Earlier (16.4 hours ago roughly) I talked about the NHS, more important about an article in the Guardian about the NHS. An article that I considered to be a sales pitch, and not a very good one. Today I am taking a look at the ‘Smarter Health: boosting analytical capacity at NHS‘ pdf. First we need to take a look at the players, you see when we read a story we need to know what the story is about, the first step here is to take a look at who is telling us the story, because that matters, especially in the world we see today. The front page gives us the people involved:
Beth Simone Noveck – a Professor at NYU, director and co-founder.
Stefaan Verhulst – Chief Research and Development Officer of the Governance Laboratory at New York University.
Andrew Young – Associate Director of Research at The GovLab, and in addition, he is NOT the Managing Director of APY Consulting Ltd, who has the same name, but looks a few decades older.
Maria Hermosilla and Anirudh Dinesh are also directly linked to Govlab and it is Juliet McMurren who is a bit of a mystery. Perhaps she is merely a minion in that digital publishing machine, but her name pops up in a few papers, but only at the introduction, perhaps she is the one putting the reports together.
You see, if the story is everything, we need to see the storytellers in their environment, because they are setting the stage on how we should see the information given to us. This is given to us from the very beginning as the headline in the executive summary tells us: “It would be impractical and prohibitively expensive to fulfill all of NHS’s analytical needs through more hiring“, which is exactly what I raised yesterday, in addition, I did mention that part as well, just the pathway that follows is not the pathway I trust. I believe that some skills should be managed inside the NHS. I do not trust the outsider telling me how it is, especially after consultants walked off with a large slice of £11.2 billion whilst the NHS has nothing to show for it. In addition, having outsiders access to NHS data is even scarier to me than losing those billions. You see, once the data is out there, the people whose data is out there get to be the victims of dangers they never signed up for. As I see it, once copied the NHS becomes useless to them and whomever walks away with that ends up diminishing the value of the NHS and those people even more. That is in my view a big no-no! (I am not accusing Govlab of having done anything immoral or questionable), I am merely raising the issue.
I notice that part of the paper is what we read yesterday in the Guardian. It makes it easier for me as I had already crossed off several issues on my list and I stand by these elements. Yet, the report is not all bad, it is illustrative in parts, but also suggestive and that lack of clarity is never part of a good paper. The reader tends to go from assumptions and that goes to either too positive or too negative, there is never ever a balanced in-between there.
When they give us: “The NHS should also create a variety of online knowledge networks of those inside and outside of NHS, especially in UK universities, who possess the skills and willingness to help the NHS with their data analytical questions. For example, last week the Rockefeller Foundation launched the Zilient platform to connect resilience practitioners, and the GovLab and Justice Management Institute launched DataJustice. Both are designed to connect networks of professionals for mutual learning“, which could be novel, were it not that I actually gave a similar (less eloquent) idea on June 29th 2014 in my article ‘What’s in a health system?‘ (at https://lawlordtobe.com/2014/06/29/whats-in-a-health-system/), where I state: “What if we take the next generation in solutions and take away 30% of that workload? When people ask which company will do this, the answer should be ‘None!’. The UK is filled with universities, some of them regarded as the most prestigious and brightest on the planet. Consider that most IT people, might claim experience, yet their drama skills are the only ones that improved for the most, is it not up to the Universities, those who are introduced to the newest ideas, design a solution that would make the work of the doctors and nurses at the NHS better, slightly more efficient and a truckload of less hassle! Is that such a tall order? We will get to the solutions if we are willing to navigate other options. We have seen that the current path is not a success; new methods might not be a failure. It is a road that politicians should be willing to go, if only to make sure that a possible solution was not overlooked“, I admit, not as eloquent, but pretty much the same story and it came whilst the NHS was 2.5 years further away from the ‘Abyss of non-existence’ it finds itself in front of at present.
So let’s take a look at the recommendations, and the first one gives us: “The NHS should build a Project Marketplace like the environmental protection agency’s one EPA Skills Marketplace and help supply find the demand“, so how do these outside talents not cost the NHS? Even as it hides behind ‘help with specific assignments, the skills marketplace helps to match talent to those opportunities to use it‘, you see, the more specific a need is, the more expensive such an expertise tends to get. The more generic and shallow, the less need for such a marketplace, more important, as a little more is asked on what needs to be done, the costs of these people will rise substantially and it tends to rise fast.
