Inclement Weather and Risk of Missing Scheduled Hemodialysis Appointments among Patients with Kidney Failure

Visual Abstract

Abstract

Background

Nonadherence to hemodialysis appointments could potentially result in health complications that can influence morbidity and mortality. We examined the association between different types of inclement weather and hemodialysis appointment adherence.

Methods

We analyzed health records of 60,135 patients with kidney failure who received in-center hemodialysis treatment at Fresenius Kidney Care clinics across the Northeastern US counties during 2001–2019. County-level daily meteorological data on rainfall, hurricane and tropical storm events, snowfall, snow depth, and wind speed were extracted using National Oceanic and Atmosphere Agency data sources. A time-stratified case-crossover study design with conditional Poisson regression was used to estimate the effect of inclement weather exposures within the Northeastern US region. We applied a distributed lag nonlinear model framework to evaluate the delayed effect of inclement weather for up to 1 week.

Results

We observed positive associations between inclement weather and missed appointment (rainfall, hurricane and tropical storm, snowfall, snow depth, and wind advisory) when compared with noninclement weather days. The risk of missed appointments was most pronounced during the day of inclement weather (lag 0) for rainfall (incidence rate ratio [RR], 1.03 per 10-mm rainfall; 95% confidence interval [CI], 1.02 to 1.03) and snowfall (RR, 1.02; 95% CI, 1.01 to 1.02). Over 7 days (lag 0–6), hurricane and tropical storm exposures were associated with a 55% higher risk of missed appointments (RR, 1.55; 95% CI, 1.22 to 1.98). Similarly, 7-day cumulative exposure to sustained wind advisories was associated with 29% higher risk (RR, 1.29; 95% CI, 1.25 to 1.31), while wind gusts advisories showed a 34% higher risk (RR, 1.34; 95% CI, 1.29 to 1.39) of missed appointment.

Conclusions

Inclement weather was associated with higher risk of missed hemodialysis appointments within the Northeastern United States. Furthermore, the association between inclement weather and missed hemodialysis appointments persisted for several days, depending on the inclement weather type.

Introduction

Missing scheduled hemodialysis appointments can result in higher risk of morbidity, hospitalization, and mortality.¹ Nonadherence to scheduled hemodialysis can vary across age groups, race/ethnicity, and educational status. Previous studies have reported a higher prevalence of missed treatments among Black, Hispanic, and Native American patients within the United States.² In addition, male patients and patients with lower attained education status (a high school education or below), on average, were less likely to adhere to scheduled appointments.³ Other factors such as short hemodialysis duration, unreliable transportation, holidays, gastrointestinal upset, chronic pain, and psychiatric illness were also barriers to hemodialysis adherence.⁴ Patient autonomy and comfort may also influence nonadherence among US in-center hemodialysis patients.⁵

Although multiple studies have explored determinants for nonadherence,^1–5 thus far, very few studies have investigated weather as a predictor for missed hemodialysis treatments.⁴ One such study considered same-day snowfall above or below 1 inch (25.4 mm) as a representation of inclement weather but did not investigate other severe weather types that could potentially influence treatment nonadherence.⁴ Inclement weather can interrupt electrical power, damage infrastructure, and overwhelm health care systems, hindering dialysis treatment.^6,7 A 2009 study reported increased missed appointments and hospital admissions among dialysis patients in New Orleans after Hurricane Katrina.⁷ Such disturbances can lead to unintended complications, morbidities, and premature death from missed treatments.^8–10 Others have reported higher hospitalization risk among patients with endocrine disorders and genitourinary diseases residing in counties the week after a hurricane event was observed.¹¹ In light of this, there is a need to characterize how inclement weather affects dialysis treatment adherence. Such data can enhance preparedness for severe weather events, especially with ongoing climate change.

In this observational study, we examined the association between different types of inclement weather (rainfall, snowfall, snow depth, wind advisory, and hurricane and tropical storm) and hemodialysis appointment adherence among patients with kidney failure in the Northeastern United States. We hypothesized that the short-term risk of missed hemodialysis appointments would be higher after inclement weather events compared with noninclement weather days.

