HCI 5073 Module 5 Informatics Improvement Proposal Example

Finding the Heat Cases We Miss: An Informatics Improvement Proposal for County Heat-Illness Surveillance Covering Case Definition, Data Quality Monitoring, Data Sharing and Decision Support

Student Name

American College of Education

HCI5073: Public Health Informatics

Module 5 Assignment

Instructor Name

November 2, 2026

The Problem This Proposal Solves

Across four modules, this project has examined how a composite Midwestern county watches for heat-related illness. The emergency department syndromic system is fast and dependable, but a chart review found that its query identified only about two thirds of heat cases, and the missed cases were concentrated among older patients triaged for vague complaints such as a fall or feeling weak. The data quality analysis traced part of that gap to the feed itself: late diagnosis codes at one hospital, triage notes from only two of four hospitals, a query that reads only the first three diagnosis fields, unreported interface outages and visits across the state line that never arrive. The dashboard evaluation found that staff had no seasonal baseline against which to judge a day's count and that small-number maps misled users.

This proposal brings those findings together into one plan. Its aim is a surveillance system that finds more of the heat illness that occurs, especially among older adults, tells staff clearly when conditions are unusual and can be trusted on the hottest days. It is designed to be carried out within two heat seasons, using mostly existing systems.

Component One: A Broader, Tested Case Definition

The first change is to the query. Harduar Morano and Watkins (2017), studying heat illness in Florida's emergency, hospital and death records, concluded that searching all diagnosis fields and combining data sources may make heat surveillance more sensitive. The county will ask the state to extend its heat query to every diagnosis field in the record rather than the first three. It will also test a second, broader definition for use on heat advisory days only: visits by adults 65 and older with chief complaints of weakness, dehydration, syncope, falls or altered mental status. On ordinary days such visits are mostly unrelated to heat, but on advisory days an excess above the expected count is a useful signal of hidden heat illness.

Both definitions will be validated before use. An analyst and a nurse reviewer will compare a sample of 400 visits from the previous summer against clinicians' full notes to estimate sensitivity and predictive value for each definition, so that the county knows what it gains in cases found and what it loses in false positives.

Component Two: Automated Data Quality Monitoring

The second component turns the one-time quality audit from Module 2 into a routine. Using the conformance, completeness and plausibility categories from Kahn et al. (2016), the epidemiologist will build an automated daily report, run before the dashboard refreshes, that checks each hospital's feed for expected visit volume compared with its recent average, the share of visits with diagnosis codes within 24 hours, the presence of triage notes and chief complaints, and valid ZIP codes. When a hospital falls below a set threshold, for example visit volume more than 20% below its four-week average, the report will send an alert to county staff and to the hospital's interface contact. A weekly summary will go to all four hospitals so that each can see its own performance next to the others, which in similar programs has encouraged improvement without new mandates. The county will also ask the two hospitals that do not send triage notes to add them, since the state platform can already receive them.

Component Three: Filling Gaps With Other Data

The third component adds data sources to reach people the emergency department feed does not. First, the county will request daily emergency medical services run data from the regional ambulance providers, which already report to the state's electronic records system. Calls coded for heat exposure, and calls to older adults found unresponsive at home on hot days, would capture people who never reach a hospital or who die before arrival. Second, the county will work with the state to negotiate a formal exchange arrangement with the adjoining state so that residents' visits to hospitals across the border are included, beginning with the two border hospitals that account for most of those visits. Third, the medical examiner's office will be asked to send weekly counts of deaths in which heat was a contributing factor during the heat season. Each source has delays and limits, but together they give a fuller picture than any one feed.

Component Four: Decision Support Instead of More Charts

The fourth component rebuilds the dashboard around decisions, following the redesign in Module 4. Rather than eleven charts, the internal view will open with a single status indicator showing whether heat-related visits are above the expected level for the date and weather, calculated from at least three prior seasons of data. An alert will appear only when visits exceed that level for two consecutive days, reducing the false alarms that led staff to ignore the tool. In a study of 1,617 counties, Vaidyanathan et al. (2019) found hospital admissions linked to heat climbing well before the most extreme days, and argued that warning systems work better when tuned to local data, so the county will also analyze its own multiyear data to learn at what heat index its visits begin to rise and share that threshold with the local weather office and emergency management. For residents, the public page will open with one short paragraph on last week's heat visits, then the nearest cooling centers, and it will be built for phones first.

Governance, Privacy and Cost

The plan requires changes to data agreements. The existing state-county agreement will be amended to allow the broader query, the addition of emergency medical services data and access to a limited set of fields for validation reviews. All new data will be stored in the county's secure analytic environment, accessed only by named staff, and reported publicly only in aggregate with counts under five suppressed. The cross-border agreement will require both states' legal review and may take more than a year.

Costs are modest because most components use existing systems. The largest expense is staff time: an estimated 0.4 full-time analyst position for the first year to build the quality report, validate definitions and develop the baseline model, falling to 0.2 thereafter, plus nurse reviewer time for validation. Software licenses for the dashboard are already paid. The county will seek a state public health infrastructure grant for the first-year analyst time.

Evaluation

Success will be judged with the same attributes used in Module 1, drawn from German et al. (2001), which lets the county set this season's results beside the baseline without translation. Sensitivity and predictive value positive will be re-estimated by chart review after the first improved season, with a target of identifying at least 80% of heat cases while keeping predictive value above 75%. Timeliness will be measured as the share of visits with diagnosis codes within 24 hours at each hospital, with a target of 85% at every facility. Data quality will be tracked through the automated report, representativeness through the share of heat cases among adults 65 and older and the number identified through ambulance data, and usefulness through the number of outreach actions triggered by dashboard alerts and staff ratings of the tool at the end of each season. Results will be reviewed with hospital partners and emergency management each October.

Conclusion

The county's heat surveillance is fast but incomplete, and the incompleteness falls hardest on older adults who are most likely to be harmed. This proposal addresses the problem at each point where the earlier modules found a weakness: the definition used to find cases, the quality of the data feeding it, the populations it cannot see and the way results reach decision-makers. None of the changes requires a new system. Together, they would give the county a surveillance program that finds more heat illness, signals clearly when action is needed and can show, with measured evidence, that it has improved.

References

German, R. R., Lee, L. M., Horan, J. M., Milstein, R. L., Pertowski, C. A., Waller, M. N., & Guidelines Working Group Centers for Disease Control and Prevention. (2001). Updated guidelines for evaluating public health surveillance systems: Recommendations from the Guidelines Working Group. MMWR Recommendations and Reports, 50(RR-13), 1-35.

Harduar Morano, L., & Watkins, S. (2017). Evaluation of diagnostic codes in morbidity and mortality data sources for heat-related illness surveillance. Public Health Reports, 132(3), 326-335. https://doi.org/10.1177/0033354917699826

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), 18. https://doi.org/10.13063/2327-9214.1244

Vaidyanathan, A., Saha, S., Vicedo-Cabrera, A. M., Gasparrini, A., Abdurehman, N., Jordan, R., Hawkins, M., Hess, J., & Elixhauser, A. (2019). Assessment of extreme heat and hospitalizations to inform early warning systems. Proceedings of the National Academy of Sciences, 116(12), 5420-5427. https://doi.org/10.1073/pnas.1806393116