A plant passes its walkaround with nothing more than a loose extension cord flagged. Two weeks later, the same site gets cited — not for a hazard, but for how three cases were logged. That gap is the story. OSHA recordkeeping has stayed among the more frequently cited categories in general industry inspections for years, and the regulation behind it, 29 CFR Part 1904, hasn't changed in any way that explains it. What has stayed constant is the structure of the job itself: judgement calls made under time pressure, by different people, at different sites, applying the same words differently. This piece lays out why that keeps happening and what closes the gap.
Why does OSHA keep citing recordkeeping when the rule hasn't changed?
Because the rule was never built to be mechanical. 29 CFR Part 1904 asks employers to answer three questions for every injury or illness: is it work-related under 1904.5, is it a new case under 1904.6, and does it meet a recording criterion under 1904.7 — death, days away from work, restricted work or job transfer, medical treatment beyond first aid, loss of consciousness, or a significant diagnosis. Two of those tests, restricted work and medical treatment beyond first aid, don't resolve on their own. They depend on what actually happened to a specific worker on a specific day, and that's exactly where two reasonable people can land on different answers.
Compare that to a machine-guarding citation. A guard is either there or it isn't. A recordkeeping determination doesn't work that way. It runs through a chain of small judgements — was this a restriction on a routine job duty, was the treatment given more than first aid — and each link is a place the classification can slip. Inspectors know this. When they check a 300 Log against underlying medical records, incident reports and witness statements, they're not hunting for a missing signature. They're re-running the same judgement call the employer already made, and checking whether it holds up.
Why recordkeeping stays structurally hard to get right
None of this is about employers being careless. It's about a set of features baked into how recordkeeping actually gets done inside a real company.
| Structural factor | Why it produces inconsistency |
|---|---|
| Judgement-dependent criteria | Restricted work and medical treatment beyond first aid both hinge on facts, not fixed thresholds, so two coordinators can read the same case differently |
| Multi-site consistency | Each establishment keeps its own log, so a company with ten sites effectively runs ten separate recordability programmes unless someone actively aligns them |
| Staff turnover in the safety-coordinator role | The person making the call changes every year or two, and training rarely covers the closed first aid list or 1904.7 in real depth |
| Form ambiguity | Free-text fields on Form 301 and the 300 Log invite paraphrase rather than a fixed classification, so wording drifts case to case |
| No documented decision trail | Without a record of why a case was classified a certain way, nobody — including the person who made the call — can check it later |
| Time pressure at the point of entry | Cases get logged in the days after an incident, often by whoever is available, not necessarily whoever understands the rule best |
Table: structural reasons OSHA recordkeeping stays error-prone, independent of any change in the underlying regulation.
Take the first aid list on its own. 1904.7(b)(5)(ii) names fourteen specific treatments that count as first aid and nothing else. Anything outside that closed list is medical treatment, and medical treatment beyond first aid is a recording criterion on its own. Fourteen items sounds manageable until you watch it applied in practice. A new safety coordinator sees a worker prescribed an over-the-counter strength ibuprofen at a higher dose and assumes that's still first aid because it's "just ibuprofen." It isn't — prescription-strength dosing of an OTC medication falls outside the list. Multiply that kind of near-miss judgement across shifts, sites and a rotating cast of coordinators, and the pattern OSHA sees on audit stops looking random and starts looking structural. Our first aid versus medical treatment guide walks through the full fourteen-item list line by line, because "closed list" only helps if you actually know what's on it.
Does decentralised recordkeeping make this worse across multiple sites?
Yes, and it's probably the single biggest multiplier. The 300 Log, 300A summary and ITA submission cycle is a per-establishment obligation. The log runs all year at each site. The 300A gets posted from 1 February to 30 April at each site. The ITA submission is generally due 2 March. None of that is centralised by design — OSHA built the rule around the establishment, not the company. That means a business with five locations has five people, or five rotating shifts of people, independently deciding what counts as a restriction or what counts as medical treatment. Without a shared, written standard for how those calls get made, five sites will produce five slightly different interpretations of the same rule. It's not that any one coordinator is wrong. It's that nothing forces the five answers to agree with each other, and OSHA can visit any one site and find the local version doesn't hold up.
Is high turnover in safety-coordinator roles really a factor in citations?
It shows up constantly in how logs actually get built. The person filling out the 300 Log this year is often not the person who filled it out three years ago. Recordkeeping training is frequently a short module bundled into a broader safety orientation, not a standalone deep dive into 1904.7's criteria. A new coordinator inherits a log, inherits informal habits from whoever trained them for twenty minutes, and starts making calls without ever having read the regulation closely. That's not a knock on the individual — it's what happens when a judgement-heavy compliance function sits on a role with a short average tenure. The fix isn't better people. It's a documented, repeatable process that doesn't reset every time the role changes hands.
A worked example: passing the walkaround, failing the log
Here's a realistic version of the pattern inspectors describe seeing. A mid-sized manufacturer runs a clean floor: guards in place, lockout-tagout followed, PPE compliance high. The physical inspection turns up almost nothing. Then the inspector pulls the 300 Log and asks to see the medical documentation behind three entries from the past year.
