# One Health Surveillance Data Integration: Challenges and Solutions


## Key Takeaways

- **Semantic misalignment of diagnostic codes and case definitions is the primary integration barrier**, with human health using ICD-based coding and animal health employing syndrome or species-specific classifications, hindering direct comparability of events like "encephalitis" versus "neurologic disease."
- **Divergent surveillance objectives create data asymmetry**, as human health prioritizes individual case detection and blood safety (incentivizing detailed clinical encounters), while animal health focuses on trade compliance and production loss prevention (aggregating herd-level events), exemplified by West Nile virus surveillance where equine vaccine availability reduces animal health prioritization.
- **Institutional fragmentation and governance gaps impede integration**, with separate agencies for human, animal, and environmental health lacking a unified mandate and shared vision, necessitating formal interagency working groups with defined terms of reference to foster collaboration.
- **Data structure and scale mismatches pose significant technical challenges**, including differing spatial resolutions (administrative boundaries vs. ecological ranges) and temporal aggregation cycles (daily human reporting vs. annual animal certification), requiring the definition of common units for effective signal preservation.
- **Incentive asymmetry and funding instability risk sustainability**, as sectors contribute data only when perceiving direct benefit, and passive notification systems often fail when initial project funding concludes, underscoring the need for integration to be a continuous institutional commitment rather than a one-time technical fix.
- **A reference model approach to data harmonization is most robust**, mapping each source system to a canonical data model rather than direct pairwise mapping, and employing a two-level mapping for coded fields (source to common vocabulary, then confidence level) to preserve uncertainty and facilitate analysis.

---

Zoonotic pathogens do not respect administrative boundaries between human health, veterinary medicine, and environmental monitoring. Surveillance data generated within each sector frequently remain trapped in separate databases, coded with different terminologies, collected at different spatial scales, and governed by different legal frameworks. This article examines the technical and institutional obstacles to integrating these streams and reviews practical approaches for overcoming them. It serves veterinary researchers designing cross-sectoral surveillance studies, diagnosticians interpreting multi-source data, and policymakers evaluating surveillance infrastructure options. The central question addressed is how disparate data systems can be made interoperable without sacrificing the analytical rigor each sector requires.

Integration failures carry measurable consequences. When influenza A viruses circulate in swine, gaps in knowledge about their ecology and evolution impede understanding of how these viruses affect human health, a limitation underscored by the 2009 H1N1 pandemic and the subsequent call for greater surveillance and data sharing on swine influenza [Vincent et al., institutional publication on influenza A virus in swine](https://pubmed.ncbi.nlm.nih.gov/23556412/). Similarly, the emergence of low pathogenic avian influenza H7N9, which produced more zoonotic infections than H5N1, demonstrates that surveillance optimized for one sector can miss threats arising in another [Naguib et al., institutional publication on global patterns of avian influenza A H7](https://pubmed.ncbi.nlm.nih.gov/31381759/). The rationale for integration is therefore not administrative convenience but earlier detection and more efficient resource allocation.

## At a Glance

| Parameter | Consideration |
|---|---|
| Primary integration barrier | Semantic misalignment of diagnostic codes and case definitions across sectors |
| Data structure mismatch | Human health uses ICD-based coding, animal health uses syndrome or species-specific classifications |
| Governance gap | No single authority holds mandate across human, animal, and environmental surveillance |
| Incentive asymmetry | Blood safety drives public health investment, animal health priorities differ when vaccines exist and infections are subclinical |
| Most accessible veterinary data source | Diagnostic laboratory data for syndromic surveillance |
| Critical enabling step | Formal interagency working groups with defined terms of reference |
| Sustainability risk | Passive notification systems fail when funding cycles end |
| Evidence base | Qualitative expert interviews identify political will as essential for integration |

## Conceptual Foundations of Cross-Sectoral Surveillance

### Divergent Surveillance Objectives

Human health surveillance prioritizes case detection, outbreak response, and blood product safety. Animal health surveillance emphasizes trade compliance, production loss prevention, and population-level disease freedom certification. These objectives generate different data: human systems record individual clinical encounters with demographic detail, while animal systems aggregate herd or flock events with production context. The West Nile virus experience in Europe illustrates the resulting asymmetry. Blood safety provides a strong incentive for public health authorities to fund surveillance, whereas the availability of an effective equine vaccine and the predominantly subclinical course of infection reduce the priority assigned by animal health authorities [Gossner et al., institutional publication on West Nile virus surveillance in Europe](https://pubmed.ncbi.nlm.nih.gov/28494844/). Integration therefore requires aligning systems whose stakeholders have different tolerance for false positives, different reporting timelines, and different legal obligations.