Methods

Study Population

We obtained deidentified electronic health records of patients with kidney failure 19 years or older and receiving in-center hemodialysis treatment at Fresenius Kidney Care facilities across the Northeastern United States between 2001 and 2019. The eligible study participants included patients with kidney failure (19 years or older) who had hemodialysis treatment during the study period. There were a total of 60,717 patients with kidney failure from 99 clinics within 27 counties (Figure 1), of which of 60,135 met the inclusion criteria and were included in the analysis. We used clinic ZIP codes where patients received treatment as a proxy for a residence to assign county-specific exposures. We defined missed appointments as events when a patient failed to show up for a scheduled treatment without an excuse. These exclude missed treatment due to prearranged travel or hospitalizations. Counts of missed appointments were aggregated for each day and by county. This study was considered exempt from human subjects review by the University of Maryland Institutional Review Board.

Figure 1.

Map of focused counties within Northeastern United States.¹² A total of 99 Fresenius Kidney Care (FKC) clinics within 27 counties. Figure 1 can be viewed in color online at www.cjasn.org.

Exposure

On the basis of the ZIP codes of the Fresenius Kidney Care clinics, we verified the county of each clinic by using ZIP Code Tabulation Areas.¹³ Subsequently, we selected the nearest weather station from each county's centroid (longitude and latitude) through the Global Historical Climatology Network (GHCN) database maintained by National Oceanic and Atmospheric Administration. Using the rnoaa R package, we extracted the following data elements for each patient from the GHCN database: total daily precipitation (amount of all types of precipitation, including melted and frozen) from April to October, total daily snowfall (amount of snowfall since the previous 24 hours observation) of November–March, total daily snow depth (depth of the new and old snow remaining on the ground at 24 hours observation time) of November–March, sustained wind speed (the fastest speed for the day that is sustained for at least 2 minutes), and peak wind speed (the fastest speed for the day that is sustained for at least 5 seconds) from 2001 to 2019.^14–16 Rainfall-related analyses were restricted to warmer months (April–October) since the total daily precipitation combined melted and frozen types of rainfall, while snowfall- and snow depth–related analyses were restricted to colder months (November–March). Owing to the limited availability for snow-related data, we restricted the study area to 13 counties for snowfall-related and snow depth–related analysis (Supplemental Table 1). Missing values for each inclement weather data were <9%, and we used all available data in our analysis.

Daily rainfall, snowfall, and snow depth were analyzed as continuous variables. We applied formalized wind categories using the National Weather Service definitions for wind advisory on the basis of sustained wind speed >31 mph or 26.9 knots (13.86 m/s) and peak wind speed >46 mph or 40 knots (20.56 m/s).¹⁷ We used the hurricaneexposuredata R package to identify hurricane and tropical storm events from May to October that affected our study area from 2001 to 2018.¹⁸ The package includes processed hazards-related data on county-level Atlantic basin tropical storms along the Eastern portion of the United States for counties within at least 250 km from the storm track. For our main analysis, we defined a county as exposed to a hurricane and tropical storm event when gale force sustained wind speeds reach 34 knots (17.49 m/s) or above brought on by a hurricane storm event.^11,18,19 As part of a sensitivity analysis, we explored distance-based thresholds in defining tropical storm–related exposures by including hurricane storms between a county's population-weighted center and storm track distances of 150 and 200 km.

