Last Friday, CMS released the first round of long-anticipated Medicaid unwinding data. The main focus has been on the renewal outcome data submitted in state monthly reports, which unfortunately was only available for the 18 states who began terminating coverage as of March 31 or in April 2023 due to a lengthy lag – and a lot of that we had already seen. However, CMS also posted call center statistics for almost all states (49 and DC), including average call wait time and abandonment rate. We’ve taken a close look at those.
As our colleague Tricia Brooks has said for years, call center statistics are the “canary in the coalmine.” States have been required to report call center data monthly as part of CMS’s performance indicator project; however, most of these data have never seen the light of day. Call center statistics can provide insights into states’ unwinding and provide an early warning that things are going wrong. Take call abandonment rates – when we see call abandonment rates increase, it is likely that enrollees are not getting the assistance they need to complete their renewal forms.
So what can we glean about the unwinding so far from the first months of call center data? Well, let’s focus in on the 18 states that began terminations for the months we have data:
When call wait times increase, call abandonment rates also increase; as discussed above, this means that individuals are unable to receive needed assistance to complete renewals forms or provide necessary documentation to maintain Medicaid coverage. In the data from April, a few states saw increases in call wait times and abandonment rates (e.g. Arizona, Florida, Indiana, and Utah). Overall, call wait times for states that started terminations in April were largely under ten minutes (eleven out of seventeen states), but three states had waits of over 30 minutes (New Mexico, Florida and Idaho). We’ll get into the nuances and context for how these data are being reported in a minute.
On the other end of the spectrum, there were actually states that saw declines in their call center metrics. Declines in wait times and associated abandonment rates themselves are potentially a good sign, indicating that the state may have taken steps to increase capacity and improve call center efficiency. However, when these declines parallel declines in call volume, like the data for New Mexico and West Virginia shows, that is not necessarily encouraging. Instead, a drop in call volume occurring at the same time disenrollments began may indicate people were not aware of the end of the continuous enrollment condition, which aligns with Kaiser Family Foundation survey findings from May.
One issue that unfortunately cannot be analyzed at all from the available data is the difference in wait times for non-English speakers. Anecdotally, longer wait times for Spanish speakers have been raised as a problem in a number of states – so this is an important question. Families with limited English proficiency are a high risk group for inappropriate terminations during the unwinding process so special attention must be paid to language access.
In the detailed spreadsheet CMS released, the call center data includes data notes that reflect some important nuances. For example, some state data include calls for other benefit programs, so call volume related to Medicaid may actually be lower than the number reflects. In some states call center data do not include calls made after hours or only include calls transferred to a live agent. The latter in particular may throw the numbers off. One particularly startling footnote: South Dakota does not operate a call center (and therefore there are no statistics for the state)!
It is worth noting that some states are posting call center data separately from CMS data on their own dashboards. Five of the eighteen states in the table above also post call center data themselves (KS, NE, NM, PA, and UT). Call center data published by states may not match what is published by CMS due to differences in how the metrics are reported (daily or weekly vs. monthly; queue wait time vs. call time overall; abandonment rate vs. abandonment rate for queue calls over 60 seconds). In the case of Utah, for example, the state’s data differed from that posted by CMS across all metrics, including a difference of around 16,000 calls.
Analyzing call center data can help stakeholders monitor the unwinding and understand the high rates of procedural terminations that many states are experiencing. Long wait times or high call abandonment rates suggest that individuals are not getting through to get support with their renewals.
[Editor’s Note: This is the 40th blog in the Unwinding Wednesday series. For more information, visit our PHE Unwinding resource page where you’ll find other blogs in this series, reports, webinars and the 50-state tracker.]