The Barriers for Digital Health Startups

It’s over forty years since the first personal wireless telecare products came to market.  Over the years, along with many others, I’ve been writing about their potential and the opportunity they present to save healthcare costs and by extension, our healthcare systems.  Five years ago, many of us got excited when the Tricorder Prize was announced, with the promise of a Star Trek-like device that would diagnose multiple conditions being demonstrated by 2015.  That deadline has now slipped to 2017, but it’s not stopped a plethora of new healthcare devices being announced in the meantime, helped along by the twin vogues of crowdfunding and lifestyle.

So where are all of these digital health devices?  If you visit a hospital or GP, they’re mostly noticeable by their absence.  Startups are coming and going with ever greater rapidity, whilst healthcare costs grow relentlessly.   What is stopping digital health devices fulfilling their potential?  At the recent Future of Wireless International conference, I chaired a session with speakers from within the medical device community and working at the sharp end of healthcare, who shared their views about the challenges.  It was one of the most brutally honest and candid discussions I’ve come across, which deserves to be heard by anyone entering this market.  So here is a precis of their essential advice for any digital health startup.

I’ll split this into two articles, and start with a salutary lesson in health economics based on Shamus Husheer’s experience.  Shamus has been through a couple of digital health startups: his previous two looking at fertility – Temperature Concepts and Duofertility.  Both look excellent propositions on paper, as they target a point of desperation when couples are trying to get pregnant.  Shamus claims that Duofertility has a comparable success rate to traditional IVF, required no drugs and cost only 5% of the IVF alternative.  Yet despite getting into High Street pharmacies, and an encouraging boost in sales resulting in hundreds of pregnancies, systemic barriers remained.  The problem it hit, although it seemed to avoid the traditional valley of death within the digital health startup community, was that by the point couples needed fertility treatment they were desperate, and fearing they might only have one chance, chose the “tried and tested” route of IVF.  Saving money paled into insignificance against the fear of losing out.

That experience got Shamus thinking about the hard economics, which he shared as the three key lessons of digital health.  The conundrum he saw was that if you couldn’t make money from the discretionary spend of conception, then where could you turn a profit?  His first lesson is that:

Diagnostics are not economically important.

The perceived wisdom is that digital health is the best means of saving costs within healthcare.  But if you look at where money is spent, that wisdom rapidly looks dubious.  Taking the NHS as an example (because they helpfully publish real data about their costs), spending is not linear.  The brutal reality is that as we get older, it gets more expensive to keep us alive.  But keeping us alive against all odds is what we’ve come to expect from our healthcare systems.  The graph below shows hospital costs that were calculated by the Institute of Fiscal Studies from the NHS data, showing the average healthcare spend per person for each year of our lives.

There’s an early blip for babies, loaded by the relatively high costs of premature care, but then spending per person remains fairly flat until we reach our forties, other than a blip for pregnancy.  After that, things change.  From our fifties we start collecting long term chronic conditions; past seventy other parts of us start to go wrong, increasing the cost to the health system.  Within the NHS, around 80% of costs are attributable to the over sixties.  That’s a well-known figure.  What is not well-known is that 90% of those costs are associated with treatment, not diagnosis.  (In fact that 90% is probably an underestimate.  In recent years, the enthusiasm for “big ticket” diagnostics, using CT and MRI scanning has pushed up diagnostics costs, particularly in the US.  Take them out and treatment costs could account for up to 99% of the total.)

Regardless of the exact number, the important point to realise is that diagnostics are the minor cost of healthcare systems.  Whilst early diagnosis may help prevent subsequent treatment costs, what will save the NHS or any other healthcare system is not more or better diagnosis, but lower cost treatment, or a way of avoiding treatment altogether.  So all of the enthusiasm surrounding the Tricorder probably had more to do with us being latent Trekkies than anything to do with improving health.  Or to put it another way, if you’re concentrating on developing better diagnostic hardware, your startup may be doomed.  Here endeth the first lesson.

Hardware is not very profitable

Shamus’ second lesson is more practical.  It considers the cost of making and selling medical hardware.  He reiterated the basic economics of hardware manufacture.  If you make it for $1, you’ll probably sell it for $2 to a distributor, who will sell it on to a retailer for $4.  For this type of product, they’ll add a 50% margin, taking it to $8, which, with VAT added, gives it a retail price of $10.  Out of that, your gross profit will be just $1.  For mass market consumer medical products there a friction point in pricing that’s around $99.  Above that, it starts to become a specialist product, which means you will have a limited market, so you probably won’t sell a lot of them.  At $99, you may sell lots, but you’ll be making about $10 gross profit, which means you’re only making $2 of net profit on each device.

This is a problem.  Assuming you’re a VC funded startup, your VCs will expect to see you making a net profit of around $1 million a year once you’re up and shipping product.  If it’s less than that you won’t be an attractive acquisition for a larger medical company, and if you’re not an attractive acquisition for a large medical company you certainly won’t be an attractive investment proposition for a VC – you’ll be shown the door.  But getting to that $1 million is difficult.  With a $2 net profit per product, to reach it you’ll need to sell to half a million new patients each year.  Assuming you can attract 20% of the people who develop the condition you’re targeting, which would be very good going, you’d need to focus on a disease which is contracted by at least two and a half million people every year.  Choose a less popular disease and the funding numbers don’t add up.  There are around ten such diseases, but the diagnostic market for them is already sewn up by the large medical manufacturers.

