Introduction
1. | Suboptimal metabolic control (due to under- or overtreatment with carbohydrates) still occurs, associated with co-morbidity and long-term complications. |
2. | There is a gap of knowledge between clinical guidelines and management in daily practice. |
3. | There is large heterogeneity between individual patients with identical GSD subtypes and genotypes |
4. | There is a discrepancy between prescribed diets and actual used diets. |
5. | Clinical parameters are mostly measured in the hospital on relatively random moments. |
6. | Traditional biomarkers are suboptimal and biochemically distant from the primary metabolic block. |
7. | Patients with rare diseases usually do not live close to so-called centers of expertise, which challenges ‘shared care models’. |
Methods
-
Phase I — Prototyping and software design:The GCP was designed during monthly meetings with input from software developers, (parents of) GSD patients, researchers, and healthcare providers from June 2014 till present. The GCP was composed of two web applications; the GSD App for patients and their caregivers and the GSD clinical dashboard for healthcare providers. The GSD App was developed in Dutch, the native language for most of our patients, whereas the GSD clinical dashboard was developed in English. Figure 1 presents the detailed architecture of the GCP.
-
Intended uses — The intended use of the GSD App is to allow individuals with hepatic GSD to support their dietary management provided by healthcare providers, under normal circumstances and intercurrent illness, by monitoring and sharing home site collected data with healthcare providers. The intended use of the GSD clinical dashboard is to integrate data collected by hepatic GSD patients, either at home or during a hospital admission, and to provide subsequent GSD dietary management advice for normal circumstances, intercurrent illness, and those situations where the emergency protocol is applicable.
-
Data security and management — For both applications, the Hypertext Transfer Protocol Secure (HTTPS) and the Open Web Application Security Project Top Ten awareness document (https://www.owasp.org/) was adopted to secure integrity and privacy of online communication within the GCP. Access to the applications required a username and password. Passwords were stored with a salted and one-way encryption method to protect from accidental or unlawful loss. Software test plans and complaints management procedures were set up for post-market evaluation. The GCP was hosted on Microsoft Azure (refer to Azure subscription agreement: https://azure.microsoft.com/en-us/support/legal/subscription-agreement/). For the GSD App development, the open source frameworks AngularJS, JavaScript jQuery, and Bootstrap were chosen. A non-commercial license from Highcharts (https://www.highcharts.com) was used to display the graphs in the GSD clinical dashboard.
-
-
Phase II — Software development and retrospective clinical data entry:Features and functionality were reviewed and issues on usability were managed, processed, and documented with the use of a Jira issue tracker from Atlassian® by software developers, researchers, and healthcare providers.The following software functionalities were included:
-
Dietary registration and prescription module — The Dutch Food Composition Table (NEVO table) (National Institute for Public Health and the Environment 2012) was used for the development of the dietary registration module in the GSD App and the prescription module for dieticians in the GSD clinical dashboard, respectively. The NEVO table contains information on macro- and micronutrients content and total kilocalories of all food items frequently eaten by the Dutch population. The NEVO table also includes data on the medical formulas, dietary supplements, maltodextrin products, and uncooked cornstarch, such as Glycosade®.
-
Emergency protocol module — The local hospital emergency protocol guideline was used as a template for the emergency protocol module in the GSD clinical dashboard. The emergency protocol module was designed in such a way that it could automatically generate an emergency letter with the use of the patient’s GSD type and actual body weight. After generation, the emergency letter was shared with the corresponding patient in the GSD App.
-
Data import functions — In the GSD clinical dashboard, an import function for CGM data (Dexcom G4/G5 CGM system, Dexcom Inc., San Diego, CA) was created. An Application Programming Interface (API) was acquired by Fitbit, Inc. to import data from the activity wearable in the GSD App.
