Real World Evidence & Market Access Summit USA

Dec 3, 2015 - Dec 4, 2015, Philadelphia

Leverage Real Life Data & Analytics for Value-based Market Access

Rules-Based Data Analysis and Predictive Modelling: Stemming the Tide of Non-Adherence

Getting ahead of the problem is key when it comes to tackling non-adherence.



Express Scripts, which boasts around 85 million patient-members across the US, is the largest Prescription Benefit Manager (PBM) in the United States and is ranked #22 on the Fortune 25 list. Its Senior Director of Advanced Analytics, David Tomala, tells us about how PBMs use rules-based data analysis and predictive modelling to prevent drug waste, prescription drug abuse, and promote patient medical adherence.

Prescription benefit managers

In the US, medical and health insurance companies absorb healthcare costs for patients. Health insurance covers the cost of most diagnostic consultations and examinations, surgery, hospitalization and, often, the cost of prescription drugs. Most insurance policies provide what is called a “prescription benefit” – a fixed amount that can be used to reimburse patients for purchasing prescription drugs. When patients are prescribed medications, they usually pay using the prescription benefit under their health or medical insurance.

Since it is costly for insurance companies to pay for experts who can determine if claims for prescription benefits should be settled or rejected, PBMs have been created solely for the task of processing, evaluating and settling prescription benefit claims. Insurance companies can then enroll their health policy holders in PBMs.

The country spends about $300 billion USD yearly on prescription drugs. I always find it fascinating that, essentially, the same amount spent on prescription drugs is roughly equivalent to the money that the country wastes on patients not taking those prescription drugs.

PBMs use rules based on published research evidence to create a formulary or list of prescription drugs that are covered by insurance policies. When a patient is prescribed medication for a condition that is covered by the insurance, patients who are members of a PBM, can have their prescriptions filled by the PBM or through pharmacies affiliated with the PBM. Patient-members can always purchase medication from a local drugstore, but if they wish for the purchase to be paid for by their insurer, they will necessarily have to obtain them through PBMs - who can provide the drugs at lower cost for their members; PBMs buy drugs in bulk for a lower negotiated price.

Solving a problem

In 2013, the US healthcare insurance system spent $317.4 billion USD to treat complications attributable to medical non-adherence. Tomala says, “The country spends about $300 billion USD yearly on prescription drugs. I always find it fascinating that, essentially, the same amount spent on prescription drugs is roughly equivalent to the money that the country wastes on patients not taking those prescription drugs.”

“In layman’s terms,” he explains, “when a patient is prescribed anti-diabetic medication and the patient fails to take the medication, it’s not just expensive in the short-term for the patient and his insurance carrier.  It has wider ramifications on the patient’s health and quality of life, as the patient will need eye surgery or leg surgery in the future.”

The solution to this problem could be rules-based analytics and predictive modelling.

Benefits of rules-based analytics

PBMs, such as Express Scripts, gather large amounts of transactional data, including information on how members get their prescriptions filled, medical billing data, and recorded telephone conversations between members and PBM representatives. Electronic communications on the website, including streams of clicks and messages, are also gathered. All this data is analyzed to determine what medications members are being prescribed, how often they take those medications, which new medications may interact with other medications already prescribed to their patients, and when patients don’t take those medications. The analysis of this data relies on rules-based analytics.

“As I understand it,” says Tomala, “rules-based analytics is based on simple rules such as: don’t dispense a diet pill with high blood pressure medication.  These rules are based on clinical research evidence, and are useful in handling queries as to contraindications of drugs, drug interactions, and which drugs on the formulary shouldn’t be filled.”

Rules-based analytics employs clinical tools, such as scales and questionnaires, to reduce inappropriate prescribing and medication errors that result in adverse reactions. It is also used to improve adherence to medication and thus improves patient outcomes. On the whole, Tomala observes, “It is helpful, but it is also limiting as it doesn’t help generate new knowledge.” For instance, “Rules-based analytics can enable a computer system to label a patient-member as ‘late to fill’ when the patient was dispensed 30 pills for high blood pressure and at the end of 30 days failed to ask for a re-fill,” Tomala adds. The company that uses a purely rules-based analytics scheme will then call that individual patient to remind them to have their prescriptions re-filled.

Predictive modelling is 9.8 times more accurate than patient self-reports about their own medication adherence levels.