### The Semantic Interoperability Problem

Data integration fails most often at the level of meaning. A human clinician records "encephalitis" using an ICD code. A veterinary diagnostician records "neurologic disease" using a syndrome code. A vector ecologist records "positive pool" with a cycle threshold value. These records may describe the same epizootic event, but no shared ontology links them. Veterinary syndromic surveillance systems have addressed this by developing standardized syndrome classifications from clinical and laboratory data, yet these classifications rarely map cleanly onto human health terminologies [Dórea, Sanchez, and Revie, institutional publication on veterinary syndromic surveillance](https://pubmed.ncbi.nlm.nih.gov/21640415/). The consequence is that even when data are technically shareable, they are not semantically comparable.

### Spatial and Temporal Scale Mismatch

Human surveillance data cluster at administrative boundaries such as postal codes or health districts. Animal surveillance data follow production system geography, which may cross those boundaries. Wildlife surveillance data follow ecological ranges that ignore both. Temporal resolution differs as well: human notifiable disease reporting operates on daily or weekly cycles, while animal health certification may operate on annual cycles tied to trade schedules. Any integrated system must define a common spatial unit and temporal aggregation that preserves signal without distorting any single sector's data.

## Institutional Barriers to Integration

### Siloed Governance and Mandate Fragmentation

In most jurisdictions, human health, animal health, and environmental monitoring are managed by separate agencies with distinct legal mandates, funding streams, and accountability structures. Qualitative research with Australian experts in zoonotic disease surveillance identified the absence of a clear definition and shared vision for One Health as a barrier to interdisciplinary collaboration, with siloed approaches restricting the ability of professionals to work across disciplines [Johnson, Hansen, and Bi, institutional publication on integrated One Health surveillance in Australia](https://pubmed.ncbi.nlm.nih.gov/29226606/). The same study found that participants considered political will an essential requirement for integration, a finding consistent with the European experience where the creation of a formal interagency working group was identified as a crucial step toward integration [Gossner et al., institutional publication on West Nile virus surveillance in Europe](https://pubmed.ncbi.nlm.nih.gov/28494844/).

### Incentive Misalignment and Funding Instability

Sectors contribute data to integrated systems only when they perceive benefit. Public health agencies may resist sharing case-level data due to privacy law. Veterinary agencies may resist sharing premises-level data due to producer confidentiality concerns. Wildlife agencies may lack laboratory capacity to generate sequence data at the volume human health systems expect. Syndromic surveillance systems based on passive notification or data transfers have faced sustainability issues once initial project funding ended [Dórea, Sanchez, and Revie, institutional publication on veterinary syndromic surveillance](https://pubmed.ncbi.nlm.nih.gov/21640415/). Integration is not a one-time technical fix but a continuing institutional commitment.

## Data Governance and Legal Frameworks

### Privacy, Confidentiality, and Data Ownership

Human health data are protected by privacy regulations that restrict secondary use. Animal health data are protected by commercial confidentiality agreements with producers. Environmental data may be subject to open-access mandates. An integrated surveillance platform must reconcile these regimes, typically through tiered access controls, de-identification protocols, and data use agreements that specify permitted analyzes. The international standards for animal health surveillance and trade-related disease control published by the World Organization for Animal Health provide a framework for data sharing obligations among veterinary authorities [WOAH terrestrial animal health code](https://www.woah.org/en/what-we-do/standards/codes-and-manuals/terrestrial-code-online-access/), but these standards do not address human health data or environmental data.

### Data Quality and Provenance

Integrated systems inherit the quality characteriztics of their weakest source. Veterinary data may lack denominator information, making incidence calculations unreliable. Human data may lack exposure information, making zoonotic linkage impossible. Provenance tracking, the recording of where each record originated and what transformations it underwent, becomes essential for interpreting integrated analyzes. Without provenance metadata, a researcher cannot distinguish a true absence of cases from a reporting gap.

