NURS 732 Week 5 IT and Quality Improvement Paper Example

IT and Quality Improvement: What Our Registry, Alerts and Text Messages Did and Did Not Do for Blood Pressure Control in a Safety-Net Clinic

Introduction

When our community health center began working to raise its blood pressure control rate from 58%, I assumed technology would do much of the work. We had an electronic health record, a population health module the clinic had paid for but barely used, and a patient portal. Over nine months, our hypertension team used four technologies: a registry of patients with hypertension, a monthly dashboard, a decision support alert for high readings and an evening text message program that brought home readings into the chart. This paper reflects on what each did and did not do, using the improvement literature and Sollecito and Johnson's (2013) discussion of information systems in continuous quality improvement to interpret the experience. My conclusion, stated at the outset, is that technology magnified our team's work but never replaced it.

The Registry: Defining the Population

Sollecito and Johnson (2013) emphasize that improvement depends on data that define the population and measure performance over time. Our registry did that. Built from diagnosis codes and problem lists, it identified 1,412 adults with hypertension and pulled each patient's last recorded blood pressure. For the first time, the team could see the whole population rather than the patients who happened to be on today's schedule.

The registry also had flaws I had not anticipated. About 9% of patients on it had no visit in eighteen months, so their last reading was stale, and some patients with clearly elevated readings had never been diagnosed. Building the registry required a medical assistant and me to review the codes by hand for two weeks. It defined our population well enough to begin, but it was not as clean as the vendor's demonstration suggested.

The Dashboard: Feeding Data Back

The second tool was a monthly dashboard showing the control rate overall, by provider and by care team, with a run chart. The dashboard made progress visible, and the team looked forward to it. Evidence suggests feedback of this kind can support improvement: in an academic medical center's department-wide effort across 61 clinics, a common data infrastructure, a dashboard for local use and data feedback to clinical leaders accompanied a rise in repeat blood pressure measurement from a mean of 41% to 61% (Murphy et al., 2026). But I learned that the dashboard could also mislead. Our control rate rose in the first two months before we had changed anything, probably because the registry cleanup removed patients who no longer came to the clinic. The run chart, read carelessly, would have credited the team for a data correction.

The Decision Support Alert: Prompting Action

The third tool was an alert that fired when a blood pressure of 140/90 or higher was entered, prompting the medical assistant to repeat the reading and the provider to consider treatment intensification. I expected the alert to drive our results. It did change behavior, but not as much as I hoped, and not always in the way that mattered most.

The evidence helps explain why. In a post hoc analysis of a cluster randomized trial in urban primary care practices with 4,612 patients whose blood pressure was uncontrolled, practices using a real-time, guideline-based decision support system intensified treatment at 47.3% of visits with uncontrolled readings compared with 11.6% in usual care practices; each step up in treatment intensity was associated with lower systolic pressure, yet overall control rates were similar between the groups (Song et al., 2026). The alert changed what clinicians did, but that did not translate into control, perhaps because patients did not fill or take the added medications. That pattern matched what we saw: providers acted on more alerts, but patients' readings at the next visit improved less than the number of medication changes suggested.

Implementation also mattered. In a cluster randomized trial of a pediatric hypertension alert in 40 rural primary care clinics, elevated readings were repeated during the visit in 51.5% of cases at clinics with high-intensity implementation (face-to-face training, a check-in every month and performance feedback), compared with 23.6% at clinics that received only online training and 6.2% under usual care (Kharbanda et al., 2026). Our alert was turned on with an email announcement and a one-page guide. In the first month, medical assistants repeated only about a third of high readings when it fired. Repeats increased only after we added the alert to the daily huddle and the dashboard. An alert is a sentence the system says to a busy person; whether it is heard depends on everything around it.

Text Messages and Home Readings: Bringing the Patient's Data In

The fourth tool was an evening text from our community health worker asking patients on the cuff-lending program to reply with their average home reading, which staff entered into the record. It was the simplest technology we used and arguably the most useful. Home readings showed that a quarter of patients counted as uncontrolled by their last clinic reading were controlled at home, and they gave providers information between visits.

Patient-facing tools work best when they are designed around users. A team developing a patient-facing blood pressure decision support application translated feedback from 70 care team members and 18 patients into programming requirements, and the themes included timely, actionable alerts, messaging for follow-up, reduced data entry burden, workflow integration, and equity and accessibility (Canfield et al., 2026). Our text program, though far simpler, succeeded for the same reasons: it fit patients' evenings, required one reply and asked staff for one data entry. It also failed in ways the same themes predict. Patients who sent readings and heard nothing back stopped replying, so in the third month the community health worker began answering every reply with a short message, either thanks or a note that a nurse would call. Response rates recovered within weeks.

