Winter Storm Uri Reduced Availability of Daily Direct-Care Nurse Staff in Texas Nursing Homes

Abstract

We examined the effect of Winter Storm Uri on daily direct-care nurse staffing levels in 1,173 Texas nursing homes (NHs) from February 13^th to 19^th, 2021. We used data from Payroll-Based Journaling and the Care Compare website. Linear mixed effects models were used to estimate the linear and non-linear change in staffing. During Winter Storm Uri, Texas NHs experienced a linear decrease in daily staffing levels with subsequent non-linear increase for registered nurses (RNs; p < .001) and certified nursing assistants (CNAs; p < .001), and staffing increased linearly for licensed practical nurses (LPNs; p < .001). Compared to 1 week before the storm, Texas NHs reported lower staffing levels across all three types of staff, but most dramatically among LPNs and CNAs. In supplemental analyses, urban and lower quality NHs fared slightly poorer than rural and higher-quality NHs. Winter storms pose a particular challenge to NHs and their ability to maintain direct-care nurse staffing levels.

Keywords

disaster medicine staffing shortages disaster preparedness

What this paper adds

• This paper adds winter storms to a growing list of natural disasters that affect the availability of direct-care nurse staff in nursing homes.

• In areas where winter storms are uncommon, their effects on staffing may be most dramatic in the first few days, and a shortage of staff may exacerbate the consequences for residents.

Applications of study findings

• Given the intensification of climatological disasters due to global warming, nursing homes’ disaster plans should include response to sudden onset disasters that are uncommon for their geographic area.

• Southern cities and states must consider infrastructure factors (eg transportation and power supplies) that affect a nursing home’s ability to secure sufficient staffing during winter storms.

Introduction

Beginning on Saturday, February 13th, 2021, the state of Texas suffered through a prolonged exposure to unseasonably low temperatures fueled by an upper-level polar vortex of frigid air that descended south from Central Canada to encompass the entire state (West, 2021). During the ensuing 6 days of Winter Storm Uri, Texas experienced record low temperatures, freezing rain, sleet, and significant snow accumulations that led to widespread road closures and deadly interruptions to the power grid that did not fully resolve until February 19th, 2021 (Soergel, 2021; West, 2021). Estimates from the State of Texas suggest that 246 lives were lost as a result of the storm although some outlets have suggested mortality rates several magnitudes higher (Svitek, 2022). More than 570 of the 1,200 nursing homes (NHs) in Texas reported emergencies to the Texas Health & Human Services Commission, with more than 400 NHs losing power or access to potable water (U.S. Senate, 2023). Given the mortality rates and the chaos caused by the storm, Winter Storm Uri posed a unique threat to long-term care residents in the state.

One avenue by which residents of NHs were likely affected is through the disruption of staffing levels, particularly licensed staff such as registered nurses (RNs) and licensed practical nurses (LPNs) and unlicensed staff such as certified nursing assistants (CNAs). Prior research has shown that direct-care nurse staffing levels are associated with higher NH quality (Antwi & Bowblis, 2018; Clemens et al., 2021; Dellefield et al., 2015; Jutkowitz et al., 2022), thus it is crucial to maintain nurse staffing levels during disasters when residents are medically and psychologically more vulnerable than usual due to disruptions in regular operations. Analyses conducted during hurricane disasters suggest that retaining staff in NHs can be difficult during a crisis (Hyer et al., 2009), and that the staffing-related expenses borne by NHs are rarely reimbursed by state or federal entities before or after the disaster (Thomas et al., 2010). During Hurricane Irma in 2017, many Florida NHs increased their direct-care nurse staffing levels in response to the storm (Jester et al., 2021). The magnitude of the direct-care staffing increase was influenced by evacuation status (ie evacuated NHs increased their staffing levels more than NHs that sheltered-in-place) and NH quality (ie higher-quality NHs increased their staffing levels more than lower quality NHs) (Jester et al., 2021; 2022). Unlike most hurricanes, which allow for advance preparation because of current forecasting technology (eg Saffir-Simpson Winds Scale Rating, daily reports from the National Oceanic and Atmospheric Administration), Winter Storm Uri was unexpected for Texas in its severity and duration, and is an unfortunate but increasingly likely result of global warming and the warming of the Arctic (Cohen et al., 2021).

Building on previous research, we set out to examine the specific patterns of daily staffing loss attributable to Winter Storm Uri across RNs, LPNs, and CNAs in NHs. The overarching goal of this paper is to provide new information that can help NHs plan for similar disasters in the future in terms of patterns and likely duration of staffing loss, as well as understand factors linked with greater or lesser disruptions to staffing in the face of a similar natural disaster. We hypothesized that Texas NHs would report lower staffing levels from February 13^th to 19^th, 2021 during Winter Storm Uri than 1 week prior (February 6^th to 12^th, 2021), suggesting that Winter Storm Uri adversely affected the availability of staff. Supplemental analyses compared rural to urban NHs and higher quality to lower quality NHs to further characterize facilities that were especially vulnerable during the storm.

