Risk of Hospitalization and Mortality Following Medically Attended Norovirus Infection—Veterans Health Administration, 2010–2018

Abstract

Background

While prior studies have suggested a role for norovirus gastroenteritis in contributing to severe morbidity and mortality, the importance of norovirus as a causal pathogen for hospitalization and mortality remains poorly understood. We estimated the effect of laboratory-confirmed norovirus infection on hospitalization and mortality among a national cohort of veterans who sought care within the Veterans Affairs health care system.

Methods

We analyzed electronic health record data from a cohort study of adults who were tested for norovirus within the Veterans Affairs system between 1 January 2010 and 31 December 2018. Adjusted risk ratios (aRRs) for hospitalization and mortality were estimated using log-binomial regression models, adjusting for age, Clostridioides difficile, underlying medical conditions, and nursing home residence.

Results

In total, 23 196 veterans had 25 668 stool samples tested for norovirus; 2156 samples (8.4%) tested positive. Testing positive for norovirus infection, compared with testing negative, was associated with a slight increased risk of hospitalization (aRR, 1.13 [95% confidence interval, 1.06–1.21]) and a significant increased risk of mortality within 3 days after the norovirus test (2.14 [1.10–4.14]). The mortality aRR within 1 week and 1 month were reduced to 1.40 (95% confidence interval, .84–2.34) and 0.97 (.70–1.35), respectively.

Conclusions

Older veterans with multiple comorbid conditions were at a slight increased risk of hospitalization and significant increased risk of mortality in the 3 days after a norovirus-positive test, compared with those testing negative. Clinicians should be aware of these risks and can use these data to inform clinical management for veterans with norovirus.

Norovirus infection was associated with a slight increased risk of hospitalization and significant increased risk of mortality within 3 days of norovirus diagnosis among a large national cohort of US veterans.

Noroviruses cause acute gastroenteritis (AGE) across all age groups and result in a substantial disease burden of an estimated 21 million cases of AGE annually in the United States [1]. In the majority of cases, norovirus-associated AGE is mild and self-limiting, but older adults, particularly those with underlying medical conditions, are vulnerable to more severe complications [2, 3]. Prior studies have indicated that immunosenescence or the presence of underlying medical conditions could exacerbate or extend AGE symptoms. While the exact causal mechanisms for morbidity are unknown and likely multifactorial, studies have reported severe complications following norovirus infections, such as severe dehydration, aspiration pneumonia, sepsis, severe malnutrition, bowel perforation, acute renal failure, and ventricular arrythmias [2, 4–6]. These complications may necessitate hospitalization of individuals with norovirus-associated AGE and in some instances may result in death. Each year, norovirus is estimated to cause approximately 900 deaths and 109 000 hospitalizations in the United States, with 86% of norovirus-associated deaths and 43% of norovirus-associated hospitalizations occurring in adults aged ≥65 years [7].

Strong evidence around the risk of mortality associated with norovirus is lacking, given that death is a rare outcome for this illness and that clinician-directed laboratory testing for norovirus is underused [8, 9]. Prior studies of norovirus-associated mortality have relied on indirect modeling, focused on outbreaks among nursing homes, or evaluated case reports where norovirus-associated symptoms were reported at the time of death [7, 10–12]. However, systematic, large-scale, laboratory-confirmed evaluations assessing the effect of norovirus on severe outcomes, such as hospitalization and mortality, are warranted to strengthen this evidence base and inform prevention efforts.

The objective of the current study was to estimate the effect of laboratory-confirmed norovirus infection on the risk of hospitalization and mortality, particularly short-term mortality, among a national cohort of US veterans. US veterans represent a potentially high-risk population for severe norovirus outcomes, given their older age and high rates of underlying medical conditions [13]. Strengthening the evidence base for the association of laboratory-confirmed norovirus with severe morbidity outcomes will help guide norovirus prevention, testing, treatment, and control efforts, in particular guiding future norovirus vaccination policy decisions.

