HIM 220 Module 4 Project One Example

Project One: One Measure, Written Down, a Readmission Data Dictionary for Kettle River Health

[Student Name]

Southern New Hampshire University

HIM 220: Healthcare Data Management

Project One

[Instructor Name]

[Date]

The organization, setting and figures below are a composite written as a model document. No real employer, client, colleague or patient is described.

Project One: One Measure, Written Down, a Readmission Data Dictionary for Kettle River Health

Kettle River Health's leaders saw two readmission rates for the same months and could not tell which to believe. The first module traced the gap to definitions nobody had written down. This project writes them down. It describes who uses readmission data and why, defines three labeled measure versions, documents each data element and sets rules for maintaining the dictionary.

What a Data Dictionary Does

A data dictionary is a shared reference that defines each data element and measure: its name, meaning, format, allowed values, source, owner and the rules used to calculate or derive it. Kahn et al. (2016) emphasized that data quality depends on conformance to agreed definitions; a dictionary is where those definitions live. It also makes reports reproducible, so two analysts following the same entry should get the same number.

Users and Purposes

Three groups use readmission data for different purposes. The quality department wants to find every patient who returns to any Kettle River site so care transitions can improve. Finance wants to anticipate penalties and payer contract measures, which count only certain returns. The board wants a single headline figure it can track over time. One definition cannot serve all three, so the dictionary defines three measures, each labeled with its purpose.

Measure A: System All-Cause Return

Measure A, for quality improvement, counts any inpatient admission or observation stay at either Kettle River hospital within 30 days after an index inpatient discharge, for patients of any age and payer. It excludes index stays ending in death, transfer to another acute hospital or discharge against medical advice. Planned returns are included, since quality staff review them separately. Zuckerman et al. (2016) showed that observation stays rose as readmissions fell under Medicare's penalty program, which is why this measure counts observation returns.

Measure B: Unplanned Inpatient Return, Same Hospital

Measure B, for finance and contract monitoring, counts unplanned inpatient readmissions to the same hospital within 30 days after an index inpatient discharge. It excludes observation stays, planned readmissions identified by a documented scheduled procedure or chemotherapy list, index stays ending in death, transfer or discharge against medical advice and returns to the other hospital. This matches how the finance team calculated 11.6%.

Measure C: Board Headline

Measure C, for the board, counts unplanned inpatient or observation returns to either hospital within 30 days, excluding planned returns and the same index exclusions. It sits between the other two and changes little when coding practices shift between inpatient and observation status. Joynt and Jha (2012) cautioned that readmission rates reflect patient and community factors as well as hospital care, so the board report will show Measure C alongside the share of index patients living in high-deprivation neighborhoods.

Reconciling the Two Figures

Recalculating last quarter's data under each definition reproduces both original numbers and explains the gap. Measure A yields 14.2%, matching the quality report. Measure B yields 11.6%, matching finance. Measure C yields 12.9%. About half the difference between A and B comes from observation returns, a third from planned readmissions and the rest from returns to the other hospital.

Table 1. Three Readmission Measures, Same Quarter

Note. Recalculated by the author from warehouse data for the most recent quarter.

Element-Level Entries

Each measure depends on data elements that also need definitions. Table 2 shows selected entries. Every entry lists the element's name, definition, data type, allowed values, source system, owner and any business rule. Owners are roles, not individuals, so entries survive staff turnover.

Table 2. Selected Data Dictionary Entries

Note. Selected entries; the full dictionary contains 24 elements.

Rules That Prevent Drift

Definitions drift when people change them quietly. Every change will be requested in writing, reviewed by the data governance committee and recorded with a version number, effective date and reason. Reports will display the measure name and dictionary version beneath each chart. Retired definitions stay in the dictionary so historical reports can be reproduced. New analysts will be trained on the dictionary during orientation, and any report built outside it will carry a warning label until it is reviewed.

Validating the Definitions

To confirm the dictionary works as intended, a sample of 50 index discharges was reviewed against the full record. Measure logic matched the chart in 47 cases. The three mismatches involved observation stays later converted to inpatient status, which the warehouse had captured before utilization review finished. The encounter type entry now specifies using the final status, and the warehouse load will wait 72 hours after discharge before classifying encounters.

Connections to Data Quality

The dictionary also depends on the quality problems identified in Module Two. Duplicate records can hide readmissions when a patient's two stays sit under different identifiers, which is why the patient identifier entry requires the merged enterprise number. Implausible timestamps can create negative intervals, so the return encounter rule rejects them for review.

Conclusion

Writing down three purpose-specific definitions ended the argument over Kettle River's readmission rate: both numbers were right for their purposes, and a third serves the board. Element-level entries with owners and business rules, version control and validation make the measures reproducible. The next module calculates core healthcare statistics using the same discipline.

References

Joynt, K. E., & Jha, A. K. (2012). Thirty-day readmissions: Truth and consequences. New England Journal of Medicine, 366(15), 1366-1369. https://doi.org/10.1056/NEJMp1201598

Kahn, M. G., Callahan, T. J., Barnard, J., Bauck, A. E., Brown, J., Davidson, B. N., Estiri, H., Goerg, C., Holve, E., Johnson, S. G., Liaw, S.-T., Hamilton-Lopez, M., Meeker, D., Ong, T. C., Ryan, P., Shang, N., Weiskopf, N. G., Weng, C., Zozus, M. N., & Schilling, L. (2016). A harmonized data quality assessment terminology and framework for the secondary use of electronic health record data. eGEMs, 4(1), Article 18. https://doi.org/10.13063/2327-9214.1244

Zuckerman, R. B., Sheingold, S. H., Orav, E. J., Ruhter, J., & Epstein, A. M. (2016). Readmissions, observation, and the Hospital Readmissions Reduction Program. New England Journal of Medicine, 374(16), 1543-1551. https://doi.org/10.1056/NEJMsa1513024