Under PDPM, the Minimum Data Set (MDS 3.0) is more than a clinical record. The way an assessment is completed, coded and submitted shapes the payment a building receives for a resident's stay. That means a lot of small, human steps carry real financial weight, and each is a place where information can quietly slip.
This post walks through the path from assessment to payment and looks at where analytics can give leadership early visibility. The point is not to chase higher codes. It is to make sure the record accurately reflects the care and needs of the resident, and that nothing falls through the cracks between departments.
The path in five steps
- Information gathering. Nursing, therapy, dietary, social services and the physician all contribute observations and documentation.
- Assessment completion. The MDS coordinator and team code items based on that documentation and the assessment look-back periods.
- Submission. The completed assessment is transmitted within required timeframes.
- Classification. The data is grouped into PDPM components that drive the payment rate.
- Billing and reconciliation. The claim goes out and the business office compares what was expected with what was paid.
Where the gaps tend to appear
Between departments
Therapy may document a functional change that nursing documentation does not reflect, or a diagnosis may be in the physician's notes but not on the problem list the MDS team uses. Neither department did anything wrong. The information simply did not meet in time.
A useful dashboard compares sources side by side and flags mismatches for a human to review. Think of it as a prompt to ask a question, not a verdict.
Around timing
Assessment windows are strict. A schedule that slips by a day or two, a weekend admission, or a coordinator out on leave can create late or missed assessments. Tracking assessments by due date, status and owner makes the upcoming week visible instead of leaving it in someone's head.
Between expected and actual payment
When the business office sees a payment that looks off, the investigation often starts weeks after the assessment. Analytics can shorten that loop by comparing the expected classification at the time of assessment with what the claim reflected, and highlighting residents where the two diverge.
Metrics worth watching
- Assessments due in the next seven days, by type and owner.
- Assessments completed late or at risk of being late, trended over time.
- Time from admission to first assessment completion.
- Case mix distribution for the building, trended month to month, to catch unexplained shifts.
- Open questions or queries waiting on a clinician response, and how long they have been open.
Notice that none of these ask whether the building is coding high or low. They ask whether the process is running on time and whether information is flowing. Unexpected movement in case mix should prompt a conversation about whether it reflects real changes in the residents you serve, for example a shift in admission sources, rather than an assumption either way.
Guardrails matter
Any reimbursement analytics program should be built on documentation integrity. Coding should follow what is supported by the clinical record, and dashboards should be designed to surface questions to clinicians and compliance staff, not to push a number. Many operators pair a monthly review of case mix trends with their compliance committee so that unusual patterns get a second set of eyes.
A practical first step
If you want to start without any new software, pick one week and trace five resident stays from admission to claim. For each, note who touched the information, how long each handoff took, and where anyone had to chase a missing piece. The friction you find is your list of things to measure.
Where CarePulse fits
CarePulse can pull assessment schedules and classification data from your EHR, such as PointClickCare, into a view that shows due dates, late risks and case mix trends in one place. If you would like to see your own assessment pipeline mapped this way, we are happy to walk through it with you using your data.