Statistical Analyses

We used a time-stratified case-crossover study design to investigate the association between inclement weather events and missed hemodialysis appointments. Case-crossover methods are consistently used in epidemiological analyses involving acute exposures and clearly defined event-based outcomes. In our case-crossover design, exposure immediately preceding the missed appointments (case period) was compared with exposure during multiple control periods that included 7, 14, 21, or 28 days before or after the case period. The use of time-stratified referents helped us avoid overlap biases often associated with control periods adjacent to case periods. Inclement weather exposures for case and control periods were compared using stratum indicators. Self-matching is a unique feature of the case-crossover design that eliminates the need to adjust for individual-level time-invariant confounders, including age, sex, race, and socioeconomic status.²⁰ For these reasons, case-crossover design is ideal for investigating acute outcomes related to short-term exposure. Stratum indicators were based on year, month, day of the week, and county.^21,22 We used conditional Poisson regression with aggregated daily missed appointments as the outcome and inclement weather type as the main exposure.²³ We included the day of the week as the covariate and an offset variable equaling the natural log of the monthly average number of scheduled appointments for each county to account for varying populations.²³ We adopted a case-crossover analysis within a distributed lag nonlinear model (DLNM) framework to characterize lag effects for up to 7 days. DLNM provides the flexibility to simultaneously model potential nonlinear associations between inclement weather and missed appointments, and how this association changes over time, that is, time elapsed since exposure onset. As such, we used DLNM to obtain risk for individual lag days (e.g., risk at lag 0 referring to the risk of missing scheduled hemodialysis appointment during the day of inclement weather exposure, and lag 7 referring to the risk of missing scheduled hemodialysis appointment 7 days after the inclement weather exposure) and cumulative effect up to 7 days (lag 0–6), which is computed by factoring all the contributions across lags.²⁴ We also conducted stratified analyses by sex (female and male) and race/ethnicity (Hispanic, non-Hispanic Black, non-Hispanic White, Asian, and other) to investigate whether risk associated with inclement weather varied by sex and race/ethnicity.

All analyses were conducted using R statistical software version 3.6.1 with dlnm, gnm, and dplyr packages.^25–28 All statistical tests were two-tailed and based on a significance level of 0.05.

Results

A total of 60,135 patients with kidney failure visited 99 Fresenius Kidney Care facilities in the Northeastern United States from 2001 to 2019 (Table 1, Supplemental Tables 2 and 3). Most patients were male (57%) and were either non-Hispanic Black (40%) or non-Hispanic White (40%). The study population's total visits and missed appointments were 16,612,373 and 454,932, respectively. Overall, 28,495 (47%) patients reported missing at least one hemodialysis session, and 29% reported missing three or more sessions (Table 1).

Table 1.

Characteristics of adults receiving in-center hemodialysis treatment at Fresenius Kidney Care facilities across the Northeastern United States between 2001 and 2019

Characteristics
Counties, n	27
Clinics, n	99
Patients, n	60,135
Total number of visits to dialysis clinics, n	16,612,373
Average number of visits per patient	274 (385)
Average number of visits per patient per year (SD)	94 (61)
Total number of missed dialysis clinic appointments, n	454,932
No. of missed appointments per patient, n (%)
0	31,640 (53)
1	7081 (12)
2	3752 (6)
≥3	17,662 (29)
Age at initial treatment, n (%)
<40	5224 (9)
40–49	8123 (14)
50–59	13,272 (22)
60–69	14,929 (25)
70–79	12,146 (20)
≥80	6441 (10)
Sex, n (%)
Female	25,158 (42)
Male	34,598 (57)
Not reported	379 (0.6)
Race/ethnicity, n (%)
Hispanic	4666 (8)
Non-Hispanic Black	23,810 (40)
Non-Hispanic White	23,813 (40)
Asian American	935 (2)
Other	465 (0.8)
Not reported	6446 (10)

Summary statistics for inclement weather types across focused counties within the Northeastern United States from 2001 to 2019 are presented in Table 2. Regionally, we observed averaged daily rainfall of 3.1 mm and a maximum daily measure of 221.2 mm. During the colder months (November to March), daily average snowfall and snow depth recorded were 6.6 and 22.5 mm, respectively. Similarly, the average daily sustained wind speed was 8.4 m/s and ranged from 1.3 to 28.6 m/s. The daily peak wind speed ranged from 1.3 to 61.7 m/s, with an average wind speed of 11.0 m/s. Both wind advisories on the basis of sustained wind and peak wind were the most frequent during the winter (7% and 3%, respectively). In 2001, 2002, 2009, 2010, 2015, and 2016, none of the 27 counties experienced hurricanes and tropical storms with gale force wind speed ≥34 knots (17.49 m/s). We observed that 2012 had the highest number of counties affected by hurricanes and tropical storm events (n=15).

