The More You Know, The More You Can Help Patients

Many people suffer from a plethora of undiagnosed conditions – meaning they are suffering in vain and could be entering onto a slippery slope toward poor overall health.

Introduction

Of the 29.1 million people with diabetes in the United States, 8.1 million of them are undiagnosed. That’s a lot of people who are unknowingly heading toward potential complications, such as nerve damage, kidney damage, eye damage, foot damage, hearing impairment, skin conditions, Alzheimer’s disease – just to name a few. In fact, diabetes remains the 7th leading cause of death in the United States, with nearly 70,000 death certificates listing it as the underlying cause of death, and a total of over 234,000 death certificates listing diabetes as an underlying or contributing cause of death.1 Therefore, the importance of identifying and treating the condition early cannot be overstated.

This is the potential fallout from just one undiagnosed condition. The unfortunate fact of the matter is, however, that many people suffer from a plethora of undiagnosed conditions – meaning they are suffering in vain and could be entering onto a slippery slope toward poor overall health. Undiagnosed patients are individuals who either have the condition but have not had it detected (often patients who are overdue for a physical exam or laboratory work), or those who do not have the condition yet but are trending towards it. Patients often are not diagnosed with conditions due to the absence of overt symptoms, the lack of awareness about the disease or the rarity of the specific condition.

Discovering these undiagnosed patients could result in a variety of benevolent and bottom-line advantages. Of course, getting treatment to patients early on could help to curb the progression of many diseases and conditions – and it could also help to create healthier populations. What’s more, informing physicians about these undiagnosed patients would create a financial windfall for them and for life sciences organizations. Conversely, not knowing about these undiagnosed patients could make it difficult for healthcare providers and life sciences companies to reach financial goals.

This data can be turned into “patient personas” that represent new and unique patient segments and their propensity toward certain undiagnosed conditions.

The big question

How can a life sciences company identify these undiagnosed patients when they are typically hidden from view through conventional means? To that end, here are some concrete steps that life sciences companies can take to uncover previously undiagnosed patients.

#1: Collect a broad array of data.

To identify undiagnosed patients, life sciences companies need to understand patient populations. While clinical data is important in this pursuit, so too is socioeconomic and demographic data, such as income, education, spending patterns, ethnicity, home values, age and gender. EXL draws upon the following data types to better understand patients – and their propensities for various diseases:

  • Behaviors & attitudes
  • Demographic & attribution
  • Clinical factors
  • Cost & quality
  • Utilization

#2: Crunch data to create patient personas.

By leveraging predictive and prescriptive analytics methodologies, this data can be turned into“patient personas”that represent new and unique patient segments and their propensity toward certain undiagnosed conditions. In addition, these analytics can be applied to identify patients who may not be classified as having a condition such as diabetes or heart disease yet, but are trending (or likely to trend) in that direction.

These patient personas include clinical and financial risk scores that can potentially help to identify each patient’s propensity for chronic and other conditions. Indeed, with this information, life sciences companies can create a variety of patient personas that can be used to predict the likelihood of future health utilization. For example, some patients can be categorized as “balanced adults.” Patients in this category are well managed and middle-aged adults – primarily female, predominantly with a college-level education and white-collar employees with medium incomes. They are a healthy population with low utilization and per member per month (PMPM) costs, which is aligned to an average number of chronic conditions. Because they are low risk and well managed individuals, there is no need to allocate costly resources on them.

With other personas, however, the risk is much greater. The chart below illustrates a variety of personas and their relative health risks:

It can be very valuable to the life sciences organization to understand their current patient population, their treated patient population and possibly their competitor’s patient population.

#3: Match the patient personas with demographic information about patients in a specific area, the area that a physician or hospital serves, to demonstrate which conditions are likely to be prevalent.

It can be very valuable to the life sciences organization to understand their current patient population, their treated patient population and possibly their competitor’s patient population in an area that a physician or hospital serves, to demonstrate which conditions are likely to be prevalent. This information can enable life sciences companies to address questions such as: Who are these folks? Where are more undiagnosed patients with similar attributes located in the United States down to the zip-code and physician level? And then, how can that information be used with healthcare providers in terms of messaging in the field?

To make data more actionable, healthcare organizations need to determine if patients in specific populations share the same demographic attributes, potential clinical factors, behaviors and attitudes as patients who are suffering from a specific disease. This “propensity matching” enables healthcare organizations to understand the undiagnosed or under-treated patient populations. For example, an area that has a high number of low-income people is more likely to have a high prevalence of diabetes, as access to healthy food is often limited In essence, life sciences companies need to examine medical claims data-based patient personas to identify current patient populations, then augment that data with socioeconomic and eligibility data, as well as newer third party data, such as daily fitness data, to determine if these patients have a propensity for a certain disease or condition, and how active or non-active they are in potentially treating it on their own.

Propensity matching enables healthcare organizations to understand the undiagnosed or under-treated patient populations.

#4: Further leverage the patient personas to understand just how likely it is that the undiagnosed patients will be to accept and act upon the medical advice received from providers – or how “impactable” these patients actually are.

Data can also be leveraged to determine how compliant or noncompliant various personas are likely to be. Indeed, after identifying patients who have a propensity for a certain disease or condition, the next question becomes: Is itpossibletohaveasignificantimpactonthesepatients? What is the likelihood that the potential patients will listen to the messaging – and more important act upon it? For example, when working with patients who show a propensity for diabetes, socioeconomic data might reveal that the patients are open to changing their diets and lifestyles to help curb the progression of the disease. However, if the potential patients live in a food desert with no access to healthy food options, it’s unlikely that they will be able to act upon this potential intervention.

