Q&A: Sham Sokka on how radiologists can leverage AI to minimize patient no-shows

Sham Sokka, PhD, has spent the bulk of his career in radiology, where he’s worked for 15 years with a range of clients to shape and customize imaging modalities, workflows and software. Now the vice president and head of Radiology Solutions at Philips, he sees firsthand how patient no-shows can affect a radiology practice—especially in a field where quality and efficiency have been key to cutting costs and improving service.

Sokka, whose research has led him to believe artificial intelligence could be a potential solution to the no-show problem, answered some of Radiology Business’s questions on the topic.

Radiology Business: Why are no-show volumes of concern in radiology? What’s different about the specialty that makes appointment failure particularly burdensome?

Sham Sokka, PhD: One of the main concerns is simply cost and efficiency. Imaging exams are very expensive. In some cases, they can be $600 or more per exam. This is much costlier than other medical appointments, and radiology patients demonstrate different predictors of no-show occurrence than patients in a primary care setting. This is due in part to the referral nature of the radiology business. Accordingly, the lost revenue opportunity of patient no-shows in terms of workflow and equipment utilization for radiology departments has significant financial impact for healthcare providers. 

The second concern is around patient care. Patients who are asked to undergo imaging exams are suspected to have a potentially serious issue or disease. In these cases, the imaging exam will help us see what’s happening or not happening with the patient more clearly, so we can pinpoint the problem and take action to correct the ailment. When patients don’t show up, it delays a definitive diagnosis and possibly life-saving treatment. So, it’s important from a clinical standpoint and population health view as well to make sure we reduce the overall no-show rate. 

What’s the current state of no-shows across U.S. radiology practices? Are those rates similar elsewhere?

Appointments for which patients fail to attend without prior notice represent a widespread problem in radiology. We’ve done in-depth research, looking at no-show appointments for more than eight million radiology patient visits over a 15-year period in a multi-hospital setting. Overall, no-show rates are about 8-9 percent on average. However, anecdotal evidence tells us that rates can be much higher—between 20 to 30 percent—and, in some cases, as high as 40 percent.

What’s your insight into the main reasons for patient no-shows? How do missed appointments affect workflow in radiology departments? 

Based on research in this area, we found a wide variety of reasons for patient no-shows, ranging from cost or lack of insurance to fear of the imaging exam itself to simply not getting a reminder notice or not having transportation to the imaging center.  

There’s a very interesting and significant connection between patient no-show and workflow. We’ve found that one of the biggest correlations for patient no-shows can be attributed to scheduling lead time for appointments. If the appointment is scheduled for less than two weeks away, patients are more likely to show up. But if the appointment is booked for more than two weeks away, patients are less likely to show up. 

When patients don’t show up for appointments, it takes away an opportunity for another patient to have an exam done that day, and it puts the patient who “no-shows” back in the queue, thereby extending the waitlist for other patients. Essentially, there’s a problematic ripple effect. 

Patient no-shows can contribute to longer lead times for appointments, which we know is a factor for patients not showing up in the first place or choosing to go to a different hospital or imaging center for the exam.

How can no-show volumes be monitored across practices and imaging centers?

Healthcare informatics and radiology information systems can track which patients didn’t show up for appointments, and multi-site institutions can determine if the reason for the no-show was that they were double booked at another site. Using these tools, healthcare providers can identify what types of imaging exams, CT versus MR for example, have a higher rate of no-shows. They can also track how many patient no-shows they had each month or year, and easily calculate the loss of revenue or opportunity cost. But, beyond that, it’s difficult to find out why patients didn’t show up for appointments. There is no systematic way to capture that type of data easily, and that is probably the most telling factor.

How can AI contribute to closing the gap?

This is where artificial intelligence can really make a big difference, because it helps us narrow down the variables we know contribute to a patient not showing up for an appointment. Hospital scheduling procedure characteristics and patient demographics are shown to be important predictors of compliance. However, there are some other subtle nuances as well. For example, we know patients who are over 65 years of age and may be retired are more likely to keep their appointments than younger patients who are often working during the day. Other factors, such as repeat patients versus first-timers, income, insurance or even the distance of the person’s home from the hospital or imaging center play a factor as well. 

The important thing to realize is that individually these variables or predictors are not enough to effectively predict patient no-show rates, but collectively they can help us do a much better job. This is where AI is most helpful. Based on the variables and other types of data such as age, gender or demographics, AI can help us predict how likely it is that a particular patient will show up for an appointment. With artificial intelligence, we can close the gap by aggregating large volumes of retrospective data, demographic data, census data and constructing quantitative models to both predict no-show occurrence and highlight features that are informative in the prediction.

What kinds of future solutions could you see addressing high no-show rates in radiology? 

The first step is always getting the operational intelligence by investigating what the extent and implication of the problem is in that particular hospital or imaging center, because while patient no-shows are a common problem, the root cause and solution for each hospital is often slightly different. 

That being said, we think there are a few general areas, such as scheduling models and reminder systems, that radiology departments can focus on to increase patient compliance. Scheduling models can make the process itself more automated and assign appointments to patients based on risk level. In this way, patients with a high risk of not showing up can be scheduled later in the day when it would have less impact on workflow and other patients if they don’t show up for their appointment. We know reminders work in general, but different types of reminders may be needed for different patients. For example, many older patients want a call from a person and younger people usually respond better to text messages. 

Additionally, any efforts to improve the overall patient experience of imaging are helpful. For instance, accessibility is a big factor in patients showing up for appointments. Efforts to reduce anxiety or stress about imaging exams are also essential and include keeping patients informed and assured of their comfort and safety, communicating in advance what the procedure will be like and helping them prepare mentally and physically for the process. 

We’ll see solutions continue to advance in this area as it’s a problem worth solving. Radiology administrators are taking the patient no-show rate seriously and want to mitigate the impact on their department.

""

After graduating from Indiana University-Bloomington with a bachelor’s in journalism, Anicka joined TriMed’s Chicago team in 2017 covering cardiology. Close to her heart is long-form journalism, Pilot G-2 pens, dark chocolate and her dog Harper Lee.

Trimed Popup
Trimed Popup