Methods

Data

This study used data from Payroll-Based Journaling (PBJ) and the Care Compare website. Briefly, Centers for Medicare & Medicaid Services provides daily staffing data through a database called PBJ. Unlike the prior reporting system, in which staffing was often inflated in anticipation of an inspection (Geng et al., 2019), PBJ allows payroll to be used to validate staffing data. PBJ reported unadjusted hours by staff type, including employed and contract staff, and a rolling resident census to standardize staffing levels across NHs of different sizes. The Care Compare website reported whether a NH was part of a continuing care retirement community (CCRC), profit status (for profit, not for profit, government), an overall quality star rating (1 [worst]–5 [best]), a weighted health inspection score based on prior survey performance, the number of citations from infection control inspections, and the number of substantiated complaints from any source (eg residents, family, staff, ombudsman).

Analytic Sample

As Winter Storm Uri affected the majority of the state, our sample included all Texas NHs reporting non-administrative direct-care nurse staffing information in PBJ (n = 1,174). One facility was excluded because it could not be matched to the Care Compare data. This left 1,173 unique NHs in the final analytic sample, with 1,140 being retained for linear mixed models due to missing data on the NH quality covariates (overall star rating, weighted health inspection score, number of citations from infection control inspections, number of substantiated complaints). In order to compare across facilities of different sizes, hours were converted to hours per resident-day (HPRD) by dividing by the resident census on the given day. Staffing data from February 6^th to 12^th was matched to data from February 13^th to 19^th to calculate differences in staffing levels, matching each day of the week. For example, February 6^th and February 13^th were the first and second Saturday of February, respectively. This matching was necessary to capture the fluctuation in staffing typically seen across each week (ie on average, weekends have lower staffing levels than weekdays) (Geng et al., 2019).

Facility Quality and Characteristics

Given that prior work has shown that quality star rating influences the direct-care staffing response to natural disasters (Jester et al., 2022), we controlled for several forms of quality: (1) the overall star rating indexed general quality (ie developed from health inspections, quality metrics, and staffing deficits), (2) the weighted health inspection score indexed quality of care (ie includes data from the three most recent survey cycles, complaint surveys, focused infection control surveys, and revisit points), (3) the number of citations from infection control inspections indexed the impact of COVID-19 on staff retention (ie inspections developed during the COVID-19 pandemic), and (4) the number of substantiated complaints indexed complaints from any source (eg resident, family, staff, ombudsman) found by surveyors to be valid (Peterson et al., 2021a; 2021b).

Structural characteristics such as CCRC and profit status, resident census, and county fixed-effects were adjusted for. County fixed-effects were especially relevant due to geographic variability in the effects of the COVID-19 pandemic and Winter Storm Uri. RN or CNA staffing levels were included as covariates in each model (CNA staffing levels as a covariate for the RN and LPN models; RN staffing levels as a covariate for the CNA model) to control for the general availability of staff in each NH during Winter Storm Uri, which is consistent with prior work (Jester et al., 2021).

Statistical Analyses

Descriptive statistics were used to report Texas NHs’ characteristics in February 2021. Linear mixed effects models estimated the unique effect of Winter Storm Uri on daily change in staffing levels over time from February 13th to 19th, 2021. Here, the effect of time provided an estimate of the daily change in staffing, measured in HPRD (ie slope). The effect of time² provided an estimate of the quadratic daily change in staffing, measured in HPRD (ie slope²). If the quadratic slope was not statistically significant, it was removed from the model for parsimony. Wherever possible, 95% confidence intervals were provided alongside conventional p-values. Random effects included an estimate of within-group variance (σ²), between-group variance (τ₀₀), and the intraclass correlation coefficient (ICC; ie the proportion of variance explained by clustering NHs). These effects provide nuance to the analyses, but do not inferentially test a hypothesis. We also report two estimates of effect size. The marginal R² considers the variance explained in the model by fixed effects, while the conditional R² considers the variance explained in the model by both the fixed and random effects. All analyses were completed in R 4.2.0 (lme4 and sjPlot packages).

Supplemental Analyses

In order to determine whether a facility resided in a rural or urban location, 5-digit zip codes were matched to Rural-Urban Continuum Codes provided by the U.S. Department of Agriculture’s Economic Research Service. Detailed methodology is available online: https://www.ers.usda.gov/data-products/rural-urban-continuum-codes/documentation/. The Rural-Urban Continuum Codes classify counties as metropolitan (urban) or nonmetropolitan (rural) by their population size and adjacency to urbanized centers. For example: Dallas County would be considered metropolitan because of its large population size of 2.6 million; Kaufman County has a significantly smaller population of 158 000, but it is considered metropolitan because of its close proximity to the city of Dallas.

Higher quality NHs were defined as having an overall quality star rating of four or five, while lower-quality NHs were defined having an overall quality star rating of one, two, or three on the Care Compare website. Additional information is available online: https://data.cms.gov/provider-data/archived-data/nursing-homes.

We provide RN, LPN, CNA, and total staffing deficits among rural and urban NHs and higher quality and lower quality NHs. Deficits were measured by taking the staffing levels during each day of Winter Storm Uri and subtracting the staffing levels of the prior week. Because facilities differ in their baseline staffing levels (eg higher quality NHs provide higher staffing levels), deficits during Winter Storm Uri were also reported as a percentage of the prior week’s levels.

