When you buy a new smartphone and it doesn't work, you can return it and it will be replaced free of charge. A new car comes with some sort of comprehensive warranty that covers the parts. Both these scenarios strive to demonstrate a simple concept: buyers need protection against goods and services that do not work as advertised. Otherwise, unscrupulous individuals might run around unfettered to sell goods and services without having ensured that they actually work or without having a mechanism in place to ensure that, if they don't work, the consumer is appropriately compensated with replacement goods or a refund.

So how come this doesn't apply to healthcare? Or more precisely, how come this doesn't apply to the drugs and devices that we use? Let's leave aside the 'services' part of the equation and just focus on the 'goods'. How come the buyer doesn't get a replacement product if it didn't work the first time? That's easy. If I took my Crestor as prescribed and it didn't work the first time, chances are very likely that it won't work the second time either. But how come the healthcare buyer doesn't get a refund if the product he buys doesn't work? A patient takes his statin and his high cholesterol simply isn't controlled. Who comes forward with a chequebook in hand and calculates the amount spent out-of-pocket by the patient and the amount covered by private or public insurance and then proceeds to write a cheque? Shouldn't someone stand behind this product and the science?

Part of the answer as to the 'why doesn't this apply to healthcare' question lies in the fact that healthcare is uncertain and part of this uncertainty has nothing to do with the product itself. Despite the best designed RCTs, the plethora of real world experience, the input of peers and the development of evidence-based guidelines, the truth is that sometimes we simply don't know if an intervention is going to work. Does this matter? Well it does if the product doesn't work because of a factor unrelated to the product itself. But it's exceedingly hard to tease out lack of efficacy due only to the product or only to something else.

In other words, part of the answer lies in the fact that sometimes it's not the product but the 'consumer'. If you use your smartphone in the bathtub, you are unlikely to have it refunded. If you drive your car with the handbrake on, you are unlikely to have the damages refunded under the manufacturer's warranty. If you don't take your medication as prescribed, it is unlikely to deliver the intended efficacy and you shouldn't be in line for a refund.

Healthcare is uncertain and part of this uncertainty has nothing to do with the product

But what does 'don't take your medication as prescribed really mean'? This is where it gets complicated. If a patient is non-adherent to the medication because of intolerable adverse events, that is an acceptable reason to give the buyer his money back. We simply cannot ask people to put up with adverse events that reduce their quality of life for the sake of not wanting to give someone a credit note - especially if the outcomes are suboptimal. If the patient is non-adherent by virtue of his personality traits (ie he's forgetful) or ignorance (ie he simply doesn't believe he's got high blood pressure and he flushes his pills down the toilet), then he should not receive any refund. How do you know which is which? You don't. But we know that clinicians can tell at the point of care when they don't see any clinical improvement.

Look, this column does not provide the space or length to go into all the scenarios. It is meant to serve as the framework for an idea. We all understand that a 'refund' model will not be perfect. We all understand that some buyers will get refunds when they shouldn't and others will not get refunds when they should. We all understand that we can come up with at least 10 'outlier' scenarios that poke holes in the 'refund' model. We all understand that some clinicians at the point of care may get fooled by patients' reasons for discontinuation of therapy. But here's the upshot: with new therapies on the horizon that are aiming to cost upwards of $14,000 per year to treat high cholesterol and therapies already here that cost six figures to treat cancers and hepatitis C, the model has to change. We can no longer afford to simply shrug our shoulders when treatments don't work and continue to pay for them. And I don't think manufacturers would want us to either. We need a starting point - no matter how imperfect - that strives to balance the cost burden between those who benefit, those who pay and those who produce.