METHODS

The Veterans Health Administration (VHA) encompasses the largest integrated health care system in the United States, providing care to approximately 6.81 million patients across 1297 care sites nationwide [14]. Electronic health record (EHR) data were obtained from the Corporate Data Warehouse (which includes inpatient and outpatient medical encounters) as well as VHA's Praedico public health surveillance system, which compiles public health data across multiple EHR domains, including diagnostic testing [15]. We conducted a cohort study of adults (aged ≥18 years) who were tested for norovirus within the VHA between 1 January 2010 and 18 December 2018.

Norovirus infection was confirmed by multiple polymerase chain reaction assays conducted by both internal Department of Veterans Affairs (VA) and commercial laboratories. Diagnostic tests for a patient conducted within 30 days of one another were deduplicated by retaining the earliest test result or the norovirus-positive result. The 2 primary outcomes were acute hospitalization and all-cause mortality. Acute hospitalization was defined as an inpatient admission within 3 days before or after the date of stool sample collection for norovirus testing. Patients admitted to the hospital for >3 days before their norovirus test were excluded to minimize nosocomial infections and to restrict that population to patients at risk of requiring hospitalization. Sensitivity analyses were conducted varying the definition of acute hospitalization to inpatient admissions within 7, 10, or 14 days before or after norovirus diagnostic testing. All-cause mortality was defined using the date of death following a norovirus diagnostic test and was assessed at 3, 7, and 30 days after a norovirus diagnostic test. Deaths occurring both inside and outside the hospital were documented in the EHR [16].

Other covariates of interest extracted from the EHR included age, sex, race, ethnicity, Clostridioides difficile testing, underlying medical conditions, and nursing home residence. C difficile is a bacterial cause of AGE that is also associated with severe outcomes, such as hospitalization and, in rare instances, death, and it has been codetected with norovirus in other studies, suggesting that it may be a potential confounder in this analysis [12]. C difficile testing within 7 days before or after norovirus diagnostic testing was ascertained, including molecular polymerase chain reaction, antigen, toxin, and culture testing (test types vary across VA medical centers). Recurrent C difficile was defined as repeated positive C difficile test results (by any test type) within 2–8 weeks [17]. Nursing home residence was defined based on the facility location or location for source of admission.

Underlying medical conditions were based on codes from the International Classification of Diseases, Ninth Revision, Clinical Modification and International Classification of Diseases, Tenth Revision, Clinical Modification pulled from 2010–2018 inpatient records dated before the norovirus test date or from active conditions on a patient's problem list. The codes were grouped into categories based on Charlson comorbidity index (CCI) classifications and into other relevant conditions based on the Healthcare Cost and Utilization Project clinical classification [18]. The CCI is a weighted score of the quantity and severity of comorbid conditions that has widely demonstrated associations with health outcomes in both general and VA populations [19, 20]. Immunocompromising conditions included human immunodeficiency virus infection, cancers and metastatic solid tumors, history of transplantation, and immunity disorder. Death notes, discharge summaries, autopsy reports, and death certificates for patients who died within 7 days of a positive norovirus test result were reviewed and abstracted for details on the cause of death and contributory factors.

Data were obtained for the purpose of public health operations in VHA and did not include factors necessitating patient consent. No additional analyses were performed outside of public health operational activities; thus, the activity was determined to meet the requirements of public health surveillance, as defined in 45 CFR §46.102(l)(2). This project was approved by the Stanford University Institutional Review Board (protocol 47191; “Public Health Surveillance in the Department of Veterans Affairs”), and the requirement written informed consent was waived.

Statistical Analysis

Characteristics were compared between patients who tested positive for norovirus and those who tested negative, using χ² test for categorical or Wilcoxon rank sum test for continuous variables. Stratified cumulative incidence curves for all-cause mortality, comparing patients who tested positive for norovirus and those who tested negative, were estimated using the Kaplan-Meier estimator. The Supremum test for proportional hazards was calculated to confirm that the proportional hazards assumption was not met for calculating a Cox proportion hazard ratio. Given the typical clinical course of immediate symptom onset and clinical impact within the first 3 days after norovirus infection, the primary outcome of interest was immediate mortality risk within 3 days after a norovirus test.

