This FQHC slashed its patient no-show rate with AI in 3 months

Must read

The New York City-based Urban Health Plan used artificial intelligence to improve operational efficiency and patient care and got a handle on its annually high patient no-show rates with cost-effective patient interventions.

Higher-than-average missed appointments

No-shows are due to multiple factors, including social determinants of health like transportation

But they are also a long-standing challenge to healthcare organizations that can affect the patient’s care quality, increase healthcare organization costs and present unnecessary reductions in overall patient access to healthcare appointments.

In March, Urban Health Plan had 42,000 health visits – the highest in its history, according to a presentation at an eClinicalWorks health summit in Boston last week. 

“We had a multifaceted approach to just addressing patient access and engagement overall,” Alison Connelly-Flores, chief medical information officer of Urban Health Plan, told Healthcare IT News

That achievement is significant because annually, UHP experiences a high number of missed appointments and appointments for patients waiting are sometimes “booked out way too far.” Providers are typically overbooked to accommodate for no-shows, said Connelly-Flores.

The federally qualified community health center organization, one of the largest in New York state, provides primary care, 18 specialties, diagnostics and other services for approximately 86,000 patients. The no-show rate was very high across UHP’s 12 sites in South Bronx, Corona Queens and Central Harlem neighborhoods, its 12 school-based health centers and behavioral health services.

While UHP scheduled 794,322 visits in 2022, only 57.6% were completed compared to the national average in eClinicalWorks EHR data – 71%. 

With 336,600 no-shows and overbooking to compensate, the results can range from long wait times, patient dissatisfaction and stress on providers, all of which can exacerbate burnout.

The organization needed to change things around to keep health centers open. 

Leadership wanted to know if their no-show rate numbers were above or below the national average and who are the patients that kept missing their appointments. 

Through the pilot, UHP learned its no-show rate was 16.52% higher than its EHR peers.

It also learned that despite attrition among the low-, moderate- and high-probability groups, UHP’s show rates were consistent for each group from 2019-2022 – even through the pandemic.

“It was just interesting that those groups sort of behaved the same way. I didn’t anticipate that … it was definitely a pattern.” Connelly-Flores said after the session.

Machine learning analysis of network-level data

The no-show algorithm is a screening analysis enabling systematic and consistent data sorting that allows data analysts and users to see network-level data.

“I can essentially have a conversation with this data,” said Sameer Bhat, cofounder and vice president of sales at eClinicalWorks, ahead of the discussion about UHP’s pilot test of the algorithm, which is part of a larger eClinicalWorks study with numerous customers.

Bhat demonstrated how to look at demographics like ethnicity and poverty level on a healow dashboard and noted the platform is EHR-agnostic. 

The platform can also aggregate discrete EHR data to identify gaps in care, according to the company’s website.

As he introduced Connelly-Flores and the pilot team, he said, “we are blown away with some of the findings.”

They say the algorithm can find the needles in the haystacks and can help identify patients with high no-show probability with 85%-90% accuracy.

Missed appointment interventions that worked

In addition to working with the no-show prediction model, the UHP team adjusted its outreach process using eClinicalMessenger.

UHP already manages more than a million yearly voice messages, secure text messages and email reminders, according to  eClinicalWorks’ announcement about the pilot.

After the model identified patients with the highest and moderate risks for missing their appointments, UHP tested new focused interventions to help make sure that patients made the appointments they scheduled.

Would a telephone call help? With 3,000 appointments per day, UHP cannot call all of their patients scheduled, Connelly-Flores said.

eClinicalWorks kept 38,431 with a high no-show probability in a control group. The company shared 18,061 with a high no-show probability, as well as 908 with a medium no-show probability with UHP in order to test focused interventions.

UHP distributed a high-risk no-show report to designated associates and provided a script to make the calls and messaging as consistent as possible. The designated associates documented call outcomes.

Patients who had missed their appointments that day were offered an opportunity to switch to a same-day virtual visit by initiating telemedicine visits or rescheduling appointments. UHP doctors will call the patients that missed their appointments themselves.

Connelly-Flores said that if a patient is reached by a doctor, nearly 100% will accept the same-day virtual rescheduling option. They use the video app to call those patients, and just switch to a telehealth visit once a patient accepts.

“If you get a ‘last-minute’ cancelation or rescheduling, you may still be able to recover the appointment,” an eClinicalWorks blog dated March 2023 says. 

“If the last-minute change is due to a transportation or travel issue, perhaps a televisit will suffice. This can save the appointment and even encourage more frequent visits due to its convenience.”

For scheduled telehealth visits with patients that have a high or moderate probability to miss their appointment and haven’t been seen in 15 months or more, UHP sent additional text messages.

UHP also increased access to virtual care extending hours to 89 per week. 

To better support providers, the healthcare organization revised their workload analysis templates to factor in each provider’s no-show rate and added same-day slots to accommodate a switch to virtual care. 

While these adjustments can require monitoring, overall, UHP found that the strategies could reduce the rate of no-shows even further.

The algorithm was implemented in January 2023, and the intervention resulted in 4,432 more visits during the three-month pilot.

The low no-show probability rate for 2023 thus far is more than 5% higher than the previous four years.

The outcome between the two pilot patient groups also showed a 24.14% increase in the likelihood to make their appointments for those patients at high risk for no-shows and an 8.08% improvement for those with moderate risk.

While the show rate for the patients most likely to miss their appointments increased by 154%, UHP’s interventions also increased the show rate by 19.17% for the moderate-risk no-show patients.

UHP added a specific full-time role and adjusted one existing role to provide part-time support to call only those the algorithm identified as highest risk for a no-show.

During the pilot, the targeted telephone calls were about 400 per day, Connelly-Flores said.

“It wasn’t a huge lift.” 

She said the next steps for UHP include integrating the algorithm into its EHR, involving case management, addressing barriers to care, sending more customized appointment reminders to patients at high-risk for missed appointments and taking a deeper dive into the data to learn more. 

“By leveraging the power of data and machine learning, we can help providers like Urban Health Plan deliver more effective care to their patients and reduce the burden of missed appointments,” said Girish Navani, CEO and cofounder of eClinicalWorks, in the company’s statement.

“This ultimately helps reduce the cost of healthcare and aid better patient outcomes.”

“When patients receive timely care, they see better health outcomes.” Connelly-Flores said.

Andrea Fox is senior editor of Healthcare IT News.
Email: [email protected]

Healthcare IT News is a HIMSS Media publication.



Source link

- Advertisement -spot_img

More articles

LEAVE A REPLY

Please enter your comment!
Please enter your name here

- Advertisement -spot_img

Latest article