How AI in the hiring process reduces nurse turnover
Using data science to predict which RNs are the right fit reduces turnover, vacancy, and reliance on agency labor
Nursing Workforce Challenges: Turnover, Vacancy, Shortage
Excessive turnover across healthcare, and especially among RNs, has been an issue in health care for many years. In 2014, a 10-year study of new nurses by the RN Work Project, backed by the Robert Wood Johnson Foundation, reported that, on average, 17.5% of nurses in the study group left their first job within a year of being hired.1Kovner, CT, Brewer, CS, et al. What Does Nurse Turnover Rate Mean and What Is the Rate? Policy Polit Nurs Pract. 2014;15(3-4):64-71. Recent comprehensive survey data from the Advisory Board revealed even higher rates of <1 year turnover among nurses.
Today, the situation seems to be growing more acute as novice nurses outnumber experienced nurses, and turnover is highest among new nurses. As a result, an increasing number of hospitals (55.3%) had an RN vacancy rate higher than 7.5%. 2Bureau of Labor Statistics, Employment Situation Summary, May 2019, accessed at www.bls.gov/news.release/empsit.t14.htm
Meanwhile, healthcare stood very close to full employment in June 2019, according to the U.S. Bureau of Labor Statistics (BLS). By 2026, the BLS projects the baby boom generation will create an influx of older patients and a simultaneous retirement of those available to care for them – resulting in a need for 15% more RNs than today.3Bureau of Labor Statistics, Occupational Outlook Handbook, accessed at www.bls.gov/ooh/healthcare/registered-nurses.htm#tab-6
The Impact of Nursing Workforce Challenges
A large body of research has tracked a direct effect of turnover on several measures of patient care. For example, a greater number of patient-care hours performed by RNs per day is associated with better care for hospitalized patients. 4Needleman, J., Buerhaus, P. Nurse-Staffing Levels and the Quality of Care in Hospitals. N Engl J Med 2002; 346:1715-1722 Higher RN staffing correlates to reductions in hospital-related mortality, cardiac arrest, hospital acquired pneumonia, and other adverse events.5Kane R.L., Shamliyan, T., et al. Nurse staffing and quality of patient care. Evid Rep Technol Assess. 2007 Mar;(151):1-115. Nurse satisfaction with the work environment is closely aligned with patient satisfaction measures.6Peršolja M. The effect of nurse staffing patterns on patient satisfaction and needs: a cross-sectional study. J Nurs Manag. 2018 Oct;26(7):858-865. In addition, significant association exists between the actual vs. needed staffing levels and patient satisfaction.7Kutney-Lee A, McHugh MD, Sloane DM, et al. Nursing: a key to patient satisfaction. Health Aff. 2009;28(4):w669–w677
Regional One Health
Like many healthcare providers, Regional One Health in Memphis, TN, faces increasing difficulties in hiring and retaining qualified employees. The labor market continues to tighten, nurses and other staff switch jobs frequently, and competition from neighboring providers, who have more financial security, looms large.
As a stand-alone, safety-net hospital serving a high-risk underserved patient population, Regional One is an essential part of the community. The system encompasses a 326-bed acute-care hospital with a regional Level I Trauma and Burn Center; Level 3 NICU; high-risk OB; a rehab hospital; a long-term acute care hospital; and a range of ambulatory clinics and services throughout the city. It competes for talent with two large multi-state health systems in the area, both with flagship hospitals nearby, along with a VA hospital, and two hospitals owned by national for-profit Tenet Healthcare Corp.
Despite its urban setting, Regional One’s service area extends 150 miles into western Tennessee, eastern Arkansas, part of northern Mississippi and even a little bit of Missouri, so it’s both urban and rural. When nearly 25% of a hospital’s patients have no payer source and another 50% are covered by Medicaid and Medicare, its work is cut out for it. Memphis has struggled economically, with significant pockets of poverty. Though it is on the mend, many cities have more attractive living situations, making it harder to recruit from outside.
The Challenge of Turnover
In 2016, some nursing departments at Regional One had annual turnover in excess of 40%. Nearly 30% of nurses were leaving in the first year. Turnover plagued other departments in the hospital, with housekeeping, for example, experiencing 360-day turnover in excess of 50%.
Taking advantage of a number of nursing schools in the metro area, Regional One had implemented a nurse residency program, to add support for the new graduates in their first year of practice. The turnover was equally high for those participating in that program as for the more experienced staff.
