My View - Hospital in the home is not the model we should be scaling

VIEW - Karl Cole, Fellow of HiNZ 

A few weeks ago I posted on LinkedIn that I was sick — not as a patient, but as a GP — of watching AI and digital health being used to make the old hospital model run faster. The post struck a nerve. The reaction told me what I already suspected: a lot of people in digital health quietly disagree with the direction we are heading. 

The wrong unit of care 

Every week another announcement: virtual wards, hospital in the home, remote monitoring “replacing admissions.” These are framed as transformation, but they are not. They take the most expensive, most clinician-intensive, least patient-controlled unit of care we have - the hospital - and re-export it into people’s homes via apps and sensors. 

And they will inevitably re-export the hospital’s organisational shape with it. Hospitals are organised by organ system: cardiology, renal, psychiatric, surgery. A hospital-led virtual ward will be a cardiology virtual ward or a renal virtual ward and inherit the constraints, each running its own monitoring stream into the same patient’s home, blind to the others.  

Compare that with how a blood test or an X-ray works, they are services organised around the patient’s need, not around a hospitals department’s territory. Primary care is built on the same patient-centric logic: the integrating layer that already holds the whole patient. Putting RPM only in the hospital reproduces the silo. Putting it in primary care does not. 

Hospitalisation is what happens when life has broken down enough that autonomy disappears. It is sometimes necessary, but is never the goal. We should be aspiring to a system in which it is needed less often, not delivering more of it more efficiently in more places. 

The leverage is on the other side 

Vote Health for 2024/25 is around $29.6 billion or roughly $5,500 per New Zealander per year. Of that, only $300 to $500 reaches primary care. We spend five to nine percent of the per-person health dollar on the part of the system that keeps people well, and the rest on what happens when that fails. Hospital in the home does not move that ratio, it just changes the address at which the ninety percent is delivered. 

Take blood pressure as the worked example. Around a million New Zealanders live with hypertension. Lifting national control rates from 10 percent would put 100,000 more people in target range, most concentrated where the inequity sits. Māori carry 1.5 times and Pacific peoples 2 times the stroke risk of NZ Europeans, and present at mean ages of 60 and 62 against 75. The Ettehad meta-analysis (pooling trials with mean follow-up of around 4 years) found that a sustained 10 mmHg drop in systolic BP delivers, over that period, a 27 percent reduction in stroke risk and 13 percent reduction in all-cause mortality. Applied to the 100,000 cohort over five years, that implies of the order of 1,000 to 1,500 cardiovascular hospitalisations and 200 deaths averted. 

Compare the two paths. A reactive Hospital-in-the-Home model wraps acute services around the 1,500 patients who event, at a cost of roughly $35,000 to $40,000 each, after the damage is done. A preventive primary-care model touches all 100,000 with a three-month intensive programme at around $500 to $700 per patient. The five-year programme cost is similar in both, between $50 to $70 million, but one model manages disease the system has already failed to prevent, and the other reaches the populations carrying one-and-a-half to two times the risk before the event. Beyond five years, the preventive path compounds across CVD, CKD, diabetes and dementia. The reactive path does not. 

A familiar referral, brought into the present 

What is being described here is what would be possible as “Better healthcare in the home”, brought into the present: a cuffless device worn for three months, the patient seeing their own data shift in response to sleep, salt, stress, exercise and medication. The 24-hour ABPM is a measurement. Three months of engaged monitoring is a behaviour-change programme that happens to start with BP, is proactive and evidence based. 

And it does not stop at BP. Combine continuous BP with continuous ECG, movement, sleep and weight, and the model stops being disease management. It becomes proactive risk detection across the patient as a whole - atrial fibrillation found before the stroke, deconditioning found before the fall, sleep apnoea found before the cardiac event. One platform, several pathways, and the underlying logic is anticipation rather than reaction. Not stopping an organ getting sick, but keeping the person well.  

Patients who learn to interpret their own data and adjust their habits do not unlearn those skills when the device comes off; they redeploy them, to glucose, weight, medication adherence, and the next chronic disease conversation. Heart rate variability is a good example: a single reading in clinic tells you very little, but a personal baseline trended over months, and then available the day someone presents unwell, turns the same data into a genuinely informative health measure. 

Prevention is currently a luxury good 

People with discretionary income already buy prevention: they wear the Garmin, they pay for the gym, they take half a day off for the ABPM, they have the kitchen and the time and the health literacy to act on the results. The existing paradigm quietly assumes all of these things and the people without discretionary income end up presenting at ED with the disease the discretionary-income population prevented years earlier. 

The equity case for primary-care RPM is that it turns prevention from a luxury good into a public good. There is no version of “close the equity gap” that does not run through structured, publicly funded, primary-care-delivered prevention. 

Four things have to change 

Move the unit of measurement upstream. Bed days avoided is the wrong KPI when the goal is fewer bed days needed. Population-level BP control by ethnicity, ED avoidance and continuity of care should sit alongside HitH bed-day numbers in any digital health dashboard. 

Move the money with the work. If primary care is to absorb monitoring at scale (and it should), the capitation envelope has to reflect it. RPM-enabled targeted titration of interventions and self management should be funded as a distinct activity. The 1 July 2026 capitation reweighting is the obvious moment. 

Move the data with the patient. A virtual ward without the primary care AND consumers trend of data is just an inpatient unit with a longer cable. The Shared Digital Health Record is a start; bidirectional flow is the play. 

Move the time horizon. Today’s ED presentations are not failures of last month’s primary care funding, they are failures of the prevention investment we did not make a decade ago. We have to fund both the strokes happening now and the prevention that bends the curve in 2031. Different time horizons need different funding streams. 

Where this lands 

Digital health has a once-in-a-generation chance to redesign the system. We will waste it if we use AI and remote monitoring to make a 1950s hospital model run faster. The future we should be scaling is not hospital in the home, it is life in the home - supported by primary care, enabled by good technology, with hospitalisation reserved for the moments when everything else has genuinely failed. That is the harder system change. It is also the only one that meaningfully closes the equity gap.

Read more VIEWS

Return to eHealthNews.nz home page