Dairy Herd Data Management: Using Records for Health Decisions

By Dr. Zubair Khalid, DVM, MS, PhD · 2026-08-11

Key Takeaways

• Dairy herd data management has transitioned from retrospective record-keeping to prospective decision-making tools, emphasizing the integration of multiple data streams (milk production, somatic cell count, body condition score, metabolic tests like BHB and NEFA, disease/culling records, and sensor-derived alerts) to identify problems early and monitor intervention efficacy.
• The transition period (three weeks pre- and post-calving) is critical, as physiological imbalances (insulin resistance, negative energy balance, hypocalcemia, reduced immune function) concentrate preventable diseases; monitoring should specifically target this period using metrics like daily milk weights, body condition changes, and metabolic profiles.
• Production records, particularly daily milk weights in early lactation, serve as early indicators of transition failure and subclinical disease, but must be interpreted alongside body condition scores (proxy for energy balance over weeks) and metabolic tests (snapshots of mobilization and oxidation) to assess subclinical disease prevalence.
• Sensor technologies (activity monitors, rumination collars) offer continuous data but require validation against clinical outcomes; validated traits include activity, feeding/drinking behavior, and physical condition, with accelerometer-based systems showing higher validation rates.
• A structured herd-level diagnostic sequence involves establishing baselines, comparing current values to benchmarks and published thresholds (e.g., prepartum NEFA > 0.4 mmol/L), analyzing temporal patterns, and stratifying data by parity and lactation stage to pinpoint specific management or disease issues.
• Decision-making for intervention is guided by thresholds for specific findings (e.g., subclinical ketosis prevalence >15%, negative energy balance indicated by mean BCS loss >0.5 units in first 30 days), with responses ranging from ration adjustments to reviews of milking routines and environmental hygiene.

Herd records have moved from retrospective ledgers to prospective decision tools. For the practicing veterinarian, the question is no longer whether data can improve herd health, but which data streams, at what frequency, and interpreted against which benchmarks, actually change clinical outcomes. This article provides a framework for using dairy herd records to identify problems early, monitor intervention efficacy, and allocate veterinary effort where it produces the greatest health and economic return. It is written for veterinarians who advise dairy clients and who need a structured approach to data interpretation that moves beyond single-cow diagnostics.

The clinical questions addressed here are practical. How do you distinguish a seasonal blip in milk production from a developing transition cow problem? Which metabolic tests add information beyond what production records already show? When does a sensor-derived alert warrant a farm visit? The answers depend on integrating multiple data types, understanding the physiological basis for each indicator, and applying monitoring criteria that have published evidence behind them. This article covers the conceptual foundation of herd-level monitoring, the data types available, the analytical logic for interpreting them, and the decision frameworks that turn records into health interventions. Later sections address specific disease syndromes and monitoring protocols.

At a Glance

The Physiological Basis for Herd-Level Monitoring

The transition period, defined as the three weeks before and three weeks after calving, concentrates most of the preventable disease in a dairy herd. Essentially all dairy cattle experience insulin resistance, reduced feed intake, negative energy balance, hypocalcemia, reduced immune function, and bacterial contamination of the uterus in this window. One-third of dairy cows may be affected by some form of metabolic or infectious disease in early lactation. These disturbances are not independent events. They share a common root in physiological imbalance, a state where regulatory mechanisms are insufficient for the animal to function optimally, leading to a high risk of a complex of digestive, metabolic, and infectious problems.

The clinical implication is that monitoring should target the transition period specifically, not the whole lactation uniformly. Health problems during the periparturient period relate largely to cows having difficulty adapting to the nutrient needs of lactation. When adaptation fails, the consequences cascade: reduced feed intake worsens negative energy balance, which suppresses immune function, which increases susceptibility to metritis, mastitis, and retained placenta. A monitoring program that captures this cascade early, through production records, body condition, and metabolic tests, allows intervention before clinical disease becomes established.

Data Types and Their Interpretive Logic

Production Records

Milk production is the most accessible and frequently recorded health indicator. Deviations from expected yield, whether at the individual cow or group level, often precede clinical disease by days. The interpretive challenge is establishing the expected baseline. Rolling herd average, mature equivalent projections, and lactation curves each provide a different reference. For transition cow monitoring, daily milk weights in the first 30 days in milk are more informative than monthly tests because they capture the rate of rise toward peak, which reflects metabolic adaptation.

Group-level production data can mask individual problems. A pen average that drops 5 percent may represent one severely affected cow or a uniform mild problem across the group. The monitoring strategy should therefore include both individual cow flags, such as a drop of more than 10 percent from the previous day, and group-level trends. The herd health approach to dairy cow nutrition and production diseases emphasizes that production records must be interpreted alongside body condition, dietary analysis, and metabolic testing to assess subclinical disease, also clinical cases.

