Our last post talked about the Timing of messages, the first of the “Four-T’s” of effective patient messaging. In this post we cover “Targeting”, the second “T”, the most important of the bunch.
“Targeting” refers to using analytics to identify a specific group of members to receive a communication. If there is one “T” to focus on…this is the one. The return-on-investment in messaging is driven more by this targeting process than by any other single factor. There are two main risks of poor targeting:
- “False-Negatives”. This is when members are missed by targeting criteria. If we fail to identify the members that need the message, it’s kind of hard to imagine a successful program.
For example, if we intend to notify patients about the recall of Vioxx (a drug that was famously recalled in 2004), but we fail to identify all the patients on Vioxx, we are not messaging effectively.
At an absolute minimum, the system must be able to capture the patients that satisfy the message criteria.
- “False-Positives”. Almost as important as false-negatives is the flip-side – false-positives – patients that might look like the target group, but for some reason don’t qualify and shouldn’t get the message.
For example, if we intend to remind patients to get their annual mammogram (a STAR Rating metric), we might target women between ages 52 and 74. But what about the women within that group that have already had a mammogram this year? Or more importantly, what about those that have had a double-mastectomy? Those patients are false-positives and should be removed from our target list.
At best, messaging to false-positives is pointless. At worst, we risk offending or alienating the member. In any event, false-positives are viewed as “noise” and they will erode confidence in your messaging program.
Remember, the goal is to target every patient that meets the criteria for a specific intervention….and not a single patient more.
Now that we know what we want to avoid, let’s talk about how to do it. To understand what makes up a quality member targeting capability, look to the following key points:
- Integrated Data. Data siloes do not make for good member targeting. We want a wholistic data-view of the patient for optimal targeting. This means integrating demographic, pharmacy and medical data (and potentially other data sets – like lab values, HRA results, etc.).
The ability to target or exclude members using both pharmacy and medical criteria is very powerful, and is the only way to optimize targeting. Using a partial data-set (like pharmacy claims-only) can be effective, but when the patient’s clinical context is lost, targeting quality suffers.
For example, one of the HEDIS metrics calculates a health plan’s percentage of post-heart attack patients that are taking a beta-blocker medication (recommended therapy after a heart attack). The only way to target this population for messaging is to identify patients who’ve had a heart attack using clinical codes (ICD-10), and then identify the ones that are not filling their beta blocker prescriptions using pharmacy codes (NDC, GPI). An integrated data set, in this case, is a must.
- Flexible / Iterative Analytics. Flexibility is also key, and hard-coded or “canned” analytics can be a challenge to quality member targeting because targeting logic must be regularly updated. New clinical codes are published each year (and must be added to the targeting and exclusion criteria), and new drug codes enter the market on a daily-basis. At the same time, NCQA and CMS are updating their quality metrics every year. Analytics that can’t be frequently updated (ideally automatically), become less useful over time. In fact, they can get so stale that they eventually become a liability. Keeping up with all these changes is critical to quality, and a system that makes updates difficult is an impediment to the goal.
Also, the ability to “iterate” with these analytics is crucial. We say that “output drives the inputs”. Often, the volume or character of member targeting output will prompt the user to change or “tweak” the criteria to get a slightly different output.
For example, a review of a targeted patient list might reveal that a key patient exclusion was missed. Or perhaps the output is too large for a message campaign. The user will often tighten or change criteria to yield an output that’s closer to what they expected. For this process to be efficient and practical, the system must allow rapid iteration – or the ability for the user to run, adjust, and rerun the analysis until they get the output that works for them.
- Temporal Effect. Another thing to consider is the timing of the member targeting analysis. By this, we mean the effect of time on your targeting efforts. When member targeting criteria are applied to a population, the result is a “snapshot in time”. As soon as that snapshot is taken, it begins to lose fidelity. For optimal member targeting, the time between your data analysis, and your message, must be as small as possible.
For example, let’s say we targeted a list of members yesterday for a flu shot reminder (to be sent in one week). Things change. By the time that message goes out, some of the targeted group will likely be false-positives. Even in that short time, some will have disenrolled, some might opt out, and some will simply have already received the flu shot. Do you really want to ask them to do something they’ve already done? No, and a quality member-targeting system will allow you to re-run criteria on your targeted population immediately prior to messaging – to weed out those emerging false-positives. Minimizing false-positives is one of the most worthwhile efforts in messaging.
Remember, the ROI is found in the quality of member targeting. Target all the members who need a message…and not one member more. Can your messaging platform do that? Ours can….please contact us to learn more!
- November 2018
- July 2018
- April 2018
- March 2018
- February 2018
- November 2017
- October 2017
- September 2017
- August 2017
- July 2017
- June 2017
- May 2017
- April 2017
- March 2017
- December 2016
- November 2016
- October 2016
- September 2016
- August 2016
- July 2016
- June 2016
- May 2016
- April 2016
- March 2016
- February 2016
- January 2016
- December 2015
- November 2015
- October 2015
- September 2015
- August 2015
- July 2015
- June 2015
- May 2015 View More