In the first phase of this study [9], we performed a systematic review to identify and assess a set of clinical indicators that can be used for monitoring chronic diseases in primary care. In this study, we selected clinical indicators for five diseases that have high prevalence globally and are associated with multimorbidity and polypharmacy including type 2 diabetes mellitus, arterial hypertension, chronic heart failure, asthma, and osteoarthritis [10,11]. The review consisted of the appraisal of clinical guidelines and primary peer-reviewed studies of any design that were carried out mainly in primary care.

The second phase in this study was an adapted Delphi procedure conducted to evaluate the usability of the indicators in practice and to develop a monitoring tool by integrating the indicators in a user-friendly EMR layout. Figure 1 presents the three phases of the study and the associated research steps. The steps for the development of the tool included consensus from general practice experts, an evaluation by specialists, and a draft of a tool that included the relevant indicators. In the following sections, we present the methods in more detail.

Experts from the region of Zurich were contacted by email. We identified “well-renowned” expert physicians in the region of Zurich that had a special interest in chronic care and work experience in an outpatient setting. The experts were either well-known by some members of our team, who are clinically active physicians, or were identified by asking colleagues. All contacted experts agreed to participate. The study team included two experts in primary care and EMRs (CC and MZ) and an expert in systematic reviews (NM-G) as well as a medical student who was not considered as an expert (LF). For asthma and arterial hypertension, two experts in outpatient care with a special interest in chronic care, Dr. Claudia Steurer-Stey and Dr. Paolo Suter, participated in the study. For type 2 diabetes mellitus, chronic heart failure, and osteoarthritis, Dr. Henryk Zulewski, Dr. Tobias Höfflinghaus, and Dr. Lukas Wildi, PD participated, respectively. We contacted five additional experts in total, one for each chronic disease. Multimedia Appendix 1 shows a list of all experts and affiliations.

Experts in any of the five chronic diseases evaluated the set of indicators by means of an adapted Delphi procedure. The Delphi procedure is a structured communication method consisting of a panel of participants and experts on a certain topic [12]. Multiple rounds are conducted on a specific topic, and in every new round the decision of one expert is influenced by the anonymous decisions of the rest of the expert panel in the previous round. The Delphi procedure has been proven to be a feasible method in evaluating indicators for chronic diseases and for generating a consensus [13,14]. We use the term adapted Delphi procedure, because the evaluation rounds were not anonymous and consisted of only a few participants. In total, we performed three rounds.

The first round consisted of face-to-face meeting sessions with our study group of four, including two experts in primary care and EMRs, one expert in systematic reviews, and a doctoral student (Multimedia Appendix 1). Each indicator was categorized into one of four types of data elements: 1) to be part of the monitoring at least annually, 2) data normally included in EMRs, 3) data not to be collected at a regular basis, and 4) data that should not be collected at all. Table 1 exemplifies this approach for diabetes mellitus. Before the second discussion round, we excluded data the experts indicated should not be collected.

In a second round, the study group and one consulting expert of each specialty discussed the condensed set of indicators resulting from the first round (Multimedia Appendix 1). The discussion led to an even more condensed set of indicators, which the study group further re-evaluated in a third round to focus on feasibility and exclusion of redundant indicators.

Based on a table of the condensed set of indicators, we developed a framework table for each condition, including 1 to 4 sublayers, to provide a structure for the indicators. We introduced the different layers to optimize usability and improve the overview. Layer one is the first visible layer when the tool opens. Each indicator represents itself or its own category of sub-indicators. When an indicator is selected from the first layer, the subsequent layer becomes visible. We designed the monitoring tool in a layout format that enables its integration into an EMR.

