Development of the Houston–Apollo model for older people living in remote areas in Taiwan

Senior healthcare is challenging in remote areas, particularly in an economically disadvantaged population. This study examined the benefits of a combined healthcare system (Houston–Apollo model) in improvements of physical performance and medical care utilization of local older people.


Introduction
Taiwan has officially become an aged society since 2018, as over 14% of its population is ≥65 years old. Yunlin County is the second oldest county in Taiwan. Despite having more aged people in this agricultural county, seeking health services is a daunting task for them. Insufficient health and transportation services in rural areas have created barriers for these old people to seek medical help. [1][2][3] Although the National Health Insurance of Taiwan has lowered the financial burden of medical cost, the geographically uneven distribution of medical facilities caused inequality of medical accessibility. To bring down the barriers, telemedicine could be an effective approach, 4 given that broadband access had reached 96.7% coverage in rural and remote areas of Taiwan by 2017. 5 To meet the nutritional needs of older people and offer opportunities for social engagement, the government in Taiwan has provided a social welfare system in setting congregate meal services (CMS) in every township. Local organizers offered meals for people ≥65 years. Half of the cost (30 NTD/meal, which equals about 1 US dollar) was paid by the diners, and another half was paid by the government. With the program, older people were motivated to go out from their homes and gathered to improve interpersonal relationships and socialization. There were approximately 120 CMS in Yunlin County. Each would serve 30-150 (with an average of 50) older people, and about 10 000 older people received this service regularly.
Participation in CMS can enhance nutritional status, social interaction and functionality of older adults. 6 Before this project, the CMS did not view health improvement as a major goal. An evidence-based model of healthcare in such community services was also lacking. The purpose of this project is to describe a geriatric health service model to integrate CMS, local general physicians and hospitals.

Houston-Apollo model
To establish this integration, since November 2018, the National Taiwan University Hospital Yunlin Branch (NTUHYL) has initiated the Houston-Apollo model (HAM). HAM selected the wellknown Apollo space project of the United States as the blueprint. When astronauts are in trouble, the base at Houston will provide full support for the astronauts to complete the mission. In HAM, NTUHYL worked as Houston to guide and support general physicians (as the crew of the Apollo spacecraft). The latter provided free health consultation in remote villages via telemedicine. If there were difficulties in dealing with the health problems of the old people, the local general physicians could call for help from NTUHYL. We collaborated as a team to deliver healthcare services for the rural old population.
To build the collaboration, NTUHYL first invited local general physicians practicing in rural communities to participate in HAM. They were eligible to join the project if they were willing to spend 60 min a day, 5 days a week, using video-conferencing to provide free health consultations to old people in the CMS. NTUHYL then provided video-conferencing devices for every primary care setting and CMS to deliver telemedicine consultation services and health education. Each doctor served 1-2 CMS at noon, when the old people would gather and wait for their lunch. A device, nicknamed Baby Machine, was also set up in each CMS to measure and monitor the vital signs and physical parameters of older people.
The older people, general physicians and NTUHYL formed a bidirectional triangle (Fig. 1). The older people received consultations from general physician if they had: (i) sickness or medical problems, (ii) abnormal vital signs measured by the Baby Machine, and (iii) potential of polypharmacy misuse. In response to the consultation, local doctors had the following strategies: (i) providing advice directly, (ii) asking older people to visit their clinic if medication is needed, (iii) consulting doctors in NTUHYL if medical specialty needed, and (iv) arranging transferal to NTUHYL. The interaction between local doctors and older people was focused on education (such as hypertension monitoring and exercise) rather than formal diagnostic behavior, and older people were asked to visit doctors directly once any illness needed treatment.  In addition, the physical status of the older people was transferred to the data center in NTUHYL in a timely manner, which would then generate intervention strategies, for example, cooking guidance to service meal providers and physical training via real-time video. A 1-year community-based study comparing preintervention and post-intervention was performed to examine the effectiveness of HAM.

Subjects
We recruited people aged ≥65 years, who received a regular meal service in 12 CMS. People who had registered with the CMS but received only home meal service due to ambulation disability or were working were excluded. The participants were offered both written and verbal explanations of the study before we obtained formal consents from them. They were also assured the confidentiality of their information.

