This is an open-access article distributed under the terms of the Creative Commons Attribution License (https://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work, first published in the Journal of Medical Internet Research, is properly cited. The complete bibliographic information, a link to the original publication on http://www.jmir.org/, as well as this copyright and license information must be included.
MonashWatch is a telehealth public hospital outreach pilot service as a component of the Government of Victoria’s statewide redesign initiative called HealthLinks: Chronic Care. Rather than only paying for hospitalizations, projected funding is released earlier to hospitals to allow them to reduce hospitalization costs. MonashWatch introduced a web-based app, Patient Journey Record System, to assess the risk of the journeys of a cohort of patients identified as frequent admitters. Telecare guides call patients using the Patient Journey Record System to flag potential deterioration. Health coaches (nursing and allied health staff) triage risk and adapt care for individuals.
The aim was a pragmatic controlled evaluation of the impact of MonashWatch on the primary outcome of bed days for acute nonsurgical admissions in the intention-to-treat group versus the usual care group. The secondary outcome was hospital admission rates. The net promoter score was used to gauge satisfaction.
Patients were recruited into an intention-to-treat group, which included active telehealth and declined/lost/died groups, versus a systematically sampled (4:1) usual care group. A rolling sample of 250-300 active telehealth patients was maintained from December 23, 2016 to June 23, 2019. The outcome—mean bed days in intervention versus control—was adjusted using analysis of covariance for age, gender, admission type, and effective days active in MonashWatch. Time-series analysis tested for trends in change patterns.
MonashWatch recruited 1373 suitable patients who were allocated into the groups: usual care (n=293) and intention-to-treat (n=1080; active telehealth: 471/1080, 43.6%; declined: 485, 44.9%; lost to follow-up: 178 /1080, 10.7%; died: 8/1080, 0.7%). Admission frequency of intention-to-treat compared to that of the usual care group did not significantly improve (
Clinically and statistically meaningful reductions in acute hospital bed days in the intention-to-treat group when compared to that of the usual care group were demonstrated (
Potentially preventable hospitalizations or potentially avoidable admission costs are of significant interest, not only to governments and hospitals, but to individuals, their families, the community, and general practice [
HealthLinks: Chronic Care (HLCC) is a voluntary, funding-neutral reform that aims to support the Australian State of Victoria’s public health services in adopting outcome-based, rather than activity-based, funding [
The MonashWatch telehealth and coaching model used design principles to establish a collaborative patient-journey approach responding to broad social determinants beyond disease management and the boundaries of hospital, primary, home, and social care.
Laypersons called
This paper reports a pragmatic summative evaluation of the MonashWatch service. We compared bed days for an intention-to-treat group versus a usual care control group for 30 months from the MonashWatch service commencement. The intention-to-treat group included a MonashWatch active telehealth group consisting of those who used the telehealth service.
The Australia-wide universal free public health system has both federal and state/territorial governance. Parallel private health systems exist. Medicare is the Australian federal government’s scheme to give universal public access to health care (funded by taxation—the Medicare levy) through (1) direct clinical service funding to general practitioners and specialists in all states and territories and (2) indirect financing, with the states and territories administering public hospital and most community services.
Most services, including social services and welfare, aged care, education, and employment, have split funding and administration across federal and state/territory systems.
There have been multiple initiatives to address avoidable admission costs across the jurisdictions for more than 20 years. The Australian Institute of Health and Welfare (AIHW) annually reports on 18 International Statistical Classification of Diseases, Tenth Revision (ICD-10) diagnostic codes of chronic, acute, and vaccination-preventable admissions for national performance monitoring by local area [
Victoria is the second most populous state of Australia with a population of 6.25 million. The Victorian Department of Health and Human Services (DHHS) funds and administers 85 public health service organizations. It has had activity-based funding (fee-for-service) for all acute admissions with accelerating demand growth in hospital separations per 1000 individuals. Victoria has deployed hospital readmission prevention programs since 1996, which have improved satisfaction with care but have had little impact on cost containment [
The HLCC initiative began in 2016, identifying patients at-risk of ≥3 repeat hospitalizations and their admission costs in the subsequent 12 months [
Monash Health is the most extensive public hospital and community care system in Victoria. Its 15,000 staff (with a large hospital readmission prevention program base) work at more than 40 sites, providing more than 3 million occasions of service, admitting more than 238,000 hospital patients, and handling more than 206,000 emergency presentations per year. HLCC data indicated that in 2017 and 2018, Monash Health had more than 3000 patients with >4 acute medical admissions and more than 12,000 with >3 acute medical admissions (30% of which were potentially preventable hospitalizations). It was an early adopter of the HLCC service initiative MonashWatch in 2016.