Table 2.

Daily measures of inclement weather in 27 counties in the Northeastern United States (2001–2019)

Inclement Weather Type	Minimum	Median	Mean (SD)	Maximum
Rainfall, mma	0.0	0.0	3.1 (9.2)	221.2
Snowfall, mmb	0.0	0.0	6.6 (29.5)	769.6
Snow depth, mmc	0.0	0.0	22.5 (66.1)	863.6
Sustained wind speed, m/sd	1.3	7.6	8.4 (2.8)	28.6
Peak wind speed, m/se	1.3	10.3	11.0 (3.8)	61.7

Seasons	Frequency by Season (%)
	Wind Advisory (Sustained Wind >13.86 m/s)	Wind Advisory (Peak Wind >20.56 m/s)
Spring (March–May)	5% (1835/36,887)	2% (724/36,887)
Summer (June–August)	2% (617/37,102)	0.8% (283/37,102)
Fall (September–November)	3% (1116/36,983)	1% (443/36,983)
Winter (December–February)	7% (2385/36,105)	3% (931/36,105)

No. of Hurricane and Tropical Storm‒Affectedf Counties by Year
Year	No. of Counties	Year	No. of Counties
2001	0	2010	0
2002	0	2011	4
2003	2	2012	15
2004	2	2013	12
2005	3	2014	1
2006	6	2015	0
2007	6	2016	0
2008	11	2017	4
2009	0	2018	1

^aRainfall: daily amount of precipitation for the day from April to October.¹⁶

^bSnowfall: daily amount of snowfall from November to March. Record of the snowfall (snow and ice pellets) since the previous snowfall observation (24 hours).¹⁶

^cSnow depth: daily reading of snow on the ground from November to March. Depth of the new and old snow remaining on the ground at observation time (24 hours).¹⁶

^dSustained wind speed: the fastest speed for the day that is sustained for at least 2 minutes.¹⁶

^ePeak wind speed: the fastest speed for the day that is sustained for at least 5 seconds.¹⁶

^fHurricane and tropical storm: storms with wind speed ≥34 knots (from May to October).¹⁸

The percentage of missed appointments by inclement weather type is presented in Table 3. Overall, the average rates were higher on the exposure days compared with nonexposure days for all the inclement weather types. The 27 counties recorded 35,858 county days with rainfall >0 mm between 2001 and 2019. The average percentage of missed hemodialysis appointments during those days was 2.5% and was not different from days with no rainfall. There were 67 county days with hurricane and tropical storm events, and the average percentage of missed appointments during such days was 7.8% compared with 2.4% for the nonhurricane and tropical storm days. During the cold season (November to March), there were a total of 3634 county days with snowfall >0 mm. The average percentage of missed appointments during these days was 4.9% compared with 3.4% for the nonsnowfall days. Similarly, the percentage of missed appointments was higher for days with snow depth >0 mm. Similarly, the average rate of missed appointments was higher on wind advisory days, regardless of the wind speed expression (sustained wind speed or peak wind speed).

Table 3.

Average percentages of daily missed hemodialysis appointments by inclement weather for 27 counties in the Northeastern United States

Inclement Weather Type	County-Days	Average % of Missed Appointments (SD)
Rainfall >0 mm	35,858	2.5% (6.1)
Rainfall=0 mm	62,880	2.5% (6.5)
Hurricane and tropical storm	67	7.8% (19.2)
No hurricane and tropical storm	69,832	2.4% (6.4)
Snowfall >0 mm	3634	4.9% (10.4)
Snowfall=0 mm	21,858	3.4% (8.3)
Snow depth >0 mm	4720	4.3% (9.2)
Snow depth=0 mm	20,126	3.5% (8.6)
Wind advisory sustained	5953	3.6% (6.5)
No wind advisory sustained	140,987	2.6% (9.0)
Wind advisory gusts	2381	4.0% (6.5)
No wind advisory gusts	143,509	2.6% (9.9)

