No one would fault a parent for taking time off work to care for their sick child. Most companies provide benefits to make sure employees can do just that. However, most wellness and care management programs would miss the opportunity to help that parent through a difficult time. The programs typically focus on a single disease and the person who has it. As a result, their view of the physically healthy parent’s challenges is, at best, myopic.
But what if companies could understand the true risk of an employee’s dependents—and how that risk affects the employee’s paid time off patterns?
Logically, one would expect parents of children with higher levels of illness to have more absences. However, to quantify the increase in lost work days and the costs to employers, one must have the proper data and the ability to make the necessary connections among that data.
HCMS Group LLC has found a way to pinpoint this information. To do it, they had to address several factors.
To acquire this level of insight, one must be able to calculate absence at the person level and connect dependents’ benefit experiences to an individual employee. One must also have the correct absence data, which includes short- and long-term disability, workers’ compensation, and sick leave. In addition, all these data types must connect seamlessly to health plan data. Without this unique collection of data sets and the ability to connect them at the person level, it would be difficult to quantify incremental lost work days.
HCMS Group LLC used its Research and Reference Database (RRDb) to analyze employees who were eligible for all health plan and absence benefits in 2016 and 2017. You can find more details about the parameters in the Methods section.
Not surprisingly, employees with higher-risk dependents had more total lost work days than employees who had no dependents or low-risk ones. The high-risk group had an extra three days of absence per year (Figure 1) compared to employees with low-risk dependents, mainly due to increases in STD and sick leave. (See Table 1 for a detailed comparison of the three groups.)
Figure 1 – Differences in total annual absence days
Table 1 – Detailed comparison of the three groups
Performing a simple calculation shows how this can be quantified into incremental spend for an employer. Figure 2 shows the expected increase in annual labor costs that results from these excess absence days.
Figure 2 – Estimated Costs = Incremental Lost Days x Number of Employees x Average Daily Salary
Knowing employer costs, no matter how exact the numbers, reveals only part of the picture. When a child has medical problems, their family faces astronomical costs—if not financial, certainly emotional. So how can employers help employees who have sick children? Without integrated data at the person level and a whole-person focused approach to care, these cases would generally be overlooked or mishandled.
KnovaSolutions, the clinical program offered by HCMS Group LLC, is designed for this type of situation. Identifying affected employees is the first step. Outreach and enrollment are then completed, and a nurse begins to work with the covered employee. The nurse can support the employee and provide advice on the child’s health issues. In many cases, this can alleviate some of the employee’s burden.
Using a person-centric, integrated view of benefits data and clinical care is the best way to reduce employee absences and provide support during difficult times.
Eligible employees were classified into three groups: those without dependents, those with low-risk dependents, and those with high-risk dependents. To minimize the impact of maternity-related short-term disability absence, only dependents ages 6 months to 18 years were analyzed. Risk was quantified using the patent-pending Human Capital Risk Index (HUI) score from HCMS Group LLC.
Results in Figures 1 and 2 and Table 1 were calculated using employees who were employed and enrolled in all absence types for the full 24-month time period (spanning 2016 and 2017). Statistical significance tests were done using tests with a significance level of p < 0.05.
A longitudinal regression modeling approach was used to ensure that differences between the groups’ results were not attributable to differences in the demographics of the groups. The results held each year from 2013 to 2017. Note that this population was required to be employed and enrolled in all absence types for the years 2013 to 2017.
Figure 3 – Regression adjusted results by year
Two-part regression models were used to produce adjusted results. Generalized linear models were used to account for the non-normal distributions associated with absence days. The first model assumes a binomial distribution with a logit linking function and is used to estimate the probability of having absence days greater than zero. The second model assumes a gamma distribution and a log link function and is used to estimate absence days for those employees with absence days greater than zero. The estimates of the two models are multiplied together for each employee to get a final estimate of expected total absence days. Finally, because each employee has multiple records (one per year), a repeated measures option was specified along with an autoregressive error structure.