Results

Characteristics of Texas’ Nursing Homes

Regarding NH characteristics, 81% were for-profit, 7% were not for-profit, 12% were government-run, and 6% were part of a CCRC. Nearly half of NHs were rated as 1-star or 2-star (18% 5-star, 17% 4-star, 19% 3-star, 22% 2-star, 25% 1-star), and the mean total weighted inspection score was 73.13 (SD = 61.31). On average, Texas NHs received 1.07 citations during focused infection control inspections (SD = 1.95; Range [0,17]) and 3.87 substantiated complaints (SD = 5.93; Range [0,50]).

Effect of Winter Storm Uri on Staffing Hours

After controlling for facility, quality, and county characteristics, Texas NHs decreased staffing levels linearly, with a subsequent non-linear increase over time for RNs (p < .001) and CNAs (p < .001), and the levels increased linearly over time for LPNs (p < .001) during Winter Storm Uri (Table 1). This suggests that NHs’ staffing levels were lower over the weekend when Winter Storm Uri initially hit Texas (ie Saturday, February 13th, 2021), but direct-care staffing of all types increased over the work week as the State of Texas responded to the storm.

Table 1.

Linear Mixed Effects Models Showing Change in Direct-Care Nurse Staffing During Winter Storm Uri.

Variables	Beta	95% CI	p	Beta	95% CI	p	Beta	95% CI	p
Fixed effects	RN HPRD			LPN HPRD			CNA HPRD
Time (linear slope)	−0.13	−0.16–0.09	<0.001	0.02	0.02–0.02	<0.001	−0.48	−0.54–0.42	<0.001
Time² (quadratic slope)	0.00	0.00–0.00	<0.001	—	—	—	0.02	0.01–0.02	<0.001
CNA HPRD	0.03	0.02–0.04	<0.001	0.11	0.09–0.12	<0.001	—	—	—
RN HPRD	—	—	—	—	—	—	0.09	0.05–0.13	<0.001
Resident census	−0.00	−0.00–0.00	<0.001	−0.00	−0.00–0.00	<0.001	−0.00	−0.01–0.00	<0.001
CCRC	−0.04	−0.15–0.06	0.39	0.15	0.06–0.24	0.001	0.24	0.11–0.38	<0.001
For-profit (ref)	—	—	—	—	—	—	—	—	—
Not for-profit	0.25	0.15–0.34	<0.001	−0.09	−0.18–−0.00	0.05	0.28	0.15–0.41	<0.001
Government-owned	0.00	−0.08–0.08	0.98	0.01	−0.06–0.08	0.82	0.04	−0.06–0.15	0.41
Overall star rating [1] (ref)	—	—	—	—	—	—	—	—	—
Overall star rating [2]	0.05	−0.02–0.13	0.17	0.01	−0.06–0.07	0.83	0.14	0.04–0.24	0.004
Overall star rating [3]	0.04	−0.05–0.13	0.38	0.01	−0.07–0.09	0.78	0.13	0.02–0.25	0.02
Overall star rating [4]	0.06	−0.04–0.16	0.22	0.02	−0.07–0.10	0.72	0.24	0.11–0.37	<0.001
Overall star rating [5]	0.26	0.15–0.36	<0.001	0.06	−0.03–0.15	0.19	0.29	0.15–0.42	<0.001
Weighted inspection score	0.00	−0.00–0.00	0.63	−0.00	−0.00–0.00	0.85	0.00	0.00–0.00	0.004
# Citations from IC inspections	−0.01	−0.02–0.01	0.28	−0.01	−0.02–0.00	0.07	−0.00	−0.02–0.02	0.90
# Substantiated complaints	−0.00	−0.01–0.00	0.67	0.00	−0.00–0.01	0.17	−0.00	−0.01–0.00	0.66
Random effects	RN HPRD			LPN HPRD			CNA HPRD
Within-group variance (σ²)		0.02			0.05			0.08
Between-group variance (τ00)		0.12			0.09			0.21
Intraclass correlation (ICC)		0.83			0.64			0.71
# Nursing homes		1,140			1,140			1,140
# Observations		7,964			7,964			7,964
Marginal R²/Conditional R²		0.193/0.865			0.200/0.711			0.273/0.787

Note. Marginal R² includes the fixed effects while the conditional R² includes the fixed and random effects. In addition to the covariates listed above, county fixed-effects were included.

Abbreviations. 95% CI = 95% Confidence Interval. CCRC = Continuing Care Retirement Community. CNA = Certified Nursing Assistant. HPRD = Hours Per Resident-Day. IC = Infection Control. LPN = Licensed Practical Nurse. RN = Registered Nurse.

Although increases in staffing do suggest a return to normalcy over the work week, compared to 1 week before the storm (February 6^th–12^th, 2021), Texas NHs reported lower staffing levels across all three types of direct-care staff overall. This was most dramatic among LPNs and CNAs (see Figure 1; Table 2), and was largest for Days 3–5 (Monday–Wednesday) of the winter storm, suggesting that Winter Storm Uri affected the availability of staff.

Figure 1.

Difference in direct-care nurse staffing during Winter Storm Uri and 1 week prior in hours per resident-day. Note. Winter Storm Uri began on February 13th, 2021. Day 1 was calculated as the mean staffing levels by type on February 13th, 2021 minus the mean staffing levels by type on February 6th, 2021. Abbreviations. CNA = Certified Nursing Assistant. HPRD = Hours Per Resident-Day. LPN = Licensed Practical Nurse. RN = Registered Nurse.

Table 2.

