Aggregated workforce and safety data dictionary template
Give every aggregate workforce or safety field a meaning that survives an export change. This editable dictionary records the population, counting unit, date basis, source mapping and permitted use beside a stable metric ID. It helps a reporting team notice when two numbers with the same column heading count different things.
The seven-page pack contains instructions, four reusable blank sheets and two fictional worked pages. Edit the DOCX or print the matching PDF. Neither file calculates totals or rates, checks identities, classifies safety cases, tests privacy risk or approves a disclosure. Its fields record decisions made by authorised owners.
Download the editable aggregated workforce and safety data dictionary (DOCX).
Download the print aggregated workforce and safety data dictionary (PDF).
Define what one counted item represents
Start with the reporting question and an internal metric ID. Keep the requested output label separate from the source column name. Write a plain-language definition and identify the unit counted: people, contracts, assignments, events, case records, hours or another specified object. A column called “count” does not supply that definition.
Record the entity and location coverage, included and excluded populations, and any required category decision. Distinguish the time the observation describes from the date the file was extracted. An updated extract can change a number’s meaning or correct an earlier record without describing a later operating period.
Use the definition card for the selected measure, not as a universal classification system. Obtain the applicable reporting definitions separately. The pack does not determine employee or worker status, whether an incident is legally recordable, or which reporting obligation applies. Keep the owner’s decision reference and any unresolved classification question visible.
Map the field rather than trusting its label
For each source version, record the exact file, table and column, the custodian and the extraction date. Describe the selection, grouping, deduplication or other transformation needed to produce the aggregate. Link to the controlled calculation or extraction specification instead of pasting personal records into the dictionary.
A mapping needs a validity interval or an explicit version boundary. If a system migration reuses the same column heading for a different count, retain two mappings. Do not overwrite the old definition and leave earlier figures appearing comparable. A familiar heading is only an alias until its meaning has been checked.
For a rate, identify the numerator and denominator definitions, their populations and periods, and the selected scaling basis. BLS occupational-safety definitions explicitly describe the count and hours used in its incidence measures. This is narrow US methodological context; the template does not adopt those measures or their classification rules.
Keep an unavailable denominator unavailable. A headcount cannot stand in for recorded hours merely because both relate to the same workforce. If the intended rate’s basis has not been selected, mark the rate definition unresolved and assign that decision. The blank forms contain no rate formula or default multiplier.
Follow two fields through a fictional migration
All organisations, figures and decisions here are invented. Maple Works asks for an aggregate snapshot of its selected employees at 23:59 IST on 30 September 2026 and a separate September safety summary. The authorised owners supply the complete miniature evidence packet in the download. It includes no personal or medical records.
Workforce packet W1 confirms sixteen people with one active contract each and two people with two active contracts each. These groups are disjoint and exhaustive for the selected snapshot. They represent eighteen people and twenty contracts. The custodian has already established the person links; the reporting coordinator is not asked to reconstruct them.
Legacy export L1 has column “count” equal to twenty, defined as active contracts. Replacement export N1 uses the same heading for eighteen distinct selected employees. Both describe the same snapshot and underlying population. Treating twenty to eighteen as a workforce reduction would mistake a changed counting unit for a change in people.
The dictionary therefore retains separate IDs HC_PERSON for the person count and CT_ACTIVE for the contract count. N1 maps to HC_PERSON; L1 maps to CT_ACTIVE. Neither is a correction to the other’s arithmetic. The proposed employee trend remains on hold until each period is supplied on one approved person-count basis.
Keep event counts separate from case counts
Safety packet S1 uses an owner-approved internal scheme for this fictional exercise. Event group A contains one event and two case records; group B contains one event and one case record. Together they contain two events and three case records. No claim is made that these are statutory recordable injuries or that the three records represent three distinct people.
Legacy export L2 labels its two-event total “incidents”. Replacement export N2 uses that same heading for the three case records from the same September packet. The new number does not establish a rise in events or a deterioration in safety. Create SFT_EVENT and SFT_CASE definitions, with their separate counting units and source mappings.
The safety owner’s record defines a case record for this exercise as one separately retained internal case entry linked to an event; multiple entries can share an event. Whether entries concern the same person across events is not supplied. Do not infer a distinct-person count, a severity category or an incident cause from the aggregate bridge.
No hours-worked schedule or rate specification accompanies S1. The dictionary records hours as unavailable for this request and leaves the proposed frequency-rate field unresolved. Zero would mean a measured quantity, not the absence of the required evidence. The event and case counts can still be described within their stated internal scope.
Specify value states and sharing conditions
Use separate fields for value and status. Define what measured zero, unavailable, not applicable and withheld mean in this particular output. Preserve the source token and the owner’s explanation. Do not invent a zero to make an import accept a blank, or change a withheld figure into an arithmetic estimate from neighbouring cells.
Record the approved recipient, output granularity and access route for evidence questions. Aggregation is not an automatic release decision. The template supplies no universal small-group threshold and performs no privacy test. Keep names, health details and matching keys with the authorised custodians; obtain the review needed for the actual output and audience.
In the example, the dictionary is a restricted internal working record for the reporting team. The fictional decision supplies no approval to publish real employee or safety aggregates. A later public table needs its own review of the exact fields, combinations, explanations and recipients. The manual release sheet preserves that separate decision.
Test the dictionary before handing it over
Ask a second reader to map L1, N1, L2 and N2 without seeing the intended dashboard. They should obtain twenty contracts, eighteen people, two events and three case records, not a single before-and-after series. If the reader needs to guess what “count” or “incidents” means, the mapping is not ready.
Try three deliberate failures. Reuse the latest definition for L1 and the historical contract count becomes a false person count. Join the two safety exports as successive periods and the migration becomes a false trend. Replace unavailable hours with zero and an evidence gap becomes a misleading measurement. Retain the failed test and correction reference.
Freeze the dictionary version with the exact aggregate output and source mappings. When an owner changes a definition, record which measures and periods need reconsideration and who will supply a comparable restatement, if possible. Keep earlier versions available; a new definition does not automatically make every earlier value convertible.
Use the HR aggregate handover guide for controlled access arrangements and the safety data guide for population and hours checks. The source document register helps preserve evidence versions. This dictionary adds reusable field definitions and migration mappings; it does not replace those operational reviews.
Source: US Bureau of Labor Statistics, BLS OSH Definitions, incidence-rate definitions. Checked 15 October 2026.
Version 1.0. Original Nitium Advisory forms, diagram and fictional migration case. No source endorsement, legal conclusion, clinical assessment, independent assurance or automated data validation is supplied. Apply the separately agreed reporting definitions and access decisions to real work.
Comments