Rather than adding a new layer, the MonashWatch model was intended to be a catalyst within a working health system, to begin a transition of acute services to outcome-based funding.
Patient journeys involve physiological, psychological, social, and environmental issues. Disturbances in any or combinations of domains including housing, food security, support with daily living, access issues, as well as biology, medication issues, or clinical deterioration may lead to tipping points into acute admissions [
Clinicians—nursing or allied health—enable others in the goals and health journeys which they chose to follow [
Following HLCC algorithm identification and allocation, the intention-to-treat group were invited to initially consent and agree to a home visit for enrollment, formal consent, baseline assessment, and induction. Telecare guides, then made conversational phone calls to enrolled people to track their health and needs, in accordance with personal preference and previous Patient Journey Record System flags.
Regular audiotaped calls between 1 to 5 times per week (median 1), depending on risk level, were conducted by telecare guides. They used the Patient Journey Record System semistructured monitoring app, which began with open-ended narratives and included directed questions [
Patient Journey Record semistructured monitoring app that begins with open-ended narratives and includes direct questions.
The MonashWatch model (reproduced from Martin et al [
The health coaches were all recruited from within the Monash Health community and acute services to develop the new service. Telecare guides were recruited from the local or adjacent community. Care pathways to existing services were mapped and tested before the service commenced. The pathways were to the general practitioners, hospitals’ emergency, inpatient and outpatients, hospital readmission prevention programs, and other community health, social, and welfare services including housing, legal, financial, employment, education, and voluntary organizations.
MonashWatch participants were selected in a highly disadvantaged catchment area adjacent to the Dandenong hospital. Keeping the MonashWatch team local to patients was a concept borrowed from Buurtzorg: the Dutch neighborhood care model [
The HLCC web-based algorithm incorporated a wide range of conditions in adults >18 years old [
AIHW potentially preventable hospitalizations diagnostic codes only accounted for 18% of HLCC admissions. Only approximately 20% of HLCC-identified hospitalizations in Monash Health, and other Victorian health services had ever accessed hospital readmission prevention programs or other hospital admission prevention services.
The DHHS provides continuously updated HLCC-eligible cohort lists to hospital groups and funds care improvement initiatives based on projected reductions in admission costs. Once the patient was deemed eligible, when they had their next acute admission, they could be enrolled. Enrollment commenced with a gradual ramp-up from December 2016 and continued beyond the evaluation cutoff point. Pragmatic screening by health coaches excluded those who were not suited to a self-rated phone-based health model (eg, nursing home, necessitated use of an interpreter, and patients who would pose a high risk to staff visiting at home). Patients were considered candidates to be entered into the MonashWatch evaluation pool before allocation, based on a ratio of 4:1. There was minimal chance of bias because the health coaches and team performing the assignment had no idea who would benefit in advance in this pilot service, and the allocation was conducted using hospital unit numbers from a list without patient details.
The primary outcome metric was bed days (ie, length of stay related to emergency nonsurgical admissions) derived from the Victorian Admitted Episode Data from the Victorian Emergency Minimum Dataset [
A secondary outcome metric was rate of emergency nonsurgical admissions. This was initially considered as the primary outcome; however, capitation costs being the biggest driver of the HLCC program led to bed days being more critical. Net promoter score was also a secondary outcome.