## Technical Architecture Options

### Federated Versus Centralized Models

A centralized data warehouse requires all sectors to transmit data to a single repository, simplifying analysis but creating governance problems around data ownership and access. A federated model keeps data in sectoral systems and runs queries across them through standardized interfaces, preserving local control but requiring sophisticated query translation. The federated approach aligns with the observation that no single surveillance model fits all contexts, and that countries should implement the integrated approach that meets their needs [Gossner et al., institutional publication on West Nile virus surveillance in Europe](https://pubmed.ncbi.nlm.nih.gov/28494844/). Veterinary researchers should evaluate which model their institutional relationships can sustain before committing to technical specifications.

### Standards for Data Exchange

Minimum data sets, controlled vocabularies, and machine-readable case definitions are prerequisites for automated integration. The international standards for animal health surveillance published by the World Organization for Animal Health specify reporting requirements for notifiable diseases [WOAH terrestrial animal health code](https://www.woah.org/en/what-we-do/standards/codes-and-manuals/terrestrial-code-online-access/), and these can serve as a foundation for veterinary contributions to integrated platforms. Human health systems maintain their own reporting standards. The gap between these standards is the technical space where integration work must occur.

## Practical Data Harmonization: A Working Framework

Harmonization begins with a formal inventory of what each sector actually collects. A surveillance system cannot integrate data it has not first characterized. For each data stream, record the purpose, the legal basis for collection, the unit of observation, the case definition, the sampling strategy, and the data entry environment. This inventory becomes the reference document against which all subsequent mapping decisions are made. The exercise is often more revealing than expected. Veterinary practitioners record breed, age, production stage, and vaccination history. Human health systems record occupation, travel history, and comorbidity. Environmental monitoring records location, date, and analyte concentration. These fields rarely align without deliberate transformation.

The harmonization process proceeds through three stages: syntactic alignment, semantic alignment, and analytical alignment. Syntactic alignment standardizes formats, units, and coding systems. Semantic alignment reconciles the meaning of terms across sectors. Analytical alignment ensures that the integrated data can answer the question that motivated integration in the first place. Each stage has distinct failure modes, and skipping a stage produces data that is technically merged but practically unusable.

### The Harmonization Sequence

Stage one requires a data dictionary for every source system. Document the field name, data type, allowed values, and the clinical or administrative meaning of each entry. This step exposes inconsistencies that are invisible when systems operate independently. One system may record age in years, another in months, and a third as a date of birth. One may use free text for breed, another a closed list. One may record body temperature in Celsius, another in Fahrenheit. These are syntactic problems, and they are the easiest to solve.

Stage two addresses semantics. The term "case" is a useful example. A human influenza case may be laboratory-confirmed, clinically diagnosed, or syndromic. A swine influenza case may be detected through passive reporting of respiratory disease, active sampling at slaughter, or routine diagnostic submissions. The same word describes fundamentally different events. The [review of influenza A virus in swine worldwide](https://pubmed.ncbi.nlm.nih.gov/23556412/) notes that relatively little is known about IAV circulating in swine compared with the avian and human knowledge base, and this gap impedes understanding of how viruses adapted to swine or humans impact the ecology and evolution of IAV as a whole. That gap is partly a semantic problem. Until each sector defines what it means by a case, integration produces false confidence.

Stage three requires a clear analytical question. Integration is not an end in itself. The question determines which fields must be harmonized and which can remain sector-specific. A system designed to detect early signals of zoonotic spillover needs different fields than one designed to measure the burden of endemic disease. The [European experience with West Nile virus surveillance](https://pubmed.ncbi.nlm.nih.gov/28494844/) demonstrates that no single model fits all settings, and that countries should implement the integrated approach that meets their needs. The same principle applies at the level of individual data streams. Define the question first, then decide what must be harmonized.

