was associated with lower BRD prevalence. Ground and road surfaces adjacent to the calf raising area were also variable by region, and paved surfaces were associated with lower BRD prevalence. Management practices associated with BRD varied across the state and may be addressed to inform the adoption and implementation of potentially protective management decisions on California dairies and other regions with similar dairy systems.

Association of plasma haptoglobin concentration and other biomarkers with bovine respiratory disease status in pre-weaned dairy calves

We conducted a nested, case-control study of pre-weaned dairy calves ( n = 477; 4 California dairy farms) to assess the association between bovine respiratory disease (BRD) and hematologic biomarkers, including plasma haptoglobin (Hp) and plasma bactericide (PB). At each location, heifer or bull dairy calves were observed 2-4 times per week until confirmed as BRD-positive using parallel interpretation of thoracic ultrasound examination and auscultation. In addition, control calves were enrolled after being confirmed as BRD-negative using ultrasound and auscultation. Complete blood counts (CBC), PB, and Hp concentrations were measured. Hp values were higher in calves with confirmed BRD than in controls ( p < 0.01). The area under the curve (AUC) for the various biomarkers was obtained from the corresponding receiver operating characteristic curves. The AUC for Hp was 0.68, a value greater than those for PB or the remaining CBC parameters, indicating that Hp may be the most useful biomarker of BRD in pre-weaned dairy calves. The cutoff value for Hp was 0.195 g/L.

Markov Networks of Collateral Resistance: National Antimicrobial Resistance Monitoring System Surveillance Results from Escherichia coli Isolates, 2004-2012

Surveillance of antimicrobial resistance (AMR) is an important component of public health. Antimicrobial drug use generates selective pressure that may lead to resistance against to the administered drug, and may also select for collateral resistances to other drugs. Analysis of AMR surveillance data has focused on resistance to individual drugs but joint distributions of resistance in bacterial populations are infrequently analyzed and reported. New methods are needed to characterize and communicate joint resistance distributions. Markov networks are a class of graphical models that define connections, or edges, between pairs of variables with non-zero partial correlations and are used here to describe AMR resistance relationships. The graphical least absolute shrinkage and selection operator is used to estimate sparse Markov networks from AMR surveillance data. The method is demonstrated using a subset of Escherichia coli isolates collected by the National Antimicrobial Resistance Monitoring System between 2004 and 2012 which included AMR results for 16 drugs from 14418 isolates. Of the 119 possible unique edges, 33 unique edges were identified at least once during the study period and graphical density ranged from 16.2% to 24.8%. Two frequent dense subgraphs were noted, one containing the five β-lactam drugs and the other containing both sulfonamides, three aminoglycosides, and tetracycline. Density did not appear to change over time (p = 0.71). Unweighted modularity did not appear to change over time (p = 0.18), but a significant decreasing trend was noted in the modularity of the weighted networks (p < 0.005) indicating relationships between drugs of different classes tended to increase in strength and frequency over time compared to relationships between drugs of the same class. The current method provides a novel method to study the joint resistance distribution, but additional work is required to unite the underlying biological and genetic characteristics of the isolates with the current results derived from phenotypic data.

Sensitivity and specificity of on-farm scoring systems and nasal culture to detect bovine respiratory disease complex in preweaned dairy calves

The California (CA) and Wisconsin (WI) clinical scoring systems have been proposed for bovine respiratory disease complex (BRDC) detection in preweaned dairy calves. The screening sensitivity (SSe), for estimating BRDC prevalence in a cohort of calves, diagnostic sensitivity (DSe), for confirming BRDC in ill calves, and specificity (Sp) were estimated for each of the scoring systems, as well as for nasal swab cultures for aerobic bacteria and mycoplasma species. Thoracic ultrasound and auscultation were used as the reference standard tests interpreted in parallel. A total of 536 calves (221 with BRDC and 315 healthy) were sampled from 5 premises in California. The SSe of 46.8%, DSe of 72.6%, and Sp of 87.4% was determined for the CA system. The SSe of 46.0%, DSe of 71.1%, and Sp of 91.2% was determined for the WI system. For aerobic culture, the SSe was 43.4%, DSe was 52.6%, and Sp was 71.3%; for Mycoplasma spp. culture, the SSe was 57.5%, DSe was 68.9%, and Sp was 59.7%. The screening and diagnostic sensitivities of the scoring systems were not significantly different but the Sp of the WI system was greater by 3.8%. Scoring systems can serve as rapid on-farm tools to determine the burden of BRDC in preweaned dairy calves. However, users may expect the SSe to be less than the DSe when confirming BRDC in an ill calf.