Analysis of covariance (ANCOVA) is a statistical technique that adjusts for covariates in determining the outcomes of an intervention. Least square means are an acceptable method to calculate the means adjusted for covariates [
Quality assurance of the data analysis was performed in several ways. One author conducted the ANCOVA using XLSTAT software (version 2020:4.1; Addinsoft). Two other authors independently analyzed the same dataset, after rechecking the download from Victorian Admitted Episode Data/Victorian Emergency Minimum Dataset [
Ethics approval for low-risk clinical research was obtained from the Monash Health's Health Research Ethics Committee. The Australian government’s main research and development agency, the Commonwealth Scientific and Industrial Research Organisation, is conducting an external evaluation of the diverse state-wide HLCC initiatives in Victoria, and this also has ethics approval.
The HLCC clinical algorithm identified 2502 patients as having a high risk of repeat admissions within the period of December 23, 2016 to June 23, 2019.
MonashWatch identified 1373 suitable HLCC patients: usual care (n=293) and intention-to-treat (all: n=1080; active telehealth: 471/1080, 43.6%; declined: 485/1080, 44.9%; lost to follow-up: 116/1080, 10.7%; died: 8/1080, 0.7%;
MonashWatch pragmatic clinical evaluation participants. Intention-to-treat, including active telehealth, and usual care cohort allocation in the MonashWatch pragmatic evaluation. MW: MonashWatch.
Mean participant age for usual care was 64.3 (SD 17.6; median 71, IQR 19) years, and mean participant age for intention-to-treat was 68.3 (SD 16.8; median 71, IQR 19) years. The number of effective days active for usual care was 756.2 (SD 180.5; median 1003, IQR 44), and the number of effective days active for intention-to-treat was 624.2 (SD 269.2; median 908, IQR 278).
In the usual care group, 293 patients had 639 admissions, and in the intention-to-treat group, 1080 had 934 admissions from the time they joined the MonashWatch program (effective days active) until June 23, 2019 (
Descriptive statistics on admission rates in MonashWatch intention-to-treat and usual care.
Statistic | Usual care | Intervention | ||
Admitted at least once, n (%) | 163 (55.6) | 549 (50.8) | .054 | |
|
|
|||
|
Median (third quartile) | 3 (5) | 2 (4) | .05 |
|
Mean (sample SD) | 3.9 (4.3) | 3.6 (4.6) | .056 |
|
Pearson skewness | 3.1 | 6.8 | <.001 |
Admissions were highly skewed in the intervention group with several outliers with frequent short admissions to the emergency department for chest pain and abdominal symptoms which may account for the highly skewed profile of the intention-to-treat group. The raw median and mean number of admissions per person were higher in the control group (
AIHW potentially preventable hospital admission codes were present in 18.3 % (117/639) of all acute admissions in usual care and 16.4% (153/ 934) of all acute admissions in intention-to-treat (
Most common diagnoses for potentially preventable hospitalizations as defined by the AIHW within each group of MonashWatch patients.
