Redesigning telemedicine: preliminary findings from an innovative assisted telemedicine healthcare model | BMC Primary Care

Intervention design

The design and development of the programme were carried out in phases. The formative phase included informal interviews with various stakeholders, including patients, physicians, and policy makers, to identify barriers to access to and quality of telemedicine care. This was followed by the development phase in which a telemedicine platform and training modules for health workers were created to help address the barriers identified. The deployment phase included identifying and training health workers, testing the technology platform, establishing telemedicine clinics, dry running the clinic workflow, refining standard operating procedures, monitoring, quality assurance and data analysis.

The ‘Digisahayam’ telemedicine technology platform was designed to be inter-operable and integrates task-shifting enabling mechanisms, electronic health records (EHR), point-of-care diagnostics, electronic clinical decision support systems (eCDSS) [17], and state-of-the-art health technologies [18, 19]. Technologies with Application Programming Interfaces (APIs), such as a digital stethoscope for remote auscultation and a physician-controlled camera (Table S1), were identified and integrated into the platform. All data collected on the platform was stored securely on an Amazon Web Services server.

The telemedicine system and implementation plan were designed to address some of the barriers identified in the formative phase, such as digital literacy, technology and language barriers among patients; lack of insight into the clinical history, physical examination and laboratory findings for physicians; and poor interoperability of digital technologies and poor integration of telemedicine models in the care pathway and clinic workflow.

The eCDSS has been developed and tested extensively by the Centre for Chronic Disease Control and All India Institute of Medical Sciences, New Delhi [17, 18]. The eCDSS software uses thousands of embedded case studies and evidence-based guidelines to generate clinical decision recommendations. The recommendations include optimal drugs and dosages, follow-up plans, and personalised lifestyle advice for diabetes mellitus and hypertension. The eCDSS was incorporated into the telemedicine platform after formative interviews with physicians that identified barriers and facilitators to its use [20]. Efforts were made to enhance its ease of use and interoperability by ensuring appropriate task-shifting in patient data collection via integrations in the nurse’s history-taking template, easy accessibility of clinical recommendations on the physician-facing interface and tailoring the drug recommendations according to local availability.

Telemedicine nurses were trained to collect patient history using a customised symptom-based structured template, perform physical examinations, and conduct lab investigations before initiating tele-consultations whenever necessary. All patients were screened for medical or surgical emergencies using a triaging template and referred to a tertiary care centre if there were symptoms or signs of a medical or surgical emergency. The telemedicine nurse was trained to connect the patient to the physician, convey the findings, facilitate physician-patient interactions, and prevent unnecessary visits. A laboratory technician was appointed to perform the relevant point of care testing of 13 common laboratory investigations and electrocardiograms (ECGs). In addition to the detailed history and examination findings provided by the trained nurses, the data from the integrated devices provided the physician with information on vital signs, basic laboratory results, ECG with artificial intelligence-based interpretation and the ability to auscultate the patient remotely. Consultations were first done by a primary care physician and, if further referred, by a specialist or sub-specialist physician.

The telemedicine nurses were also trained to make reminder calls and send messages to patients with diabetes mellitus and hypertension due for follow-up visits. The telemedicine platform and connected devices were both 4G enabled and battery-powered, allowing portability. Therefore, in addition to the assisted telemedicine consultations facilitated at the clinics, doorstep consultations were carried out at the homes of elderly and bedridden individuals living in the areas surrounding the clinic. The clinic processes were monitored and supervised by an experienced senior nurse designated as a quality assurance officer.

A comprehensive community program, including outreach telemedicine clinics and health promotion campaigns in the areas surrounding the clinics, was also instituted fortnightly. Regular household visits were also done to educate the community on healthy lifestyle practices. Every month was dedicated to raising awareness of a specific disease entity.

