Health Care Provider Knowledge and Perceptions of FDA-Approved and Do-It-Yourself Automated Insulin Delivery

Abstract

Background:

Automated insulin delivery (AID) technology may reduce variability in blood glucose, resulting in lower risk for hypoglycemia and associated complications, and by extension improve quality of life. While clinical trials, research, and patient experience have consistently demonstrated the value of AID, this technology is still inaccessible to many patients. Patient-driven innovation has resulted in alternative do-it-yourself (DIY) solutions to available off-the-shelf AID devices.

Method:

This two-phase cross-sectional observational study addressed health care provider (HCP) perceptions of AID as well as the perceived need for, development of, and evaluation of an AID fact sheet comparing the most commonly used Federal Drug Administration approved AID and DIY AID devices.

Results:

Negative attitudes toward the use of DIY AID were low. The majority of HCPs saw their lack of knowledge about how DIY AID work to be the greatest barrier to answering patient questions about what is available (74.4%). Additionally, the majority of HCPs (64.5%) indicated they were either “likely” or “very likely” to use the fact sheet when answering patient questions about AID options.

Conclusion:

Increased awareness and utilization of AID technology offer hope to further reduce the burden of diabetes, but there is a need to bridge the knowledge gap about DIY AID. A fact sheet provides a way to facilitate discussions of this emerging technology between HCPs and patients. Next steps could investigate additional ways to put needed information in the hands of HCPs.

Introduction

Type 1 diabetes is a burdensome and complex disease affecting 1.2 to 1.5 million children and adults in the United States.^1-3 Automated insulin delivery (AID) is one of the ways technology can be used to manage type 1 diabetes.^4-7 Automated insulin delivery combines insulin pump therapy, continuous glucose monitoring (CGM), and an algorithm to adjust insulin dosing automatically based on glucose levels.^4-7

While clinical trials, research, and experience have consistently demonstrated the value of AID,^4-7 this technology remains inaccessible to many.^8-13 Inaccessibility can be due to cost barriers, Federal Drug Administration (FDA) approval processes, and manufacturer proprietary designs.^8-13 Furthermore, AID devices are often created without end-user involvement, leading to engineered products that overlook the designs and features desired by patients, which can preclude them from using these devices.^11-13 In recent years, do-it-yourself (DIY) AID emerged through patient-driven innovation, offering an effective, affordable, and customizable solution for diabetes management.^14-16

The purpose of this study was threefold: (a) to assess the perceived need among health care providers (HCPs) for a comparison fact sheet of FDA-approved and DIY AID technology; (b) to develop an updated and relevant fact sheet of most commonly used FDA-approved and DIY AID technology; and (c) to assess the relevance of content and usefulness of fact sheet to HCPs. This study has the potential to reduce barriers to AID technology uptake by increasing HCP’s awareness and understanding of AID options.

Do-It-Yourself Automated Insulin Delivery

Do-it-yourself automated insulin delivery systems (eg, OpenAPS, AndroidAPS, Loop, and Omnipod Loop) are open-source software programs available through a coding process that can be downloaded or “built” as an application by the patient (using instructions readily available online). Once built, these applications can be deployed for use on one’s phone or minicomputer, with the end result of automating an insulin pump’s temporary basal rates (eg, Medtronic Minimed or Omnipod pumps). This automation of temporary basal rates is guided by glucose rise and fall. Consequently, DIY AID is only as effective as the accuracy of the CGM used. It is worth noting that if there is a failure in communication, the pump reverts back to its preset basal rates as an embedded safety feature. Do-it-yourself automated insulin delivery users have been able to utilize Dexcom G4, G5, G6, Medtronic, and Freestyle Libre sensors (as examples), which allow for greater user flexibility given various insurance and cost barriers.

Methods

The study was deemed Health Care Improvement by the University of Utah Institutional Review Board.

Design

This cross-sectional observational study was conducted in two phases and included the perceived need for, development of, and evaluation of an AID comparison tool as described in Table 1. In Phase 1, the main objective was to assess HCP technology experience, AID knowledgebase, comfort levels answering patient questions about AID technology, and desire for further education on the subject. In Phase 2, the primary objective was to evaluate the usefulness of a developed AID comparison fact sheet for use by HCPs with their patients.

Table 1.

Summary of Phase 1 and Phase 2 Methods.

	Phase 1	Phase 2
Objective	• Assess technology (insulin pump, CGM, FDA-approved AID, and DIY AID) experience • Assess AID knowledge, comfort levels, and need for additional education	• Assess past FDA-approved and DIY AID education • Evaluate usefulness and acceptability of an AID comparison fact sheet
Sample	• Cross-sectional sample of local HCPs (N = 43)	• Cross-sectional sample of national HCPs (N = 137)
Approach	• In-person, HCPs attending a state-chapter American Association of Diabetes Educators (AADE) conference	• Email to American Association of Diabetes Educators (AADE) membership • Email to Diabetes Advanced Network Access (DANA) membership • One in-person presentation, pediatric endocrinology clinic
Data Collection	• Paper survey	• Online survey delivered via RedCap

Abbreviations: AID, automated insulin delivery; CGM, continuous glucose monitoring; DIY, do-it-yourself; FDA, Federal Drug Administration; HCPs, health care providers.