Measures and data collection
Demographic data were collected at the beginning of the intervention, including age, sex and underlying chronic diseases. The vital signs (height, weight and blood pressure [BP]) and physical parameters (grip strength, five times sit-to-stand test [FTSST] and 6-m walking speeds) were collected monthly under the direction of staff members in the service. Only subjects who had Baby Machine-measured data at the beginning of the study and 1 year later were used for longitudinal comparison (Fig. 2). Furthermore, these data were also compared with previous studies in one rural county (Hualian) and urban county (Taichung) of Taiwan, 7 and another urban area in Asia region (Hong Kong). 8 Hypertension was determined according to the 2018 guideline of the American Heart Association. 9 Hypertension stage 1 was defined by systolic BP ≥130 mmHg or diastolic BP ≥80 mmHg, and stage 2 was systolic BP ≥140 mmHg or diastolic BP ≥90 mmHg. Body mass index (BMI) was determined according to the World Health Organization. 10 BMI <18.5 was defined as underweight. BMI 18.5-24.9 was defined as normal weight. BMI  7 For FTSST, the score was assigned from 1 (the slowest) to 4 (the fastest). The length of time required for the measure was used for scoring, with a score of 1 = ≥16.7 s, 2 = 13.7-16.6 s; 3 = 11.2-13.6 s and 4 = ≤11.1 s. 11 The staff at each CMS helped older people to use the telemedicine devices correctly and guide them to perform various physical activities for measuring their physical function. Medical utilization was also calculated to evaluate change in medical accessibility and equality. The medical resource access toward the NTUHYL in 2018 and 2019 of the 470 participants were retrospectively retrieved via electronic medical record system. The access records were classified into four categories: dental outpatient visiting, outpatient visiting to other departments, hospitalization and emergency department visiting.

Statistical analysis
Demographic data were summed by descriptive analysis. Continuous variables were expressed with mean ± SD. Categorical variables were expressed with numbers and percentage (n, %). In comparison with previous studies, the t-test was used to compare specific summary statistics. 7,8 McNemar's test was used to test whether or not disease status was consistent across the two sampling times in grip strength and walking speed. The chi-squared test was used to compare changes in hypertension and FTSST due to multiple groups of disease status. The medical resource utilization was evaluated using the access frequency of dental outpatients, outpatients for other departments, hospitalization and emergency department.

Results
There were eight primary care doctors and 727 older people who had registered in the 12 CMS (Fig. 2). Of these, 244 people did not actually have lunch in the CMS and received home meal services due to ambulation disability or work. Of the rest of the 483 people, 470 agreed to join the project under inform consent. These included 187 men and 283 women ( Table 1). The mean ± SD age was 75.55 ± 6.5 years. The parameters were measured at the beginning of the project. According to American Heart Association guidelines, 65.95% of participants had hypertension, and 98 (24.75%) belonged to hypertension stage 1 and 192 (48.48%) to stage 2. Fourteen participants (3.84%) were underweight, 107 (29.32%) were overweight and 22 (6.03%) were obese. Two hundred and nine (50.36%) participants had a weak grip strength. In FTSST, 71 (16.82%) had the worst performance (score = 1); 82 (19.43%) had score = 2; 92 (21.8%) had score = 3; and the rest 177 (41.94%) had the best performance (score = 4). Disability in walking speed was seen in 307 (77.92%) participants.
Of the 97 participants who had repeated hypertension evaluation 1 year after the beginning of the project (Table 2), the mean ± SD systolic and diastolic BP (mmHg) decreased from 137.4 ± 20.1 to 133.3 ± 19.6 (P = 0.019) and from 76.9 ± 12.9 to 74.4 ± 12.5 (P = 0.008), respectively. However, the numbers of diagnoses of hypertension had no change. The mean ± SD grip strength (kg) of 88 older people was 22.3 ± 8.42 and 23.0 ± 7.39 before and after the intervention (P = 0.107), respectively. Of 96 people who received repeated FTSST, the mean ± SD exercise time (s) decreased from 11.3 ± 3.92 to 10.4 ± 3.08 (P = 0.011). There was no significant change in score groups (P = 0.125). The mean ± SD walking speed (m/s) of 81 people who received repeated evaluation increased from 0.71 ± 0.25 to 0.74 ± 0.23 (P = 0.039), but there was also no significant change in numbers of disability (37 to 31, P = 0.109). In comparison with the retention group and drop-out group, there were no significant differences in age, sex, BP, BMI, grip strength and walking speed, but the FTSST was significantly longer in the drop-out group (Table S2). The evaluation results on the medical resource utilization before and after the project launched were summarized in Table 3. Both the outpatient visits to dental and other departments were increased significantly. The number of visits to dental clinics increased from 16 to 32, which corresponded to a two times increase (95% CI = 1.1-3.64). The visits to non-dental clinics also increased from 1193 to 1362, which corresponded to the 1.14 times increase (95% CI = 1.06-1.23). In contrast, although use of the emergency room declined in 2019, the rate ratio did not achieve a statistical significance (0.8 [0.56-1.15] for a decrease of 19.7%). For hospitalization, the use pattern is relatively stable: reducing slightly and non-significantly merely in amounts of use (for a 3% reduction). Four cases had newly diagnosed cancer (acute myeloblastic leukemia, malignant neoplasm of bladder, malignant neoplasm of sigmoid colon, and malignant neoplasm of cecum, respectively) found by the referral system of HAM.