The second recommendation gives us “The NHS should construct an NHS Data Lab modelled on the Ministry of Justice Data Lab to make better use of sensitive administrative data“, which also sounds nice, but data labs cost, there is hardware, there is software and hiding it all in some cloud with questionable security will not be a solution, in addition, IT gets stretched even further and there is a difference between ‘better use of sensitive administrative data‘ and ‘safe and responsible use of sensitive administrative data‘, they are not the same, not by a long shot. In addition, I already made that point effective enough in the previous blog. The third one counts “The NHS should build an employee expert network like health and human services’ HHS Profiles and help the demand find the supply across the NHS“, which is what I partially raised when I opted for ‘low level‘ people in my previous blog. Yet this issue is correctly raised and if recommendation is started on yesterday by the NHS HR, than that would be an excellent idea as well. Adding a training path where some can learn skills like Q-Software, data cleaning skills and data capture skills, that might not be the worst idea either as it allows for an IT growth path. There are plenty of NHS volunteers, who are now retired, but still desire to be engaged, not everyone is doing that holding a set of people skills, some are happy to do other tasks. In addition there is the Employer-supported volunteering path, where Market Research and Data employees could spend a few hours a week working on the NHS systems, helping and teaching to create dashboards. In that HQIP Director Dr Danny Keenan might hold his hand up as high as possible to get started on the communication issues he currently states to have. Recommendations 4, 5 and 6 are basically skill finders in another dimension, so having that crossed off with HR to set a proper visibility path should solve all these issues. Option 7 gets us “bring in more talent from outside, including from the private sector“, which is a cost and not a small one either, that is depending on the needs there are. Recommendation 8 is the one that stopped me. You see ‘Open Data Learning Hub‘ I almost like I mentioned earlier, but here we see ‘data scientists to grow its data science community‘ with the added quote “Each toolkit includes the original data and step-by-step instructions for using the data to conduct sample analyses, create visualisations from the data and connect interested data scientists“, which makes me wonder, have they considered that the cost here will increase dramatically fast? In addition, how many data scientists are there now? If there are a few, why a training growth path was not initialised a few weeks after the first data scientist was hired. Because from some of the required trivial reports, those people are really expensive to use for making the ‘basic’ reports. Now, there is a part that I am not aware of, but recommendation 8 is leaving me with questions. If there are proper data scientists, why was growth not acted on long ago, if there are no Data scientists, that open learning hub will attract experts stating that ‘it is a complex issue‘ on too many projects, making this a marketing jump to hiring people, which is not the path we want to go through. Recommendation is just dangerous. As I stated, I have no faith in certain groups and ‘exchange data to help solve public problems. These collaborations focus both on sharing data but also on sharing talent‘, yet when certain ‘assisting‘ experts in the insurance world have been accessing the data sets, once they have the aggregated data they need, they will fall from vision like snowflakes in a summer sunshine, this recommendation is one that should be rejected as soon as the rejecter has read the words. It reads like Recommendation 10 is a potpourri of some of the previous recommendations, yet again we see “a “conversational” infrastructure to a secure physical infrastructure for managing data to tap the best know-how outside NHS on an ongoing basis“, which is pretty much an introduction towards hiring external consultants, which was a bad idea from the very beginning. The entire papers is followed by a score of issues, some I blew away in the previous blog article, some are there to (as I personally see it) illustrate fortune cookie wisdom, which is always debatable and will always be used out of context. An examples is: ‘We are achieving 1% of the potential to improve people’s lives‘. When we see ‘Healthcare data in England is collected, published and used by a variety of institutions, each of which has its own cadre of statisticians, analysts, and economists‘, there is the implied part, but the actual scope who is collecting what and who is collecting it for private organisations and insurances is equally left out there. As we see the groups we also see quotes like “Among its staff are approximately 150 people who hold the title of analyst for NHS England, including those in an operational research and evaluation unit created in 2014. NHS England publishes indicators on performance and satisfaction data about patient experience, bed availability, and wait times, and administers friends and family satisfaction testing“, yet when we see the group ‘Clinical Commissioning Groups‘, we get to see “We know there is lots of information on quality out there but don’t know if all NHS staff, such as clinicians, commissioners and service managers, are equipped to access and analyse it for both operational and improvement purposes, nor do we know if it meets people’s requirements, says one of NHS England’s clinical leads“, which beckons the question, did this clinical lead not know who to ask, what to ask and when to expect an answer? This reads like a ‘let’s set the premise of ignorance‘, whilst the systems in place would (read: should) have been there to inform the participants. This is not a paper on informing, this reads like a paper on creating doubt, preferably a paper pushing