The time-dependent association between inclement weather types and missed hemodialysis appointments is shown in Figure 2. In general, the risk of missing scheduled appointments tended to be highest during the day of inclement weather. A 10-mm higher rainfall was associated with a 2.6% higher risk of missed appointments the same day (lag 0 rate ratio [RR], 1.02; 95% confidence interval [CI], 1.02 to 1.03), and the risk declined over the subsequent 7 days (Figure 2). For every 10-mm greater snowfall and snow depth, the RR of same day (lag 0) missed appointments was 1.02 (95% CI, 1.01 to 1.02) and 1.02 (95% CI, 1.01 to 1.02), respectively. The presence of wind advisory on the basis of sustained wind speed (>13.86 m/s) was associated with a 5.3% higher risk of (lag 0) missed appointments (RR, 1.05; 95% CI, 1.03 to 1.07). Similarly, days with wind advisory on the basis of wind gusts of >20.56 m/s was associated with a 9.6% higher rate of missed appointments (RR, 1.10; 95% CI, 1.07 to 1.12). Higher hurricane and tropical storm–related risk of missed appointments was seen at lag 0 (RR, 1.38; 95% CI, 1.17 to 1.63). The snowfall-related risk persisted for up to 7 days, while snow depth‒related risk lasted for up to 2 days. Significant wind advisory‒related risks persisted for up to at least 2 days for peak and sustained wind speeds (Figure 2).

Figure 2.

Lag-specific association between exposure to inclement weather type and risk of missed hemodialysis. Regression models included the day of the week as the covariate and an offset variable equaling the natural log of the monthly average number of scheduled appointments for each county. CI, confidence interval; RR, rate ratio.

For the continuous inclement weather types, we observed positive exposure-response linear trends when considering associations between 7-day cumulative rainfall, snowfall, snow depth, and missed hemodialysis appointments (Figure 3). A ten-unit higher rainfall, snowfall, and snow depth was associated with 3.8%, 5.2%, and 2.7% higher risk of missed appointments, respectively.

Figure 3.

Association between 7-day cumulative (lag 0–6) exposure to inclement weather type and risk of missed hemodialysis appointment. Inclement weather types were analyzed as a continuous variable. Incidence RRs are presented as black lines with 95% CI as gray region.

Table 4 presents the association between 7-day cumulative exposure (lag 0–6) to hurricane and tropical storm, sustained wind advisory, and wind gust advisory (as categorical variables) on missed hemodialysis appointment. In this analysis, hurricane and tropical storm event was associated with a higher risk of missed appointment (RR, 1.55; 95% CI, 1.22 to 1.98). Both wind advisories on the basis of sustained winds and wind gusts were associated with a significantly higher risk of a missed appointment. Wind advisory on the basis of sustained wind speed showed overall cumulative risk of 1.29 (95% CI, 1.25 to 1.31), while wind advisory on the basis of wind gusts was associated with a 34% higher risk (RR, 1.34; 95% CI, 1.29 to 1.39). When stratified by sex (female and male) and race/ethnicity (Hispanic, non-Hispanic Black, non-Hispanic White, Asian, and other), we did not observe notable effect modification (Supplemental Figures 1–7 and Supplemental Table 4). When redefining tropical storm–related exposures in a sensitivity analysis, we observed a significant positive association with same-day (lag 0) hurricane and tropical storm within 150 km (RR, 1.20; 95% CI, 1.07 to 1.36) and with 5-day lag (RR, 1.27; 95% CI, 1.09 to 1.47). Hurricanes and tropical storms within 200 km and within 250 km showed attenuated risk at lag 0 but higher risks at lag 5 (RR, 1.27; 95% CI, 1.09 to 1.47) and lag 2 (RR, 1.17; 95% CI, 1.05 to 1.30), respectively (Supplemental Figure 8).