Body Condition Scoring

Body condition score is a proxy for energy balance over weeks, not days. It changes slowly, so it is not useful for acute disease detection, but it is valuable for assessing whether the herd is entering the transition period with appropriate fat reserves. Cows that calve overconditioned are at higher risk of negative energy balance and its sequelae. Cows that calve underconditioned lack reserves to support early lactation. The monitoring of metabolic health in transition dairy cattle describes body condition as one of the routine, proactive observations intended to provide early detection of problems and an opportunity for investigation before costs escalate.

Scoring protocols should be standardized. The same veterinarian or trained staff member should score the same cows at defined time points, typically dry-off, calving, and 30 days in milk. Change in body condition between these points is more informative than a single absolute score.

Metabolic Testing

Blood metabolites provide a snapshot of metabolic status at a specific moment. Non-esterified fatty acids reflect the magnitude of fat mobilization, and beta-hydroxybutyrate reflects the completeness of hepatic oxidation of those fatty acids. The review of transition cow metabolic monitoring identifies high NEFA in the last 7 to 10 days before expected calving as associated with increased risk of displaced abomasum, retained placenta, and other periparturient disease. Sampling strategy matters. A single cow-side test on a sick cow confirms a diagnosis. Herd-level testing requires a sampling protocol, typically 8 to 12 cows in a defined stage of lactation, to estimate prevalence and identify whether a problem is individual or systemic.

The evidence base for metabolic thresholds comes from studies that relate metabolite concentrations to subsequent disease risk. These thresholds are population-level estimates, not individual diagnoses. A cow with elevated NEFA is at increased risk, but many cows with elevated NEFA remain clinically healthy. The value of metabolic testing is therefore in herd-level surveillance, not individual prediction.

The Role of Sensor Technologies

Precision livestock farming technologies add continuous data streams that complement periodic measurements. Activity monitors, rumination sensors, and milk conductivity probes generate alerts that can identify illness earlier than visual observation. The systematic review of validated sensor technologies for dairy cattle welfare assessment found that only 18 of 129 commercially available sensors had external validation records, with accelerometer-based systems having the highest validation rate at 30 percent. Cameras, load cells, milk sensors, and boluses had validation rates of 10 percent or lower.

This validation gap matters clinically. An unvalidated sensor may generate alerts that are not specific to disease, producing alarm fatigue and eroding trust in the system. The veterinarian should ask what trait the sensor measures, how it was validated, and against what reference standard. Validated traits in the literature include animal activity, feeding and drinking behavior, physical condition, and health. Sensors that measure these traits directly are more likely to support clinical decisions than those that infer health from indirect signals.

The Herd-Level Diagnostic Sequence

Herd record analysis follows a defined sequence that moves from signal detection to hypothesis testing to intervention design. The sequence begins with screening data that are already collected routinely, then progresses to targeted diagnostics only when screening identifies a problem. This staged approach prevents both over-investigation of normal variation and delayed response to genuine deterioration.

The first step is to establish the current baseline for each key metric. For production records, this means calculating rolling 30-day averages for milk yield, fat and protein percentage, and somatic cell count, stratified by parity and days in milk. For reproduction, the baseline includes conception risk, submission rate, and calving interval. For health events, the baseline is the incidence rate of each recorded condition per 100 cow-months at risk. Without a baseline, no deviation can be interpreted.

The second step is to compare current values against the baseline and against published reference thresholds. The transition period literature provides one well-validated example: elevated prepartum non-esterified fatty acids above 0.4 mmol/L in the final 7 to 10 days before calving are associated with increased risk of displaced abomasum, retained placenta, and other periparturient disease, as reviewed in the monitoring strategies described by LeBlanc. Herd-level interpretation requires calculating the proportion of cows exceeding such thresholds, also the mean value. A herd mean within range can conceal a substantial subpopulation at risk.

The third step is temporal pattern analysis. A single month's deviation may reflect random variation, a data entry error, or a change in laboratory or equipment calibration. A deviation that persists for two consecutive months, or that shows a monotonic trend across three months, warrants investigation. Seasonal patterns should be expected for heat stress, with rectal temperature during heat stress being heritable in Holstein cattle, as demonstrated by genome-wide association mapping. A herd with a known heat stress problem should show predictable summer deterioration in production and reproduction records, the absence of such a pattern may indicate that cooling interventions are working, or that the records are not capturing the true effect.