Difference in Direct-Care Nurse Staffing During Winter Storm Uri and 1 Week Prior.

Variable	Mean RN HPRD	Mean LPN HPRD	Mean CNA HPRD
One Week before the Storm
Feb. 6th, 2021	0.321	0.925	1.654
Feb. 7th, 2021	0.320	0.907	1.610
Feb. 8th, 2021 (Monday)	0.284	1.040	1.786
Feb. 9th, 2021	0.284	1.054	1.838
Feb. 10th, 2021	0.287	1.066	1.829
Feb. 11th, 2021	0.276	1.045	1.780
Feb. 12th, 2021 (Friday)	0.278	1.025	1.770
Week of winter storm Uri
Feb. 13th, 2021	0.316	0.912	1.631
Feb. 14th, 2021	0.302	0.876	1.480
Feb. 15th, 2021 (Monday)	0.254	0.900	1.399
Feb. 16th, 2021	0.268	0.959	1.570
Feb. 17th, 2021	0.264	0.969	1.609
Feb. 18th, 2021	0.268	0.994	1.679
Feb. 19th, 2021 (Friday)	0.277	1.009	1.708
Difference in staffing
Feb. 13th–6th, 2021	−0.005	−0.013	−0.023
Feb. 14th–7th, 2021	−0.018	−0.031	−0.130
Feb. 15th–8th, 2021	−0.030	−0.140	−0.387
Feb. 16th–9th, 2021	−0.016	−0.095	−0.268
Feb. 17th–10th, 2021	−0.023	−0.097	−0.220
Feb. 18th–11th, 2021	−0.008	−0.051	−0.101
Feb. 19th–12th, 2021	−0.001	−0.016	−0.062
Percent change in staffing
Feb. 13th–6th, 2021	−2%	−1%	−1%
Feb. 14th–7th, 2021	−6%	−3%	−8%
Feb. 15th–8th, 2021	−11%	−13%	−22%
Feb. 16th–9th, 2021	−6%	−9%	−15%
Feb. 17th–10th, 2021	−8%	−9%	−12%
Feb. 18th–11th, 2021	−3%	−5%	−6%
Feb. 19th–12th, 2021	0%	−2%	−4%

Note. Winter Storm Uri began on February 13th, 2021. Sample included 1,173 NHs from February 6th to 12th, 2021, but ranged from 1,168–1,173 from February 13th to 19th, 2021. Percent change is in reference to the staffing levels reported 1 week before Winter Storm Uri.

Abbreviations. CNA = Certified Nursing Assistant. HPRD = Hours Per Resident-Day. LPN = Licensed Practical Nurse. RN = Registered Nurse.

Over the 7 days, Texas NH residents were given approximately 6.06 minutes less time with RNs, 26.58 minutes less time with LPNs, and 71.46 minutes less time with CNAs on average. In a typical Texas NH that cared for 64 residents throughout the week of Winter Storm Uri (state average), this would represent a deficit of 6.46 RN hours, 28.35 LPN hours, and 76.22 CNA hours. At the storm’s worst, Texas NHs experienced a 22% reduction in CNA, 13% reduction in LPN, and 11% reduction in RN staffing levels on average compared to 1 week prior (Table 2).

Supplemental Analyses by Rurality Status and Quality Rating

On average, rural NHs and higher-quality NHs reported slightly fewer staffing deficits than urban NHs and lower-quality NHs, respectively (Figure 2; Figure 3). At the storm’s worst, urban NHs experienced a 24% reduction in CNA, 14% reduction in LPN, and 12% reduction in RN staffing levels on average compared to 1 week prior, while rural NHs reported a 16%, 13%, and 10% reduction, respectively (Table 3). Similarly, lower-quality NHs experienced a 22% reduction in CNA, 14% reduction in LPN, and 9% reduction in RN staffing levels on average compared to 1 week prior, while higher-quality NHs reported a 20%, 10%, and 11% reduction, respectively (Table 4).

Figure 2.

Effect of Winter Storm Uri on rural and urban nursing home staffing levels in hours per resident-day. Note. Winter Storm Uri began on February 13th, 2021. Nursing homes were labeled as urban if they resided in a metropolitan county and rural if they resided in a nonmetropolitan county as defined by the Rural-Urban Continuum Codes. Methodology can be viewed here: https://www.ers.usda.gov/data-products/rural-urban-continuum-codes/documentation/.

Figure 3.

Effect of Winter Storm Uri on higher quality and lower quality nursing home staffing levels in hours per resident-day. Note. Winter Storm Uri began on February 13th, 2021. Higher-quality nursing homes were defined as having an overall quality star rating of 4 or 5 and lower-quality nursing homes were defined as having an overall quality star rating of 1, 2, or 3 from the Care Compare Website. Abbreviations. CNA = Certified Nursing Assistant. LPN = Licensed Practical Nurse. RN = Registered Nurse. HQ = Higher-Quality. LQ = Lower-Quality.

Table 3.

Difference in Direct-Care Nurse Staffing During Winter Storm Uri and 1 Week Prior by Rurality Status.

