Two years ago, NHS England launched its national virtual ward programme. This has shown promise in a range of priority areas, from staff productivity to hospital flow, and there has been a proliferation of virtual wards in many shapes and sizes since 2022.

Though the national programme was premised on the known benefits of keeping patients at home, comprehensive evaluations of the impact of fully-fledged virtual wards on other outcomes have only started bearing fruit recently. There remains much more to be done, but there is a growing body of experience that shows that if properly designed virtual wards can be very effective.  It also turns out that there are a number of ways to get things wrong.

Drawing on our experience at PPL in designing, delivering, and evaluating virtual wards, on existing guidance, and on reviews related to older people and more general evaluation, we have identified five key issues leaders will face when designing and implementing virtual wards. For each, there are lessons to be learnt from existing evidence, and there are potential unintended consequences that need to be accounted for.

‘Step-up’ or ‘step-down’

‘Step-up’ models – those focused on patients who would have likely ended up in hospital – can help reduce pressure on emergency departments and prevent unplanned admissions. Developing systems to identify patients with the appropriate level of acuity can be tricky, but if done right the benefits in avoided hospital activity can be significant.

The target cohort for ‘step-down’ models that look to accelerate discharge and prevent readmission is more easily identified. Step-down models are potentially more likely to help with problems of flow and exit block by getting long-stay patients out of hospital, and they can also help prevent readmission. Again, identifying the right patients is not easy – stepping patients down into a virtual ward who would otherwise have been discharged home is a risk.

There is a risk in both cases that patients end up staying longer in the VW than they would have done in hospital.  Setting clear criteria and careful monitoring of admission rates and lengths of stay can safeguard against this.

‘Generalist’ or ‘specialist’

We have worked to deliver and evaluate both centralised, ‘generalist’ virtual ward programmes and disaggregated ‘specialist’ programmes.

The generalist approach allows staff to be deployed more flexibly, simpler administration and benefits monitoring, and less duplication. Generalist programmes are almost always primarily focused on the frail and those with respiratory conditions.

The specialist approach can bring in specialists to target smaller cohorts with more specific needs, work to tailored operating procedures, and flexibly collaborate with other sectors including social care.


The field is too new to be able to definitively say what the optimal scale is, and the answer will inevitably depend on the service’s focus, delivery model, intended outcomes, and other local factors.

A recent study that suggested that VW was significantly more expensive than traditional inpatient care seems have evaluated delivery costs at a point in the intervention’s development where it was still operating at a very sub-optimal scale. Wider evaluations of these types of approaches show that they often take place too soon and fail to consider that complex interventions of this type take time to bed in.


Keeping patients at home is not a new idea but recent developments in technology, particularly remote monitoring, could improve both patient outcomes and staff productivity.

Though the technology is still in its infancy, a competitive provider market means improvements are inevitable. The basic tech-enablement is the same regardless of the other considerations we have set out, so it is worth starting to trial the technology now as not all providers are equal. As is usually the case with technology, you should look to commission at scale.

In the short run higher delivery costs are likely, though this should be seen as an investment. More significant is the potential risk to patient safety – careful and considered piloting, and effective management of relationships with providers can help mitigate this.

The interplay between tech-enablement and digital exclusion is under-explored, so organisations for whom tackling population health inequalities is a strategic priority should embed mechanisms to monitor differential impact into their programmes.

Benefits monitoring

Working through the above issues before you start implementation and clearly recording decisions will help shape the questions any future evaluations will seek to answer.

Robust and reliable answers to these questions depend on the quality of the data used. Information on individual patient journeys is indispensable, and so is clarity on delivery costs. The latter can be difficult for virtual ward programmes that draw on multiple funding pots which is common in cases where existing programmes of work, such as ‘Hospital at Home’ have been adapted post-April 2022.

Though we have set out five distinct areas to consider, they are all interdependent and must be considered holistically. Other factors such as the intended outcome, target condition, age of target cohort, and local population context must also be taken into account.

There remain unanswered questions and gaps in evidence when it comes to virtual wards. Different staffing models may yield different outcomes. There may be unintended ‘shunting’ of care on to unpaid carers. Different referral and triage processes may affect the characteristics of patients admitted to virtual wards and some inequalities in access may develop.

Despite the uncertainty, evidence is mounting that, if done right, virtual wards can help the health system begin to address some of our most intractable problems.