We also evaluated 1-week and 1-month mortality risk to assess how the relative risk associated with laboratory-confirmed norovirus infection changed over time. Adjusted risk ratios (aRRs) were estimated using log-binomial regression models, with adjustment for age (using quadratic splines) [21], C difficile detection and recurrence, immunocompromising conditions, CCI (using quadratic splines), and nursing home residence. Because all-cause mortality was a rare outcome, inverse probability of exposure weights, which are less affected by rare outcomes [22], were used to weight the mortality log-binomial models for the confounders listed above. Generalized estimating equations were used to account for correlation of repeat norovirus tests. A sensitivity analysis was conducted that stratified the effect of norovirus on acute hospitalization by age (<65 vs ≥65 years). Another sensitivity analysis was conducted excluding clusters of >10 patients with norovirus positive test results within 3 days of each other at the same facility, to assess the sensitivity of results to large norovirus outbreaks.

RESULTS

Study Population and Descriptive Characteristics

From 1 January 2010 to 31 December 2018, a total of 23 196 adults received 25 668 diagnostic tests for norovirus infection within the VHA, and 2156 (8.4%) had positive results. Approximately 5.8–6.8 million patients were served by these medical facilities over this period. The median age at testing was 65 years, although a larger proportion of patients who tested positive for norovirus, compared with those who tested negative, were ≥85 years old (12.3% vs 6.6%, respectively) (Table 1). A larger proportion of patients who tested positive for norovirus also had documentation of nursing home residence (24.9% vs 15.8% of those who tested negative; P < .001). Approximately one-quarter of patients had a CCI severity score of ≥5 (Table 1).

Table 1.

Characteristics of 23  196 Veterans Receiving 25  668 Tests for Norovirus Infection, Stratified by Result—Veterans Health Administration, 2010−2018

Characteristic	Patients, No. (%)a		P Valueb
	Norovirus Tests With Positive Results (n = 2156)	Norovirus Tests With Negative Results (n = 23 512)
Unique patients, no.c	2107	21 257	…
Demographics
Age, median (range; IQR), y	65 (18–101; 51–75)	65 (18–101; 54–72)	.90
Age at time of test, y
18–24	77 (3.6)	315 (1.3)	<.001
25–44	312 (14.5)	3190 (13.6)
45–64	671 (31.1)	7602 (32.3)
65–84	830 (38.5)	10847 (46.1)
≥85	266 (12.3)	1558 (6.6)
Male sex	1981 (91.9)	21 094 (89.7)	.001
Race and ethnicity
American Indian/Alaskan, non-Hispanic	22 (1.0)	217 (0.9)	.001
Asian, non-Hispanic	15 (0.7)	167 (0.7)
Hawaiian/other Pacific Islander, non-Hispanic	14 (0.6)	183 (0.8)
Black, non-Hispanic	246 (11.4)	2592 (11.0)
Hispanic or Latino	137 (6.4)	1205 (5.1)
White non-Hispanic	1454 (67.4)	17 409 (74.0)
Multiple, non-Hispanic	85 (3.9)	149 (0.6)
Unknown/missing	183 (8.5)	1590 (6.8)
Nursing home resident	536 (24.9)	3711 (15.8)	<.001
Concurrent positive Clostridioides difficile test resultd	148/1651 (9.0)	2963/21 159 (14.0)	<.001
Recurrent C difficile	6/148 (4.1)	226/2963 (7.6)	.15
Data on underlying medical conditions available	2078 (96.4)	23 175 (98.6)	…
Comorbid conditions included in the CCIe
AIDS/HIV	53 (2.6)	514 (2.2)	.33
Any cancer	334 (16.1)	4093 (17.7)	.08
Metastatic solid tumor	71 (3.4)	1199 (5.2)	<.001
Renal disease	480 (23.1)	5151 (22.2)	.36
Hemiplegia or paraplegia	99 (4.8)	839 (3.6)	.008
Diabetes without chronic complications	381 (18.3)	4055 (17.5)	.34
Diabetes with chronic complications	329 (15.8)	4243 (18.3)	.005
Peptic ulcer disease	139 (6.7)	1642 (7.1)	.50
Chronic pulmonary disease	691 (33.3)	7751 (33.4)	.86
Rheumatologic disease	65 (3.1)	813 (3.5)	.37
Peripheral vascular disease	312 (15.0)	3933 (17.0)	.02
Congestive heart failure	373 (17.9)	4197 (18.1)	.86
Myocardial infarction	209 (10.1)	2561 (11.1)	.17
Cerebrovascular disease	388 (18.7)	3593 (15.5)	<.001
Moderate or severe liver disease	52 (2.5)	728 (3.1)	.11
Mild liver disease	222 (10.7)	3019 (13.0)	.002
Dementia	348 (16.7)	1854 (8.0)	<.001
CCI
0	513 (24.7)	5639 (24.3)	.19
1–2	554 (26.7)	6482 (28.0)
3–4	435 (20.9)	4441 (19.2)
≥5	576 (27.7)	6613 (28.5)
CCI, median (IQR)	2 (1–5)	2 (1–5)	.47
Other comorbid conditionsf
Crohn's disease, regional enteritis, or ulcerative colitis	50 (2.4)	1092 (4.7)	<.001
IBS	66 (3.2)	1621 (7.0)	<.001
Diverticulosis/diverticulitis	242 (11.6)	3224 (13.9)	.004
Transplant receipt	114 (5.5)	621 (2.7)	<.001
Immunity disorders	31 (1.5)	349 (1.5)	.96
Hypertension	1460 (70.3)	16250 (70.1)	.89
Traumatic brain injury or spinal cord injury	158 (7.6)	1451 (6.3)	.02