	Urban NHs (n = 857)			Rural NHs (n = 316)
Difference in staffing	Mean RN HPRD	Mean LPN HPRD	Mean CNA HPRD	Mean RN HPRD	Mean LPN HPRD	Mean CNA HPRD
Feb. 13th–6th, 2021	−0.006	−0.014	−0.027	−0.002	−0.011	−0.011
Feb. 14th–7th, 2021	−0.019	−0.031	−0.141	−0.017	−0.031	−0.100
Feb. 15th–8th, 2021	−0.039	−0.143	−0.422	−0.007	−0.134	−0.288
Feb. 16th–9th, 2021	−0.014	−0.100	−0.283	−0.021	−0.083	−0.226
Feb. 17th–10th, 2021	−0.023	−0.093	−0.223	−0.020	−0.106	−0.213
Feb. 18th–11th, 2021	−0.009	−0.038	−0.102	−0.005	−0.081	−0.101
Feb. 19th–12th, 2021	−0.006	−0.016	−0.072	0.013	−0.016	−0.034
Difference in staffing	% change	% change	% change	% change	% change	% change
Feb. 13th–6th, 2021	−2%	−2%	−2%	−1%	−1%	−1%
Feb. 14th–7th, 2021	−6%	−3%	−9%	−7%	−3%	−6%
Feb. 15th–8th, 2021	−12%	−14%	−24%	−4%	−13%	−16%
Feb. 16th–9th, 2021	−4%	−9%	−15%	−10%	−8%	−12%
Feb. 17th–10th, 2021	−7%	−9%	−12%	−10%	−10%	−12%
Feb. 18th–11th, 2021	−3%	−4%	−6%	−3%	−8%	−6%
Feb. 19th–12th, 2021	−2%	−2%	−4%	7%	−2%	−2%

Note. Winter Storm Uri began on February 13th, 2021. Percent change is in reference to the staffing levels reported 1 week before Winter Storm Uri.

Abbreviations. CNA = Certified Nursing Assistant. HPRD = Hours Per Resident-Day. LPN = Licensed Practical Nurse. RN = Registered Nurse.

Table 4.

Difference in direct-care nurse staffing during Winter Storm Uri and 1 week prior by quality status.

	Higher Quality NHs (Overall Star Rating 4, 5) (n = 391)			Lower Quality NHs (Overall Star Rating 1, 2, 3) (n = 749)
Difference in staffing	Mean RN HPRD	Mean LPN HPRD	Mean CNA HPRD	Mean RN HPRD	Mean LPN HPRD	Mean CNA HPRD
Feb. 13th–6th, 2021	0.003	−0.017	−0.027	−0.008	−0.008	−0.019
Feb. 14th–7th, 2021	−0.014	−0.041	−0.152	−0.016	−0.019	−0.112
Feb. 15th–8th, 2021	−0.045	−0.107	−0.385	−0.020	−0.146	−0.379
Feb. 16th–9th, 2021	−0.019	−0.069	−0.252	−0.013	−0.100	−0.266
Feb. 17th–10th, 2021	−0.027	−0.083	−0.183	−0.021	−0.096	−0.235
Feb. 18th–11th, 2021	−0.005	−0.045	−0.093	−0.008	−0.051	−0.106
Feb. 19th–12th, 2021	0.000	−0.026	−0.053	−0.001	−0.007	−0.068
Difference in staffing	% change	% change	% change	% change	% change	% change
Feb. 13th–6th, 2021	1%	−2%	−2%	−3%	−1%	−1%
Feb. 14th–7th, 2021	−3%	−4%	−9%	−6%	−2%	−7%
Feb. 15th–8th, 2021	−11%	−10%	−20%	−9%	−14%	−22%
Feb. 16th–9th, 2021	−5%	−6%	−13%	−6%	−10%	−15%
Feb. 17th–10th, 2021	−7%	−8%	−10%	−9%	−9%	−13%
Feb. 18th–11th, 2021	−1%	−4%	−5%	−4%	−5%	−6%
Feb. 19th–12th, 2021	0%	−2%	−3%	0%	−1%	−4%

Note. Winter Storm Uri began on February 13th, 2021. Percent change is in reference to the staffing levels reported 1 week before Winter Storm Uri.

Abbreviations. CNA = Certified Nursing Assistant. HPRD = Hours Per Resident-Day. LPN = Licensed Practical Nurse. RN = Registered Nurse.

Discussion

After controlling for facility, quality, and county characteristics, we found significant changes in direct-care staffing levels across Texas NHs during Winter Storm Uri. Specifically, as hypothesized, RN, LPN, and CNA staffing increased over the work week, suggesting a return to normalcy (Table 1). We also found that during the days after storm onset, even before the storm dissipated, the rate of RN and CNA staffing began to increase quadratically while LPN staffing increased linearly. Quadratic changes in staffing reflect a dip in the first few days of the winter storm (eg over the weekend) followed by a recovery toward the prior week’s levels. This relationship likely reflects the facilities’ efforts to restore their staffing levels and the work of road crews to remove the accumulated ice and snow, enabling staff members to safely travel to and from work. Despite these recorded staffing increases, Texas NHs experienced a significant staffing deficit during the storm for RN, LPN, and CNA staff when compared to the week prior on average (Figure 1). This supports our hypothesis that Winter Storm Uri caused major disruptions to staff availability. Overall, our results highlight the contrast between this winter storm and hurricane disasters in which NHs had ample time to make preparations, including ensuring the availability of additional staff (Jester et al., 2021).