After identifying where potential patients might be, it’s important for life sciences companies to work directly with providers to help identify these patients.

#5: Work with providers to find undiagnosed patients.

After identifying where potential patients might be, it’s important for life sciences companies to work directly with providers to help identify these patients. For example, if a certain area is expected to be home to a large number of diabetic patients, life science companies might want to encourage physicians and other care providers to perform glucose tests in an effort to identify diabetics.

So, instead of sending a blanket message to providers across the United States, life sciences companies would send specific, relevant messages to targeted providers in certain areas. These more sophisticated messages could point directly toward working with future patient populations in an effort to generate more patient starts. In essence, the messaging would be tied to the understanding of the various patient personas that are prevalent within certain zip codes. In addition, the messaging could also take into account if the potential patients will be receptive to various treatment options.

The risk score is an indication of the likelihood of these patients total cost of care increasing, decreasing or staying the same over the next 12 months.

Case Study: Anti-Opiate Abuse Medication

Discovering undiagnosed patients became a pressing priority for a large life sciences company in the United States when the company developed a new anti-opiate abuse medication. The challenge for the company was to market this product to as many potential patients as possible.

As such, when the life sciences company came to EXL, leaders had three objectives. First, they wanted to risk stratify and better understand the current patient population of high opiate user patients. Second, they wanted to identify patients who are showing up on the claim form as having high opiate use. Third, the life sciences company wanted to understand the risk stratified universe and the undiagnosed patient population and share this information with payers.

To satisfy these objectives, EXL mined data within its rich claims databases and built a“market basket”around patients who are high opiate users. More specifically, EXL collected data on about 5,000 patients and then examined the comorbid conditions of these patients, such as mental/behavioral disorders, back pain, hypertension, asthma, cardiovascular, Hepatitis C, eye disorder, ear disorder, or depression.

For each of these comorbidities, a risk score was generated. The risk score is an indication of the likelihood of these patients total cost of care increasing, decreasing or staying the same over the next 12 months. The risk score of an average patient across the health care spectrum is generally one. The average risk score of an opioid population is 2. The average risk score, however, for patients with comorbidities grew to 3.5 (see chart).

By leveraging socioeconomic, clinical, and risk attribute data, EXL identified the undiagnosed patient population down to the zip-code level.

Case Study: Anti-Opiate Abuse Medication

  • Average risk for general patient population is 1.00
  • Average risk for Opioid patient population
    is double to 2.00
    • This forecasts that patients to spend 100% more in the next 12 months
  • Average risk for this comorbid patient population is significantly higher @ 3.5!
    • This forecasts that patients will spend 350% more in the next 12 months


As such, the patient’s average cost of care for the next twelve months would be estimated at a thousand dollars. With a risk score of 3.5, however, the cost of care would be estimated to be about 350 times higher. Some of the reasons that the cost of care incurred such increases included lack of office visits, lack of medication compliance, and lack of lab visits.

Once EXL began to understand the reasons for these increases, attention turned to the current patient population. By leveraging socioeconomic, clinical, and risk attribute data, EXL identified the undiagnosed patient population down to the zip-code level.

The analysis shows the state of opioid use in today’s universe, what opioid use might look like in the future and the likelihood of specific patient populations being compliant with medication recommendations.

Top Core Based Statistical Area by Prospective Risk

Knowing where opioid patients by risk are located is just the beginning though. EXL then used data analytics to examine the compliance of the current treated population, categorizing patients as either opiate dependent-compliant or opiate dependent-noncompliant. As such, the life sciences company could then determine the chance of the undiagnosed patients being compliant or noncompliant when treatment options are presented to them.

Such information becomes valuable as the company strives to expand its marketing efforts. In essence, the analysis shows the state of opioid use in today’s universe, what opioid use might look like in the future and – perhaps most importantly -- the likelihood of specific patient populations being compliant with medication recommendations that they receive from their care providers. More specifically, the analysis enables the life science company to answer pertinent questions such as:

  • Where do the undiagnosed patients exist – by core-based statistical area (CBSA)?
  • What do these patients look like?
  • Why do these patients act the way do?
  • Will these patients be receptive to the life sciences company’s messaging regarding the importance the use of anti-opiate abuse medication?

Life sciences companies are helping to increase quality of care and reduce medical costs across the board.

Using data to answer such questions put the life sciences company in a position to better market their products by:

  • Identifying where highest at-risk patients are geographically.
  • Aligning regional marketing program(s) around care plan messaging.
  • Strategically identifying where to focus co-pay assist programs.
  • Understanding untreated population and how to gain access.
  • Making it possible to support co-promotion partner alignments.
  • Developing effective brand strategy and accurate forecasting.
  • Adding insight to messages to healthcare providers on patient personas with greatest risk.
  • Adding a new measure of value on the healthcare providers with the most at-risk patients, and the healthcare providers with the most personas that fit the brand’s target audience.
  • Promoting a general awareness and understanding of regional (zip- code level) variations and heat mapping.

Indeed, such analysis helps life sciences companies more effectively market their products to reach larger patient populations. The effort is not entirely self-servicing though. By identifying these patients and getting them needed treatments early on in the disease cycle, life sciences companies are helping to increase quality of care and reduce medical costs across the board. Indeed, it’s a win-win-win – for life sciences companies, providers and, most importantly, patients.


Written by EXL Health Team

Contact US