Abbreviations: CCI, Charlson comorbidity index; HIV, human immunodeficiency virus; IBS, irritable bowel syndrome; IQR, interquartile range.

^aData represent no. (%) of patients unless otherwise specified, with percentages are based on the number of tests (results), not the number of unique patients.

^b P values were calculated using χ² test for categorical and Wilcoxon rank sum test for continuous variables.

^cThere were 168 patients who had both a positive and negative norovirus test over the 9 year time period (separated by at least 30 days).

^d C difficile testing was conducted within 7 days before or after the norovirus test date and included molecular polymerase chain reaction, antigen, toxin, and culture testing. Recurrent C difficile was defined as repeated positive C difficile results within 2–8 weeks.

^eThe CCI is a summary of comorbid disease calculated as a weighted score that accounts for the quantity and severity of comorbid conditions. The following conditions were mutually exclusive: diabetes with chronic complications and diabetes without chronic complications; mild liver disease and moderate or severe liver disease; and any cancer and metastatic solid tumor.

^fCoding of these conditions was based on Healthcare Cost and Utilization Project Clinical Classification Software (CCS) categorization, as follows: Crohn's disease, regional enteritis, or ulcerative colitis,

International Classification of Diseases, Tenth Revision, Clinical Modification (ICD-10-CM) K50 and K510 and International Classification of Diseases, Ninth Revision (ICD-9-CM) 555, 556, and CCS 9.6.2; IBS, ICD-10-CM K58 and ICD-9-CM 564.1; diverticulosis/diverticulitis, ICD-10-CM K57 and ICD-9-CM 562 and CCS 9.6.4; transplant recipient, ICD-10-CM D89.81, T86, Z94, Z98.85, and Z48.2 and ICD-9 V42 and 996.8; immunity disorders, ICD-10-CM CCS BLD008 and ICD-9-CM CCS 3.10; hypertension, ICD-10-CM CCS CIR007 and CCS CIR008 and ICD-9-CM CCS 7.1; traumatic brain injury or spinal cord injury, ICD-10-CM CCS INJ008, INJ045, and Z87.820 and ICD-9-CM CCS 16.3, CCS 16.4, 310, and 854.