Consistent with prior work on poorer staffing response to a major hurricane in lower quality Florida NHs (Jester et al., 2022), we found that lower-quality Texas NHs fared slightly worse during Winter Storm Uri. And while differences between higher quality and lower quality NHs may appear small on the overall scale, these reductions in staffing levels represent a significant issue when considering that lower-quality NHs typically have deficiencies in staffing levels even during ideal circumstances (CMS, 2022). In other words, disaster-related deficits borne by a higher-quality NH may be less likely to affect quality of care due to their pre-storm above-standard staffing levels, whereas deficits borne by a lower-quality NH are likely to exacerbate pre-existing issues with quality of care due to their below-standard staffing levels.

Of note, CNA staffing levels were most affected by Winter Storm Uri, followed by LPNs, and finally RNs. NHs generally operate with more CNAs than LPNs and RNs to be cost-effective, therefore a greater number of CNAs would be at-risk of absenteeism during a natural disaster. When NHs have too few CNAs attending to residents’ needs, quality of care is impacted (Hyer et al., 2011) and resident-level outcomes such as hospitalization are worse (Cherubini et al., 2012). CNA absenteeism and turnover have previously been tied to poorer resident-level outcomes such as greater use of physical restraints and catheters, and a larger proportion of residents with moderate to severe pain and pressure sores (Castle & Ferguson-Rome, 2015). We do not know the level of NH morbidity and mortality related to the events of Winter Storm Uri. However, our results suggest that the risks to residents were likely elevated on the days when many of those they depend on for care were absent (Svitek, 2022). Seeing that direct care is primarily delivered by CNAs, NH administrators should prioritize planning to retain CNAs during winter storms if travel is suspected to be impacted. While solutions require some creativity (eg cost-sharing with an assisted living community if the facility is part of a CCRC, offering over-night compensation for staff and their families, including meals and shelter), they may ultimately prevent the downstream weather-related morbidity and mortality associated with prior disasters (Dosa et al., 2010; 2020; Skarha et al., 2021).

Rural NHs fared slightly better than urban NHs with regard to staffing deficits during Winter Storm Uri. It may be that staff from rural NHs were more likely to rely on personal transportation, whereas staff from urban NHs were more likely to rely on public transit. Texas’ cities only provide above-ground public transportation options (eg bus, trolley, light rail), which are significantly more affected by all types of weather when compared to below-ground public options (ie underground railway or “subway”) (Singhal et al., 2014). Another facet of urban transportation is its reliance on interstate and arterial roadways that provide large volumes of travelers and are therefore subject to multiple-vehicle chain-reaction crashes during winter storms (Call et al., 2018). These travel-related issues may explain some level of the increased absenteeism in urban NHs during Winter Storm Uri.

These findings build on the NH disaster-staffing literature that has focused on hurricanes (Hyer et al., 2009; Jester et al., 2021; 2022; Thomas et al., 2010), the COVID-19 pandemic (Werner & Coe, 2021; Xu et al., 2020), and nuclear disasters (Murakami et al., 2015; Uekusa, 2019), by including an analysis of a winter storm. Winter Storm Uri brought historic levels of ice and snow to Texas, amid persistent low temperatures, but it was uniquely deadly due to the associated power grid collapse (Soergel, 2021; West, 2021). Our work emphasizes the dependence of NHs and other healthcare providers on the resources of their states, cities, and communities to deliver services amid a resource or infrastructure failure. Northern states that routinely experience frigid temperatures and snow accumulation may be better prepared to manage the effects of winter storms, partly due to their infrastructure (eg pipes developed for freezing weather) and availability of services (eg deployment of roadway salt, use of snow plows). However, the North American winter storm of December 2022 (Elamroussi et al., 2022) suggests that the usual steps taken in northern states to prepare for winter weather may not be enough to manage events of the future. As the climate continues to change (Cohen et al., 2021), and with the intensification of both winter and tropical storms in addition to worsening heat and drought (Smith, 2021), preparedness means being ready for unprecedented and extreme events, including events that overwhelm existing infrastructure. Future research is needed to discern how Texas NHs improved their staffing levels after the initial drop and whether the experience influenced their disaster planning or subsequent morbidity and mortality rates.

While these results may be expected, reports following Winter Storm Uri show that some NHs were not prepared for the staff absences and other storm effects, such as power loss, that would be exacerbated by a shortage of staff. Accounts described numerous disruptions, including hurried evacuations of fragile residents on school buses (U.S. Senate, 2023), suggesting that residents of short-staffed NHs may have been at a higher risk of morbidity or mortality during the storm. Considering the stakes, our research reinforces the urgent need for NHs and other care facilities to expect the unexpected in their risk assessments and develop procedures to ensure adequate staffing during disasters in the future. Our findings also add to the age-friendly health system 4M + paradigm (What Matters, Medication, Mentation, Mobility) for disaster preparedness in NHs (Dosa et al., 2023). The fifth “M” proposed by Dosa and colleagues (2023) - Marshalling Resources–highlights the crucial role of direct-care staff in prevention (e.g., determining how resident acuity and characteristics affect sufficient staff levels), response (eg deploying staff hourly to assess for heat- or cold-related illness), and recovery (eg identifying gaps in planning and response) during a natural disaster. We provide the estimated staffing deficit for a week-long winter storm and the expected time until recovery (ie staffing levels began to improve after 72 hours) that may be used by administrators for more accurate planning. We additionally show that the higher-quality NHs exposed to Winter Storm Uri experienced initial staff losses, but recovered faster than lower-quality NHs.

