The Fields Behind the Most Common SART CORS Validation Errors

When most reporting teams worry about validation, they worry about the objective data: a wrong retrieval date, a miscounted embryo, an outcome recorded incorrectly. CDC's annual validation of ART data shows almost the opposite. Those fields are reported nearly perfectly. The fields that most often disagree with the medical record are the patient's diagnosis codes, the reason for ART. That is where the discrepancies cluster, and it is worth understanding why.

How the validation works

Each year, CDC validates a sample of reporting clinics against the medical record. In recent years about 35 clinics, roughly 7 to 10 percent of all reporting clinics, are selected through stratified random sampling, with larger clinics more likely to be chosen. Validation usually runs May to June, and the team compares what the clinic reported with what the chart actually says.

For the 2022 reporting year, 35 of 457 reporting clinics were selected, and 2,073 cycles were fully validated along with 301 fertility-preservation banking cycles partially validated. For each clinic, the fully validated sample includes up to 40 cycles resulting in pregnancy, up to 20 not resulting in pregnancy, and up to 10 using donor eggs or embryos, plus up to 10 banking cycles. From that comparison, CDC calculates a discrepancy rate for each field: the share of cycles where the reported value differs from the record.

The data your clinic enters in SART CORS is the same clinical data this national validation measures. So these discrepancy rates are a good map of where reporting most often drifts away from the chart.

Where the discrepancies actually are

Here are the discrepancy rates for the 2022 reporting year. Notice the split.

Every objective event field, the dates, the embryo count, the outcome, the number of infants, sits at or below 0.6 percent, most of them near zero. The four highest discrepancy rates in the entire table are all patient-diagnosis fields, the reason for ART.

The four fields that cause the trouble

The direction of each error is as telling as the rate:

• Ovulatory dysfunction, 2.1 percent, overreported. In 60 percent of the discrepancies, the clinic reported it but it was not found in the medical record.
• Other factors, 2.0 percent, underreported. In 61 percent of discrepancies, it was in the record but not reported.
• Diminished ovarian reserve, 1.3 percent, underreported. In 84 percent of discrepancies, it was in the record but not reported.
• Unknown factors, 1.3 percent, underreported. In 74 percent of discrepancies, it was in the record but not reported.

So one diagnosis gets added when the chart does not support it, and three get left off when the chart does. That pattern points straight at how these fields are entered.

Why diagnoses drift and dates do not

A retrieval date is a fact. It is pulled from a structured record, and there is nothing to interpret, which is why it almost never disagrees with the chart. A diagnosis is a decision. Whether ovulatory dysfunction or diminished ovarian reserve gets recorded depends on how the chart is read and coded, and that is where reported data separates from documented data.

Overreporting ovulatory dysfunction looks like a field being set when it is assumed rather than documented. Underreporting diminished ovarian reserve, other factors, and unknown factors looks like the opposite problem: a single primary reason gets entered and the contributing factors that are in the chart never make it into the report. Both are reason-for-ART fields being treated as a quick selection rather than a faithful transcription of the record.

How to keep these fields clean

The fix is narrow, because the problem is narrow:

• Code the reason for ART from what the chart documents, not from assumption or a default selection.
• Capture every contributing factor, not just the headline diagnosis. Diminished ovarian reserve, other factors, and unknown factors are underreported precisely because they are secondary reasons that get dropped.
• Reconcile the reported diagnosis against the chart before the cycle is submitted, while the record is still in front of you.

The objective fields will mostly take care of themselves. The diagnosis fields are where attention actually pays off.

How EmbryoLogix catches it before submission

EmbryoLogix pulls the diagnosis and reason-for-ART fields from the EMR record rather than leaving them to a manual pick, which removes the guesswork that drives both overreporting and underreporting. It flags cycles where a reported diagnosis is not supported by the record, and where factors documented in the chart are missing from the report, and it surfaces those fields for review before the file is submitted. Because these are the fields the national validation weighs most heavily, getting them to match the chart is where reporting accuracy is actually won.

The quick version

• The dates, embryo counts, and outcomes are almost never the problem. They sit near zero percent.
• The discrepancies live in the reason-for-ART diagnosis fields. Ovulatory dysfunction is overreported, while diminished ovarian reserve, other factors, and unknown factors are underreported.
• Code the diagnosis from the chart, capture every contributing factor, and reconcile before submission. That is where the most common validation errors are made and where they are prevented.