Voice

Andy Callow, CDIO, Kettering General Hospital: Embracing the potential of the Cloud

By Andy Callow, Chief Digital Information Officer at Kettering General Hospital (Twitter: @andy_callow)

As part of the HTN Health Tech Trends Series, Andy Callow, CDIO at Kettering General explores the potential of the Cloud in healthcare and argues why it makes sense.

What is the NHS Waiting For?

In behavioural science there is a concept known as the intention-action gap. This refers to the difference between what people say they plan to do and what they actually do. It would seem to be that when it comes to Cloud and the NHS, there’s a lot of talk and very little action. In 2013, the Government produced a Cloud First policy for all technology decisions, giving public sector organisations the permission they presumably said was holding them back from embracing the potential of the Cloud. And then very little happened. Many organisations then continued to build data centres on premise or even use off-site hosting, and were unable to tackle the barriers of making the jump to the Cloud.

Seven years later, most of the NHS has made little progress. There are of course good stories but they are exceptions rather than a rule. In 2015 I joined NHS Digital as the Head of Technology Delivery for the NHS Website NHS.UK (known as NHS Choices back then). The site had moved out of local hosting in 2014 and we were able to make considerable transformation of the technology stack, staying within the same budget envelope over the next three years. We put a case study together in 2016 showing the cost savings of moving  onto the cloud and the benefits that had been realised in terms of flexibility, availability, site capacity and user-experienced site speed during that time. I remember sharing the case study with numerous people, including with NHS and Local Government colleagues and then watching nothing happen.

There are also some examples of externally hosted web-based systems, but these are generally purchased in that form rather than a dedicated approach to move from on-premise to cloud hosted services. What I’m interested in is a concerted effort to remove or minimise on-premise data centre footprint.

There are plenty of examples where Trusts refresh their on-premise data centre, without seeming to pause to consider other options. In my own organisation, Kettering General Hospital, a couple of years before I joined, we moved our data centre out from a basement prone to flooding into a shiny container off the car park. There will be hundreds of NHS organisations up and down the country that will have done similar things over the past 7 years without considering anything but on-premise. We desperately need to close the intention-action gap.

Why Public Cloud?

It’s almost embarrassing to even have to make the argument for Public Cloud these days, given 2013 is a long time ago and the hyperscalers have grown massively in this time. For example, platforms for others are now a third of Amazon revenue[i] and Microsoft Azure has grown from a $5.5bn business in Q2 of 2015 to $44bn in Q4 of 2019[ii]. I’m not saying it is easy, by any means. This is a multi-year endeavor, requiring a lot of planning and continual attention, but I’ll be really depressed if I’m looking back in 2027 and collectively across the NHS we’ve not made substantial progress.

Why the Cloud Makes Sense:

  • Security – hyperscalers can do security much better than any Trust. They have the scale, expertise and financial incentives to apply the best cyber security practice. No Trust, no matter how good is ever going to come that near.
  • Stability – Public cloud offers the ability to provide high availability, with resilience and auto repair with ease
  • Scalability – You pay for what you use. Tear down and scale up to meet the needs of the business. The need to get physical tin into a data centre, which only comes out when it has died cannot compare with that level of scale. Are you really happy to be waiting 4 weeks for a server to arrive, installed in your racks and get configured?
  • People – This is a biggie for me. Firstly, the availability of skills to manage bare tin will diminish rapidly over the next 10 years, so that means it will cost a lot or you’ll asking people with limited skills to do ever more complex stuff. Plus we are not going to attract the brightest Ops graduates/talent who’ve spent the last three years having never touched physical servers, to on-premise data centre management.
  • Technologies – The pace of change in the Public cloud is faster than you can reasonably consume it. The pace of change in a physical data centre is the rate that you can move tin in and out. How many clinical systems does the NHS run that are on old software versions because the overhead of upgrading is just too massive? I shudder to think.
  • Cost and Carbon Footprint – Building that data centre in the car park was a considerable investment. In fact the minute it was built it is a considerable liability, and what’s more we have to refresh it every 3 to 5 years just to keep up.
  • From an environmental point of view, hyperscalers have much more sophisticated cooling and power arrangements that any one Trust can set up. Just think of the combined net effect of turning off all those individual air con units across the NHS estate.

The Common NHS Barriers

  • Funding Models – The NHS runs on an annual capital cycle, which lends itself to physical investments. Moving to a Cloud subscription requires hard-pressed revenue. This is where NHSE/I steered by NHSX could make a massive difference here and help support this change in funding models. However, many people have made the case that the depreciation of the data centre is cost comparator that can be exploited when making the case, which is something we’re looking to use here at Kettering.
  • Our Suppliers Aren’t Ready – This is a common plea, but have you considered having a conversation with them about how you can work in partnership in your pursuit of moving to the cloud? Perhaps they are waiting for the first customer to push this?
  • Clinical Systems Are Different – There are examples of PAS systems that are externally hosted by some of the larger suppliers. That should tell you that it can be done. But it does rely on the next barrier…
  • Our Internet Connection Isn’t Fast Enough – When was the last time you looked at your internet growth forecast? It will be growing massively year on year, so you’ll need to make plans for a resilient internet connection anyway. Once you have that, it will pave the way for your next move regarding cloud.
  • We move too much data around, it will cost too much – It is true, if you don’t pay attention to your architecture, you can burn a lot of money by moving data around, but that doesn’t mean it can’t be done. You just need a different mind set, and one that is aware of the cost implications of each decision, which in my view is a better position to be in.
  • We won’t have control of our costs – You have very little control of your costs in a physical data centre. You may have done a virtualisation project and improved the utilisation of the servers, but your cost base is now fixed. However, your Ops Engineers will have little or no ideas of how much any one server costs to run per month/year. In my experience with NHS.UK, if you give talented people the information about the cost of the Cloud services and ask them to get the best performance within a cost envelope, they can be incredibly innovative in finding ways to squeeze the best out of each pound.
  • We’ve just invested in new physical infrastructure – The good old sunk-cost bias fallacy. That kit is not an asset, it is a growing liability. Plus if you don’t make some plans now, it will be too late to do anything by the time the next hardware refresh date comes around.

