Information technology-supported integrated health service for older adults in long-term care settings | BMC Medicine

Study design overview

The Health-RESPECT study is a pragmatic, multicenter, cluster randomized controlled trial. Cluster randomization was performed using the institutional identification number. The LTCH and NH were recruited and randomly allocated to the intervention or control group. The LTCH and NH located at Seoul or Gyeonggi-do were recruited and randomly allocated to the intervention or control group. We started recruiting participants in September 2019. The details of the Health-RESPECT study were published in a previous protocol paper [12]. The present study was registered with the Clinical Research Information Service (CRIS, trial registration number: KCT0004360) (initial submission date: September 4, 2019). The study protocol was reviewed by the Seoul National University Bundang Hospital institutional review board (IRB No. B-1908/556–304). The study design adhered to the principles of the Declaration of Helsinki.

Participants

Patients aged over 65 years, who were admitted to LTCHs or NHs for the management of multiple medical conditions such as dementia or functional disability, who had stayed at the facility for at least two weeks before enrollment in the study, and who had at least one chronic disease (hypertension, diabetes, heart failure, etc.) were included in the study. We excluded patients whose life expectancy was less than 3 months, as assessed by physicians at the institution. Patients who were scheduled for discharge within 3 months or refused to participate in this study were also excluded. Written informed consent was obtained from study participants with intact cognition, or from a health care proxy on behalf of participants with impaired cognition.

Health-RESPECT platform and data acquisition

Data of vital signs (blood pressure, heart rate, and body temperature), laboratory findings, and prescribed medications are collected and updated from the electronic medical records (EMR) of the LTCH every month by nationwide health information exchange (HIE) system [16,17,18] (Additional file 1: Fig S1). Through the collected data, an individualized management plan for chronic disease, medication, and rehabilitation is provided according to the algorithm within the platform. Health RESPECT program provides management goals and specific comments based on a determined decision support algorithm, considering the patients’ frailty and disease status. For the Health-RESPECT clinical study, we established a new HIE system for the participating LTC institutions. We ensured that all vital signs were uploaded from the participating sites to our Health-RESPECT platform.

Measurement

Frailty status was evaluated at enrollment using the Korean version of the FRAIL (K-FRAIL) scale [19, 20]. Patients with scores ≥ 3 were classified as frail, those with scores of 1 to 2 were pre-frail, and 0 were robust. Cognitive and physical function were evaluated using the Korean version of the Mini-Mental State Examination (MMSE-K) and functional ambulation category (FAC), respectively. Activities of daily living (ADL) were evaluated using the ADL Hierarchy Scale, which tests four items (personal hygiene, toilet use, locomotion, and eating) and ranges from 0 to 6, with higher scores indicating increased ADL dependency [21].

Treatment targets for hypertension and diabetes differ according to frailty status. The target blood pressure for hypertension was 140/90 mm Hg in the robust and prefrail groups and 150/90 mm Hg in the frail groups. The target hemoglobin A1c (HbA1c) for diabetes was < 7.5% in robust groups, < 8.0% in prefrail groups, and < 8.5% in frail groups, or for random glucose levels, the target was ≤ 190 mg/dL in robust groups, ≤ 210 mg/dL in prefrail groups and ≤ 230 mg/dL in frail groups [12]. Data on patients’ vital signs, laboratory findings, diagnosis, and prescribed medication were collected from the electronic medical records of the LTC institutions and accumulated in Health-RESPECT. Potentially inappropriate medication (PIM) and polypharmacy were determined based on medications from the prescribed record.

Health-related quality of life (HRQoL) was assessed using the Korean version of the EuroQol instrument (Korean preference weighted EuroQol 5-Dimensions (EQ-5D) [22, 23]. The EQ-5D comprises five dimensions (mobility, self-care, usual activities, pain or discomfort, and anxiety or depression), each of which can be rated at three levels (no problem, some problems, severe problems) [24]. For the EQ-VAS, patients were asked to rate their current health state on a scale of 0 (worst imaginable) to 100 (best imaginable).

The list of PIMs was defined by referring to the Beers Criteria 2019 and considering the medical environment of the LTC institutions in Korea [25]. A detailed list of medications was presented in the protocol paper and Additional file 1: Table S1 [12].

The composite quality-of-care indicator (CQI) computed from the interRAI assessment system (interRAI CQI) measures the effects of intervention by the overall change in participants’ functional status [25]. It comprises a set of individual QIs (inadequate pain management, little or no activity, physical restraint use, etc.). The CQI score was calculated as the sum of 33 individual QIs indicating whether an older adult is at risk for functional decline or needs clinical interventions [26, 27].

