The COVID-19 pandemic has turned our collective attention to healthcare systems around the world. Here in Canada, we have witnessed it suffer tremendous strain. One problem in particular has been exacerbated and stands out among many as requiring immediate attention: severe healthcare staffing issues.
Staff shortages and scheduling challenges existed in the health workforce well before the pandemic, but the impact of COVID-19 on the workforce combined with the surge in demand for health services led to a very dire situation.
Finding solutions that fit
Sajjad Dehnoei is one of many people who have worked to create tools to help alleviate the problem long before the COVID-19 pandemic. As a student in Telfer’s Master of Science in Health Systems program, under the supervision of Professor Antoine Sauré, Sajjad worked with staff and researchers at the Children’s Hospital of Eastern Ontario (CHEO) to develop a predictive tool that can determine more accurate staffing requirements. The project ran from August 2018 to September 2020.
One of the main difficulties in managing hospital staff schedules for the health workforce is the unpredictability of patient flux. Different types of patients require different levels of care, sometimes from different types of healthcare providers (nurses, counselors, aides, physicians).
This is certainly the case for CHEO’s inpatient mental health unit, confirms Roxanna Sheppard, Clinical Manager, Mental Health Acute Care Service. “Many factors affect healthcare staffing: union rules, specific staffing needs per unit, challenges in staffing on a shift-by-shift basis, staff shortages.” Where most solutions for this issue use fixed patient arrivals and resource requirements, Sajjad’s research focused on developing a method that incorporated the uncertainty of the inpatient unit.
Working with Roxanna to navigate administrative hurdles and access relevant data from CHEO’s inpatient mental health unit, the study determined how many shifts were over or understaffed, created patient types, and leveraged cutting-edge machine learning and optimization techniques to classify patients in meaningful groups. The use of artificial intelligence was particularly impactful, as the complexities of mental health care combined with the unique nature of pediatric care added to the volume of data, making it challenging to analyze with traditional methods.
Read Sajjad's Dehnoei's full paper: A stochastic optimization approach for staff scheduling decisions at inpatient units
Ultimately, this innovative tool allowed staff at CHEO to better understand staffing requirements based on real patient needs, and plan staff schedules more efficiently. This results in enhanced patient care, shorter wait times, and better-managed budgets.
Building a business case for more efficient healthcare staffing
According to the Canadian Institute for Health Information (CIHI), hospitals in Canada spent an average of 63.3% of their budgets on compensation in 2020-2021. Included in total compensation is the cost of overtime.
“In the past, we had chronic understaffing and regularly had to scramble to fill shifts. There were many requests for staff to work overtime.” shares Roxanna. “The process of working with Sajjad to build this model has been very validating; it has backed up what managers have been saying for a long time.”
Managers in the health workforce have been requesting staff increases for several years. The predictive healthcare staffing model created by Sajjad allowed CHEO administrators to determine baseline staffing requirements more accurately. This provided a concrete foundation to build a business case to request more staff. Timed to coincide with a hospital-wide staffing project, the increase in staff for CHEO’s inpatient mental health unit was approved, resulting in the hiring of 5.7 FTE child & youth counselors and 3.6 FTE nurses in July 2022, merely two years after the model was created.
Research with real-life impact
Not only does inefficient staffing increase overall compensation cost, but it can also potentially cause negative outcomes for patients. Though this predictive model was created specifically for CHEO, it can be applied to healthcare staffing more broadly to help reduce wait times and provide adequate care while respecting staffing ratios.
“The model is generalizable, applicable to different care providers such as nurses or doctors.” explains Sajjad. “It’s applicable to different care ratios and easily adjustable. Similar models have been implemented in other clinics.”
Both Sajjad and Roxanna expressed their satisfaction with the process and the results. “It was interesting to be part of the project, I don’t often work with researchers” says Roxanna. “Sajjad was lovely to work with, very communicative and receptive. I would be very open to working with a Telfer researcher again.”
For his part, Sajjad was grateful for the opportunity to do his internship at CHEO. So much so that he became a volunteer once the project was over, and subsequently worked there as a data analyst.
Successful cross-disciplinary collaborations
Rarely does research translate into implementation so quickly. According to Dr. Kathleen Pajer, Medical Lead, CHEO Precision Child Mental Health Initiative, “It typically takes 7 to 14 years for research to get to the clinical setting. This project took less than 4 years, it was extremely successful. These collaborations are extremely valuable to patient care.”
This particular collaboration between Telfer and CHEO came to be thanks to knowledge mobilization activities at Telfer, in the form of a symposium on artificial intelligence in healthcare. William Gardner, a researcher at CHEO’s research institute and a colleague of Dr. Pajer, was in attendance and subsequently reached out to one of the speakers, Telfer Professor Jonathan Patrick, with a request to help solve a well-defined problem. Professor Patrick turned to Sajjad, and the rest, as they say, is history.
Dr. Pajer attributes the success of this project to the mix of business mindset, clinician knowledge and innovative research. The results already show that the proposed approach significantly decreases the expected cost of staffing, compared to the traditional approach of staffing all shifts the same way. Due to the success of this collaboration with Telfer researchers, CHEO administrators look forward to future partnerships.