Hospitalization

Of patients who tested positive for norovirus, 691 (41.9%) were defined as acutely hospitalized, compared with 7280 (34.7%) of those who tested negative for norovirus. There were 3039 patients who were excluded as not at risk for acute hospitalization because they were already hospitalized for >3 days before receiving a norovirus test, among whom 507 (16.7%) tested positive for norovirus, suggestive of nosocomial infection. Volume depletion (including dehydration) was recorded slightly more frequently among acutely hospitalized patients who tested positive for norovirus than among those who tested negative (21.7% vs 17.8%, respectively; P < .05) (Supplementary Table 1). Several other diagnostic codes were slightly less common among patients who tested positive for norovirus than among those who tested negative, including hypokalemia (11.4% vs 13.2%, respectively), hypo-osmolality and hyponatremia (4.1% vs 9.3%), acute renal failure (25.5% vs 30.4%), sepsis (6.1% vs 10.6%), and arrythmias (18.8% vs 21.1%) (all P < .05). Acutely hospitalized patients who tested positive for norovirus had a lower rate of intensive care unit (ICU) admission than those who tested negative (5.9% vs 8.4% for those testing negative), shorter stay (2 [interquartile range, 1–5] vs 4 [2–7] days), and a lower rate of hospital readmission within 30 days of discharge (17.1% vs 21.2%) (Supplementary Table 1).

Testing positive for norovirus infection, compared with testing negative, was associated with a slightly, but statistically significant, increased risk of acute hospitalization (aRR, 1.13 [95% confidence interval [CI], 1.06–1.21]) (Table 2). This relative risk was slightly modified by age, with a higher increased aRR among patients <65 versus ≥65 years of age (aRR, 1.22 [95% CI, 1.10–1.37] vs 1.08 [1.01–1.18]; Likelihood ratio test P = .002). Results were consistent when varying the definition of acute hospitalization and when excluding norovirus clusters (Supplementary Table 2).

Table 2.

Relative Risk of Acute Hospitalization and Mortality Within 3, 7, and 30 Days of a Norovirus Test, Stratified by Result—Veterans Health Administration, 2010−2018

	Patients, No. (%)a		Crude RR (95% CI	Adjusted RR (95% CI)b
	Tests With Norovirus-Positive Results (n = 2156)	Tests With Norovirus-Negative Results (n = 23 512)
Acute hospitalizationc	691/1649 (41.9)	7280/20 980 (34.7)	1.22 (1.15–1.30)	1.13 (1.06–1.21)
Mortality
d 3 (Immediate risk)	15 (0.7)	67 (0.3)	2.44 (1.40–4.37)	2.14 (1.10–4.14)
d 7	22 (1.0)	164 (0.7)	1.46 (.94–2.28)	1.40 (.84–2.34)
d 30	56 (2.6)	617 (2.6)	0.99 (.76–1.30)	0.97 (.70–1.35)

Abbreviations: CI, confidence interval; RR, risk ratio.

^aPercentages are based on the number of tests (results), not the number of unique patients.

^bLog-binomial models were adjusted for age (using quadratic splines), Clostridioides difficile detection and recurrence, immunocompromising conditions (human immunodeficiency virus infection, cancers and metastatic solid tumor, history of transplantation, or immunity disorder), nursing home residence, and severity of comorbid conditions, as defined by the Charlson comorbidity index (using quadratic splines). Given sample size constraints, regression models for mortality used inverse probability of exposure weights, which are less affected by rare outcomes, to adjust for confounders.

^cAcute hospitalization was defined as an inpatient admission within 3 days before or after the date of stool specimen collection for norovirus diagnostic testing. Patients who had been admitted for >3 days before their norovirus-positive test were excluded from the denominator and analysis.

Mortality

The cumulative incidence of mortality over 30 days following a norovirus diagnostic test, comparing patients who tested positive versus negative for norovirus infection, is displayed in Figure 1. Testing positive for norovirus was associated with an increased risk of short-term mortality at day 3 (aRR, 2.14 [95% CI, 1.10–4.14]), but the risk ratios gradually decreased toward null associations over time (day 7, (1.40 [.84–2.34]; day 30, 0.97 [.70–1.35]) (Table 2).

Figure 1.

Kaplan-Meier crude cumulative mortality risk among patients who tested positive for norovirus infection compared with those who tested negative—Veterans Health Administration, 2010−2018.