Survey of management practices related to bovine respiratory disease in preweaned calves on California dairies

In the spring of 2013, a survey of California (CA) dairies was performed to characterize management practices related to bovine respiratory disease in preweaned calves, compare these practices across geographic regions of the state, and determine the principal components that explain the variability in management between herds. The questionnaire consisted of 53 questions divided into 6 sections to assess management practices affecting dairy calves from precalving to weaning. The questionnaire was mailed to 1,523 grade A licensed dairies in CA and 224 responses (14.7%) were collected. Survey response rates were similar over the 3 defined regions of CA: northern CA, northern San Joaquin Valley, and the greater southern CA region. The mean size of respondent herds was 1,423 milking cows. Most dairies reported raising preweaned calves on-site (59.7%). In 93.3% of dairies, preweaned calves were raised in some form of individual housing. Nonsaleable milk was the most frequent liquid diet fed to preweaned heifers (75.2%). Several important differences were identified between calf-raising practices in CA and practices reported in recent nationwide studies, including herd sizes, housing practices, and sources of milk fed to heifers. The differences between the CA and nationwide studies may be explained by differences in herd size. Regional differences within CA were also identified. Compared with the 2 other regions, northern CA dairies were found to have smaller herds, less Holstein cattle, calves remained with dams for longer periods of time after calving, were more likely to be certified organic dairies, and raised their own calves more often. Principal component analysis was performed and identified 11 components composed of 28 variables (questions) that explained 66.5% of the variability in the data. The identified components and questions will contribute to developing a risk assessment tool for bovine respiratory disease in preweaned dairy calves.

Development of a novel clinical scoring system for on-farm diagnosis of bovine respiratory disease in pre-weaned dairy calves

Several clinical scoring systems for diagnosis of bovine respiratory disease (BRD) in calves have been proposed. However, such systems were based on subjective judgment, rather than statistical methods, to weight scores. Data from a pair-matched case-control study on a California calf raising facility was used to develop three novel scoring systems to diagnose BRD in preweaned dairy calves. Disease status was assigned using both clinical signs and diagnostic test results for BRD-associated pathogens. Regression coefficients were used to weight score values. The systems presented use nasal and ocular discharge, rectal temperature, ear and head carriage, coughing, and respiratory quality as predictors. The systems developed in this research utilize fewer severity categories of clinical signs, require less calf handling, and had excellent agreement (Kappa > 0.8) when compared to an earlier scoring system. The first scoring system dichotomized all clinical predictors but required inducing a cough. The second scoring system removed induced cough as a clinical abnormality but required distinguishing between three levels of nasal discharge severity. The third system removed induced cough and forced a dichotomized variable for nasal discharge. The first system presented in this study used the following predictors and assigned values: coughing (induced or spontaneous coughing, 2 points), nasal discharge (any discharge, 3 points), ocular discharge (any discharge, 2 points), ear and head carriage (ear droop or head tilt, 5 points), fever (≥39.2°C or 102.5°F, 2 points), and respiratory quality (abnormal respiration, 2 points). Calves were categorized “BRD positive” if their total score was ≥4. This system correctly classified 95.4% cases and 88.6% controls. The second presented system categorized the predictors and assigned weights as follows: coughing (spontaneous only, 2 points), mild nasal discharge (unilateral, serous, or watery discharge, 3 points), moderate to severe nasal discharge (bilateral, cloudy, mucoid, mucopurlent, or copious discharge, 5 points), ocular discharge (any discharge, 1 point), ear and head carriage (ear droop or head tilt, 5 points), fever (≥39.2°C, 2 points), and respiratory quality (abnormal respiration, 2 points). Calves were categorized “BRD positive” if their total score was ≥4. This system correctly classified 89.3% cases and 92.8% controls. The third presented system used the following predictors and scores: coughing (spontaneous only, 2 points), nasal discharge (any, 4 points), ocular discharge (any, 2 points), ear and head carriage (ear droop or head tilt, 5 points), fever (≥39.2°C, 2 points), and respiratory quality (abnormal respiration, 2 points). Calves were categorized “BRD positive” if their total score was ≥5. This system correctly classified 89.4% cases and 90.8% controls. Each of the proposed systems offer few levels of clinical signs and data-based weights for on-farm diagnosis of BRD in dairy calves.