Randomization and intervention

In our study, one university hospital, seven LTCHs, and six NHs were recruited through a medical referral team of a university hospital where they communicate and transfer patients to LTC facilities. University hospital and LTC use HIE systems to exchange medical information. Although there are various EMR companies and systems in Korea, we could only carry out additional development with the two companies with the highest market share in LTC facilities to connect HIE systems with the Health-RESPECT system. So, LTC facilities were recruited when (1) expressed interest in our study when the medical referral team contacted (2) used one of the two major EMR company systems, and (3) agreed to participate in this study through a face-to-face meeting.

To account for the potential impact of institutional factors on the efficacy of the intervention, the age distribution of patients was utilized as a proxy for disease, clinical condition, and functional status. Accordingly, institutions were randomized after the proportion of older adults aged over 75 years was matched. The LTCH and NH were allocated into the intervention and control groups at a 1:1 ratio. Given the introduction of Health-RESPECT as an intervention, the study participants could not be blinded. To maintain anonymity, the assessors and the statistician were blinded to the allocation of facilities until their role was complete.

The intervention comprised a CGA; evidence-based management algorithm for hypertension, diabetes, and heart failure (Additional file 1: Fig S2); screening for PIM; tailored cognitive, physical, and swallowing rehabilitation program; and alarm for adverse events. Access to medical information within Health-RESPECT was granted only to medical staff. The CGA of Health-RESPECT consisted of a digitalized “patient evaluation form,” which must be completed and submitted periodically for reimbursement from LTCHs, along with additional assessment from nurses at LTCHs and NHs (ADL, IADL, K-FRAIL, and FAC). Nurses at LTCHs and NHs conducted CGA. The usage of the Health-RESPECT and the measurement method was described in detail in a manual book and distributed to each LTCH and NH, and the research nurse visited the site in person to provide education if needed. The medication management program screened all prescribed medications and provided the number and specified PIM list corresponding to the absolute or potentially inappropriate drug list. Patients randomized to the control group were managed with usual care without any exposure to assessment methods and intervention tools. Fig. 1.

Study timeline for the Health RESPECT study. This figure illustrates the timeline of the Health RESPECT study from initial enrolment to follow-up assessments. Baseline assessments (T1) are conducted at the start, followed by subsequent evaluations at 3 months (T2) and 6 months (T3)

Outcome and follow-up

Primary and secondary outcomes were evaluated at 3 months after the intervention or observation period by independent assessors trained by the research team. The co-primary outcomes were: (1) control rate of hypertension and diabetes, (2) medication adjustment (PIM prescription rate, the proportion of polypharmacy discontinuing), and (3) combination of potential quality-of-care problems (CQI) from the interRAI Long Term Care Facility (LTCF) assessment system, which is widely used in about 35 countries; the system consists of a psychometrically sound CGA instrument, a manual, and applications including scales and quality indicators [26, 27]. The hypertension and diabetes control rate was determined considering the frailty status [28, 29]. Detailed descriptions of primary and secondary outcomes were attached to the supplementary file (Additional file 1: Table S2).

Statistical analysis

The sample size was calculated based on the reduction of PIM prescriptions. We expected that the intervention would reduce PIM by 20% as compared to the control group. Assuming a 5% significance, 80% power, and an intra-cluster correlation coefficient of 0.01, we calculated that at least 480 participants were needed. The primary analysis was intention-to-treat (ITT) and the secondary analysis was per-protocol (PP).

Continuous variables were expressed as mean ± standard deviation or median and interquartile range and compared by unpaired Student’s t-test. Categorical variables were expressed as counts and percentages, and the chi-square test was used to compare proportions. We analyzed the difference in outcomes between the intervention and control groups by fitting a linear mixed effects model (LMM) and a generalized linear mixed effects model (GLMM) to account for the correlation between patients hospitalized in the same LTCH or NH and repeated observations of the same patient. Specifically, two random intercepts were introduced, one at the LTC institution level using an identification number (ID) and the other at the patient level. The LTC institution-ID random intercepts capture the variability in the outcome among different LTC institutions. Furthermore, the random effects adjust for the difference in outcomes between LTC institutions at the baseline. Statistical analyses were performed with SAS (version 9.4, SAS Institute, Cary, NC). All statistical analyses were two-tailed, and p-values < 0.05 were considered statistically significant.