Of 22 individuals who died within 7 days after a positive norovirus test result, 21 (95.5%) had more detailed information on cause of death (Supplementary Table 3). Common immediate or contributing causes of death included cardiopulmonary arrest (9 [42.9%]), aspiration pneumonia (7 [33.3%]), and sepsis or septic shock (5 [23.8%]). Nine patients (42.9%) had gastroenteritis, viral colitis, or norovirus mentioned as an immediate or contributing cause of death; 8 of these 9 deaths occurred within 3 days after a positive norovirus result. All of these patients had underlying medical conditions, with a median (interquartile range) CCI of 4 (2–7), although the CCI distribution among these patients did not differ significantly from that in patients who died within 7 days of a negative norovirus result (6 [3–8]; P = .10). Five (23%) of the 22 individuals who died within 7 days of a positive norovirus result were admitted to the ICU, compared with 57 (35%) of the 165 who died within 7 days of a negative result (P = .27).

DISCUSSION

Among a large, national cohort of patients tested for norovirus within the VHA, we identified a slightly increased risk of hospitalization and a significantly increased risk of mortality shortly after norovirus detection, compared with outcomes in patients who tested negative for norovirus. While death following a norovirus test was a rare outcome, patients who tested positive for norovirus were twice as likely to die within 3 days of testing, compared with those who tested negative. Older adults may benefit from timely norovirus diagnosis and may require hospitalization to treat subsequent complications and reduce the risk of mortality. Candidate norovirus vaccines in development may be beneficial for reducing this norovirus-associated morbidity and mortality, particularly in this population and in the current context of no antiviral treatments.

Our finding that norovirus infection is associated with a slight increased risk of hospitalization among adults seeking care at the VA is consistent with the prior literature. An analysis of nursing home outbreaks in 3 US states from 2009 to 2010 found a significantly increased hospitalization rate during norovirus outbreak compared with nonoutbreak periods (aRR, 1.09 [95% CI, 1.05–1.14]) [11]. Despite the different study populations and study designs, the magnitude of association between norovirus and hospitalization was similar between our study (aRR, 1.13) and the study of nursing home outbreaks (aRR, 1.09). Our results are consistent with those indirect modeling studies estimating that approximately 109 000 hospitalizations due to norovirus occur annually in the United States [7].

Hospitalization may be necessary in certain instances of norovirus infection where the risk of dehydration or complications relating to underlying frailty and comorbid conditions may be elevated. Interestingly, dehydration was more common among patients who tested positive for norovirus compared with those who tested negative (as was aspiration pneumonia, though only slightly), but other relevant complications—including hypokalemia, hypo-osmolality and hyponatremia, acute renal failure, sepsis, and arrhythmias—were more common among patients who tested negative. These latter complications may be more common among patients with other causes of AGE; alternatively, this finding could reflect diagnostic coding practices unrelated to causative agents. Notably, we did find evidence in the detailed chart abstraction of sepsis and aspiration pneumonia as immediate or antecedent causes of death among patients who died within 7 days of a positive norovirus test. Also of note, this analysis identified approximately 500 likely nosocomial norovirus infections over the 9-year study period, highlighting the importance of infection control measures to minimize transmission of norovirus in health care settings.

The highest risk of mortality associated with norovirus infection was observed within the acute time period after diagnostic testing, with patients who tested positive for norovirus twice as likely to die within 3 days of testing as those who tested negative. The shift in the relative risk toward the null within 1 week and return to baseline mortality risk within 30 days following AGE testing further suggests that the negative impact of norovirus infection may be concentrated in the short term. This could be due to challenges in early recognition of illness given the multiple comorbid conditions and medications in many of these patients, or it could also be due to the lack of specific treatments available for norovirus, other than supportive care.

Furthermore, the increased risk of short-term mortality associated with norovirus infection was observed despite evidence suggesting that the comparison group of patients who tested negative for norovirus may have been quite ill (based on increased rate of ICU stays, longer hospital stays, and higher proportions of patients with sepsis and acute renal failure), strengthening the evidence that norovirus is a key agent in short-term outcomes. A study in Germany in 2008–2009 also found an increased relative risk of mortality following community-onset norovirus AGE, compared with findings in non-AGE controls, which was greatest soon after infection, with a diminishing relative risk over time [23]. Furthermore, the analysis of nursing home outbreaks in 3 US states reported a significantly increased mortality rate during norovirus outbreak versus nonoutbreak periods, concentrated in the initial week of the outbreak [11].