### Data Standards and Interoperability Issues

The table below summarizes common standards encountered in One Health surveillance work, the sectors that typically use them, and the interoperability issues that arise when they meet.

| Standard | Typical sector | Primary content | Interoperability issue |
|---|---|---|---|
| SNOMED CT | Human clinical | Clinical terms, findings, procedures | Veterinary uptake limited, mapping to veterinary terminologies incomplete |
| LOINC | Human laboratory | Laboratory test names and results | Veterinary laboratories often use local codes or free text |
| ICD-10/ICD-11 | Human health | Disease classification | Species-specific disease codes absent, zoonotic diseases coded differently across sectors |
| Systematized Nomenclature of Veterinary Medicine (SNOMED-V) | Veterinary clinical | Veterinary clinical terms | Divergence from human SNOMED CT creates mapping burden |
| OIE-WAHIS | Animal health | Notifiable disease reports | Country-level aggregation loses individual case detail |
| Darwin Core | Environmental/biodiversity | Species occurrence data | Lacks clinical and epidemiological fields |
| HL7 FHIR | Human health | Clinical data exchange | Veterinary systems rarely implement FHIR profiles |
| ISO 3166 | All | Geographic identifiers | Administrative boundaries do not match ecological or production system boundaries |

The practical consequence of this fragmentation is that integration projects spend most of their effort on mapping instead of analysis. A laboratory result in a veterinary system may be recorded as a text string, while the equivalent human system uses LOINC. A wildlife sample may be georeferenced to a national park boundary, while a livestock case is referenced to a farm address. These mismatches are also technical. They reflect different institutional histories, different regulatory mandates, and different analytical traditions.

### Mapping Strategies That Work

The most robust approach is a reference model. Create a canonical data model for the integrated system, then map each source system to that model instead of mapping systems directly to each other. Direct pairwise mapping becomes unmanageable as the number of sources grows. A reference model also provides a stable target when source systems change their internal structures, which they will.

For coded fields, use a two-level mapping. The first level maps source codes to a common vocabulary. The second level records the confidence of the mapping. Exact matches, partial matches, and uncertain matches should be distinguished explicitly. A partial match, for example mapping a veterinary code for "porcine respiratory disease complex" to a human code for "influenza-like illness," carries interpretive risk. The integrated dataset must preserve this uncertainty instead of hiding it.

Free text fields require a different approach. Natural language processing can extract structured data from clinical notes, but the accuracy varies by species, by clinical setting, and by the quality of the original documentation. [Veterinary syndromic surveillance initiatives](https://pubmed.ncbi.nlm.nih.gov/21640415/) have shown that diagnostic laboratories provide the most readily available data sources for animal health surveillance, partly because laboratory data is already structured. Clinical free text is valuable but requires validation before it can support automated integration.

### Validation and Quality Control

Every transformation step requires validation. When a value is converted from one unit to another, when a code is mapped to a new vocabulary, when a free text entry is classified into a structured category, the result must be checked against a reference standard. Build validation into the pipeline instead of applying it after the fact. Automated checks can flag out-of-range values, impossible dates, and inconsistent combinations. Manual review should be reserved for the transformations that carry the highest interpretive risk.

The [challenges of implementing an integrated One Health surveillance system in Australia](https://pubmed.ncbi.nlm.nih.gov/29226606/) include the absence of a clear definition and vision for One Health, siloed approaches by different sectors, and the need for political will. These institutional factors determine whether the technical work of harmonization is sustainable. A mapping that is technically sound but institutionally unsupported will not survive staff turnover, funding cycles, or changes in political priority.

### Documentation and Version Control

The harmonization framework must be documented as a living document. Record every mapping decision, the rationale for it, the date it was made, and the person or group that made it. Version control is essential. When a source system changes its coding scheme, the mapping must be updated, and the update must be traceable. Without this discipline, the integrated dataset becomes an archaeological site, and no one can say with confidence what a given field actually means.

The documentation should also record what was deliberately excluded. Some fields cannot be harmonized without violating privacy or confidentiality obligations. Some cannot be harmonized because the source systems do not collect them. Some should not be harmonized because the analytical question does not require it. Explicit exclusion is a legitimate outcome of the harmonization process, and it should be recorded with the same care as inclusion.

### Species and Production System Considerations

The correct harmonization approach varies by species and production system. Companion animal data is typically generated in clinical settings, with individual patient records and owner consent. Livestock data is often generated at the herd level, with production records, movement data, and slaughterhouse surveillance. Wildlife data is generated through opportunistic sampling, with variable spatial and temporal coverage. Each of these contexts changes the feasibility of integration. Herd-level data cannot be mapped to individual-level human data without a clear analytical justification. Wildlife data cannot support the same temporal resolution as clinical data. The harmonization framework must respect these differences instead of forcing a uniform structure.