Discussion
Our results detected a high prevalence of previously undiagnosed hypertension, obesity and poor physical performance among the participants of CMS, suggesting lack of health awareness. The grip strength, FTSST and walking speed were all worse than the populations of other urban (Taichung) and rural (Hualien) regions in Taiwan. 7 Despite previous studies that suggested physical parameters of the elderly population in rural regions were usually better due to participation in daily agriculture work, 7,12,13 other studies showed that they might have worse global physical performance due to a higher incidence of osteoarthritis and poor nutrition. 14 The worse physical performance of the participants of CMS as compared with those in urban areas indicated the presence of underlying health problems in Yunlin.
In this study, we showed the effectiveness of HAM in community health monitoring and intervention. We found improvements in physical parameters, including BP, standing and walking abilities by the intervention of HAM. These findings implied that the physical capabilities of the rural elderly population could be facilitated through adequate intervention. The causes of improvements were multifactorial: (i) illnesses were managed in a timely by Apollo doctors, eliminating the obstacles of performing physical activities in the first place; (ii) establishment of personal health records could enhance awareness of social and physical function capability; 15 (iii) HAM eliminated the obstacles of transportation found in rural areas, which was reflected by increased usage of outpatient clinics and dentists, and early detection of cancers; and (iv) consultations about nutrition facilitated diet and physical function. 16 Previous study showed that lower income and longer commuting distance were factors that result in the inability to attend scheduled outpatient clinics. 17 In addition, older people rarely consulted healthcare providers before they visited the emergency room. 18 The reduction of emergency services use along with increased usage of non-urgent outpatient services in participants of CMS reflected that older people have improved health literacy by using HAM. They may manage mild illness because of frequent contact with medical personnel via telemedicine, which avoids later catastrophic medical urgencies.
Technology has become a major determinant in individual health service use behavior, as proposed by the Behavioral Model of Andersen and Newman (Fig. S1). 19 Telemedicine has provided a simple and constant method of intervention for older people in rural areas. The common intervention in preventing frailty of older people usually included physical training [20][21][22][23] and nutrition assessment, 20,21,24 which was performed by a therapist within a limited time. Although this kind of intervention is usually effective, the cost-effectiveness regarding cost of medical services was either not taken into account 25 or showed greater cost. 26,27 Telemedicine in HAM decreased the cost and provided a sustainable system without time constraints. These advantages provide us with a chance for long-term follow-up.
In a systemic review, most community intervention was delivered by students, psychologists and nurses, while doctors were involved only in 19% of studies, and usually in a supervising role. 28 However, direct management by doctors is beneficial in managing multidisciplinary care for chronic diseases, 29 which are common in older people. The difficulty of incorporating local general physicians included the financial and legal concerns of telemedicine. 30 Moreover, personal data protection is also an important issue. 30 In HAM, the primary doctors served as consulting counters and provided health suggestions without receiving direct payment from the CMS. Instead, the doctors would consolidate their role of major healthcare provider in a local area and build up a sense of belonging in the country. Furthermore, the doctors would receive support from specialists in NTUHYL if there were medical problems beyond the capacity of local clinics. NTUHYL also set up a tele-communication system with the standard of electronic medical record protection, which provided a safe platform for telemedicine. The continuous record of health parameters in HAM provided a personalized health record of the participants. The concept of community health records provided a more complete view of population health, targeting community health interventions in a data-driven manner. 31 Incorporation of electronic health records could dramatically increase the timeliness of data availability. 32 This collaboration can not only compensate for the reduced access to medical care in rural areas but also help gather data to improve the quality or focus of healthcare in those areas. To manage the physical disability found in HAM, NTUHYL has launched several ongoing projects targeting nutrition, dental condition, cognitive and psychiatric status, and osteoporosis in the CMS. For example, the diets of the CMS were modified by a hospital dietician and based on the functional status and personal dietary habits. A multidisciplinary health screen focusing on physical disability was also performed using a single community-tailored strategy. The effectiveness of these interventions could be monitored by HAM with high efficiency.
The present study had several limitations. This study was a quasi-experimental design without a control group. However, such a design is common in telemedicine studies and is valid. 33 The high drop-out rate of continuous monitoring is a major limitation. The CMS is not a mandated service in Taiwan and participation in such a program is completely voluntary, which mean the attendance rates may fluctuate. In addition, the high drop-out rate may attribute to rural-urban disparities in health and social resources that are encountered by rural older adults. 34,35 Specifically, low health literacy, more limitations in activities of daily living, and the existence of cognitive impairment, which have been identified as risk factors for research attrition are more common among older adults who reside in rural areas, [35][36][37] and thus may hinder our sample from regular participation in CMS and HAM. This is shown because further investigation revealed that FTSST was worse in the drop-out group. All these rural-urban health disparities along with social barriers, such as transportation difficulties, can make our participants less motivated and less able for continuously engaging in services than the general older population. The decline in motivation of older participants in behavioral science studies is also a common reason for a low retention rate. 38 To improve compliance, we applied several methods such as the timely recommendation of monitored abnormalities to improve participants' confidence of the usefulness of health monitoring. The possible benefits of HAM were demonstrated by changes in medical resource utility. Greater effectiveness could be proved by changes in disease incidence and mortality.
In conclusion, HAM manifests a model of integrating medical resources for the benefit of the deprived and the need to counteract the inequalities in distribution of medical resources. It has been known that organizing and coordinating resources across different healthcare levels and sites is far more complex than maintaining the existing healthcare system. This model can provide an example of an effective way to consolidate limited resources and achieve a more than satisfactory outcome.