Here Francis White came in.  Francis is the VP of sales and Business development for Alivecor – an innovative ECG which works with mobile phones and tablets and is a great example of how a digital health startup can succeed.  Before joining Alivecor, Francis was a Business Director for Medtronic – one of the world’s largest manufacturers of medical equipment, which meant he could also point out the reality of the status quo in digital health.  Go into any surgery or hospital and you’ll see Medtronic, or the name of one of the half dozen other major medical equipment suppliers, on almost every product.  Each of these companies has a marketing budget many times more than the total funding of any digital health startup.  In general, they don’t spend this on promoting disruptive new products, but on selling confidence based on familiarity with their existing product range, accompanied by gradual evolution.  To put it another way, the traditional digital health business is all about selling machines with a slightly better “ping”.  It’s not that these companies have any issues with innovative startups – they acquire plenty of them, but they rely on a very different business model, which is one of not shaking the boat, but selling incremental improvements.  It used to be the accepted wisdom in corporate culture that no-one was fired for buying IBM.  The PC was the death knell for that attitude.  But in healthcare, the adage still supports the business model of the behemoths, which is a challenge most startups don’t even realise they have.  As well as having to persuade healthcare providers that they can save them money, they need to persuade them to change a cosy mindset which has evolved to ensure that no procurement manager who takes the safe option will ever be sacked.  Put all of this together and the corollary is that you won’t make money from selling hardware to consumers, and it will be an uphill challenge trying to sell it to health services, as that market’s already sewn up.  There are exceptions – Alivecor is one of the few, and there are a few interesting companies which are sidestepping the barriers by targeting non-Western markets, such as Peek Vision.  But on the basis of this analysis, it seems that the only viable option for digital health startups is to concentrate on lowering treatment costs.

Lowering treatment costs

This brings us to Shamus’ third lesson, which is all about treatment, or rather compliance.  The current state of our health care systems is really quite impressive; we can treat practically every disease.  But most of the time, to do that successfully requires a contract with the patient.  They need to visit the doctor in time, take the drugs they’re prescribed, make appropriate lifestyle changes and report any changes in symptoms.  One of the reasons that we spend so much on healthcare is that the majority of patients don’t do these things, which means that the most important problem we need to solve is compliance.  For many conditions, the bulk of treatment costs are due to patients failing to follow their treatment schedule and requiring repeated, emergency readmissions to hospital.  For example, taking the case of heart failure, it’s thought that over three quarters of these readmissions are avoidable, but they currently cost the NHS around £1 billion every year, about 1% of their total budget, which is a significant, avoidable cost.

There are plenty of companies who are designing devices to help compliance, but the vast majority are things like smart pill boxes or wireless lids for medicine bottles.  They can certainly help, but they’re nudges rather than panaceas.  Nobody knows whether the patient actually takes the medication.  In my personal experience with my father, these devices got the pills out of the packet, but as many went down the side of the sofa as went down his throat – a situation I suspect is remarkably common.  Nor do these products really address lifestyle or symptom change.  They are useful, but only help those who are actively interested in managing their condition.  Shamus’ new company – Heartfelt Technologies, takes the logic of his business model analysis to the extreme by tackling chronic heart failure in patients who won’t take care of themselves.  Heart failure occurs when the heart is incapable of pumping enough blood around the body.  It can be compensated with drugs, but any lapse in the regime is likely to lead to an emergency hospital admission, which is expensive.  However, indications of change in condition can be detected around ten days before this happens through the appearance of several symptoms – shortage of breath, excessive tiredness, irregular heartbeat, a persistent cough and swelling of the legs and feet.  The problem is that a patient who is not taking their medication, not bothering to report their symptoms, or whose medication needs are changing, is heading inexorably for an emergency admission.

Heartfelt is taking the pessimistic view that there is nothing that anyone can do that will change the patient’s behaviour.  Many others have tried to over the years with little success, so Heartfelt are trying to detect change without any input or effort from the patient.  They’ve designed a passive 3D monitor that inspects a patient’s feet and ankles as they get in and out of bed, on the assumption that getting in and out of bed is one thing that a patient will continue to do.  Data goes to the cloud and any new swelling is detected.  That prompts a nurse or doctor to attend and adjust their medication, which should reduce the swelling within a few days, avoiding an emergency admission.  If it works, it could save the NHS around £1 billion a year.

There is something rather elegant, albeit extreme about the logic driving Shamus’ approach and only time will tell whether he is right to take such a cynical view of the business models.  Essentially, his analysis concludes that doing the apparently easy stuff is in fact far too difficult, and that the only logical way to develop a successful digital health company that will save a healthcare system money is to find the most difficult problem and make it easy.  Based on his own hard experience, it has a certain, inexorable logic which is very persuasive.

If we are going to make the prophecies about digital health becoming the saviour of healthcare come true, then we have to forget the siren call of technology.  To make a difference (and hopefully some money), entrepreneurs need to look beyond any altruistic drivers to the hard economics of healthcare.  Forget diagnostics and concentrate on treatment.  We probably already have enough technology to solve our treatment problems, but we will only succeed if we acknowledge that the problem is treatment, not diagnosis.  Look for the real costs and follow them, or to put it another way, work out what the problem is that we really need to solve.  At the same time, look at the equally large problem, which is getting it adopted.  I’ve concentrated mainly on one speaker so far.  The complimentary problem of adoption was discussed by two of our other speakers – Steve Feast and Keith Grimes, prompting a debate about whether the UK is a good or bad place for innovation and whether the NHS a good or bad customer.  It provided more fascinating insight.  But that’s for the next article.