HLTH 4403 Module 1 Electronic Health Record Data Analysis Example

Controlled in the Scanned Report, Uncontrolled in the Measure: How Electronic Health Record Data Are Captured and Used for Diabetes Care at a Community Health Center Network

Student Name

American College of Education

HLTH4403: Healthcare Information Management

Module 1 Assignment

Instructor Name

October 5, 2026

The Organization and Its Record

This paper examines a composite group of a dozen community health center sites serving about 41,000 patients, most of them on Medicaid or uninsured. Around 5,900 adults in the network have diabetes. All 12 centers share a single electronic health record, which holds visit notes, problem lists, medication lists, orders and results, and which feeds the reports the network sends to its federal funder and to Medicaid managed care plans.

In its most recent annual report, the network stated that 31% of adults with diabetes had poor glycemic control, meaning a most recent hemoglobin A1c above 9% or no A1c recorded during the year. The medical director doubted the figure. Clinicians knew many patients whose diabetes was well managed but whose results came from hospital or outside laboratories. This paper examines how diabetes data enter the record, how they are used, and whether they are good enough for the uses the network depends on.

How Diabetes Data Enter the Record

Diabetes data arrive through four routes, and each produces data of a different kind. When a clinician orders an A1c from the network's contracted laboratory, the result returns through an electronic interface as a structured value, a number in a defined field with a date, which the record can sort, graph and count. When a patient has an A1c drawn at a hospital or another clinic, the result usually arrives by fax or as a portable document and is scanned into the chart as an image, which a person can read but the record cannot count. Point-of-care A1c tests performed during visits are typed into a flowsheet by medical assistants, and their accuracy depends on the person entering them. Finally, clinicians often mention results in the free text of their notes, such as A1c 7.4 at the hospital last month, which is visible to the next reader but invisible to any report.

The distinction between structured and unstructured data is the key to what follows. A human reader treats a scanned result and an interfaced result as equally real. A quality report sees only the second.

Primary and Secondary Uses

Data in the record serve two kinds of purpose. Primary use is the care of the individual patient: a clinician reviewing the trend in a patient's A1c before adjusting medication. Secondary use is everything else: quality reports, population registries, outreach lists, payer contracts and research. The federal meaningful use program, created under the incentive provisions of the 2009 HITECH Act, pushed providers to adopt electronic records and to use them for more than documentation, setting objectives that eligible professionals and hospitals had to meet to receive incentive payments, including reporting clinical quality measures (Blumenthal & Tavenner, 2010).

Secondary use has grown far beyond that program. The network's diabetes registry, built on structured A1c values, generates monthly lists of patients overdue for testing or above target, which care coordinators use for outreach. Its federal quality report is calculated from the same fields, and two Medicaid plans pay a quality bonus based on the percentage of members with controlled diabetes. Each of these uses depends on data that were entered, in most cases, for the care of one patient at one visit.

Assessing Data Quality

Weiskopf and Weng (2013), reviewing how researchers assess electronic health record data, sorted the aspects of data quality that researchers check into five groups: whether data are present (completeness), true (correctness), consistent across sources (concordance), believable (plausibility) and up to date (currency). The network's quality analyst and I used those dimensions to review a random sample of 200 patients whom the report counted as poorly controlled.

Completeness was the largest problem. Forty-six of the 200, 23%, had an A1c of 9% or lower during the year that existed only as a scanned image or in note text. Correctness problems were smaller but real: in four charts a point-of-care value had been entered in the wrong field or with the decimal misplaced. Concordance, whether different parts of the record agree, failed for 11 patients whose problem lists did not include diabetes even though their medication lists and results clearly indicated it, which could remove them from the registry entirely. Plausibility checks found two A1c values above 20%, both typing errors. Currency was a problem for problem lists, where resolved conditions remained active for years.

Why the Measure Was Wrong

If the sample is representative, roughly a quarter of the patients counted as poorly controlled are in fact controlled, and the network's true rate is closer to 24% than 31%. The error is not unique to this network. In a study at another federally qualified health center, Kern et al. (2013) compared electronically reported quality measures with manual chart review and found that how often the electronic report caught care that had truly been given swung widely from one measure to another, from under half to nearly all; electronic reports significantly underestimated care for two measures and overestimated it for one. An electronic report counts what the record can read, not what the clinic actually did.

The consequences are practical. The network is losing part of its Medicaid quality bonus, its federal report understates its performance, and care coordinators spend outreach calls on patients who do not need them while the list crowds out those who do.

Who Owns the Data

The review also exposed a governance gap. No one in the network was responsible for the quality of diabetes data as such. Clinicians owned their notes, medical assistants owned the flowsheets, the health information management staff owned scanning, and the quality analyst owned the report, but the problem sat in the handoffs among them. Scanning staff were measured on how quickly documents were filed, not on whether the values inside them reached a structured field, and the analyst had no way to see what the scanners saw. The network should therefore name a data steward for the diabetes registry, most naturally the quality analyst, with authority to set entry standards, to review the quarterly chart sample and to bring problems to the medical director. Without an owner, each of the fixes below would depend on goodwill and would likely erode as staff change.

Improving Capture and Use

The network should take four steps. First, when a scanned outside result arrives, the medical assistant who files it should also enter the A1c value and date in a structured field designated for outside results, a task of under a minute that the review suggests would correct most missing values. Second, the network should connect to the regional health information exchange so that hospital laboratory results arrive electronically, removing much of the fax and scan route. Third, plausibility rules should reject point-of-care entries outside a reasonable range at the moment of entry. Fourth, the quality analyst should repeat a 100-chart review each quarter, measuring the gap between the electronic report and manual review, so that improvements in capture can be seen in the numbers and not merely assumed.

References

Blumenthal, D., & Tavenner, M. (2010). The "meaningful use" regulation for electronic health records. New England Journal of Medicine, 363(6), 501-504. https://doi.org/10.1056/NEJMp1006114

Kern, L. M., Malhotra, S., Barrón, Y., Quaresimo, J., Dhopeshwarkar, R., Pichardo, M., Edwards, A. M., & Kaushal, R. (2013). Accuracy of electronically reported "meaningful use" clinical quality measures: A cross-sectional study. Annals of Internal Medicine, 158(2), 77-83. https://doi.org/10.7326/0003-4819-158-2-201301150-00001

Weiskopf, N. G., & Weng, C. (2013). Methods and dimensions of electronic health record data quality assessment: Enabling reuse for clinical research. Journal of the American Medical Informatics Association, 20(1), 144-151. https://doi.org/10.1136/amiajnl-2011-000681