One case was a laceration treated with steri-strips at an urgent care clinic — logged by the site as first aid because "it was just strips." Steri-strips aren't on the closed 1904.7(b)(5)(ii) list; they're medical treatment, which makes the case recordable on that basis alone, regardless of how the wound looked. A second case involved a worker moved to a different task for four days after a shoulder strain — the site's report called it a "voluntary change" but the underlying paperwork showed the supervisor had restricted the worker's normal duties, which is a recording criterion under 1904.7(b)(4), not a voluntary accommodation exempt from the log. A third was a simple miscount — a recordable case never made it onto the log at all because the coordinator who handled the incident left the company two months later and the case was never transferred into the new coordinator's tracking sheet.
None of those three facts involve a hazard on the floor. All three become recordkeeping citations, because the underlying facts were classified against the wrong standard, not because the workplace was unsafe. That's the part that catches employers off guard: a clean safety record and a flawed log are not mutually exclusive, and OSHA treats them as separate questions.
Where teams get this wrong
A few patterns recur often enough to name directly.
- Treating the log as an administrative afterthought. The 300 Log gets filled in after the fact, by whoever has a spare ten minutes, rather than as a determination that deserves the same rigor as any other compliance decision.
- No documented process for how determinations are made. If nobody can point to a written standard for how a "restricted work" call gets made, every case is decided from scratch, and consistency depends entirely on memory and habit.
- No second reviewer on judgement calls. The cases that actually need a second set of eyes — the borderline restricted-work cases, the medical-treatment-versus-first-aid calls — are exactly the cases most likely to go through on one person's read of the facts.
- Assuming the first aid list is a vibe, not a list. Coordinators often reason from what "sounds minor" rather than checking the case against the actual fourteen items in 1904.7(b)(5)(ii). That gap is where a surprising number of misclassifications start.
- Losing continuity across staff changes. When a coordinator leaves, the reasoning behind past classifications leaves with them unless it was written down. The next person inherits a log with no audit trail behind it.
One EHS manager we spoke with put it this way: "We didn't have a bad safety record. We had five different opinions about what 'restricted work' meant, spread across five plants, and nobody had ever put those opinions next to each other to see they disagreed." That's not a hazard problem. It's a consistency problem, and consistency problems are what recordkeeping citations are made of.
The fix is a repeatable process, not more vigilance
Telling coordinators to "be more careful" doesn't fix a structural problem. What fixes it is removing the judgement variance itself — running every case through the same test, citing the same provision, every time, regardless of which site or which coordinator handled it. That's the specific gap Job13's recordability check is built to close: it applies the 1904.5 through 1904.7 test consistently and tells you plainly when a case is recordable, when it isn't, or when the facts genuinely need human review — it doesn't force a guess where the regulation leaves room for judgement. It won't guarantee an inspector agrees with every call OSHA might make, and this isn't legal advice, but it does mean every site is starting from the same standard instead of five different ones. For teams managing the log itself, Job13's paid plans keep the 300 Log, 300A summary and ITA export built on those same consistent determinations, and the API and MCP docs cover how to wire that check into an existing incident workflow.
Frequently asked questions
Why is OSHA recordkeeping such a common citation category?
Because the underlying test in 1904.7 depends on judgement calls — particularly around restricted work and medical treatment beyond first aid — that different people apply differently. Recordkeeping-related standards have remained among OSHA's more frequently cited categories in general industry inspections in recent years largely because the rule itself invites this kind of variance, not because employers are careless about paperwork.
Can a company be cited for recordkeeping even if its safety record is otherwise clean?
Yes. Recordkeeping citations concern how cases were classified and logged, not whether the workplace has hazards. A site can pass a physical walkaround and still be cited if the underlying medical facts of a handful of cases were classified against the wrong standard — treated as first aid when the treatment given falls outside the closed list in 1904.7(b)(5)(ii), for example.
What is the biggest single driver of recordkeeping inconsistency across multiple sites?
Decentralisation. The 300 Log, the 300A summary and the ITA submission are all per-establishment obligations, so a multi-site company has as many independent decision-makers as it has locations, unless it deliberately builds a shared standard for how those decisions get made.
Does high turnover among safety coordinators actually affect recordkeeping accuracy?
It's a significant factor. The person filling out the log changes more often than the rule does, and recordkeeping training is frequently a short module rather than a deep read of 1904.7. Without a documented, repeatable process, each new coordinator effectively restarts the interpretation from scratch.
How can a company reduce the risk of a recordkeeping citation?
Apply the same three-part test — work-relatedness, new case, a 1904.7(b) criterion — the same way for every case, and write down the reasoning so it survives staff changes. Job13's free recordability check runs that test consistently and flags genuinely judgement-dependent cases as needing review rather than forcing an answer; see OSHA's own recordkeeping guidance and Part 1904 in the eCFR for the source text.