Study areas

Three telemedicine centres were established to provide free healthcare to disadvantaged communities in the Kodambakkam and Nanganalloor areas of the Chennai district, as well as Pasuvanthanai village of Thoothukudi district in the state of Tamil Nadu, India. The sites were selected to demonstrate the feasibility and functioning of this model in both urban (Chennai clinics) and rural (Pasuvanthanai clinic) settings. The clinics functioned between 8am to 5pm on all days of the week, except on Mondays and public holidays. The laboratory service was available from 8am onwards and consultations were available from 9am onwards. The clinics were kept open on Sundays to allow working individuals to avail services.

Study population

De-identified data from the records of all patients who visited the telemedicine centres over a span of 28 months (March 2021 to June 2023) was extracted from the telemedicine platform. Written informed consent for assisted telemedicine consultations was taken from all patients. To study the feasibility of implementation, the utilisation of services was analysed using the health records of all patients and device data extracted from the telemedicine platform. The impact of the initiative on chronic disease care was analysed using health records data of all patients above the age of 18 years with diabetes mellitus or hypertension. The Centre for Chronic Disease Control Institutional Ethics Committee (IRB00006330) gave ethics approval for the analysis.

Data collection and analysis

Consultations done by primary care physicians were labelled as “general” consultations, and those by a physician with a specialized degree were labelled as “specialist” consultations. Previous history of one or more “chronic conditions”, namely, diabetes mellitus, hypertension, stroke, vascular diseases, coronary artery disease, chronic obstructive pulmonary disease, asthma, liver failure and chronic kidney disease, were collected from all patients during their first visit. The demographic profile of the patients who attended the clinics, utilisation of laboratory services, and follow-up visit rates were analysed.

Sub-group analysis of outcomes in patients with diabetes mellitus and hypertension

Patients were considered to have hypertension if they either had a known history of hypertension and were on treatment or if their Systolic Blood Pressure (SBP) was above 140 mmHg and/or Diastolic Blood Pressure (DBP) was above 90 mmHg during the visit. Patients with a known history of diabetes mellitus or Fasting Blood Glucose (FBG) value of > 126 mg/dL or random/post-prandial blood glucose value > 200 mg/dL were considered to have diabetes mellitus. Those with the precursory diagnostic cut-offs but without a known diagnosis of hypertension or diabetes mellitus were labelled as “newly detected hypertension” and “newly diagnosed diabetes mellitus”, respectively, after opportunistic screening. The patients detected with diabetes or high blood pressure were treated as per the national guidelines in the National Programme for Non-Communicable Disease (NP-NCD). The eCDSS embedded in the telemedicine platform also follow the same guidelines. The guidelines take into consideration individuals who suffer from both diabetes and hypertension, and recommends simple yet effective drugs accordingly. The data on blood pressure readings in patients with hypertension and FBG in patients with diabetes mellitus were cleaned before analysis to drop outliers. The patients aged < 18 years and those with SBP < 40mmHg and DBP < 35mmHg were excluded. The duration between the baseline and subsequent visits was calculated in months and days. The follow-up duration was categorised as baseline visit (first visit), < 1-month visit, ≥ 1-month and < 2-month visit, and so on. If a patient had multiple visits in the same time duration, the mean value of SBP, DBP, and FBG for that patient during the specified time period was included. The follow-up duration was limited to ≥ 8 months and < 9 months, as beyond these time durations, there were not enough observations to make meaningful interpretations.

A descriptive analysis was done to determine the number of visits each patient had and the number of patients in different follow-up durations. Mean (Standard Deviation (SD)) and median (Inter-Quartile Range (IQR)) values of SBP, DBP and FBG were calculated for follow-up duration. Box plots were used to show the data distribution of SBP, DBP and FBG across follow-up visits. Generalized Estimating Equation (GEE) analysis accounted for the correlation between different time points. The model included outcome variables (SBP/DBP/FBG) and time as the covariate. The analyses were conducted using the Stata (17.0) version, and plotting was conducted using PYTHON 3.10.