The objective of this study is to develop and implement a clinical decision support (CDS) tool for identifying patients with an underlying PI. The Jeffrey Modell Foundation has published a set of 10 warning signs for PIs for both adults and children [27]. We will adapt these warning signs to a CDS tool to help stratify the risk of a patient having an underlying PI. This CDS tool will enable primary care providers within the University of Iowa to refer patients to the University of Iowa’s Immunology Clinic, and guide laboratory investigations prior to referral to ensure readiness prior to the visit.

Toward that end, there are two specific aims:

CDS tools for PIs have not yet been developed and assessed in peer-reviewed literature. This intervention would represent the first attempt to address this critical need. However, we build upon several lines of evidence that bolster the rationale that a CDS tool is needed to identify PI patients:

University of Iowa Hospitals and Clinics (UIHC) represents the only comprehensive academic medical center within the state of Iowa. Between July 1, 2019, and June 30, 2020, there were 1,039,681 clinic visits, 32,872 inpatient admissions, and 50,468 emergency department visits. In the UIHC system, there are 77,779 patients that have a primary care provider in Family Medicine, General Internal Medicine, and General Pediatrics.

Patients with suspected immunodeficiencies are evaluated at the Allergy and Immunology Clinic located at the University of Iowa’s clinics in Iowa City. If patients are determined to have an immunodeficiency, then they return to the Allergy and Immunology Clinic to obtain specialized care. Per year, there are 1863 half-day clinics staffed by 12 physicians serving 8963 patients. There are approximately 1684 patients with a diagnosis of a PI whose management is overseen by physicians and other practitioners in the Immunology Clinic. The average wait time to a new appointment is 5.46 weeks for adults and 13.16 weeks for pediatric patients.

Inclusion criteria include patients who receive their care from one of the General Internal Medicine, Pediatrics, or Family Medicine clinics through the UIHC system. Exclusion criteria include patients with an established secondary or acquired immunodeficiency, and patients with an established PI diagnosis.

We will use Lean Six Sigma principles and the Define, Measure, Analyze, Improve, and Control (DMAIC) framework to implement this intervention. The Lean Six Sigma approach has been previously described in health services research literature as a methodology to improve health care delivery [31]. It combines two distinct process improvement methodologies: Lean and Six Sigma. The Toyota Motor Corporation initially developed Lean to improve manufacturing quality through the elimination of waste and creation of value for customers [32]. Lean complements Six Sigma, a data-driven management approach aimed at the elimination of defects and reduction of unwarranted variability [33].

In line with the DMAIC framework (Figure 1), we have begun defining the scope of the project. We have developed a project charter to articulate the problem at hand, our project goals, and customer requirements. The project charter also delineates the time frame for completion of tasks (Figure 2).

Additionally, to understand the needs of the customers, we are using Voice of the Customer to clarify experiences, expectations, and needs. We are currently conducting semistructured interviews with 20 patients with PIs to understand their experiences. EK, a core member of the team, is a patient with PI and is helping to formulate questions and analyze data.

During the Measure phase, we anticipate collection of data regarding our primary and secondary outcomes. The primary measure, as articulated in the project charter, is the number of newly diagnosed PI patients per month. We will measure the rate over the past 24 months to obtain a baseline, and then monitor it throughout the course of the protocol.

We will also measure the following five secondary measures:

Administrative data will be used for the primary measure and secondary measures 1-4. For secondary measure 5, we will conduct semistructured interviews and surveys to obtain the Voice of the Customer.

Throughout the analyze phase, we will use the data being obtained as part of the Measure phase to assess the efficacy of the intervention and to guide modifications as necessary.

For quantitative data, we will construct both run and control charts. The specific type of chart will be dependent on the number of evaluated patients and specific outcome measures. For example, if the CDS tool is unable to find many patients with PI, then a T-chart may be more appropriate than an X-bar chart.

For qualitative data, we will continue to use Voice of the Customer. Once the CDS tool is implemented, we will conduct semistructured interviews with primary care providers to better understand their perspectives and opportunities for further refinement so that the CDS tool is more user-friendly.

For individual cases of newly diagnosed PI, we will perform more detailed analysis to recognize causes of diagnostic delays. We will use value-added analysis to calculate the costs of diagnosis and identify causes of inefficiency in diagnosis. Additionally, we will use time analysis to chronologically map out opportunities for earlier diagnosis. Both of these analyses will then be used to further refine the CDS tool.

During the Improve phase, we will employ sequential Plan-Do-Study-Act (PDSA) cycles to refine the CDS tool to different settings, starting with Internal Medicine, then Pediatrics, and finally Family Medicine. The investigators will continue to use Voice of the Customer as a key Lean Six Sigma tool to guide these PDSA cycles.

Additionally, as we begin to finalize the CDS tool, we will use failure mode and effect analysis to proactively determine how the newly established process for identifying PIs may create inefficiencies and result in failures in diagnosis.

Toward the end of the quality improvement project, we will develop a monitoring and proactive response plan to sustain the improvement plan. Using those measures, we will establish a trigger level to ensure that new cases of suspected PIs are screened through the CDS tool. Additionally, with advances in the diagnosis of PIs, such as genetic testing, we will incorporate newer modes of evaluation into the CDS tool.

Finally, we intend on publishing our findings and presenting them at national meetings. Thorough documentation of the process, including this protocol, will be essential for transfer of knowledge to other settings and implementation in different settings.

This study protocol was submitted to the University of Iowa Institutional Review Board and determined not to be Human Subjects Research.