The [global patterns of avian influenza A (H7)](https://pubmed.ncbi.nlm.nih.gov/31381759/) illustrate the value of integrating data from wild bird reservoirs, poultry, and humans, and the authors emphasize the need for a One Health approach in controlling emerging viruses. The integration succeeded because the analytical question was clear, the data streams were well characterized, and the mapping preserved the distinct ecological and epidemiological meaning of each source. That is the standard to which harmonization work should be held.

## Recognized Complications and Failure Modes

Integration initiatives fail in predictable patterns. The most common is the "two-database" outcome, where agencies agree to share data in principle but continue maintaining parallel systems because neither side trusts the other's data quality or timeliness. This failure is detectable early when integration working groups spend more time negotiating access permissions than defining shared case definitions. A second frequent failure is the "dashboard without decisions" outcome, where integrated visualizations are produced but no agency has the mandate to act on the signals they reveal. This occurs when integration is treated as a technical exercise instead of a governance reform, and it becomes apparent when surveillance outputs are not referenced in any agency's standard operating procedures.

A third failure mode is semantic drift, where the same code or term acquires different meanings across sectors after implementation. This is particularly insidious because it produces apparently consistent data that are not comparable. Detection requires periodic cross-sector audits in which the same record is independently coded by each participating agency and the results compared. A fourth mode is alert fatigue, where integrated systems generate so many signals that analysts begin ignoring them. This typically follows overly broad case definitions adopted to accommodate the least specific data source in the network.

| Observation | Likely cause | Discriminating check |
|---|---|---|
| Parallel databases persist after integration | Mandate or funding does not require consolidation | Review whether any agency's performance metrics reference the integrated system |
| Dashboards produced but no decisions change | Governance gap, no agency owns the response | Ask which named official is accountable for acting on each alert type |
| Codes appear consistent but results diverge | Semantic drift after implementation | Re-code a shared sample of records in each sector and compare |
| Analysts ignore most alerts | Overly broad case definitions | Compare alert rate against confirmed event rate per quarter |
| Data flow stops after initial funding ends | Sustainability not budgeted | Check whether integration costs appear in base operational budgets |

## Common Errors and Corrective Actions

Less experienced analysts commonly treat data harmonisation as a one-time mapping exercise instead of a continuous process. The corrective action is to schedule re-validation whenever any participating agency changes its data collection instruments, laboratory methods, or reporting software. A second error is assuming that the most detailed data source should define the shared standard. In practice, the shared standard must accommodate the least granular reliable source, with optional extensions for richer data. A third error is conflating data availability with data quality. A laboratory that reports results promptly may still have incomplete species attribution or missing denominator data, and these deficiencies propagate silently through integrated analyzes.

A fourth error is designing integration around current disease priorities instead of durable infrastructure. When priorities shift, the system must accommodate new data types without redesign. The corrective action is to build around stable entities such as location, species, time, and diagnostic result, instead of disease-specific fields. Finally, analysts often underestimate the effort required for documentation. Version control for case definitions, mapping tables, and extraction scripts is not an administrative afterthought but the mechanism that makes integrated data interpretable by future users.

## Limitations of Current Evidence

The published literature on One Health surveillance integration is dominated by descriptive case studies and expert opinion instead of controlled evaluations. The review of West Nile virus surveillance in Europe explicitly notes that no single surveillance model fits all contexts and that countries must adapt approaches to their own circumstances, which limits the generalizability of any particular success story. Similarly, the Australian qualitative study identifies that the absence of a clear definition of One Health itself acts as a barrier to interdisciplinary collaboration, meaning that even the conceptual basis for integration remains contested.

Expert opinion diverges on several substantive points. One is whether federated models that leave data in institutional custody are genuinely superior to centralized repositories for cross-sectoral work. Proponents of federation argue that it preserves local data quality incentives, while critics contend that it merely perpetuates the access barriers that motivated integration in the first place. A second point of divergence concerns the role of diagnostic laboratory data. The veterinary syndromic surveillance review identifies laboratory data as the most readily available source for animal health surveillance, but laboratory data are biased toward cases that reach diagnostic testing, and experts disagree about how much this bias compromises population-level inference. A third area of uncertainty is the optimal frequency of data exchange. Daily transfers improve outbreak detection but increase the burden on contributing sites, while weekly or monthly transfers risk missing the temporal resolution needed for early warning. No comparative studies establish the optimal interval.