Table 4.

Association between 7-day cumulative exposure (lag 0–6) to inclement weather and missed hemodialysis appointment (incidence rate ratios and 95% confidence intervals)

Inclement Weather Type	RR (95% CI)
Hurricane and tropical storm	1.55 (1.22 to 1.98)
Wind advisory (sustained winds)	1.29 (1.25 to 1.31)
Wind advisory (wind gusts)	1.34 (1.29 to 1.39)

RR, rate ratio; CI, confidence interval.

Discussion

This study investigated the effect of inclement weather on missed hemodialysis appointments in the 27 counties in the Northeastern United States from 2001 to 2019. Overall, the rate of missed appointments was higher during inclement weather (rainfall, hurricane and tropical storms, snowfall, snow depth, and wind advisory) compared with noninclement weather days. We observed that the risk of missed appointments was highest at lag 0 and associated with rainfall, hurricane and tropical storm, snowfall, snow depth, and wind advisories on the basis of sustained winds and wind gusts. In addition, the risks of missed appointments associated with rainfall, hurricane and tropical storm, snowfall, snow depth, and wind advisory were substantial when considering a week-long cumulative lag structure (lag 0–6). Overall, we observed rainfall-related risk of missed appointments dissipating after 1 day, whereas the risk associated with snowfall, snow depth, and wind advisories persisted for several days. This is not surprising given that snow tends to remain on the ground several days after snowfall, assuming temperatures stay below freezing. At the same time, wind advisories may cause potential physical infrastructure damage leading to sustained disruption, such as road closures, building damage, and electrical outages. Hurricane and tropical storm–related risk of missed appointment was not significant after a hurricane and tropical storm event (lag 1–4). This might be explained by potential prescheduling ahead of the predicted severe storm. According to a Fresenius Kidney Care Disaster Response representative, missed appointments after a storm are primarily due to safety concerns, transportation barriers from impassable roads, displacement after an evacuation, or a preference not to visit a backup facility (B. Loeper, Disaster Response Specialist at Fresenius Kidney Care, personal communication, September 22, 2022).

Inclement weather conditions can become a major barrier to accessing necessary health care services. In particular, patients with kidney failure who regularly receive outpatient dialysis treatments are more vulnerable to potential complications from missing their regularly scheduled treatments.⁵ Previous studies have shown that missing scheduled hemodialysis appointments can result in adverse outcomes, including hyperkalemia, hyperphosphatemia, pulmonary edema, metabolic disorders, and higher risk of hospitalization and mortality.^1,3 Weather disturbances can act as a hindrance to timely dialysis treatments. Missing a single appointment may not immediately cause a life-threatening problem for some patients with kidney failure. However, multiple consecutive missed appointments because of protracted clinic closures caused by power outages, unsafe road conditions, and the shortage of available transportation can cause difficulties for patients to travel to seek treatment and for clinicians to deliver quality life-saving care. These hindrances can pose life-threatening burdens to both patients and clinicians. This challenge has been well documented in regions struck by Hurricanes Katrina, Sandy, and Maria and Winter Storm Uri.^7,8,29–31 After Hurricane Katrina in August 2005, 94 dialysis clinics closed for at least 1 week in the Gulf Coast states, including Louisiana, Mississippi, and Alabama.²⁹ A study investigated the factors associated with missed dialysis sessions after Hurricane Katrina and reported that the probability of missing three or more sessions was greater among those who lived alone before the hurricane's landfall, who did not evacuate before the storm, or who were placed in storm shelters.⁷ Similarly, patients with kidney failure in Sandy-affected areas in 2012 had more frequent emergency department visits, hospitalizations, and 30-day mortality compared with patients with kidney failure in unaffected areas.⁸ In Puerto Rico, the landfall region of Hurricanes Irma and Maria in September 2017 experienced catastrophic damage to most medical infrastructure and paralyzed dialysis operations that resulted in mandatory patient evacuations.³⁰ Winter storm Uri in 2021 caused mass power outages throughout Texas and affected approximately 54,000 dialysis patients.³¹