towards the recommendations regarding hiring and sharing, which is I admit is a speculation form my side. Yet it is strengthened when we read the premise with NHS Digital, where the language is phrased as “NHS Digital has a statutory monopoly over the collection of certain kinds of data, and over 300 professionals work on the collection and analysis” as well as “Of this group, approximately 250 are classified as analysts, whose work is focused on this routine and statutorily-mandated data collection and the publication of statistical reports“, whilst we get the implied accusation of “NHS Digital has not made much use of predictive modelling to evaluate innovations, conduct experiments or design new models of clinical care“. It seems that at New York University, the weight of a loss of £11.2 billion has no weight at all. That from a nation that has accumulated a debt of $20 trillion, should they be regarded as the experts? So when we read “no mandate—to translate that into policy recommendations nor to do research, according to Dr John Varlow, Director of Information Analysis at NHS Digital. In total, the organisation has 2000 professional staff, most of whom focus on information technology and the maintenance of NHS websites“, I need to wonder if that was the full part of the contribution of Dr John Varlow. Especially when we consider the work that maintenance of NHS websites requires. How many pages, what hardware infrastructure, more important, what other tasks are part of their workflow. An issue (as I personally see it) intentionally set outside of proper dimensionality. So when we consider the ‘produces over 250 annual official publications on topics from GP earnings to drug use among school children. These reports are accompanied by data in formatted Excel spreadsheets and machine-readable comma separated variable (csv) text files. Anonymised population-level data is available both on the NHS Digital website and on data.gov.uk‘. The reader should realise that this adds up to a little over 20 reports a month, that needs exporting, cleaning (personal data markers) set into final data forms and uploaded to the proper places. It should come down to one person does this full time, but the reality is that 1-2 other employees need to check this, to make sure identity sensitive data is not there. In addition we need to consider ‘NHS Digital publishes over a thousand indicators‘, which we can accept as a given, but based on whose recommendations? When is the policy to publish set through political means and requirements?
This gets me to the subtitle ‘The current emphasis on performance analytics needs to expand to predictive analytics, simulation and modelling‘, on what premise and whose budget? Whenever we see a statistician running around with a massive erection shouting ‘predictive analytics‘ and ‘modelling‘ it is usually because the linked implication that this additional work usually comes with a not to modest pay rise. When ‘simulation‘ is thrown into the mix, it tends to link to either resource needs, budgeting or failure analyses, which now implies that hiring becomes almost essential as these people were either hired from the start, or the skill set was not deemed essential, so raising it here raises a lot more than the reader will fathom. So when the writers decide to add Tim Kelsey on page 37, they should have considered the Guardian (at https://www.theguardian.com/society/2015/sep/17/cameron-adviser-leaves-controversial-nhs-data-scheme-private-sector), where we see: “Kelsey was appointed to the Cabinet Office in 2012 as the UK’s first transparency and open data director. However a few months later he joined the NHS, where he was the driving force behind care.data. The programme was supposed to be in place in 8,000 GP practices by 2014 but has been beset by controversies since its launch last year. It was finally put on hold after a series of blunders exposed serious problems relating to the confidentiality of patient information“, there we got the issues that the Guardian headlined with ‘NHS patient data to be made available for sale to drug and insurance firms‘ and ‘Privacy experts warn there will be no way for public to work out who has their medical records or how they are using it‘, which is actually an issue I raised more than once. So is Professor Noveck soliciting for data? Because there have been more than one marker in that paper to see as a consequence and the US in general is still massively horny for the UK NHS data sets. The report goes on for a total of 82 pages. In total it is actually a good sales pitch, but with my utterly paranoid mindset of once that data is no longer just with the NHS, the people relying on it will prefer to be dead than have to live through the aggravation that insurance companies will push them through. The page 78 mention of “the NHS could use this Brain Trust to expand its expertise outside its institutional borders. A pilot deployment would make it possible to determine empirically whether technology can bring expertise in from the outside efficiently” continues that path, because Brain Trust and expertise requires data access to set into a result. In addition ‘pilot deployment would make it possible to determine‘ tends to come with data attached, or accessible. I am not convinced that juice will be worth the squeeze and until the ghosts of data loss are set to rest (which is unlikely to ever happen), the entire sales pitch, no matter how good cloaked in 78 pages is one that the NHS needs to walk away from really fast. The PDF is at http://www.thegovlab.org/static/files/publications/nhs-health.pdf, so I will let you all decide how paranoid I am. The fact that the NHS did this with NYU and no critical report from, for example the University of Cambridge or the University of Westminster is actually an issue that a few others should make (just two of a collection of worthy investigative parties).
So is this ‘Smarter Health’ report healthy or smart? At present I fear it is neither!