The fourth step is stratification. Aggregate herd metrics obscure the differences between early lactation, peak lactation, and late lactation cows, and between primiparous and multiparous animals. A milk fat depression that appears only in second-lactation cows in the first 60 days in milk points to a different problem than one distributed across all parities and stages. Stratification by pen, feeding group, or batch can identify management-specific causes that would otherwise remain hidden.

Decision Points and Thresholds

The decision to intervene, and the intensity of that intervention, depends on the magnitude and duration of the deviation from target. Table 1 provides a framework for prioritizing herd-level findings.

The thresholds in Table 1 are pragmatic starting points, not universal standards. Herd size, production level, and regional disease pressure all shift the appropriate action level. A 50-cow herd with excellent record keeping may justify investigation at a lower prevalence than a 2,000-cow herd where the cost of sampling is proportionally smaller. The monitoring criteria described in the herd health approach to dairy cow nutrition and production diseases are intended to be applied collectively, with clinical data, milk production records, and dietary analysis considered together instead of in isolation.

Designing the Investigation

Once a deviation is confirmed, the investigation should be structured as a series of testable hypotheses. The records themselves often narrow the hypothesis space. A rise in clinical mastitis cases confined to one pen suggests an environmental or equipment issue in that pen. A decline in conception risk that coincides with a change in semen supplier suggests a semen handling or storage problem. A drop in milk protein percentage across all groups suggests a ration change or feed quality issue.

The investigation should include a farm visit with direct observation of the relevant processes. Records can identify what changed, but they rarely identify why. Observation of the milking routine, feed delivery, cow comfort, and water availability provides the context that records lack. The physiological imbalance framework described by Ingvartsen and Moyes emphasizes that periparturient health problems often reflect difficulty adapting to the nutrient demands of lactation, and that the resulting disease complex includes digestive, metabolic, and infectious components. A records-based investigation that focuses on a single disease entity may miss the underlying nutritional or management failure that predisposes to multiple conditions.

Sampling plans should be designed to answer one question at a time. For metabolic testing, the sampling window matters as much as the analyte. Prepartum NEFA should be measured in the final week before calving, and postpartum beta-hydroxybutyrate in the first two weeks of lactation, as described in the transition cow monitoring review. Samples taken outside these windows produce values that are difficult to interpret against published thresholds. The number of animals sampled should be sufficient to detect the expected prevalence, for a herd-level prevalence of 15%, sampling 12 to 15 animals per group provides reasonable confidence, but this should be adjusted for herd size and the cost of a false-negative result.

Documentation and Follow-Up

The value of herd record analysis depends on the quality of the documentation that follows it. Every investigation should produce a written record that includes the following elements: the signal that triggered the investigation, the baseline and current values for the relevant metrics, the hypotheses considered, the diagnostic tests performed, the findings, the intervention implemented, and the timeline for re-evaluation. This documentation serves two purposes. It provides the basis for assessing whether the intervention worked, and it creates a historical record that can inform future investigations when similar signals reappear.

Re-evaluation should be scheduled at an interval appropriate to the intervention. Nutritional changes may require two to four weeks to show an effect on milk composition, while changes to the milking routine may show an effect on somatic cell count within one to two weeks. The re-evaluation should use the same metrics and the same sampling strategy as the initial investigation, so that the comparison is valid. A failure to improve should trigger a second investigation instead of a simple escalation of the first intervention, because the initial diagnosis may have been incorrect or incomplete.

The documentation should also record the cost of the investigation and intervention. This allows the veterinarian and producer to assess whether the intervention was economically justified, which is a necessary component of sustained herd health programs. The interdisciplinary approach described by Mulligan and colleagues explicitly includes producer profitability as an outcome of interest, alongside animal health and welfare. A records-based intervention that improves health but costs more than it saves will not be maintained, regardless of its biological correctness.

Recognized Failure Modes in Herd-Level Monitoring

Herd data systems fail in predictable ways. The most common failure is the collection of data that is never interpreted against a defined threshold. A farm may record every treatment event, yet no one reviews the cumulative incidence of clinical mastitis by lactation number or season. The data exist, but the decision loop is never closed. Detection of this failure requires a simple audit: ask what decision was changed by last month's records. If no decision changed, the monitoring program is decorative.

A second failure mode is the use of thresholds without accounting for the population at risk. A herd with 50 cows in the close-up group and a herd with 400 cows cannot be compared using raw event counts. Incidence density, expressed as cases per 100 cow-weeks at risk, corrects for this. LeBlanc's review of transition cow metabolic monitoring emphasizes that the value of any monitoring criterion depends on knowing the denominator and the time window.