Group and diagnosis | Admission, n (%) | |
|
|
|
|
Chest pain, minor complexity | 46 (7.2) |
|
Abdominal pain and mesenteric adenitis, minor complexity | 20 (3.1) |
|
Other digestive system disorders, major complexity | 19 (3.0) |
|
Bronchitis and asthma, minor complexity | 18 (2.8) |
|
Esophagitis and gastroenteritis, minor complexity | 18 (2.8) |
|
Arrhythmia, cardiac arrest and conduction disorders, minor complexity | 16 (2.5) |
|
Other digestive system disorders, minor complexity | 16 (2.5) |
|
Kidney and urinary tract infections, minor complexity | 15 (2.3) |
|
Syncope and collapse, minor complexity | 15 (2.3) |
|
Chronic obstructive airways disease, minor complexity | 14 (2.2) |
|
|
|
|
Chest pain, minor complexity | 110 (11.8) |
|
Abdominal pain and mesenteric adenitis, minor complexity | 41 (4.4) |
|
Respiratory infections and inflammations (major complexity | 29 (3.1) |
|
Chronic obstructive airways disease, minor complexity | 25 (2.7) |
|
Syncope and collapse, minor complexity | 22 (2.4) |
|
Other digestive system disorders, minor complexity | 21 (2.2) |
|
Other digestive system disorders, major complexity | 19 (2.0) |
|
Headaches, minor complexity | 18 (1.9) |
|
Esophagitis and gastroenteritis, minor complexity | 18 (1.9) |
|
Arrhythmia, cardiac arrest and conduction disorders, minor complexity | 16 (1.7) |
|
|
|
|
Chest pain, minor complexity | 41 (10.5) |
|
Diabetes, minor complexity | 10 (2.6) |
|
Arrhythmia, cardiac arrest and conduction disorders, minor complexity | 9 (2.3) |
|
Abdominal pain and mesenteric adenitis, minor complexity | 6 (1.5) |
|
Chronic obstructive airways disease, minor complexity | 6 (1.5) |
|
Coronary atherosclerosis, minor complexity | 6 (1.5) |
|
Poisoning/toxic effects of drugs and other substances, minor complexity | 6 (1.5) |
|
Chronic obstructive airways disease, major complexity | 5 (1.3) |
|
Syncope and collapse, minor complexity | 5 (1.3) |
|
Heart failure and shock, major complexity | 4 (1.0) |
ANCOVA was conducted on bed days with admission age, gender, the presence or absence of a potentially preventable hospitalizations ICD-10 code, and effective days active as quantitative variables and with intervention versus control as qualitative variables
ANCOVA summary statistics of the impact of key variables on bed days of MonashWatch patients: standardized coefficients predicting length of stay.
Source | Value | SE | 95% CI | ||
Admission age | 0.110 | 0.019 | 5.673 | <.001 | (0.072, 0.148) |
Gender | –0.001 | 0.019 | –0.031 | .98 | (–0.038, 0.037) |
AIHWa potentially preventable hospitalizations (0, no; 1, yes) | –0.005 | 0.019 | –0.235 | .82 | (–0.042, 0.033) |
Effective days active | –0.090 | 0.020 | –4.566 | <.001 | (–0.129, –0.051) |
Control (usual care) vs intervention (intention-to-treat) | 0.086 | 0.020 | 4.348 | <.001 | (0.047, 0.125) |
aAIHW: Australian Institute of Health and Welfare.
Bed days standardized coefficients based upon ANCOVA.
Age, MonashWatch effective days active, and intention-to-treat group status predicted bed days. The usual care least square mean was 4.5 (SD 0.2, 95% CI 4.1-4.9) bed days while the intention-to-treat least square mean was 3.4 (SD 0.1, 95% CI 3.1-3.6) bed days. A statistically significant (
ANCOVA least square mean bed days comparison for usual care versus intention-to-treat groups.
Category | Mean bed days | SE | 95% CI | |
Usual care | 4.5 | 0.2 | (4.1, 4.9) | <.001 |
Intention-to-treat | 3.4 | 0.1 | (3.1, 3.6) |
|
Longitudinal tracking of average bed days per person per month nonsurgical acute admissions was conducted for 12 months before enrollment until the evaluation cutoff date (
A high net-promoter rate (satisfaction scores) of 95% was demonstrated, with common findings of about 77% in hospital evaluations [
Rate of mean bed days per person per month in the usual care group (black) and intention-to-treat (red) 12 months prior to MonashWatch enrolment (indicated by the blue arrow) and for the subsequent 30 months. (Total admissions adjusted for numbers in each group.).
This evaluation demonstrates that the MonashWatch service intervention achieved its objectives by reducing bed days and, by implication, worked within the capitated budget consistently over time.
