SOII, CFOI, and OSHA ITA answer different injury-data questions

Start with the question and target population

Use SOII when the question calls for BLS estimates of nonfatal occupational injuries and illnesses within the survey's defined population. Retain the table, series, year, industry, geography, case definition, employment or hours denominator, estimate, standard error or other published uncertainty measure, extraction time, and any suppression or comparability note. A national estimate does not establish one employer's experience, and an apparent difference needs uncertainty and coverage review before it is called a trend. ^[1]

Use CFOI when the question concerns fatal occupational injuries. CFOI is a census program built from multiple source documents, and its counts and rates have their own definitions, scope, time basis, and revision state. A fatality count should not be inserted into a nonfatal survey denominator or interpreted as a severity extension of SOII without a defined method. ^[2]

Treat ITA records as submitted establishment data

Use OSHA's establishment-specific files when the question requires records submitted through the Injury Tracking Application, with the form, establishment, calendar year, filing population, and field definitions kept intact. The public files can include Form 300A summary data and, for covered establishments, case-detail data from Forms 300 and 301. A submitted field is an administrative record, not an independently verified finding about cause, work-relatedness, severity, or compliance. ^[3]

OSHA's page cautions that submissions may include establishments that were not required to submit and that the data are not necessarily representative or fully validated. Analysis therefore needs explicit inclusion rules, duplicate and amendment handling, missing-field treatment, denominator provenance, and a statement of what the selected file can support. ^[3]

Preserve comparability before calculating a benchmark

A source-selection record should state the reader's decision, intended population, outcome measure, numerator, denominator, period, industry system and level, geography, employment threshold, coverage exclusions, collection method, revision date, and uncertainty. Map each requested comparison to one compatible source or explain the transformation. Do not silently mix survey estimates, census counts, and establishment submissions because they share a year or industry label. ^{[1] [2] [3]}

Before presenting a rate or rank, test whether definitions and coverage are stable across periods and groups. Preserve suppressed cells, revised tables, changes in coding, pandemic or other collection disruptions, small denominators, establishment entry and exit, and unmatched identifiers. A lower value can reflect population or reporting differences; these sources alone do not establish hazard control, causal improvement, underreporting, or a compliant workplace. ^{[1] [2] [3]}

Reproduce one conclusion from source to chart

A representative evaluation should answer three separate questions: a nonfatal-industry estimate from SOII, a fatal-injury count or rate from CFOI, and an establishment-submission query from the OSHA ITA file. Reviewers should reproduce the exact download, filters, population, numerator, denominator, calculation, uncertainty or limitation, chart, and narrative. Then change the year, industry level, and establishment filter to confirm that incompatible values do not flow into the same benchmark. ^{[1] [2] [3]}

The reviewed official pages establish the programs' stated data products and limitations. They do not establish the correctness of a local extraction, comparability of a chosen cohort, causation, employer performance, recordkeeping compliance, injury preventability, control effectiveness, or worker safety. Those conclusions require defined methods, complete records, and qualified occupational-safety and statistical review. ^{[1] [2] [3]}

Limitations and unknowns

• No local extraction, table selection, cohort, denominator, calculation, employer record, filing obligation, event classification, or trend was independently tested.
• The reviewed pages do not establish that separate years, industries, geographies, establishment groups, or files are directly comparable.
• No source alone establishes causation, recordkeeping compliance, underreporting, control effectiveness, or a safe workplace.

Enterprise buyer test

Translate this change into the exact population, record type, workflow stage, decision owner, effective date, and evidence that could be affected. Ask current or prospective providers to demonstrate the named workflow with representative data and an exception—not a polished feature tour. Record what official documentation establishes, what a provider states, what the team observes, and what remains unresolved.

A defensible review also identifies the dependency outside the product. Authority interpretation, policy configuration, data quality, integrations, human judgment, approval rights, release governance, training, and retained evidence may remain customer or service responsibilities. The evaluation should preserve those boundaries instead of treating a technology claim as the complete operating model.

What we will watch next

Safety Operations Standard will watch the named source and affected market records for later evidence that changes status, scope, availability, implementation timing, workflow consequence, or the limits of the initial report. A later announcement does not silently overwrite this dated account; the change ledger preserves the sequence.