Back to Kettering and the search for Clouds on a Sunday Afternoon

As the end of life date of that data centre in the car park at Kettering Hospital starts to figure on forward-look spreadsheets and we start to plan what we need from a re-built hospital as part of the Hospital Infrastructure Programme Wave 2 (HIP2) we needed to start to plan what we were going to do.

One Sunday afternoon in August 2020 I set out to draft a Cloud-First policy for the Trust. My first instinct when doing things like this is to avoid having to think, so I turned to Google. Given that KGH is an acute provider, I looked initially for similar providers. The focus was on finding Trusts who had a cloud-first strategy in place with similar wholescale ambitions as Kettering, rather than those who are using piecemeal pieces of Software as a Service (SaaS), or things like Office 365. I was looking for examples of Trusts who were planning to or already had decommissioned whole chunks of their data centres, and if I was lucky, I hoped see some who had moved their Patient Administration System (PAS) into the cloud.

My first point of call was to look at the 32 Acute Global Digital Exemplars and Fast Followers and search for Cloud strategy or policy or even just their digital strategy. Using Google, I searched against the names of these Trusts and came up with nothing. This was frustrating, but perhaps even more alarming was the absence of a published Digital Strategy for most of these. How can that even be right when we’ve spent something like £300m[iii] on these initiatives?

So Frustrated I Nearly Missed Countryfile

Having spent so much time on that Sunday afternoon trying to search for a policy that I nearly missed Countryfile, I wrote a frustrated blog about my research[iv]. This led to some interesting comments and discussions from various people and a better indication that some Trusts are starting to do something about this. But there was still no sign of a published policy.

If You Want a Job Doing…

So, in the end, we have produced a Cloud-First policy that has been approved by our Digital Hospital Committee and is now available on our website[v]. It’s not perfect, but it is short and out there. I see this as a Statement of Intent; we don’t quite know the detail of how we’re going to get there, or even how long it will take us, but it is really clear to all our technical staff and our current and future suppliers what we intend to do, and I’m really proud that we’ve made this available and enjoyed the conversations it has prompted from a number of sources. In it we’ve declared that we will have no new services in our data centre from July 2021. A Google of “cloud first policy kettering” will get you to it.

Why Are You So Secret?

The conclusion of my few hours of research made me realise how far the NHS has to go in terms of moving with the times. I also concluded that the NHS is terrible at sharing, and that makes me sad. If the purpose of the GDE programme is to exemplify good practice and share it with the rest of the NHS then it is failing at the very basic level of Making Things Open [#9 of the NHS Service Manual Principles[vi]], let along blazed the trail for full on cloud adoption. I’d love for someone to prove me wrong, but even if this is the case, I’d argue that for £300M, the folks running the GDE Programme at the very least ought to make it mandatory for each GDE and Fast follower to have a published Digital Strategy. The next step would be for them to have a published statement, updated every 6 months on progress against the Tech Vision[vii] and finally for funding to be linked to progress into the cloud.

Back to the Intention-Action Gap

Going back to the intention-action gap I mentioned at the start, humans are also greatly affected by their surrounding in terms of decision making. For example, a recent OECD study found that consumers were more willing to purchase sustainable or organic meat products when they thought that others were also doing so[viii]. Kettering General Hospital have published a cloud-first policy. That might just be the nudge you need to publish yours.

Andy Callow is the Chief Digital Information Officer at Kettering General Hospital. He’d love to hear from public sector organisations that take issue with his grumpy views. Andy can be found on Twitter @andy_callow


References:

  • [i] https://static1.squarespace.com/static/50363cf324ac8e905e7df861/t/5ed23bd3bdbbdb299cb6aa9f/1590836452410/2020+Benedict+Evans+Shoulders+of+Giants.pdf
  • [ii] https://www.statista.com/chart/15910/microsofts-annualized-commercial-cloud-revenue/
  • [iii] NHS Digital 2018–19 Annual Accounts and Report states “with £200 million funding announced by the Secretary of State in September 2018”. So an estimate of £300m for the programme to date would be fairly conservative
  • [iv] https://andy-callow.medium.com/the-dire-state-of-cloud-first-strategies-in-nhs-trusts-in-2020-773ef98e17f3
  • [v] https://www.kgh.nhs.uk/download.cfm?doc=docm93jijm4n2399.pdf&ver=4652
  • [vi] https://service-manual.nhs.uk/design-system/design-principles
  • [vii] https://www.gov.uk/government/publications/the-future-of-healthcare-our-vision-for-digital-data-and-technology-in-health-and-care
  • [viii] K. Vringer, H. R. J. Vollebergh, D. van Soest, E. van der Heijden, and F. Dietz, “Sustainable consumption dilemmas,” no. 84, 2015.