However, rules-based data analytics won’t allow systems to probe causes for medical non-adherence, nor will it initiate interventions to prevent non-adherence from occurring or from becoming a habit. To generate this kind of knowledge, Express Scripts uses data mining and predictive modelling. Under this scheme, PBMs can determine why patients haven’t come in for a refill. They probe for barriers such as cost, or concerns such as potential risk of harm associated with the drug or confusion as to the dosing regimen.

Benefits of predictive modelling

“Predictive modelling more accurately identifies patients at risk of long-term non-adherence and intervenes,” says Tomala. “From the time that the patient is prescribed medication, all the available data is used to predict the chances that a patient will fail to take the medication or fail to get a re-fill.” Unlike rules-based analytics, which needs about three to six months to observe non-adherence, predictive modelling assesses a patient’s risk of non-adherence and tailors specific interventions to promote adherence. Tomala adds, “Predictive modelling is 9.8 times more accurate than patient self-reports about their own medication adherence levels.”

Through motivational interviewing techniques, they initiate a discussion on what may be troubling patients about taking their medication.

Predictive modelling is most useful for drug utilization reviews. Medical and drug data analysis using predictive modelling allows PBMs to efficiently manage the drug plans of health insurance carriers, which ultimately benefits patients. Tomala explains that predictive modelling allows pharmacists to assist patients in their medical adherence by allowing them to make medical adherence a habit. “The proactive interventions made possible by predictive modelling encourages muscle memory to develop in taking medication,” he says.

In other words, predictive modelling identifies the targets to go after – but that’s really all it does.  It’s still up to the specialist pharmacist to connect with that individual and realize behaviour change. For instance, the online pharmacy of Express Scripts won’t call patients on the 31st day to remind them to have their prescriptions refilled and to ask why they haven’t yet come for a refill. “This triggers a defensive mechanism in patients, which are roadblocks to the discussion,” he explains.

“Predictive modelling gets ahead of problems,” Tomala explains. “At or around the time when the patient first makes contact to have a prescription filled, a trained pharmacist from Express Scripts will call the patient and explain the risks and benefits of taking the medication, and the possible side-effects and symptoms of adverse reactions. Through motivational interviewing techniques, they initiate a discussion on what may be troubling patients about taking their medication.”

Another case in point for the benefit of using predictive modelling is for patients who have been identified as having mobility or transportation issues: filling prescriptions and delivering those prescriptions every 90 days ensures more continuous adherence than if prescriptions are filled and delivered every 30 days.  The patient will only have to ask for a re-fill four times instead of 12 times in a year.

Tomala agrees that, while pharmaceutical companies already implement a mix of rules-based analytics and predictive modelling, a wider use of unstructured data analysis in real-time would be helpful in determining the existence of “pill mills” - sources of adulterated drugs which are sold in the black market.

PBMs benefit pharma

In theory, prescription benefit managers cultivate relationships with stakeholders in the pharmaceutical industry, ensuring the procurement of drugs at a lower cost for their members. Procuring drugs at a lower cost may, at first glance, seem to work against the interests of the pharma industry, which struggles to cope with the high cost of developing, trialing and marketing new drugs.

However, PBMs, through their data analyses of members’ drug adherence behavior, can make it easier for the pharma industry to establish trust and goodwill with its consumers. The analysis of member behavior, by PBMs, also allows pharma to implement pharmacovigilance - monitoring adverse drug reactions across whole populations inevitably reduces the cost of bad publicity associated with adverse drug reactions.

In the end, PBMs work to make drugs more easily accessible to their patient-members.  The large number of patients enrolled in PBMs by their insurance carriers ensures medical adherence - patients regularly purchase medications in bulk, albeit, at a lower cost.

The primary interest of PBMs is to make low-cost drugs available to their patient-members in order to make the most efficient use of insurance funds.  Subsequently, PBMs ensure that they use both rules-based analytics and predictive modelling techniques to ensure that patient-members make the best choice possible for their healthcare needs.

Promoting drug adherence and compliance doesn’t just promote the best possible health outcomes for patients, and savings for insurance carriers in terms of less money spent to treat complications resulting from non-adherence. For pharma, it promotes the safe delivery of medicines to consumers, cultivating a culture of openness, transparency, and social relevance, thereby increasing trust and goodwill among all concerned.


David Tomala will be presenting alongside 50+ industry thought leaders at the Real World Evidence & Market Access Summit, December 3-4 Philadelphia. To download the brochure, click here.

 



Real World Evidence & Market Access Summit USA

Dec 3, 2015 - Dec 4, 2015, Philadelphia

Leverage Real Life Data & Analytics for Value-based Market Access