The Intergovernmental Panel on Climate Change report suggests imminent and increased extreme weather magnitudes, duration, and frequencies with substantial alterations in regional patterns of extreme weather.³² Our findings suggest that such increases in extreme event will have a considerable effect on patients with kidney failure by disrupting their dialysis care due to unfavorable transportation conditions and/or damaged infrastructure. Clinical and public health resources are crucial to ensure continuity in delivering life-saving health care to medically susceptible communities, such as patients with kidney failure. Specifically, coordination between health care providers and dialysis centers can reinforce medical care preparedness and delivery during extreme weather events.^33,34 Future studies should consider whether peritoneal dialysis may offer benefits, particularly in areas characterized by higher frequency of extreme events.

This study has several strengths. To the best of our knowledge, this is the first study that examined the association between various inclement weather types and hemodialysis appointments among patients with kidney failure. Previous studies have focused on single hazards, such as snowfall⁴ or hurricane,^7,8,35 on hospitalization and mortality. As part of a more robust study on inclement weather exposures, we included different types of hazards, such as rainfall, hurricane and tropical storms, snow depth, wind advisories, and high wind advisories. In addition, this study included a relatively large sample of patients with kidney failure in the Northeastern United States. The dialysis treatment records were maintained by a global health care company that provides hemodialysis services. We applied a time-stratified case-crossover design, which is practical for estimating the acute effect of short-term exposure to inclement weather events.³⁶ Finally, the DLNM framework enabled us to estimate the delayed effect of exposures due to inclement weather events.

This study also has several limitations. We used each hemodialysis clinic's ZIP code as a proxy of patients' residential location. For the measurement of inclement weather, spatial heterogeneity may exist within the counties. However, potential exposure misclassification errors resulting from the use of weather stations for each county were likely to be nondifferential as the stations did not change between the case period and the control periods.³⁷ The nondifferential exposure misclassification, if existed, likely attenuated the risk estimates.³⁸ Finally, future work needs to consider the localized role of inclement weather events caused by tropical storm events, extreme precipitation, snowstorms, and severe winds and quantify its effects on health complications associated with missed appointments.

Our data suggest that inclement weather events can be a major impediment to seeking scheduled hemodialysis appointments. This study showcased the effect of inclement weather conditions at finely configured temporal scales within Eastern United States. Our findings point to the need for enhanced adaptation strategies informed by robust early warning systems to minimize treatment disruption of patients with highly vulnerable kidney failure.

Supplementary Material

cjasn-18-904-s001.pdf ^{(879.6KB, pdf)}

Disclosures

P. Kotanko reports employment with Renal Research Institute, a wholly owned subsidiary of Fresenius Medical Care; stock in Fresenius Medical Care; research funding from Fresenius Medical Care, KidneyX, and NIH; honoraria from HSTalks; multiple patents in the kidney space; and serving on the Editorial Boards of Blood Purification, Frontiers in Nephrology, and Kidney and Blood Pressure Research. R.A. Lasky reports employment with Fresenius Medical Care. F.W. Maddux reports employment with, stock in, research funding from, and patents or royalties from Fresenius Medical Care. F.W. Maddux reports advisory or leadership role for CSL-Vifor FMC Renal Pharma and other interests or relationships with AMA, ASN, and Renal Physicians Association. J.G. Raimann reports employment with Renal Research Institute, a wholly owned subsidiary of Fresenius Medical Care; stock in Fresenius Medical Care; and other interests or relationships as a member of the Board of Directors of “Easy Water for Everyone” (501c3). A. Sapkota reports ownership interest in Bloom Energy, Boeing, Carnival Corp, Lucid Motor, NCLH, Rivian, Solaredge Technologies, and Tesla and other interests or relationships with International Society for Environmental Epidemiology. All remaining authors have nothing to disclose.