Setting, Sample, and Recruitment

For both Phase 1 and Phase 2, eligible participants self-identified as HCPs interested in diabetes care and were able to read and write English. Sampling was purposive. Phase 1 participants (N = 43) included interdisciplinary HCPs who attended a state-chapter American Association of Diabetes Educators (AADE) conference in the Mountain West (United States) in November 2018. Contextual elements of this setting are worth noting; the environment where the paper survey was administered was a busy and crowded vendor booth during two separate hour-long conference session breaks, as well as at the end of the day. Phase 1 participants received a small candle as an incentive.

Phase 2 participants (N = 137) consisted of a national sample of HCPs recruited from the AADE and Diabetes Advanced Network Access (DANA) membership, and from a pediatric endocrinology specialty clinic in the Mountain West (United States). Participants recruited from AADE, DANA, and the endocrinology clinic completed an online survey assessing their diabetes technology experience and perceived usefulness of a comparison fact sheet (see Table 2). At their own request, participants from the pediatric endocrinology clinic received a 20-minute presentation explaining the background behind various AID options prior to survey completion. Phase 2 participants received no incentive for their participation.

Table 2.

Patients Cared for and Insulin Delivery Methods.

	N (%)
How Many People Have You Cared for in the Past YEAR That Use a Medtronic 670G AID?
None	25 (23.4)
1-9	39 (36.4)
10-24	20 (18.7)
25-49	17 (15.9)
>50	6 (5.6)
Unreported	3 (2.7)
How Many People Have You Cared for in the Past YEAR That Use an Open-source AID (eg, OpenAPS, Loop, and AndroidAPS)?
None	87 (80.6)
1-9	16 (14.8)
10-24	4 (3.7)
25-49	0 (0)
>50	1 (0.9)
Unreported	2 (1.8)
Likeliness or Comfort Utilizing Reviewed Fact Sheet With Patients
How Likely Are You to Use This Fact Sheet to Discuss AID Options with Your Patients BEFORE They Decide Which Device to Use?
Very unlikely	22 (20.6)
Unlikely	14 (13.1)
Undecided	23 (21.5)
Likely	32 (29.9)
Very likely	16 (15.0)
Unreported	3 (2.7)
How Likely Are You to Use This Fact Sheet to Help Answer Questions Patients May Have About AID Options?
Very unlikely	16 (14.8)
Unlikely	8 (7.4)
Undecided	14 (13.0)
Likely	45 (41.7)
Very likely	25 (23.1)
Unreported	2 (1.8)

Abbreviation: AID, automated insulin delivery.

Data Collection and Analysis

During Phase 1, a paper-based survey data were collected that took approximately two minutes to complete. Data from Phase 1 were manually input into the Statistical Package for the Social Sciences (SPSS) version 26 software and analyzed.¹⁷ During Phase 2, participants were given access to an online Research Electronic Data Capture (REDCap) survey link.¹⁸ REDCap, a web-based data management system, was also used to run statistics in Phase 2. Analysis for Phase 1 and 2 consisted of descriptive statistics such as frequencies, means, and percentages.

Results

Phase 1: Local Assessment of Artificial Pancreas Systems Knowledgebase

Health care provider participants in Phase 1 (N = 43) were predominantly Caucasian (95.3%), non-Hispanic (72.0%), and female (90.7%). Participants primarily had a bachelor’s degree (62.8%), practiced in a hospital or outpatient office (60.5%), and had less than 12 years of HCP experience (62.8%).

Participants reported varied experience managing patients who use insulin pumps, CGM, FDA-approved AID, and DIY AID. Participants reported higher comfort levels answering questions related to FDA-approved AID (34.9%) compared to DIY AID (11.6%). Barriers to answering patient questions about DIY AID were primarily related to a lack of understanding about how DIY AID work (74.4%). Some also reported barriers of fear of losing their job, license, or certification (7.0%), and a perception that DIY technology is not safe (2.3%). The majority of participants (74.3%) indicated that they were open to multiple forms of DIY AID education to enhance their knowledge, including online and in-person training, a brief comparison fact sheet, or combination of these approaches.

There were some missing data in Phase 1, especially for age (9.3%) and ethnicity (20.9%), which may be due to the structure and design of the paper survey. In addition, there was a percentage (18.6%) of unreported data for the final question asking, “What would enable you to better answer patient questions about various AID?” This is possibly related to the brief time available to complete free text questions in the context of a busy conference setting.

Phase 2: Online Review of Artificial Pancreas Systems Knowledgebase and Fact Sheet Usefulness

Phase 2 participants (N = 137) who reviewed and evaluated the fact sheet were mostly female (93.2%) and Caucasian (87.9%), with a median age of 48 years. Participants were interdisciplinary, but mostly consisted of nurses and nutritionists (91.0%) working in a hospital or specialty office (41.7%).

