Is this really that big a problem?
In the US, at least 50% of the population (133 million people) live with at least one chronic condition . Since lack of medication adherence invariably leads to problems such as unnecessary disease progression, disease complications, reduced functional abilities, a lower quality of life, and even premature death, poor adherence has been estimated to cost up to $177 billion annually in total direct and indirect health care costs. In a more recent report by the IMS Institute on avoidable healthcare costs in the U.S., it was demonstrated that six disease areas (congestive heart failure, HIV, osteoporosis, hypertension, diabetes and hypercholesterolemia) accounted for $105 billion in annual avoidable costs from non-adherence to medication treatment.
And this is not just a US only problem - the WHO estimates that, on a worldwide basis, only about 50 percent of patients typically take their medicines as prescribed.
A closer look at the cost-benefits of medication adherence
Multiple studies have confirmed the benefits of medication adherence including the ones listed earlier. A rigorous study was conducted recently looking at cost data by condition. That study calculated average annual medical savings, per patient annually, of $8, 881 in congestive heart failure, $4, 337 in hypertension, $4, 413 in diabetes, and $1, 860 in dyslipidemia. These savings were after offsetting for the increased pharmacy spend associated with being adherent; the increased pharmacy spend is because increased adherence means more medications purchased. This study showed that those who filled none of their discharge prescriptions within 120 days after myocardial infarction (MI) had an 80% increased chance of death and those who filled only some of their prescriptions had a 44% increased chance of death compared with those who filled the majority of their prescriptions.
All of which begs the questions - if the value (cost, health, life, quality of life) is clear, why does this problem exist at all? And why is this problem so incredibly prevalent?
Root causes of non-adherence
The causes of non-adherence are complex, and multifactorial. Non-adherence is broken down into two types:
- Primary non-adherence i.e. not initially filling the prescription written;
- Secondary non-adherence i.e. not taking as prescribed or not refilling the prescription.
Net net, the reasons for non-adherence can be traced back to multiple root causes but the ones that are addressable are as follows:
- Better patient education: Medication regimens for the more chronic diseases (which invariably present with additional co-morbidities) are complicated. A prescription for a complex MI patient could include aspirin, lisinopril, atenolol, atorvastatin, spironolactone, furosemide, potassium, sublingual nitroglycerin and clopidogrel. And these medications can, and often do, change. Expecting older patients and those with language barriers to understand, remember and take these medications on time (2 of this one, 1 of that one, and 1 of this other one but before lunch and the others after meals and at night) is unrealistic.
- Motivation or rather lack of immediate gratification / feedback: Drugs often don’t manifest an immediately visible benefit. For example, the use of antihypertensive agents in asymptomatic patients may be perceived by a patient as not providing benefit because they simply do not feel better as drugs often address an underlying problem and not necessarily the symptoms. Additionally, antihypertensives have side effects, many of which are immediately experienced by patients.
So how does one go about addressing this problem? The real challenge is the need for someone to have an “a-ha” moment. Multiple approaches have been tried as described below, with varying success. Even in cases where a certain approach has shown promise, the problem still seems to be one of user adoption and sustained adherence (there’s some irony for you).
Approaches to improve medication adherence
Data in healthcare is not, and to some extent cannot, be shared easily. This makes discovering patterns in data and mining data difficult since the data required to truly understand medication adherence comes from multiple sources and is in multiple formats. So analytics on medication adherence requires the following sets of data - each of which is owned by a different entity.
- The prescription: The entire flow is triggered when a physician writes an Rx (there are steps prior to this such a pre-auth etc. but is skipped for simplicity). See here for a summary view and here for a deep dive. These e-prescriptions flow from the EMR to the pharmacy via SureScripts.
- The prescription fill: A patient picks up an Rx. The Rx fill information is communicated at a summary level to SureScripts and the details are shared with Rx fill aggregators like Argus. Aggregators in turn share with this data with your health plan etc.
Given these two sets of disparate data, analytics is difficult but you can imagine triggers such as:
- Rx has not been filled after X days (of being written or procedure or discharge, etc.);
- Rx has not been refilled;
- Rx refill is due (based on Rx dosage and amount dispensed).