This study has several limitations. RN staffing may be underestimated due to the use of salaried positions (eg hours recorded as 40 per week for salaried employees who may work additional hours). Though we included county fixed-effects and a proxy for COVID-19 response and mitigation (ie results from infection control inspections), these covariates may not have adequately captured the impact of the pandemic on the availability of the direct-care nurse staff workforce. Our study did not have access to ice or snow accumulation (a potential measure of transportation disruption), evacuation, or power loss data, all of which may have impacted the staffing response of facilities (Jester et al., 2021; 2022; Skarha et al., 2021), but our comparison to staffing levels recorded 1 week prior helped to isolate the effect of the storm. That said, we encourage future work to examine these important effects. Multicollinearity was assessed with variance inflation factors, and no issues were found in the models. Given that overall star ratings are correlated with weighted deficiency scores, models were reanalyzed without the weighted deficiency score as a covariate. Study findings did not change.

Conclusion

Like hurricanes, winter storms pose a unique risk to NHs and their ability to maintain direct-care nurse staffing levels. Though Texas NHs eventually increased their RN (quadratically), LPN (linearly), and CNA (quadratically) staffing levels during Winter Storm Uri, facilities experienced a significant staffing deficit when compared to the week prior on average. These findings underscore the importance of disaster planning and mitigation strategies (eg direct-care staff staying overnight within facilities) during winter storms when transportation is expected to be disrupted.

Footnotes

Declaration of Conflicting Interests

The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The author(s) disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: This work was supported by the National Institute on Aging [grant number R01AG060581-01].

ORCID iDs

Dylan J. Jester

Lindsay J. Peterson

References

Antwi

Y. A.

Bowblis

J. R.

(2018). The impact of nurse turnover on quality of care and mortality in nursing homes: Evidence from the great recession. American Journal of Health Economics, 4(2), 131–163. https://doi.org/10.1162/ajhe_a_00096

Call

D. A.

Wilson

C. S.

Shourd

K. N.

(2018). Hazardous weather conditions and multiple‐vehicle chain‐reaction crashes in the United States. Meteorological Applications, 25(3), 466–471. https://doi.org/10.1002/met.1714

Castle

N. G.

Ferguson-Rome

J. C.

(2015). Influence of nurse aide absenteeism on nursing home quality. The Gerontologist, 55(4), 605–615. https://doi.org/10.1093/geront/gnt167

Centers for Medicare & Medicaid Services . (2022). Design for care compare nursing home five-star quality rating system: Technical users’ guide. https://www.cms.gov/medicare/provider-enrollment-and-certification/certificationandcomplianc/downloads/usersguide.pdf

Cherubini

Eusebi

Dell’Aquila

Landi

Gasperini

Bacuccoli

Menculini

Bernabei

Lattanzio

Ruggiero

(2012). Predictors of hospitalization in Italian nursing home residents: The ULISSE project. Journal of the American Medical Directors Association, 13(1), 84.e5–84.e10. https://doi.org/10.1016/j.jamda.2011.04.001

Clemens

Wodchis

McGilton

McGrail

McMahon

(2021). The relationship between quality and staffing in long-term care: A systematic review of the literature 2008–2020. International Journal of Nursing Studies, 122, 104036. https://doi.org/10.1016/j.ijnurstu.2021.104036

Cohen

Agel

Barlow

Garfinkel

C. I.

White

(2021). Linking Arctic variability and change with extreme winter weather in the United States. Science, 373(6559), 1116–1121. https://doi.org/10.1126/science.abi9167

Dellefield

M. E.

Castle

N. G.

McGilton

K. S.

Spilsbury

(2015). The relationship between registered nurses and nursing home quality: An integrative review (2008–2014). Nursing Economic$, 33(2), 95–108, 116.

Dosa

Feng

Hyer

Brown

L. M.

Thomas

Mor

(2010). Effects of Hurricane Katrina on nursing facility resident mortality, hospitalization, and functional decline. Disaster Medicine and Public Health Preparedness, 4(01), S28–S32. https://doi.org/10.1001/dmp.2010.11

10.

Dosa

Jester

Peterson

Dobbs

Black

Brown

(2023). Applying the age‐friendly‐health system 4M paradigm to reframe climate‐related disaster preparedness for nursing home populations. Health Services Research, 58(Suppl 1), 36–43. https://doi.org/10.1111/1475-6773.14043

11.

Dosa

D. M.

Skarha

Peterson

L. J.

Jester

D. J.

Sakib

Ogarek

Thomas

K. S.

Andel

Hyer

(2020). Association between exposure to Hurricane IRMA and mortality and hospitalization in Florida nursing home residents. JAMA Network Open, 3(10), e2019460. https://doi.org/10.1001/jamanetworkopen.2020.19460

12.

Elamroussi

Hanna

Sanches

(2012). December 24. A powerful winter storm claims at least 22 lives across the US as temperatures plunge, winds howl and power lines fall. CNN. https://www.cnn.com/2022/12/24/weather/christmas-arctic-winter-storm-power-outages-saturday/index.html

13.