The adapted Delphi procedure resulted in a thorough set of indicators, since only relevant and practical (ie, useful and operable) indicators were selected. The systematic review provided 1162 indicators for the five chronic conditions; however, only 25.47% (296/1162) were considered by the experts as being of high enough relevance and feasibility to be implemented in the monitoring tool, including 20.48% (51/249) of diabetes mellitus indicators (ie, 12 in additional tab, 4 in cockpit, 11 in first layer, 21 in second layer, and 3 in third layer), 26.78% (49/183) of asthma indicators (ie, 12 in additional tab, 3 in cockpit, 7 in first layer, 11 in second layer, and 16 in third layer), 14.63% (49/335) of arterial hypertension indicators (ie, 12 in additional tab, 4 in cockpit, 3 in first layer, 23 in second layer, and 7 in third layer), 33.33% (77/231) of chronic heart failure indicators (ie, 12 in additional tab, 6 in cockpit, 6 in first layer, 40 in second layer, and 13 in third layer), and 42.68% (70/164) of osteoarthritis indicators (ie, 12 in additional tab, 7 in cockpit, 7 in first layer, 17 in second layer, 20 in third layer, and 7 in fourth layer).

The face-to-face discussion sessions about the eligibility and relevance of the indicators resulted in a condensed set of relevant and practical indicators as part of an EMR monitoring tool. During the Delphi procedure, which was primarily meant to discuss the relevance and feasibility of the indicators, an additional subject that was determined to be important was the ergonomics of how the indicators should be displayed. Ergonomics within the tool were uniformly identified by all involved specialists as an essential element to achieve acceptance of a new monitoring tool. For each selected disease, the indicators were categorized and linked into sublayers, which can be accessed depending on the requirements of the user. The first layer gives an overview of the most important indicators or categories. Clicking on each layer opens a set of further indicators. For a clear design and ergonomic use, we did not exceed four layers. The tool contains an additional tab with nondisease-specific information such as allergies, smoking, or drinking habits. This tab can be accessed at any point during tool use. The tool is completed by a “cockpit”, which serves as a guidance in the process of disease management. The cockpit includes individually predefined treatment goals, and thus enables benchmarking in the monitoring process. In addition, the monitoring interval can also be documented. To guarantee individual adjustments, blank spaces described as “free text” are added in every layer. Figures 2-6 show the final design of the suggested monitoring tool that evolved during the Delphi procedure. Figure 7 shows an example of how some of the indicators of type 2 diabetes mellitus could be integrated into the EMR.

A strength of this project is the iterative Delphi process that identified the importance of the ergonomic layout of the monitoring tool. Consideration of ergonomics can enhance user-friendliness and facilitate chronic care within an EMR. This tool offers a practical approach for implementing scientific results into everyday practice. By involving generalists with extensive practice experience as well as specialists in different medical fields through an adapted Delphi procedure, a condensed set of indicators were identified as relevant for everyday use in primary care. A potential limitation of our study is that the adapted Delphi procedure did not meet all criteria of a typical Delphi procedure. Since the discussions were not performed anonymously and the persons involved might have selected the indicators they were most familiar with, this may have increased the chances of bias. Furthermore, the adapted Delphi procedure included in total only ten experts and, in certain phases, only four experts (three experts from the study group including two in primary care and EMRs and one in systematic reviews, and an additional expert in the respective field), which might reduce external validity. Additionally, the complexity of the technical integration of indicators (beyond the visual layout on the screen) is not addressed in this manuscript. Details of EMR integration will vary between the different software companies and the needs of different health care systems.

As mentioned in Figure 1, our final goal is an international consensus paper. This will be achieved by an international Delphi procedure with a larger number of experts in general practice and other specialties. To meet the demands of monitoring the rising number of multimorbid patients, the method we present here is meant to be extended to more diseases that will also be linked to each other. In the future, we hope the tool will become an integral part of the process of collecting patient data, as well as a clinical decision support system that will directly link current guidelines and algorithms with therapy suggestions.

Our generic template systematically integrates the existing scientific evidence for the standardized long-term monitoring of chronic conditions. It contains a user-friendly and clinically sensible layout and can improve the care of patients with chronic diseases by means of an EMR in primary care.