Funding

R.V. Remigio reports grant support from the Agency for Healthcare Research and Quality (Grant No. R36HS027716). R.V. Remigio was also supported by NRT-INFEWS: UMD Global STEWARDS (STEM Training at the Nexus of Energy, WAter Reuse, and FooD Systems) that was awarded to the University of Maryland School of Public Health by the National Science Foundation National Research Traineeship Program, Grant No. 1828910.

Author Contributions

Conceptualization: Richard V. Remigio, Amir Sapkota.

Data curation: Xin He, Peter Kotanko, Rachel A. Lasky, Franklin W. Maddux, Jochen G. Raimann, Richard V. Remigio.

Formal analysis: Richard V. Remigio, Hyeonjin Song.

Investigation: Xin He, Franklin W. Maddux, Jochen G. Raimann, Richard V. Remigio, Amir Sapkota, Hyeonjin Song.

Methodology: Xin He, Peter Kotanko, Rachel A. Lasky, Richard V. Remigio, Amir Sapkota, Hyeonjin Song.

Project administration: Amir Sapkota.

Resources: Richard V. Remigio, Amir Sapkota.

Software: Rachel A. Lasky, Jochen G. Raimann, Hyeonjin Song.

Supervision: Peter Kotanko, Jochen G. Raimann, Amir Sapkota.

Validation: Richard V. Remigio, Hyeonjin Song.

Visualization: Richard V. Remigio, Hyeonjin Song.

Writing – original draft: Richard V. Remigio, Hyeonjin Song.

Writing – review & editing: Xin He, Peter Kotanko, Rachel A. Lasky, Franklin W. Maddux, Jochen G. Raimann, Richard V. Remigio, Amir Sapkota, Hyeonjin Song.

Data Sharing Statement

Exposure data will be made publicly available, with no restriction. Patient data represent treatment records and will not be shared. Interested parties will need to contact Renal Research Institute directly to obtain necessary approval and sign data use agreement.

Supplemental Material

This article contains the following supplemental material online at http://links.lww.com/CJN/B760.

Supplemental Table 1. Number of county days and missingness for each inclement weather.

Supplemental Table 2. Summary of the follow-up periods of the 60,135 patients with kidney failure during 2001–2019.

Supplemental Table 3. Number of patients with kidney failure per year.

Supplemental Table 4. Association between 7-day cumulative exposure (lag 0–6) to inclement weather and missed hemodialysis appointment (incidence rate ratios [RRs] and 95% confidence intervals [CIs]) stratified by sex and race/ethnicity.

Supplemental Figure 1. Lag-specific association between exposure to inclement weather type and risk of missed hemodialysis stratified by sex and race/ethnicity.

Supplemental Figure 2. Lag-specific association between exposure to hurricane and tropical storm and risk of missed appointments stratified by sex and race/ethnicity.

Supplemental Figure 3. Lag-specific association between exposure to snowfall and risk of missed appointments stratified by sex and race/ethnicity.

Supplemental Figure 4. Lag-specific effects on missed hemodialysis appointments in incidence rate ratios (RRs) and 95% confidence intervals (CIs) for snow depth over 7 days of lag stratified by sex and race/ethnicity.

Supplemental Figure 5. Lag-specific association between exposure to wind advisory (sustained winds) and risk of missed appointments stratified by sex and race/ethnicity.

Supplemental Figure 6. Lag-specific association between exposure for exposure to wind advisory (wind gusts) and risk of missed appointments stratified by sex and race/ethnicity.

Supplemental Figure 7. Association between 7-day cumulative (lag 0–6) exposure to inclement weather type and risk of missed hemodialysis appointment stratified by sex and race/ethnicity.

Supplemental Figure 8. Lag-specific association between exposure to hurricane and tropical storm and risk of missed appointments. Exposure is based on alternative hurricane and tropical storm definitions (distance to storm track at 150 km or less, distance to storm track at 200 km or less, and distance to storm track at 250 km or less).