The percentage of patients admitted at least once was very similar (usual care: 55.6%; intention-to-treat: 50.8%). Admissions per person were nearly statistically significantly different (median,
Challenges to constrain costs while improving care are prominent in the fragmented governance and funding system of Australia. Macrolevel federal and state reforms with pooled funding have not previously proven successful. Victoria has deployed state-based local initiatives to improve hospital readmissions since before 1996 but had identified the need to further “shift the dial [
The tracking of bed days and triangulation with other findings indicated a causal association between MonashWatch and the improvement of bed day utilization using Bradford Hills criteria [
A plausible mechanism for the intervention working is addressing resilience and frailty through anticipatory care and coaching enabling stronger health networks and connections with vulnerable people. The results support a continuous adaptation to complex unstable health journey model for individuals [
The capitated structure provided very adaptable funding as needed for many issues such as transport, outpatient attendances, and home factors which underpin admissions. It provided coaches with the flexibility to go outside of health siloes working with general practitioners, hospital, alcohol, and mental health services.
Finally, this DHHS approach that is outcome and data-driven with continuous performance and costing review for teams and local initiatives keeps services from falling into complacency.
There is a range of limitations. The intention-to-treat group included 44.9% (485/1080) who declined. This arduous process may have diluted the uptake rate and thus had an impact on bed days but reflects real-world clinical service evaluations. It would be worth other methods of recruitment in the future to see if there would be increased recruitment with a more significant impact.
Pragmatic clinical evaluations in live health systems outside of research study are challenging, particularly in MonashWatch due to long-term unpredictable, complex dynamics in unstable patient journeys [
This summative evaluation was conducted in a living health system as the first phase of a government funding initiative to move from activity to value-based funding when Monash Health services were under significant funding constraints. Success was achieved in the real world without going through the traditional research route with a trial before rolling out an implementation. The positive feature of this approach is that (to date) it has not gone the way of many beautifully designed and executed pilots that never achieved implementation. In the first phase, the successful delivery of care was within existing funding. The MonashWatch-type model deployment in other health services in the second and third phases has the opportunity to improve on trial methodology. The addition of more research resources, given the current successful proof of concept, would enable the conduct of a more sophisticated randomized propensity-matched trial.
Ongoing study of the data is needed to identify who benefits from which components and how the intervention can be improved for different groups. There is a need to continue to shift current care pathways and health systems to adapt care to the needs of vulnerable MonashWatch-type patients. A whole of systems transformation is needed to respond to the dynamics of unstable health journeys, beyond the current single disease or condition siloed care. Outcome-based funding has the potential to make major inroads into fragmented care. Macrolevel health service funding changes in the Australian Health System such as in northern Spain [
The MonashWatch telehealth and coaching intervention using the HLCC innovative funding model was effective in a local catchment area of a hospital in a highly disadvantaged community and achieved its health service funding model objectives. It requires ongoing and broader implementation and evaluation. In the future, evaluation of additional teams is needed to confirm these findings in different populations and settings. Two additional teams are now in place. Ultimately, the progressive scaling up to a multisite intervention will require ongoing tailoring and evaluation with feedback for improvement.
CONSORT-EHEALTH checklist (V 1.6.1).
Australian Institute of Health and Welfare
analysis of covariance
Department of Health and Human Services
HealthLinks: Chronic Care
International Statistical Classification of Diseases
This paper acknowledges the innovative funding model (HLCC) developed by the Victorian Department of Health and Human Services, and the very generous open spirit of the HLCC team. It acknowledges the stellar work of the MonashWatch clinical team—the telecare guides and the coaches who have made the model work to date in the highly charged domain of living health services. It acknowledges the work of Kevin Smith and John-Paul Smith who have implemented and supported the Patient Journey Record System app and analytics. Kevin Smith has provided ongoing comments, editing and proofreading support during the writing of this paper. John Kellett is acknowledged as a particularly thorough reviewer. DHHS continually tracks the data and has an independent summative evaluation being conducted by the Commonwealth Scientific and Industrial Research Organization. Smart Health Solutions have developed and implemented the web-based system for supporting coaching of the patients identified by Patient Journey Record System in their PatientWatch system.
KS and CM both analyzed the data separately and independently from each other.
CM was the clinician lead in the development of the Patient Journey Record System software.