A third failure is the misinterpretation of prevalence data as incidence data. A single blood sampling event that finds 15 percent of fresh cows with elevated beta-hydroxybutyrate describes a point prevalence, not the rate of new cases. Repeated sampling across defined intervals is required to distinguish a persistent problem from a transient one.

Common Errors in Interpretation

Less experienced clinicians often treat a single abnormal value as a diagnosis. A herd-level mean for non-esterified fatty acids that exceeds a published cut-point indicates elevated risk, not a specific disease. The correct response is to examine the distribution of values, the timing relative to calving, and the concurrent disease events before designing an intervention. The herd health approach to transition cow nutrition and production disease describes monitoring criteria that are meant to be considered collectively, not in isolation.

A second common error is the failure to stratify data by parity or lactation stage. Primiparous and multiparous cows have different metabolic profiles and different disease risks. Pooling them obscures both. The corrective action is to build the analysis around biologically meaningful subgroups before any herd-level summary is produced.

A third error is the overinterpretation of sensor-derived data without attention to validation status. A systematic review of commercially available sensor technologies for dairy cattle welfare assessment found that only a minority of retailed sensors had been externally validated. An unvalidated sensor may produce consistent data that correlate poorly with the clinical condition of interest. The corrective action is to confirm the sensor measures what the clinician assumes it measures, and to corroborate alerts with direct observation.

Limitations of the Current Evidence

The evidence base for herd-level monitoring is strongest for the transition period and for metabolic disease. Ingvartsen and Moyes on nutrition, immune function and health note that disease incidence based on veterinary records has not improved despite large gains in production efficiency, which suggests that current monitoring strategies are incomplete. The physiological imbalance framework they describe is conceptually useful but does not provide a single measurable endpoint that can be monitored on farm.

Expert opinion still differs on several points. The optimal sampling frequency for metabolic tests is not established by controlled trials. The relative value of blood metabolites versus milk constituents versus sensor data is debated. Some authorities advocate for routine testing of all transition cows, while others recommend testing only when herd-level triggers are met. Both positions are defensible given the available evidence.

Referral, Specialist Consultation, and Regulatory Reporting

Most herd-level investigations can be managed by the attending veterinarian. Referral to a specialist is warranted when the investigation requires expertise beyond routine practice. Nutritional consultation is appropriate when ration formulation is implicated but the veterinarian does not provide nutritional services. A veterinary nutritionist or a diagnostic laboratory with ruminant nutrition expertise can provide ration evaluation and interpretation of forage analysis.

Laboratory involvement is required when the monitoring program depends on metabolite testing, trace element assessment, or bacteriology. The laboratory should be consulted before sampling to confirm sample handling requirements, test availability, and interpretive ranges. Some laboratories provide herd-level interpretive reports that are more useful than individual animal values.

Regulatory reporting obligations vary by jurisdiction and by disease. The WOAH terrestrial animal health code defines internationally notifiable diseases, but national authorities determine local reporting requirements. The USDA APHIS animal health information portal provides guidance for the United States, while the FAO animal production and health resources cover international livestock health service frameworks. The AVMA practice resources and the MSD Veterinary Manual offer additional professional guidance on disease reporting and diagnostic decision-making. The attending veterinarian should confirm the applicable requirements in their region before a reportable disease is suspected.

Frequently Asked Questions

How Do I Prioritize Data Collection When Budget or Time Is Limited?

Start with the measurements that carry the most decision weight for your client's production stage. For transition cows, milk production records, body condition scores, and metabolic tests for non-esterified fatty acids and beta-hydroxybutyrate provide the earliest signals of energy imbalance and disease risk. If laboratory testing is unaffordable for every cow, sample a targeted subgroup, such as cows entering second lactation or those with previous disease, and use production records to identify outliers for individual testing. Sensor systems vary widely in validation status, so select technologies with published external validation instead of unproven devices. A systematic review of commercially available sensors found that only a minority of marketed tools have been externally validated, with accelerometer-based systems showing the strongest evidence base.

What Can I Do When the Herd Has No Electronic Records?

Paper records remain workable if the data structure is disciplined. Maintain a calving book, a treatment log, and a monthly production summary as the minimum dataset. Record events on the day they occur, and assign one person responsibility for data entry. For monthly analysis, calculate rolling incidence rates for displaced abomasum, retained placenta, clinical mastitis, and culling, then compare these against the same period in the previous year. Body condition scoring and metabolic testing do not require software and can be performed during routine herd visits. The integrated herd health approach described by Mulligan and colleagues uses farm management factors, clinical data, and milk production records as its core inputs, none of which depend on proprietary platforms.