While the majority of feedback we received from Phase 1 indicated in-person training was the resource of choice for HCPs to learn more about DIY AID (30.2%), time constraints, feasibility, and lack of funding did not allow for this approach. The one exception to this was a 20-minute presentation delivered to hospital-based pediatric endocrinology HCPs (n = 15 of 137) in March 2019, prior to their completion of the survey. This presentation was delivered by request of the HCPs; therefore, these HCPs did acquire some advanced knowledge bias of AID options prior to completing the survey, when compared to other survey participants in Phase 2 who did not receive the presentation. This resulted in an inhomogeneous sample.

Given these limitations, an examination of FDA-approved and DIY AID literature was undertaken to develop an AID comparison fact sheet that included several factors: unique features of each AID (eg, compatibility and technology requirements) and cost (see Appendix 1). Although the fact sheet did not include all current variances of DIY AID (eg, Omnipod Loop and AndroidAPS), it did include examples of common systems in use at the time of data collection and analysis. The main purpose of the fact sheet was to offer a brief overview of what AID was available off-the-shelf through manufacturers (eg, Medtronic 670G) and compare it to DIY AID options as a whole.

Experience managing patients who use AID varied. While the majority of participants (n = 87, 81%) reported they had not cared for someone in the past year who used DIY APS, a similar number of participants (n = 82, 77%) reported they had cared for someone using Medtronic 670G (see Table 2 for additional details). Based on these findings, HCPs have far more experience working with patients using FDA-approved AID to manage their diabetes.

When asked, “How likely are you to use this fact sheet to help answer questions patients may have about AID options?” 64.8% indicated that they were either likely or very likely to use the fact sheet. However, when asked “How likely are you to use the fact sheet to discuss AID options with your patients before they decide which device to use,” only 44.9% indicated that they were either likely or very likely. These data suggest HCPs are willing to answer patient questions about DIY AID, but they are more hesitant to recommend it to their patients.

Discussion

To our knowledge, no previous literature exists regarding HCP perception of DIY AID or their response to an AID comparison tool, such as a fact sheet. The majority of HCPs lacked knowledge about DIY AID functionality and safety, resulting in barriers to answering patient questions about AID options. While the majority of participating HCPs in Phase 2 reported receiving education on AID systems sometime in the past (n = 70, 64.8%), it was not universal. We found HCPs are open to multiple forms of AID education and we found the AID comparison fact sheet to be acceptable and usable for electronic distribution, but not as desirable to HCPs as in-person training. Furthermore, we found that the majority of HCPs were not resistant to DIY AID, despite the lack of understanding as to how DIY AID works.

This study has strengths and limitations. The use of a national sample during Phase 2 is certainly a strength. Major knowledge gaps for HCPs were identified regarding both FDA-approved AID and DIY AID during Phase 1. This was indicated by HCPs reporting their experience with the technology as “No knowledge/Novice” for both FDA-approved and DIY AID (58.1% and 86.1%, respectively). This finding corresponds with Diffusion of Innovations theory which posits that the acceptance of new technology must undergo various stages of acceptance, associated uncertainty, and reformatted thinking.^19,20 As relates to AID, Diffusion of Innovation theory describes the bell curve process of AID uptake by innovators, patients, and HCPs over a time continuum.

We found that while the AID comparison fact sheet is usable, the feedback received from Phase 1 and the emerging field of AID suggest using different methods of electronic distribution and training to address continual updates and changes in technology (eg, the launch of DIY Omnipod Loop during our Phase 2 data collection in spring 2019). Due to our desire to keep the fact sheet concise, we chose to not include Accucheck, Dana, and AndroidAPS systems at the time of development.

Negative survey responses toward DIY AID were low, especially given the Diffusion of Innovation framework and the current “developer stage” of diffusion. For example, a relatively small percentage of participants in Phase 1 indicated their greatest barrier to answering patient questions about DIY AID technology was that they were either worried about losing their job, license, or certification (7.0%), or they felt that DIY technology was not safe (2.3%). Our data also suggest that HCPs are willing to answer patient questions about DIY AID, but they are more hesitant to recommend DIY AID to their patients.

Conclusion

Automated insulin delivery is an emerging technology that encompasses both FDA-approved and DIY options. With the increased development of DIY technology, patient input has become a more valuable resource in determining the development and acceptability of AID design. We discovered that most HCPs desire additional education concerning forms of AID, and future research should explore AID knowledge and acceptability among other HCPs such as primary care, emergency, and acute care. The knowledge gap between HCPs and their patients must be reduced. We found that electronic dissemination of updated and relevant AID educational materials to a national audience is one way to accomplish this.

Increased uptake of AID technology by HCPs and patients may ultimately result in reduced fear of hypoglycemia, reduced glucose variability, and improved quality of life. Specifically, the increased awareness and uptake of all forms of AID by HCPs may provide patients a needed sense of choice, ownership, and flexibility in managing their diabetes. The growing interoperability between devices may also result in significant cost savings for stakeholders and competition for manufacturers to provide more patient-centered technology.