ÌÇÐÄvlog

The accompanying the coronavirus disease 2019 (COVID-19) pandemic have raised concerns that the number of uninsured US residents will increase substantially in 2020—but there are no official real-time mechanisms for tracking the nation’s uninsured rate. Unlike assessing unemployment claims or Medicaid enrollment, measuring the overall share of people without insurance relies on time-consuming and imperfect . Understanding the limitations of the available data is especially critical during a fast-changing health and economic crisis like the current one.

Government-administered surveys have long been considered the gold standard for coverage estimates because they are fielded every year, have large sample sizes, achieve far higher response rates than private surveys, and often let researchers conduct analyses across smaller geographic units. But time-intensive data collection and processing contribute to significant lags in data availability.

Furthermore, the . The American Community Survey (ACS) suspended mail-outs from mid-March through June. In-person field operations for other surveys, including the National Health Interview Survey, were suspended and have not resumed in all states. The Current Population Survey (CPS) reported falling response rates: March’s 73% response rate represented a from prior months, and rates have since declined further. These issues affected its and could affect 2020 estimates, too.

The US Census Bureau began fielding a new weekly in April to rapidly measure effects of the pandemic. However, the survey has 2 major limitations—lack of any prepandemic baseline data for comparison and a dramatically lower response rate than other government surveys (2%-3%).

Private organizations can swiftly field smaller surveys, but these have their own drawbacks. Historically, population-based weighting has produced despite limited sample sizes and low response rates, but the current environment creates uncertainty regarding the adequacy of standard nonresponse adjustment through weighting. If the pandemic differentially changed response rates across subpopulations—particularly those not captured by standard demographic variables (eg, essential vs nonessential workers)—weighting may not completely eliminate nonresponse bias. Furthermore, these surveys often cannot support state or local estimates. Geographically targeted surveys have sprung up across the country but cannot be directly compared.

Available surveys can be supplemented by and benchmarked against administrative enrollment records. However, there is no comprehensive administrative data source on employer-sponsored insurance (ESI), the type of coverage most likely to be lost during the pandemic. See the on health insurance coverage in the US.

As the pandemic unfolded, researchers potential coverage losses. Early predictions suggested that as many as 27 million people in the US were at risk of losing ESI in the early months of the recession and that an estimated 5.4 million nonelderly adults would become uninsured as a result.^1,2

New surveys have offered preliminary estimates of actual pandemic-related coverage changes. An online and telephone survey of 1007 respondents estimated that 10 million adults lost ESI owing to the pandemic through late April, considerably lower than some projections.³ Job losses may have been concentrated among workers without ESI, and federal support could have helped businesses implement temporary furloughs rather than permanent layoffs. A telephone survey of 2271 adults from late May found that in households where someone had stopped working at a job through which they had health benefits, more than half had been furloughed with coverage; another 14% still had ESI through a family member.⁴

Similarly, the uninsured rate appears to have ticked upward during the first half of 2020, but less than originally forecasted. An internet-based survey of 4352 nonelderly adults found a 1.4-percentage-point increase in the uninsured rate in states that have not expanded Medicaid but no significant change in expansion states as of May.⁵ Using the Census Bureau’s new Pulse survey, another study reported 1.9 million more uninsured adults between late April and mid-July.⁶ If temporary layoffs become permanent job losses, coverage losses could worsen.

These preliminary results leave much unknown, including long-term coverage consequences of the pandemic, state-varying effects, and whether higher-quality surveys will produce similar estimates.

Federal and state governments can take several steps to enhance our ability to understand coverage changes during the pandemic. First, the Census Bureau should make publicly available the additional data already collected in the CPS and ACS. Currently, the CPS Annual Social and Economic Supplement public files report coverage at the time of interview and ever having had coverage in the prior year. The survey collects more granular data—not included in public files—about coverage in each month of the year prior to the interview. The ACS measures coverage at the time of interview and surveys respondents throughout the year but does not disclose the month of interview in its public data release. But a data set that provides a blended average of coverage rates from January, May, and December 2020 obscures the most critical effects of the COVID-19 pandemic. If the Census Bureau released the CPS longitudinal file and the ACS month of interview (or even quarter of interview), with appropriate confidentiality protections in place, this would immeasurably improve researchers’ ability to identify coverage changes before, during, and after the pandemic.

Second, the Census Bureau should extend the Pulse survey into the postpandemic period and perform necessary benchmarking and validation studies comparing Pulse estimates with the gold-standard surveys and administrative data. Furthermore, they should conduct the that are required of government surveys with low response rates and implement needed adjustments to survey weights.

Third, states can act to improve our understanding of the pandemic’s consequences. The Behavioral Risk Factor Surveillance System (BRFSS) allows states to add their own questions to survey modules, and the publicly released BRFSS data files already include monthly indicators. States should add questions of interest related to the pandemic and release coverage estimates expeditiously.

The COVID-19 pandemic raised critical questions around health insurance coverage. Timely and high-quality data are essential to guide decision-making. While survey methods and data collection are rarely in the political spotlight, policy makers must act now to maximize our ability to understand and respond to the current crisis.

Open Access: This is an open access article distributed under the terms of the CC-BY License.

Corresponding Author: Adrianna McIntyre, MPP, MPH, Harvard PhD Program in Health Policy, Harvard University, 14 Story St, 4th Floor, Cambridge, MA 02138 (amcintyre@g.harvard.edu).

Conflict of Interest Disclosures: Mr Brault reported prior employment by the US Census Bureau (2005-2014) and personal fees from the Institute for Human Centered Design outside the submitted work. Dr Sommers reported grants from Baylor Scott & White Health, Commonwealth Fund, Robert Wood Johnson Foundation, and REACH Healthcare Foundation; personal fees from American Economic Journal, Health Research & Educational Trust, Massachusetts Medical Society, the Illinois Department of Healthcare and Family Services, Urban Institute, and AcademyHealth; personal fees and nonfinancial support from Northwestern Medical Center and University of Rochester; and nonfinancial support from University of Cincinnati outside the submitted work. No other disclosures were reported.