Unifying and analyzing this data is made more difficult because of the associated challenges of uniquely identifying an individual across these data domains. Unification of these data sources is doable, but it presents additional challenges.
Expect to see advances with big data analytics in this area. Health plans are ideally situated to make inroads here given their access to most of these datasets. Population heath management vendors are also investing a lot in analytics here as this is an area where all players (including the deep pocketed) are very interested in. This approach still needs a “last mile” solution i.e. a direct patient touchpoint which brings us to…
This approach, attempted several times, uses a combination of physical devices at a patient home, some form of wireless communication, and visual or messaging cues to remind patients, or remote caregivers, to take their medication. Vitality GlowCaps is the one that comes to mind right away. Their study indicated significant improvement in adherence. But (yes, there’s always a “but” in healthcare.. ) the challenge to a large extent here is that for this approach to work well, pharmacies must be fully involved in the process. Pharmacies need to enter the Rx data into the “alerting” backend provided. I’m sure this process can be automated but that data needs to then be synced with the caps, and we have to hope that the patient doesn’t mix up the caps. In the case of Vitality, the extra work required at the pharmacy was funded by pharma companies.
The newer entrant in this space is AdhereTech, which has been getting lots of good press. AdhereTech is an entire bottle, and not just the bottle top. It also uses messaging and automated phone calls increase adherence.
It would be interesting to see some combination of analytics and device that could leverage the data being created in the Rx and use that to programmatically update and keep the device in sync.
A device centric approach (in conjunction with communications initiated by the device or data captured by the device) can go a long way towards addressing the challenges around improving patient education and reminders. However, it still doesn’t (IMHO) address the core problem of instant feedback / gratification / motivation.
We all know that mobile will play a significant role in healthcare in the coming years. However, most apps in the various app stores are rudimentary (primarily in our opinion due to lack of knowledge, fear of compliance and lack of real / test data - all of which (shameless plug) are the reasons that we founded Datica. There are about 200 apps in the app store focused on filling prescriptions and another 225 apps focused on compliance/adherence. This is a pretty small number considering the potential benefits and value. Apps in this category include those provided by CVS, Walgreens, Medicine Reminders (HD), Medisafe, MedCoach and DoseCast. Even though the app revenue itself might not be significant, there are substantial revenue opportunities to be explored around data analytics, Rx refills, refill redirection to mail order and so on. Companies like Walgreens have already made mobile app acquisitions. Walgreens acquired RxMindMe.
This is an area waiting for a solution - one which addresses both reminders (with ideally minimal patient data entry) and motivation / gratification (gamification). This is an area that we at Catalyze are very interested in as well and are working to address developer challenges around compliance, data access (real and test data), and appropriate APIs. Look for more from us on this soon.
This is the holy grail. The technology involves the incorporating an ingestible sensor (about the size of grain of sand) into the pill (or whatever form). The patient then ingests it. The sensor can wirelessly communicate with a patch worn on the patient’s skin which, in turn sends the information out via wifi. The technology can definitively determine when medications have been ingested, and can provide this information wirelessly, securely and confidentially, to patients and designated caregivers, health providers, and researchers. This data can be used to address the motivation and the reminder aspects of adherence. Not to mention that it is actual consumption data (opening a cap or tapping a button is not the same as knowing the pill has been taken).
This data in conjunction with Rx data can go a very long way in making medication adherence an addressable problem. It does require pharmaceutical company buy in but given the value and the opportunity and the passive nature of the device and data collection, it should have significant uptake quickly. Company to watch? Proteus.
So, the summary?
Why do it? - Money: very conservatively, $105 billion problem * 0.01% = $10, 500, 000 (annual) - Social impact: population health improvement, reduced spend on healthcare emergencies and so on as described earlier
What technologies to explore? - Mobile: everyone has at least one of these devices. Also look to SMS as a communication mechanism - Data analytics: what blog post would be complete without at least one reference to big data? But, on a serious note, the problem is one of scale as well and drawing intelligent conclusions and predictive insights based on Rx, claims, fills, disease and compliance is challenging.
We, at Datica, are very interested and passionate about this area and do know a few things about the space and its associated challenges. If you have an idea and are looking to understand more about data, data source, app ideas, APIs, drop us a line.