Regional One had acquired an innovative solution called Arena in 2015, but it wasn’t using it optimally, if at all. Arena brings data analytics, machine learning and predictive modeling to bear on the hiring process. It is designed to reduce the impact of unconscious human bias in hiring and expand the pool of applicants to those who might never have been considered for a job.
The previous leadership had not been sharing results from the tool with hiring managers but a new CNO/COO decided to invite the vendor in for an informational session with the HR team. Once the training was completed, the patient care and HR leadership agreed to a completely transparent process going forward, allowing each manager to see the data analytics ‘predictions’ on each candidate.
How ARENA Works
All applicants are obliged to complete the Arena questionnaire online (6-8 minutes), which produces metadata about how applicants react in different situations. Arena analyzes this data along with outcome data collected from Regional One’s HR systems, and publicly available sources such as local and regional labor statistics, job boards, the Glassdoor website, and relevant market data.
All of this data is fed into algorithms that generate predictions for the likelihood that a particular prospective employee will stay and be a productive member of a team. The algorithms produce a score, which leads a candidate to be “recommended” or “not recommended.”
These predictions are highly specific to the candidate and the role being filled. The same person who thrives in a job in one unit of a hospital may not do well in the exact same job in another unit or another location.
The use of fresh outcome data continually updates the algorithms, improving accuracy and keeping pace with changes in the organization .
As a result of integrating Arena into the hiring process, Regional One reduced 90-day turnover by 39% in the first 6 months; by one year, turnover was down 22%.
One of the strongest arguments for using the technology was revealed when Regional One continued to hire staff, despite the algorithmic predictions of unlikelihood to be retained. The turnover rate for those employees was a full 10 percentage points higher than the “recommended” pool.
“We eliminated our agency usage this year in all acute patient care units and emergency departments. We spent $7.2 million in fiscal year 2018 on traveling nurses and aggressively dropped that to $1.8 million by the end of the 2nd quarter in fiscal year 2019. For 2020, we budgeted no agency nursing expense,” explains CNO/COO Lisa Schafer.
Reducing turnover saves money. Not only did the team spend less time hiring and training new staff — which costs hundreds of thousands of dollars annually – they eliminated a major expense for agency nurses. “We eliminated our agency usage this year in all acute patient care units and emergency departments. We spent $7.2 million in fiscal year 2018 on traveling nurses and aggressively dropped that to $1.8 million by the end of the 2nd quarter in fiscal year 2019. For 2020, we budgeted no agency nursing expense,” explains CNO/COO Lisa Schafer. “NSI surveys find that for every 20 traveling nursing RNs eliminated from the budget, a hospital can save an average of $1.4 million, which is not far off from our experience.”
These new tools of artificial intelligence do not replace the human factor in hiring, but the efficacy of the retention predictions has proven to be a valuable additional insight, one that augments good hiring decisions.
There is a sea of opportunity here. As health systems fine-tune the process of measuring clinical outcomes and patient experiences and can align this data with the attending staff, predictive scores of success can be generated that relate these outcomes to job candidates.
This is the future of hiring, to ensure we have staff who are best suited for challenging work while saving our health systems significant dollars that can keep the doors open and ready to serve the community.
References [ + ]
|1.||↑||Kovner, CT, Brewer, CS, et al. What Does Nurse Turnover Rate Mean and What Is the Rate? Policy Polit Nurs Pract. 2014;15(3-4):64-71.|
|2.||↑||Bureau of Labor Statistics, Employment Situation Summary, May 2019, accessed at www.bls.gov/news.release/empsit.t14.htm|
|3.||↑||Bureau of Labor Statistics, Occupational Outlook Handbook, accessed at www.bls.gov/ooh/healthcare/registered-nurses.htm#tab-6|
|4.||↑||Needleman, J., Buerhaus, P. Nurse-Staffing Levels and the Quality of Care in Hospitals. N Engl J Med 2002; 346:1715-1722|
|5.||↑||Kane R.L., Shamliyan, T., et al. Nurse staffing and quality of patient care. Evid Rep Technol Assess. 2007 Mar;(151):1-115.|
|6.||↑||Peršolja M. The effect of nurse staffing patterns on patient satisfaction and needs: a cross-sectional study. J Nurs Manag. 2018 Oct;26(7):858-865.|
|7.||↑||Kutney-Lee A, McHugh MD, Sloane DM, et al. Nursing: a key to patient satisfaction. Health Aff. 2009;28(4):w669–w677|