Geng

Stevenson

D. G.

Grabowski

D. C.

(2019). Daily nursing home staffing levels highly variable, often below CMS expectations. Health Affairs, 38(7), 1095–1100. https://doi.org/10.1377/hlthaff.2018.05322

14.

Hyer

Brown

L. M.

Christensen

J. J.

Thomas

K. S.

(2009). Weathering the storm: Challenges to nurses providing care to nursing home residents during hurricanes. Applied Nursing Research: ANR, 22(4), e9–e14. https://doi.org/10.1016/j.apnr.2008.11.001

15.

Hyer

Thomas

K. S.

Branch

L. G.

Harman

J. S.

Johnson

C. E.

Weech-Maldonado

(2011). The influence of nurse staffing levels on quality of care in nursing homes.The Gerontologist, 51(5), 610–616. https://doi.org/10.1093/geront/gnr050

16.

Jester

D. J.

Peterson

L. J.

Thomas

K. S.

Dosa

D. M.

Andel

(2022). Nursing home compare star rating and daily direct-care nurse staffing during Hurricane Irma. Journal of the American Medical Directors Association, 23(8), 1409–1412.e1. https://doi.org/10.1016/j.jamda.2021.09.038

17.

Jester

D. J.

Thomas

K. S.

Peterson

L. J.

Dosa

D. M.

Andel

Hyer

(2021). Effect of Hurricane Irma on daily direct‐care nurse staffing in nursing homes. Journal of the American Geriatrics Society, 69(8), 2298–2305. https://doi.org/10.1111/jgs.17220

18.

Jutkowitz

Landsteiner

Ratner

Shippee

Madrigal

Ullman

Linskens

Wilt

T. J.

Duan-Porter

(2022). Effects of nurse staffing on resident outcomes in nursing homes: A systematic review. Journal of the American Medical Directors Association, 24(1), 75–81.e11. https://doi.org/10.1016/j.jamda.2022.11.002

19.

Murakami

Ono

Tsubokura

Nomura

Oikawa

Oka

Kami

Oki

(2015). Was the risk from nursing-home evacuation after the Fukushima accident higher than the radiation risk? PloS One, 10(9), e0137906. https://doi.org/10.1371/journal.pone.0137906

20.

Peterson

L. J.

Bowblis

J. R.

Jester

D. J.

Hyer

(2021a). The relationship between staffing levels and consumer complaints in nursing homes. Journal of Aging & Social Policy, 34(5), 742–754. https://doi.org/10.1080/08959420.2021.1962173

21.

Peterson

L. J.

Bowblis

J. R.

Jester

D. J.

Hyer

(2021b). US state variation in frequency and prevalence of nursing home complaints. Journal of Applied Gerontology: The Official Journal of the Southern Gerontological Society, 40(6), 582–589. https://doi.org/10.1177/0733464820946673

22.

U. S. Senate . (2023). Left in the dark: The impact of the 2021 Texas blackout on long-term care residents and the need to improve emergency preparedness. U. S. Senate. https://www.finance.senate.gov/imo/media/doc/02222023%20Left%20in%20the%20Dark%20-%20Wyden-Casey%20final.pdf

23.

Singhal

Kamga

Yazici

(2014). Impact of weather on urban transit ridership. Transportation Research Part A: Policy and Practice, 69, 379–391. https://doi.org/10.1016/j.tra.2014.09.008

24.

Skarha

Gordon

Sakib

June

Jester

D. J.

Peterson

L. J.

Andel

Dosa

D. M.

(2021). Association of power outage with mortality and hospitalizations among Florida nursing home residents after Hurricane Irma. JAMA Health Forum, 2(11), e213900. https://doi.org/10.1001/jamahealthforum.2021.3900

25.

Smith

(2021). 2020 U.S. billion-dollar weather and climate disasters in historical context. https://www.climate.gov/disasters2020?ftag=MSFd61514f

26.

Soergel

(2021). Older Texans among those hardest hit by extreme weather, power failures. https://states.aarp.org/texas/winter-storm-older-texans

27.

Svitek

(2022). Texas puts final estimate of winter storm death toll at 246. https://www.texastribune.org/2022/01/02/texas-winter-storm-final-death-toll-246/

28.

Thomas

K. S.

Hyer

Brown

L. M.

Polivka-West

Branch

L. G.

(2010). Florida’s model of nursing home Medicaid reimbursement for disaster-related expenses. The Gerontologist, 50(2), 263–270. https://doi.org/10.1093/geront/gnp132

29.

Uekusa

(2019). Exploring disasters through the eyes of residential nursing home caregivers. Social Work in Public Health, 34(6), 529–541. https://doi.org/10.1080/19371918.2019.1635942

30.

Werner

R. M.

Coe

N. B.

(2021). Nursing home staffing levels did not change significantly during COVID-19: Study examines US nursing home staffing levels during COVID-19 pandemic. Health Affairs, 40(5), 795–801. https://doi.org/10.1377/hlthaff.2020.02351

31.

West

(2021). Texas winter storm exposes gaps in senior living oversight. https://khn.org/news/article/texas-winter-storm-exposes-gaps-in-senior-living-oversight/

32.

Intrator

Bowblis

J. R.

(2020). Shortages of staff in nursing homes during the COVID-19 pandemic: What are the driving factors? Journal of the American Medical Directors Association, 21(10), 1371–1377. https://doi.org/10.1016/j.jamda.2020.08.002