## Referral, Consultation, and Reporting Triggers

Veterinary researchers engaged in integration work should seek specialist consultation when they encounter data governance questions that exceed institutional expertise. Legal review is warranted before any data sharing arrangement that involves personally identifiable information, proprietary production data, or international transfer. Laboratory involvement is required when harmonising diagnostic methods or interpreting results generated under different testing protocols, since method differences can produce apparent clusters that are artefacts of sensitivity or specificity variation.

Regulatory reporting obligations differ by jurisdiction and by pathogen. The international standards for animal health surveillance and trade-related disease control are set out in the terrestrial animal health code maintained by the World Organization for Animal Health, and veterinary researchers should consult these standards when designing surveillance outputs intended to support trade or international notification. Where a surveillance signal suggests a notifiable disease, the responsible veterinarian must follow the reporting pathway established by the relevant national authority, and this obligation takes precedence over any research or integration timeline. When a signal crosses species boundaries, such as an unusual cluster in animals that might precede human cases, the appropriate action is to notify both animal and public health authorities through their formal channels instead of relying on informal professional networks. The international framework linking human, animal, and environmental health for zoonotic disease control is articulated by the World Health Organization, and the practical guidance for zoonotic disease prioritization and cross-sector collaboration is maintained by the United States Centers for Disease Control and Prevention, both of which provide useful reference points for determining when escalation is appropriate.

## Frequently Asked Questions

### How Do We Prioritize Integration Efforts When Funding Is Limited?

Start with hazards that already have demonstrated cross-species transmission and existing reporting infrastructure. Influenza A viruses in swine illustrate this principle, as decades of documented interspecies transmission with humans justify sustained investment in shared surveillance and rapid data exchange. Formal interagency working groups offer a low-cost institutional mechanism that improves efficiency by coordinating existing activities instead of building new systems. When resources are constrained, focus on diagnostic laboratory data, which provide the most readily available and standardized source for veterinary syndromic surveillance. Prioritize data elements that serve both animal health and public health mandates, such as species, location, and clinical signs, before investing in elaborate technical infrastructure.

### What Can We Do When Standardized Data Formats Are Not Available?

Adopt a mapping layer that translates local coding schemes into a common reference vocabulary at the point of exchange. This approach preserves existing workflows while enabling cross-sector comparison. Begin with a small set of high-value variables, such as case counts, species, geographic location, and date of onset, and map those consistently before expanding scope. Document all mapping decisions and version them explicitly, since mapping rules change as local systems evolve. Validate mapped data against a sample of original records to detect translation errors. Where no formal standard exists, agree on a minimal data dictionary among participating institutions and publish it as a reference for future partners.

### How Should We Handle Data Sharing When Legal Mandates Conflict Across Sectors?

Identify the most restrictive legal constraint first and design the data flow around it. Human health data typically carry stricter privacy protections than animal health data, so de-identification should occur before any cross-sector transfer. Aggregate data to the level required by the most restrictive partner, for example reporting at regional instead of premises level. Establish a data use agreement that specifies permitted purposes, retention periods, and breach notification procedures. The World Health Organization's One Health framework recognizes that legal and regulatory differences between sectors create genuine barriers, and explicit agreements are the standard mechanism for resolving them. Review these agreements annually, as mandates and institutional leadership change.

### What Is the Minimum Data Set We Should Collect for Cross-Species Comparability?

Collect species, geographic location at the finest resolution permitted by privacy constraints, date of specimen collection or clinical onset, case definition used, and diagnostic test with result. These six elements allow temporal and spatial alignment across human, animal, and environmental datasets. Add syndrome category when a specific diagnosis is unavailable, since syndromic surveillance supports outbreak detection even without laboratory confirmation. Record the population denominator when possible, because incidence rates are more comparable across sectors than raw counts. The Centers for Disease Control and Prevention emphasizes that cross-sector collaboration depends on shared definitions and consistent data collection practices. Document any deviation from the minimum set so downstream analysts can interpret gaps correctly.