While our findings indicate an association between norovirus infection and short-term mortality in this population after adjustment for multiple confounders, given the observational nature of this study we urge caution in attributing direct causality of norovirus on mortality. However, norovirus may be a precipitating factor contributing to short-term mortality, as evidenced by the finding that approximately half of the patients who died within 3 days of testing positive for norovirus had gastroenteritis, viral colitis, or norovirus listed as an immediate or contributing cause of death. Our findings strengthen evidence that the effects of norovirus infection on severe outcomes appear soon after infection, indicating that timely identification and management of symptoms and subsequent complications, as well as infection control measures to interrupt chains of transmission, is of utmost importance to reduce morbidity and mortality associated with norovirus. It is important to note that the absolute mortality risk was only 1% (n = 22 deaths) within the 7-day window of elevated risk due to norovirus infection. Although these findings are not easily extrapolated to national-level estimates of norovirus-associated hospitalization and mortality, they are in alignment with modeling studies that have estimated approximately 109 000 hospitalizations and 900 deaths annually in the United States [7], highlighting the public health burden of norovirus that may be averted with future norovirus vaccines.

This study has several limitations, as well as notable strengths. While having laboratory confirmation of norovirus is a strength for this analysis over other studies that relied on indirect estimates or probable diagnoses in outbreak settings, there are a few limitations to relying on these laboratory results. First, tests performed in health care settings outside the VA were not captured. Second, if testing was dependent on factors such as severity of illness, underlying conditions, or location of care, certain individuals may have been less likely to be tested, thus limiting generalizability. Third, the comparison group comprised individuals who received clinician-ordered diagnostic testing, presumably for AGE symptoms, and who tested negative for norovirus; thus, our findings may be an underestimate of the causal impact of norovirus infection AGE compared with no AGE. Future work may wish to compare patients with norovirus infection and patients without acute illness, although identifying these control populations in large-scale administrative data is challenging. Despite this limitation, our results do highlight that even among patients with AGE symptoms, norovirus infection is associated with an increased risk of severe outcomes. Data were not evaluated for prior hospitalizations, medications, or other coexisting pathogens besides C difficile, which may result in residual confounding. However, key confounders—including age, C difficile, underlying medical conditions, and nursing home residence—were all adjusted for, strengthening the interpretation of our findings. Fourth, while this analysis was limited to data from 2010–2018, more recent data may have been confounded by the coronavirus disease 2019 (COVID-19) pandemic, as both exposure to norovirus, health care–seeking behavior, nonpharmaceutical interventions, and the effect of COVID-19 on hospitalizations and mortality may influence the relationship between norovirus and adverse outcomes. In addition, because Veterans are predominantly male and tend to be older with more comorbid conditions than the general US adult population, our findings may not be representative of all US adults [24]. Overall, our analysis is strengthened by using a large study population, not limited to outbreak settings, with good geographic coverage across VA health care facilities over a 9-year time period.

In conclusion, while norovirus infection is generally considered a mild acute illness, our findings highlight that severe outcomes can be associated with norovirus infection. This study found that patients in our study population who tested positive for norovirus infection, compared with those who tested negative, were more likely to be hospitalized and at twice the risk of dying within 3 days after receiving an AGE test. Veterans may benefit from public health messaging, including norovirus prevention tips and encouragement of health care consultation for testing and/or medical advice if they are experiencing symptoms consistent with norovirus. Clinicians treating older patients for AGE should be aware of these elevated risks and remain wary of dismissing norovirus as a mild illness, particularly in vulnerable patients with multiple comorbid conditions and those living in congregate settings, such as nursing homes.

Supplementary Data

Supplementary materials are available at Open Forum Infectious Diseases online. Consisting of data provided by the authors to benefit the reader, the posted materials are not copyedited and are the sole responsibility of the authors, so questions or comments should be addressed to the corresponding author.