SPARCS Data Limitations Every Analyst Should Know Before Drawing Conclusions

By Two Pillars & Co Advisory Group · Sources checked October 2026

Why do SPARCS numbers need caveats?

A figure from a large state dataset carries an air of authority. The people who produce the data are clear about its limits, and the limits are not buried, but they are easy to skip. Several of them appear in the Department’s dataset notes and reports pages, and putting them in one place helps an analyst write a fair caveat.

The Department encourages users to read the overview, data dictionary and release notes, which describe how datasets were developed, the exclusions applied and other important considerations.^[1] A short checklist of the points below is a good complement to those documents.

Is the data complete?

The dataset notes state that the data is subject to change and that, at the time of an open dataset release, some facilities are not in full compliance with reporting requirements.^[1] That means a count for one facility in a given release may be lower than the facility’s true volume, and the shortfall is not quantified in the release.

Facilities have up to 180 days after the end of a service month to submit all of that month’s data, so the most recent months in any release are the most likely to be incomplete.^[5] Analysts who compare the latest year with earlier years should expect the gap to narrow in later releases. The sources do not say how incomplete a given release is for a given facility. An analyst who needs that can ask the program or compare the figure with other sources, such as facility reports.

What does “pre-adjudicated” mean?

The Department’s submission page describes SPARCS as collecting pre-adjudicated claim data.^[4] That means records are collected before insurers have finished processing the claims. Final payment amounts, denials and adjustments may therefore not be reflected, and the charges in the data are what the facility billed, not what was paid.

This is why public hospital data cannot, by itself, tell you about profitability. A companion article in this series on that subject covers the financial side of the same caution. For this article, the key point is simple: a SPARCS charge is not revenue.

How do classification changes affect comparisons?

Two changes illustrate the issue. The notes for the inpatient file say that the expected-payer variable changed from source of payment to payment typology, effective December 2015, so payer mix before and after is not strictly comparable without care.^[1] And in July 2026, the program began classifying outpatient claims with revenue code 0361 (operating room services, minor surgery) as ambulatory surgery, without reclassifying earlier accepted records. Counts by claim type before and after that date therefore differ for reasons unrelated to patient activity.^[3]

Whenever a comparison spans years, an analyst should check the release notes for definition changes and read the forum summaries for coding changes. A trend that looks like a change in patient behavior may be a change in how records are categorized.

What should a fair caveat say?

A fair caveat for a SPARCS-based figure would say which release it came from and when, note that some facilities may be incomplete at release, state any exclusions and redactions that affect the analysis, and avoid presenting counts from the compliance dashboards. The Department is explicit that those dashboards should not be used for public reporting of admission or visit counts.^[2]

A simple discipline follows: record the dataset name, year, release date and the version of the data dictionary for each figure. If a number is later questioned, that record shows how it was produced, and it makes it easy to refresh the figure when the Department updates the data. If the figure drives a major decision, triangulate it with a second source, such as the facility’s own records, though definitions may differ.

Frequently asked questions

What to do next

1. Record the release details for every figure you calculate.
2. List the limits that apply to your question.
3. Write the caveat before you write the headline.
4. Refresh the figure when the Department updates the data.