### How Do I Explain the Value of Data Integration to a Practice Owner or Agency Director?

Frame integration as cost avoidance instead of additional expense. Integrated surveillance allows targeted measures that replace duplicative sector-specific activities, saving resources while improving detection capacity. Use a concrete local example, such as a zoonotic pathogen that already circulates in the practice area, to illustrate how earlier detection in animals could reduce human cases and associated costs. Emphasize that integration does not require abandoning existing systems, it requires coordinating them. The Australian experience shows that professionals across sectors recognize the value of integration but are constrained by institutional silos and unclear definitions. A clear, concrete proposal with named partners and defined outputs is more persuasive than a general commitment to One Health principles.

### What Record-Keeping Practices Support Future Data Integration?

Maintain a data dictionary that defines every variable, its permitted values, and its source system. Record the version of any coding standard used and the date of any mapping changes. Log all data transformations, including filtering, aggregation, and de-identification steps, so that any derived dataset can be traced back to its source records. Store raw data separately from processed data and never overwrite original files. Document data quality issues as they are discovered, including known gaps, duplicate records, and suspected misclassification. These practices support reproducibility and allow new partners to assess whether historical data are fit for their purposes. The World Organization for Animal Health terrestrial standards emphasize that surveillance data must be documented sufficiently to support trade and disease control decisions.

## Related Clinical & Scientific Guides

* [Wildlife Disease Surveillance: Designing and Implementing a One Health Program](/knowledge/veterinary-medicine/veterinary-public-health/wildlife-disease-surveillance-designing-implementing-one-health-program)
* [Biosecurity Risk Assessment for Livestock Operations: A Practical Framework](/knowledge/veterinary-medicine/veterinary-public-health/biosecurity-risk-assessment-livestock-operations-practical-framework)
* [Rabies Post-Exposure Prophylaxis in Veterinary Personnel](/knowledge/veterinary-medicine/veterinary-public-health/rabies-post-exposure-prophylaxis-in-veterinary-personnel)


## References and Further Reading

- [Review of influenza A virus in swine worldwide: a call for increased surveillance and research.](https://pubmed.ncbi.nlm.nih.gov/23556412/). 2014.
- [West Nile virus surveillance in Europe: moving towards an integrated animal-human-vector approach.](https://pubmed.ncbi.nlm.nih.gov/28494844/). 2017.
- [Global patterns of avian influenza A (H7): virus evolution and zoonotic threats.](https://pubmed.ncbi.nlm.nih.gov/31381759/). 2019.
- [Current challenges in the evaluation of cardiac safety during drug development: translational medicine meets the Critical Path Initiative.](https://pubmed.ncbi.nlm.nih.gov/19699852/). 2009.
- [The challenges of implementing an integrated One Health surveillance system in Australia.](https://pubmed.ncbi.nlm.nih.gov/29226606/). 2018.
- [Veterinary syndromic surveillance: Current initiatives and potential for development.](https://pubmed.ncbi.nlm.nih.gov/21640415/). 2011.
- [WHO One Health Initiative](https://www.who.int/health-topics/one-health). WHO.
- [CDC One Health and Zoonotic Disease Resources](https://www.cdc.gov/one-health/index.html). CDC.
- [MSD Veterinary Manual, Professional Edition](https://www.msdvetmanual.com/). MSD Veterinary Manual.

## Related Articles

- [One Health Approach to Emerging Zoonoses: Integrating Veterinary and Human Health Data](/knowledge/veterinary-medicine/veterinary-public-health/one-health-approach-emerging-zoonoses-integrating-veterinary-human-health-data)
- [Wildlife Disease Surveillance: Designing and Implementing a One Health Program](/knowledge/veterinary-medicine/veterinary-public-health/wildlife-disease-surveillance-designing-implementing-one-health-program)
- [One Health Approach to Antimicrobial Resistance: A Veterinary Perspective](/knowledge/veterinary-medicine/veterinary-public-health/one-health-approach-to-antimicrobial-resistance-a-veterinary-perspective)
- [Zoonotic Disease Surveillance in Animal Shelters: Methods and Challenges](/knowledge/veterinary-medicine/veterinary-public-health/zoonotic-disease-surveillance-in-animal-shelters-methods-and-challenges)
- [Antimicrobial Resistance Surveillance in Food Animals: Sampling Strategies and Data Interpretation](/knowledge/veterinary-medicine/veterinary-public-health/antimicrobial-resistance-surveillance-food-animals-sampling-strategies-data-interpretation)

> 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.