3 things that could make your health data unreliable
As the pandemic and the instability in the labor market continues, you should expect it to be more difficult to forecast demand and costs for health care.
The double-whammy of the Great Resignation and the COVID-19 pandemic is having unexpected effects on health data. Specifically, it has become harder to track and predict employee behavior. As a result, it’s more difficult for leaders to make adjustments that will improve benefits programs and individual health outcomes.
Here’s what you need to know to help you adapt to high employee turnover and the lingering effects of the pandemic.
High turnover makes it difficult to track effectiveness It’s well documented that staggering numbers of people in the United States are quitting their jobs. In March, for example, a record 4.5 million workers quit their jobs, with an additional 4.3 million workers quitting their jobs in April and May, the most recent months for which data is available from the Bureau of Labor Statistics.
What you may not know, however, is that the churn in the labor market is disrupting the ability of health care benefits leaders to use data to track the effectiveness of health plans and wellness programs.
That’s because companies typically track the value, ROI, and effectiveness of employer-provided benefits by analyzing metrics like adoption rates and health care utilization and costs, especially for programs designed for long-term impacts. But with so much turnover, not only could the data be unreliable, but it also means employers may have new employee populations with different issues and needs without realizing it.
Predicting behavior requires reliable data
In addition to challenges in performing backward-looking analysis, high rates of employee turnover make it difficult to use your employee health care data to make accurate predictions on future spending and health care needs due to shifting demographics. Older, more experienced people are increasingly likely to quit their jobs compared to younger ones. That could mean that over time your employee population is becoming younger, and that population has different health needs and costs.
Workers of color have stayed at their jobs at higher rates than white workers. That can mean that over time the social determinants of health of your employees are shifting.
Similarly, women are quitting their jobs at higher rates than men. That means that over time your worker base could be trending toward having more men than it did previously. Again, that changes the kinds of health care requirements that your employees may have.
These effects may be small at the individual level, but over time and at scale, they will aggregate into large effects. If you’re not tracking them now, you could be in for real surprises later.
Unexpected health needs skew data
COVID-19 has also made it difficult for health care benefits leaders to predict individuals’ behaviors and needs as they relate to wellness and benefits utilization. That’s because the pandemic has upended the prevalence and seriousness of many health conditions that workers are experiencing. For example, the number of people experiencing anxiety and depression worldwide increased by a staggering 25% during the first year of the pandemic, according to the WHO. This increased prevalence means if you rely on models that use a lot of historical data, you’ll probably find that the pandemic has introduced outliers that make accurate forecasting a challenge.
Of course, this is most fundamentally true of the coronavirus itself. But it also extends to increases in behavioral health challenges, including use rates of alcohol and other substances, as well as obesity and many other indirect effects.
Because it’s now more difficult to make predictions about costs, make sure you’re building models that account for more error variability than you otherwise would have.
Plan for the unexpected
As the pandemic and the instability in the labor market continues, you should expect it to be more difficult to forecast demand and costs for health care. That doesn’t mean it’s impossible, nor that you should stop trying. Instead, adjust expectations based on the changing demographics of your employees and be ready to revise your estimates as new data becomes available.
Related: An employer ‘How to’ for driving outcomes in today’s health benefits ecosystem
Janet Young, M.D., serves as the Lead Clinical Scientist on Springbuk’s Data Sciences and Methodologies team. Most recently, she served as a lead clinical scientist at IBM Watson Health. Janet received her M.D. from Yale University School of Medicine and also holds a masters in health service administration from the University of Michigan.