Pediatric No Show Analysis, Part II

Brief recap: last week I shared some results from an analysis we did with Rexer Analytics. Our goal was to look through nearly a million independent pediatric visits to find patterns that might help us predict no-shows.


As is often the case with analyses like this, the absence of evidence is just as important as an expected proof.  For example, we determined that the day of week, time of day (within normal business hours), or month of the year have no measurable impact on no-show rates.  That's fascinating and helpful if you are trying to make changes to your office to improve your no-show rate.

Now it's time for the punch line, though.

By far, the strongest predictor of a patient missing an appointment is the patient having a history of missing appointments.

That may not sound revolutionary, but it should make us slow down for a second.

What the data shows us is that there is, in fact, a dividing line between your patients. Patients who historically have missed 9% or more of their visits are essentially 10x more likely to miss their next appointment than those who have missed fewer than 9% of their past appointments (24.7% no-show rate vs 2.6% no-show rate).  And patients who have missed < 2.6% of their past appointments are almost certain to come to their next appointment (only 0.4% miss their next appointment).

We really need to look at what this tells us carefully.  When thinking about no-shows, the most important thing to watch out for are patients who have previously missed several appointments.  If they haven't missed an appointment before, it’s unlikely that they will miss their appointment today. If they've missed appointments before, especially several times, don’t be surprised if they fail to show up for today’s appointment.  In fact, patients with the worst no-show histories are the most likely to no-show today:  if a patient has missed 30-40% of her past appointments, there’s a 45% chance she will miss her next appointment.  And among patents that have missed 60% or more of their past appointments, there’s a 89% chance they will miss their next appointment.  

I realize that every practice I've ever worked with knows this intuitively ("Oh, that family always no-shows..."), but now we have data to support the understanding.  More importantly, now we know exactly whom we should target with our no-show policies and procedures (more than that below).

What else did we learn?

  • - The Rexer team also used decision trees and other analytic techniques to drill in more deeply into the patterns of no-show rates.  If you are scheduling a patient who has missed more than 9% of his or her appointments in the past and the appointment is more than 3 days from now, you've now increased the odds of a missed appointment to almost 40%.

    Same-day appointments, however, are rarely missed.  Among patients with a previous history of keeping appointments, same-day appointments are almost a sure thing.  And even for patients with a lot of missed appointments, same-day appointments are rarely missed (~5%).  A ha!
  • Another obvious, but important, factor: missed appointments should be tracked by family.  If you have a family with two kids and a newborn, examining that newborn for a history of missed visits isn't going to help need to check the family's history.  It's almost perfectly correlated with the results for individual patients. We would be shocked if that weren't true!

    Oh, wait, your EHR doesn't look across an entire family to show you missed visits?  Perhaps you should think about a PEDIATRIC EHR. <cough cough>

    In the average pediatric practice, about 15% of families have a history of missing 9% or more of their appointments.  These families account for over 60% of the practice’s missed appointments.  Most of these missed appointments (over 80%) were scheduled 3 or more days in advance.  If we want to reduce missed appointments, this is the place to focus our efforts.

What can we conclude so far? A policy that focuses on your chronic no-show families, and only for appointments made 3 days or more in advance, would be the most effective and efficient.  We don't need to create a blanket policy for all of your patients, we only need to address a small sub-set of your practice.   I think this is one of the reasons that so many attempts to create no-show policies fail in pediatric offices - I often see very strict policies that are unevenly implemented, making it impossible to maintain the discipline required for success.  Our conversations with the practices we monitored here supported this fact: almost none of the practices felt like their policies were consistently supported or delivered.

What else did we find?
  • Same day appointments nearly eliminate missed appointments.

    No shocker, but doesn't this suggest that practices with high no-show rates might consider adopting walk-in hours?
  • For appointments scheduled in advance, other than a patient’s historical no-show rate, the biggest predictor of a no-show coverage. Patients with Medicaid or no insurance are 3x more likely to miss a non-same day appointment.

    If you manage a pediatric office, this doesn't surprise you, but now we have the data.  It’s good to quantify the magnitude of the problem.

    Missed appointment history is still the most powerful predictor of a future missed appointment, with 48% of the predictive power in our regression analyses.
  • One piece of evidence I didn't like learning was the impact on what I call "appointment depth" on missed visit rates. In other words, does it matter how long ago an appointment was made? And the answer is...yes.  But the results are interesting.  Essentially, the missed visit rate stays relatively even for appointments that are made more than a week out up through about 6 months.  There's even a little dip in the 1-3 month section.  Once we cross the 6 month threshold, the odds of an appointment being missed essentially double.  

    Now, in our sample, those appointments made up 1% of the total volume, so there isn't a large impact.  And we didn't isolate any other practice characteristics (reminders, policy) that might affect that group.  Something for later!

There you have it.  At this point, we should move into a discussion about what to do with this information?

  • Should (can?) families with poor no-show rates be limited to same-day visits?  How do we manage well visits for them, if so?
  • Perhaps chronic no-show families should have their visits limited to certain times and days, where they will least affect your practice?
  • How does this information affect the interest or ability to deliver walk-in hours?
  • If we can identify "no-show" families, does it make it sense to double-book them? 
  • What data should my fair readers examine to support or refute this analysis?  Anyone able to confirm the lack of day/time/month bias, for example?

For our next trick, I hope to work with a few practices to apply some scientific method to see what we can do to drive no-show rates down.