Your charts are probably passing your current QA review. That's exactly the problem.
Most EMS agencies review somewhere between 5 and 10 percent of PCRs, usually the ones flagged by supervisors or pulled at random. Required fields are populated. Times are entered. The ePCR isn't throwing errors. On the surface, the chart looks done. We've written before about why this 10 percent sampling model no longer matches modern EMS documentation volume.
But "looks done" and "will hold up in an audit" are two different standards. The gaps that matter most are almost never empty fields. They're the clinical story problems: the documentation that's present but thin, the reassessments that didn't happen or weren't recorded, the rationale that's missing the one sentence that would make it defensible.
Here are the five we see most often when we run 100 percent automated review across an agency's finalized charts.
1. Missing or inadequate reassessment intervals
Protocol typically calls for reassessment every five minutes on a critical patient and every fifteen on a stable one. What we see in the charts: a set of vitals at patient contact and another at destination. Everything in between is blank.
On a 20-minute transport, that's three to four reassessment windows that have no documentation. If a patient deteriorates and the record doesn't show you were tracking them, the gap is indefensible, and auditors know exactly where to look.
What to watch for: any critical-acuity call where the time between first vitals and arrival vitals is longer than the protocol's reassessment interval and there's no documentation of what happened in between.
2. Glasgow Coma Scale recorded as a total, not as components
GCS is a three-part score: eye, verbal, motor. Many crews document the total and nothing else. A GCS of 13 is a GCS of 13 until someone asks how you got there, and if the only answer you have is a single number in the field, you can't prove it.
This matters in trauma cases, in altered mental status calls, and anywhere a provider's clinical judgment is going to be scrutinized after the fact. Deterministic, NEMSIS-native validation rules catch a truncated GCS entry automatically, without waiting for a chart to land in a manual sample.
What to watch for: GCS documented without all three components entered separately, or no repeat GCS on calls where mental status was a factor.
3. Medication administration with no documented indication or response
Drug given, dose correct, route correct. Technically the medication fields are complete. But there's no documentation of why the medication was given or what the patient's response was afterward.
This is one of the first things a medical director looks for in a chart audit. It's also what an attorney's expert witness looks for in a lawsuit. The administration alone doesn't demonstrate that your provider made a sound clinical decision. The indication and the response assessment do. This is the kind of pattern that exception-based review is built to surface instead of relying on a QA officer to catch it manually.
What to watch for: any ALS intervention where the narrative doesn't include a sentence establishing the clinical indication and a follow-up note on patient response.
4. Patient refusal documentation that won't hold up
Refusal calls carry disproportionate legal risk. You've done everything right clinically, but if the documentation doesn't capture a capacity assessment and an explanation of the risks of refusing care, the chart isn't protecting you.
Required elements vary by protocol and jurisdiction, but the common gaps are: no documentation that the patient was oriented and capable of making the decision, and no documentation that risks were explained and the patient acknowledged them. Boilerplate "patient refused transport, signed refusal form" by itself rarely satisfies a detailed review.
What to watch for: refusal calls where the narrative is shorter than the clinical situation warrants, or where there's no language establishing decision-making capacity.
5. Time stamps that create more questions than they answer
Scene times that run longer than anything in the narrative can account for. Dispatch-to-response intervals that don't match the CAD data. Patient contact times entered after transport began.
These inconsistencies are often innocent, the result of a crew entering times from memory at the end of a long shift. But they're also the first thing a billing auditor, an accreditation reviewer, or an opposing attorney checks. An unexplained time anomaly signals to a reviewer that the rest of the chart may need scrutiny.
What to watch for: any call where scene time exceeds 15 minutes and the narrative doesn't document what took that long, or where time fields are inconsistent across the record.
Why sampling misses these gaps
All five of these problems are present in low single-digit percentages of charts in a typical agency. A 10 percent sample might catch one of them occasionally, mostly by luck.
At 100 percent review, they show up consistently. You can measure them, trend them by provider and shift, and address them before they accumulate into a liability exposure or a billing audit finding. This is the same dual-engine model of deterministic rules plus AI-assisted triage that we've outlined as the path beyond sampled QA.
That's the point of automated review: not to replace a medical director's judgment, but to make sure no chart falls through the cracks between the ones your current process happens to pull.
See it on your own charts
Integritas EMSQA runs deterministic, NEMSIS-native rules across 100% of your finalized charts, no sampling required and no ePCR switch needed. See how it fits your stack on our integrations page, or how we secure agency data end to end. Book a demo to see how these five gaps show up in your own documentation.