Data Quality: What the Measure Measured

The most important lesson concerned our measure itself. Our dashboard used each patient's last clinic reading, the standard approach in performance reporting. Chaves et al. (2026), studying one health system's long-running hypertension program, found that control rates based on a single last reading paralleled those based on averaged readings over time but were consistently lower, and that agreement between the two measures was only fair. Our home reading data suggested the same. Some of our apparent improvement may have come from better measurement technique rather than better blood pressure, and some patients who were truly controlled were counted as failures. Technology made the measure easy to calculate; it did not make it accurate. If I were designing the dashboard again, I would report two control rates side by side, one from the last clinic reading and one from the average of home readings where they exist, and I would show the share of patients with any home data. That would let the team see both its progress and the limits of its measure, and it would reward efforts to bring home data into the record rather than only efforts to lower a single clinic number.

Equity: Who Could Use the Tools

Technology also reached patients unevenly. Luna et al. (2026) found home values in the charts of only about one in four hypertensive patients across an urban safety-net network, with portal users ahead, unhoused patients behind, and close to a fifth of the variation explained by which clinic a patient attended. Our portal-based entry option reached mainly younger patients with smartphones and stable addresses. The text program and paper logs reached more people, but some of our highest-risk patients had no phone or changed numbers often.

Summary of the Tools

Lessons and Christian Worldview

Three lessons stand out. Technology works through people and processes, so implementation deserves as much planning as configuration. Measures must be examined, not trusted, because the easiest number to calculate is not always the truest. And technology can widen gaps unless the team designs for patients with the least access. Proverbs observes that whoever states a case first seems right until someone else comes and examines it (English Standard Version Bible, 2001/2016, Proverbs 18:17). A dashboard always speaks first in a meeting; a prudent leader lets the team examine it before anyone believes it.

Conclusion

Our registry, dashboard, alert and text messages each contributed to improving blood pressure control, but none did what I assumed it would on its own. The registry needed cleaning, the dashboard needed careful interpretation, the alert needed implementation support and the text program needed alternatives for patients without phones. The research I read for this paper explained most of the gaps between my expectations and our experience. In future improvement work, I will plan the people and processes around each tool as carefully as the tool itself, and I will question every measure before celebrating it.

References

Canfield, S. M., Rolbiecki, A. J., Ghosh, P., Montgomery, E., Shaffer, V. A., Bobo, M., Storer, M. B., Ghumman, A., Michaels, L., Dorr, D. A., Koopman, R. J., & COACH Consortium. (2026). Qualitative user-centered design: Programming changes in a self-management blood pressure application. Applied Clinical Informatics, 17(2), 127-137. https://doi.org/10.1055/a-2823-4533

Chaves, P. H. M., DeSoria, M., Roque, L., & Mas, R. J. (2026). Hypertension control rates before and after implementation of a multicomponent quality improvement program in Medicare beneficiaries primarily of Hispanic origin in a large health system in South Florida, 2011-2018. The Journal of Clinical Hypertension, 28(8), Article e70297. https://doi.org/10.1111/jch.70297

English Standard Version Bible. (2016). Crossway. (Original work published 2001)

Kharbanda, E. O., Asche, S. E., Essien, I., Allen, C. I., Freitag, L. A., Ekstrom, H. L., Kromrey, K. A., Muthineni, A., Saman, D. M., Thirumalai, V., O'Connor, P. J., & Benziger, C. P. (2026). Implementation of a clinical decision support tool for hypertension in a rural health system. Journal of the American Medical Informatics Association, 33(5), 1000-1008. https://doi.org/10.1093/jamia/ocag031

Luna, I., Karumanchi, R., Grewal, J. S., Zhang, N. S., Rodriguez, J. A., Pletcher, M. J., Sarkar, U., & Khoong, E. C. (2026). Factors associated with evidence of self-measured blood pressure adoption in an urban safety-net health system. The Journal of Clinical Hypertension, 28(6), Article e70318. https://doi.org/10.1111/jch.70318

Murphy, K., Hood, C. J., Dirisala, K., Voit, J., Vigen, R., Bowen, M. E., & Bajaj, P. (2026). Impact of a centralized quality improvement initiative on repeat blood pressure measurement at an academic medical center. The Joint Commission Journal on Quality and Patient Safety. Advance online publication. https://doi.org/10.1016/j.jcjq.2026.08.008

Sollecito, W. A., & Johnson, J. K. (Eds.). (2013). McLaughlin and Kaluzny's continuous quality improvement in health care (4th ed.). Jones & Bartlett Learning.

Song, J., Liu, Y., Shang, Q., Liu, J., Zhang, H., Zhou, Z., Wang, W., Lu, J., & Zheng, X. (2026). Clinical decision support system, antihypertensive treatment intensification, and blood pressure control: A post hoc secondary analysis of a cluster randomized trial. JAMA Network Open, 9(5), Article e2614858. https://doi.org/10.1001/jamanetworkopen.2026.14858