How Should I Present Herd-Level Findings to a Producer Who Prefers Anecdotal Observation?

Frame the data as a check on the producer's own observations instead of a replacement for them. Show one metric that confirms what the producer already suspects, then introduce one metric that reveals a pattern they have not seen. Use a simple run chart of monthly disease incidence or average milk production so the trend is visible without statistical interpretation. Explain that routine monitoring is intended to provide early detection of problems and an opportunity for investigation before clinical consequences accumulate. Connect each data point to a management decision, such as adjusting dry cow nutrition or changing calving pen stocking density, so the producer sees the record as an action tool instead of an administrative burden.

Which Metabolic Tests Offer the Best Return for Routine Transition Monitoring?

Non-esterified fatty acids and beta-hydroxybutyrate are the most extensively studied indicators of peripartum energy status. Elevated non-esterified fatty acids in the week before calving are associated with increased risk of displaced abomasum and retained placenta, while postpartum beta-hydroxybutyrate identifies subclinical ketosis. Sample a representative group of cows, ideally 8 to 12 animals per risk group, and interpret results against published thresholds instead of laboratory reference intervals alone. Testing every transition cow is rarely necessary, targeted sampling of at-risk groups, such as overconditioned cows or those with previous metabolic disease, provides actionable information at lower cost. The evidence base for these thresholds is drawn from field studies in commercial dairy herds and is reviewed in the transition cow monitoring literature.

How Do Monitoring Priorities Differ for Grazing Herds Versus Confined Herds?

Grazing herds face different nutritional and environmental constraints, so the monitoring emphasis shifts accordingly. Body condition score management remains central, but the timing of pasture allocation and the risk of trace element deficiencies become more prominent. Confined herds allow tighter control of feed intake, making dry matter intake records and ration formulation data more informative. Heat stress monitoring matters in both systems but is more readily managed in confinement through ventilation and cooling systems. Genetic selection for thermotolerance, including markers associated with rectal temperature regulation during heat stress, may be more relevant in hot climates regardless of housing system. The physiological basis for monitoring, negative energy balance, hypocalcemia, and rumen health, applies across systems, but the practical indicators and their interpretation must be adapted to the feeding and management context.

When Should I Involve a Specialist or Report Findings to a Regulatory Authority?

Refer to a specialist when the herd problem exceeds your diagnostic resources, such as when metabolic testing suggests a nutritional interaction that requires ration formulation expertise, or when reproductive failure persists despite normal uterine health and heat detection. Regulatory reporting obligations vary by jurisdiction and disease, so consult your national veterinary authority or the relevant animal health code for current requirements. Reportable diseases, unusual mortality events, and suspected notifiable conditions should be reported promptly even while the diagnostic workup continues. International standards for surveillance and reporting are published by the World Organization for Animal Health, and national programs are described by agencies such as the USDA Animal and Plant Health Inspection Service. When in doubt about whether a condition is reportable, contact the regulatory authority before proceeding with further testing.

Related Clinical & Scientific Guides

• Rumen Health Assessment in Dairy Cows: Clinical and Subclinical Indicators
• Mastitis Control Programs in Dairy Herds: Monitoring and Prevention
• Swine Nutrition and Health: Feed-Related Disease Diagnosis

References and Further Reading

• A herd health approach to dairy cow nutrition and production diseases of the transition cow.. 2006.
• Monitoring metabolic health of dairy cattle in the transition period.. 2010.
• Nutrition, immune function and health of dairy cattle.. 2013.
• A Systematic Review on Commercially Available and Validated Sensor Technologies for Welfare Assessment of Dairy Cattle.. 2021.
• Genome-wide association mapping for identification of quantitative trait loci for rectal temperature during heat stress in Holstein cattle.. 2013.
• Consumption of Ultra-Processed Foods by Pesco-Vegetarians, Vegetarians, and Vegans: Associations with Duration and Age at Diet Initiation.. 2021.
• USDA APHIS Animal Health Information. USDA APHIS.
• FAO Animal Production and Health. FAO.
• MSD Veterinary Manual, Professional Edition. MSD Veterinary Manual.

Related Articles

• Dairy Herd Culling Decisions: Data-Driven Approaches
• Dairy Cow Monitoring Systems: Data Integration for Herd Health
• Dairy Herd Health Certification: Standards and Veterinary Audit
• Poultry Flock Health Records: Essential Data for Veterinary Decision-Making
• Dairy Calf Health: Diagnostic Approach to Neonatal Diarrhea

This article is educational professional reference material for veterinary audiences. It is not a substitute for veterinary diagnosis, individual clinical judgment, current product labeling, or applicable regulatory requirements.