-
Retrospective data entry — Data from written food and clinical measurement diaries were retrospectively collected in the GCP by the researchers to test the usability and correctness of the GCP functionalities. These data were from patients who visited the University Medical Center Groningen GSD center of expertise between March and October 2016 and who gave written informed consent for the use of their data collected during their visit.
-
-
Phase III — Implementation and pilot study prospective clinical data entry:
-
Subjects and pilot study design — Between March and July 2017, data were prospectively collected in the GCP by selected GSD patients visiting our center. Patients were introduced to the GSD App by the treating physician (TGJD) and researcher (IJH). All subjects received an up-to-date manual for the use of the GSD App. Data exchange was requested by the healthcare provider to allow individual data integration in the GCP and critical follow-up after dietary changes. Subjects were asked to use the GSD App for home site monitoring before, during, and/or after a clinic visit, according to the purpose of their visit. Furthermore, subjects could temporarily use an activity wearable with a heart rate function (Fitbit Charge HR™). A CGM system by Dexcom was used only when needed for regular care. Data from the CGM were retrospectively imported in the GSD clinical dashboard by the healthcare providers. Results of data integration in the GSD clinical dashboard (i.e., updated dietary plans and emergency protocols) were discussed with the subjects/patients and their parents during the outpatient clinic visit and/or via a telephone or videoconference consultation following the (outpatient) clinic visit.
-
Survey methods — User feedback was documented in a structured logbook and usability issues from different users (patients, caregivers, and healthcare providers) were uploaded in the Jira issue tracker for further software improvements. Subjects and/or caregivers were asked to give feedback on the GSD App via an open feedback form on paper or electronically via a SurveyMonkey questionnaire and to fill in the system usability scale (SUS) (Brooke 1996). An adjective scale was used for the interpretation of individual SUS scores (Bangor et al 2009).
-
Results
Phase I: prototyping and software design
Phase II: software development and retrospective clinical data entry
Pt nr. | Agea | GSD type | Sex | Purpose | Logbook entryb | CGM import | Activity importc | Dietary advice | |
---|---|---|---|---|---|---|---|---|---|
Phase II: retrospective clinical data entry | |||||||||
R01 | 12 | Ia | F | Nocturnal Glycosade® versus CNGDF | ✓ | ✓ | ✓ | ✓ | |
R02 | 31 | Ia | M | Regular outpatient clinic visit for metabolic monitoring | ✓ | ✓ | ✓ | ✓ | |
R03 | 36 | IIIa | F | Monitoring introduction of MCT | ✓ | ✓ | ✓ | ✓ | |
R04 | 0 | IX | M | Metabolic monitoring during breastfeeding on demand |
–
| ✓ |
–
|
–
| |
R05 | 19 | Ia | M | Metabolic monitoring | ✓ | ✓ |
–
| ✓ | |
Phase III: pilot study prospective clinical data entry | SUS score (1–100) | ||||||||
P01 | 13 | Ia | F | dNocturnal Glycosade® versus CNGDF | ✓ | ✓ |
–
| ✓ | 38 |
P02 | 0 | IX | M | dRegular outpatient clinic visit for metabolic monitoring | ✓ | ✓ |
–
| ✓ |
no response
|
P03 | 31 | Ia | F | Introduction of Glycosade® as late night drink | ✓ | ✓ |
–
| ✓ | 55 |
P04 | 1 | IX | M | Regular outpatient clinic visit for metabolic monitoring | ✓ | ✓ |
–
| ✓ | 93 |
P05 | 50 | IX | M | Monitor of sport regimen | ✓ |
–
| ✓ | – | 63 |
P06 | 4 | IIIa | F | Regular outpatient clinic visit for metabolic monitoring | ✓ |
–
|
–
| ✓ | 75 |
P07 | 20 | Ia | M | dRegular outpatient clinic visit for metabolic monitoring | ✓ | ✓ |
–
| ✓ | 45 |
P08 | 5 | IXc | M | Regular outpatient clinic visit for metabolic monitoring | ✓ |
–
|
–
| ✓ | 80 |