Mobile App Development
The agile model is a widely known software development methodology used as a guiding framework for building and testing application prototypes. The model consists of four stages (conception, initiation and analysis, design and construction, testing and deployment) that describe the overall process of software development.
Conception Stage. Theoretical and empirical knowledge demonstrates that the self-management “process” is a major factor affecting adherence to chronic disease management, medical social resource utilization, and patient quality of life. Therefore, the core design, content and function of the Suyi App are key conceptual elements of self-management. Therefore, we formed a development team with doctors, nurses, engineers, and patients to create a list of tasks and procedures for the development of apps. The list was revised and grouped according to common CKD self-management goals in the clinical and research literature (Table 1), and operation systems such as IOS and Android wereselected for prototype development.
Initiation and Analysis Stage. ThinMed medical technology brings together a software development team (including engineers and designers) by four computer science practitioners, one algorithm engineer, one architect and one graphic designer developing the APP prototype. Continuously updated according to clinical needs and the next phase of application development and testing.
Design and Construction Stage. Based on the self-management process concept, the main objectives of the app design are 1) to simplify the patient‒physician communication process and 2) to assist patients in completing their home self-management more effectively. The design features of the app include the following: 1) scheduled follow-up appointments section; 2) task management push section; 3) practical knowledge push section; 4) automatic evaluation feedback section; 5) patient education live feature section; and 6) risk warning processing section. Figure 1 shows the screen diagram of the APP, and the outline of the process of the patient using the APP under the guidance of the health care provider.
Testing and Deployment Stage. A multidisciplinary CKD management team, development engineers and solution architects reviewed the app, asked questions, and then revised it again. Next, 10 patients (5 male and 5 female adults) were selected to complete the test using the app, following which they were asked for their opinions on improvements to the APP. Based on patient feedback, the font was enlarged, the sliding bar was changed to a numeric input keyboard, and occasional disconnections and flashbacks were resolved. Prior to the feasibility study, the APP was copyrighted (2020SR0835730). Additionally, approval from the institutional review board was obtained (2018-R006-01).
Table 1 Content of Suyi App
Study design
The study was a retrospective cohort study, the primary objective was to assess the effects of two follow-up methods (" APP + outpatient follow-up "and"traditional outpatient follow-up ") improves laboratory data associated with renal function, reduces all-cause mortality, reduces the occurrence of renal replacement(hemodialysis(HD) /peritoneal dialysis(PD) / renal transplant(RT)) or avoids the initiation of urgent dialysis(defined as use of temporary catheters when initial hemodialysis).
Setting
Patients enrolled in the CKD management center received long-term follow-up and management services from a dedicated nurse based on a case management model. The center was established in 2013, headed by one nurse director and one senior physician, and staffed with five full-time nurses. A kidney disease care clinic was in operation seven days a week (concurrently with a physician clinic), where all consultations were conducted by dedicated nurses. Case management included nutrition management, relevant diseases and complications management, medication management, symptom management, exercise management, and lifestyle management. The case management process consisted of assessment, planning, implementation, feedback, and evaluation. Patients were divided into two groups: the APP + outpatient follow-up group, where patients installed a mobile application and used it at least once a month, and had additional outpatient follow-ups at least once every three months; and the traditional outpatient follow-up group, where patients did not install the app and had outpatient follow-ups at least once every three months. All patients were followed up and managed according to the standard case management model (Fig. 1).
Participants
Patients who were enrolled in the CKD management center for long-term follow-up from January 2015 to December 2019 without receiving renal replacement therapy and with CKD stages 1–5 were included in this study. The inclusion criteria were as follows: 1) diagnosis of CKD stage 1 to 5 [GFR < 90 ml/(min 1.73 m2)], 2) age 18ཞ80 years old, and 3) no cognitive impairment, 4) with an outpatient follow-up at least once every 3 months. The exclusion criteria were as follows: 1) hemodialysis (HD), peritoneal dialysis (PD), or other treatment dialysis, renal transplantation (RT); 2) acute kidney injury; and 3) recent diagnosis of cancer,4) No outpatient follow-up was conducted for a period of more than 6 months. The collection of baseline data began in January 2020, and the study endpoint was December 2022. And refer to the flowchart below for specific screening process(Fig. 2).
Outcome measurement
Laboratory data. Laboratory data associated with renal function include: eGFR, Serum creatinine, Uric Acid, Calcium, Phosphorus, Kalium, Sodium (Na), Erythrocyte, Total Cholesterol, Triglyceride, Totol protein, Albumin, Parathormone, Hemoglobin.
All-cause mortality. Refers to the ratio of the number of deaths from all causes to the total number of patients in a certain period.
Incidence of entering renal replacement therapy. Refers to the ratio of the number of incidences of patients entering HD, PD, and RT to the total number of patients. The ratio of the number of incidences of patients entering HD, PD, and RT to those with CKD stage 4–5 at baseline was further analyzed.
The use rate of temporary dialysis catheter. The ratio between the number of patients using temporary catheters at the time of entry into HD and the total number of first-time HD patients. Reducing temporary dialysis catheter usage means avoiding the initiation of urgent dialysis.
Bias
In the retrospective analysis, Patients who failed to return for follow-up at the center, lost contact, or withdrew from follow-up were considered lost to follow-up (no longer receiving CKD outpatient follow-ups at least once every three months).Baseline data analysis was performed using inverse probability treatment weighting to ensure comparability of baseline characteristics.
Study size
Extracted through the in-hospital CKD management information system, eGFR < 90 ml/min/1.73 m2 was selected by engineers entering computer language, and a total of 6015 patients who met the inclusion exclusion criteria were extracted. 2683 people in the APP + outpatient group and 3332 people in the traditional outpatient group. The occurrence of renal endpoint events in both groups was collected by HIS system extraction with telephone follow-up, and laboratory test indexes were automatically extracted by the system.
Statistical methods
The inverse probabilistic treatment weighting of the propensity score was used to balance the baseline health measures recorded in the comparison groups, including known indications for APP use[11, 12]. The propensity score was estimated by multivariable logistic regression with 164 covariates chosen a priori. Patients in the reference group were weighted. This method produces a weighted pseudo sample of patients in the reference group with the same distribution of measured covariates as the exposure group[13]. Standardized differences between unweighted and weighted samples were used to compare differences in baseline characteristics between groups (differences > 10% were considered meaningful). Weighted risk ratios and 95% CIs were obtained by modified Poisson regression, and weighted risk differences and 95% CIs were obtained by a binomial regression model with an identity link function. Two-tailed P values less than.05 were interpreted as statistically significant. Because multiple comparisons can lead to type I errors, the results of secondary, subgroup, and sensitivity analyses should be interpreted as exploratory analyses. Analyses were conducted using SAS statistical software, version 9.4 (SAS Institute Inc).