Vibrio vulnificus (Vv), a Gram-negative rod, was first isolated by the Centers for Disease Control and Prevention (CDC) in 1964 and identified as a human pathogen in 1976 [Reference Hollis1]. Blake et al. initially described the epidemiology, including mode of transmission, routes of transmission and risk factors of Vv, which cause an estimated 200 infections annually in the USA, nearly half of which are foodborne in origin [Reference Blake2, Reference Scallan3]. In the USA, Vv, which naturally occur in marine and estuarine environments and are related to a host of environmental parameters including sea surface temperatures, are found in high numbers in the Gulf of Mexico [Reference Daniels4, Reference Horseman and Surani5]. Massive flooding from the Gulf of Mexico in 2005, consequent to Hurricane Katrina resulted in 14 reported cases of Vv wound infection, probably via broken skin contact with seawater, and three Vv-attributed deaths [6]. Infection may also result from eating contaminated seafood, typically raw oysters harvested from the Gulf. Although the numbers of annual cases are relatively low, the severity of infection is high, with over 90% of infected patients requiring hospitalization and an estimated case-fatality rate of 34·9% [Reference Scallan3]. Severe disease with a fatal outcome is more prevalent in those with pre-disposing medical conditions and Vv is a leading cause of seafood-associated fatalities in the USA [Reference Iwamoto7, Reference Weis8]. Despite continued prevention efforts, data from both active and passive surveillance systems report increasing rates of infection [Reference Newton9].

Although pre-existing liver disease is a recognized risk factor for serious infection, the relative importance of other comorbidities, by route of exposure, has not been fully assessed.

The Cholera and Other Vibrio Illness Surveillance System (COVIS) was initiated by CDC, FDA, and the Gulf Coast states (Alabama, Florida, Louisiana, Mississippi, Texas) in 1988. CDC has maintained a database of Vibrio infections from humans in order to obtain data on illnesses associated with Vibrio spp. Participating health officials collect clinical data, information about underlying illness, history of seafood consumption, and exposure to seawater during the 7 days before illness. When possible, this information assists in epidemiological investigations to identify the implicated vehicle. Prior to 2007, toxigenic Vibrio cholerae O1 and O139, the causative agents of cholera, were the only nationally notifiable species. Despite sporadic reports from other states (non-coastal and Pacific and Atlantic coasts), the expansion of COVIS in 1997 marked the point when nearly all states were voluntarily reporting all Vibrio spp. Indeed, by 1997, all but one state had reported Vibrio cases to COVIS. Infection with any Vibrio spp. (vibriosis) became nationally reportable in 2007.

State and county health departments report infection of all species, along with demographic (e.g. age, sex), clinical (including pre-existing medical conditions), and exposure data (i.e. seafood consumption, travel history, skin exposure to seawater, drippings from raw or live seafood, or contact with marine wildlife) to the CDC via a standardized reporting form. Pre-existing medical conditions, including diabetes, heart disease, haematological disease (e.g. anaemia, thalassaemia), immunodeficiency, malignancy, peptic ulcer disease, gastric surgery, renal disease and liver disease are specified on the reporting form. In this analysis, reported alcoholism was coded as a proxy for liver disease. Data on the source of the specimen and the species isolated are also reported; molecular data are not collected.

We analysed reports of Vv infections submitted to COVIS from January 1988 to September 2006. The source of Vv exposure was determined by clinical signs and patient's history, and was characterized as either foodborne or non-foodborne. A non-foodborne infection required isolation of Vv from a wound or from another source (e.g. external ear, skin) with documented direct contact with saltwater, marine wildlife, raw seafood, or seafood drippings. All other infections were considered foodborne. Septicaemia was defined as isolation of the pathogen from the blood in patients with self-reported fever and hypotension (systolic blood pressure <90 mmHg). We categorized medical conditions as ‘present’ if the report indicated the patient had the diagnosis. Otherwise, the condition was considered ‘not present’. Cases with multiple Vibrio spp. were excluded from analysis.

Here we assess pre-existing medical conditions as risk factors for septic vs. non-septic Vv infection, by route of exposure (foodborne and non-foodborne). We conducted both bivariate analysis [odds ratio (OR), χ ² test for significance] and logistic regression. The initial multivariate model included all significant variables from the bivariate analysis. The final model excluded non-significant factors from the initial multivariate model. A population attributable risk percentage was calculated:

$$({\rm OR}-1/{\rm OR}) \times \%\ {\rm of}\,{\rm cases}\,{\rm with}\,{\rm attribute}.$$

A total of 1212 patients with Vv infection were reported; 86% were male, 87% were hospitalized, 72% were septic, and 33% died. The median age of patients with an infection was 56 years. Nearly half (48%) were classified as a non-foodborne infection; the remaining cases were classified as foodborne. At least one pre-existing medical condition was reported for 80% of patients; liver disease was most prevalent (40%), followed by heart disease (23%), diabetes (20%), and renal disease (10%). The prevalence of the other pre-existing conditions was each <10%. Patients who acquired infection via contaminated food were significantly more likely to be female (15% vs. 11%), younger (55 vs. 58 years), non-white (30% vs. 15%), have a greater number of pre-existing medical conditions (1·5 vs. 1·1) and become septic (87% vs. 55%) than those patients who acquired a non-foodborne infection.

Nearly 2/3 (63%) of the septic cases were acquired via contaminated food while the majority (77%) of non-septic cases were non-foodborne infections. Of the patients who were septic, 87% were male. The case-fatality rate of septic cases was 44%; of non-septic cases only 7% died.

Pre-existing medical conditions differed in septic vs. non-septic cases and by route of exposure (Table 1). On bivariate analysis of pre-existing medical conditions associated with Vv septicaemia from a foodborne exposure, patients with liver disease, haematological disease, diabetes, or peptic ulcer disease were more likely to develop septicaemia. On bivariate analysis of pre-existing medical conditions associated with Vv septicaemia from a non-foodborne exposure, patients with liver disease, malignancy, haematological disease, heart disease, or diabetes were more likely to be septic.

Table 1. Bivariate analysis of pre-existing medical conditions associated with Vibrio vulnificus septicaemia by route of exposure

Using logistic regression and adjusting for age, sex, and the number of pre-existing conditions, only patients with liver disease (aOR 5·1) were more likely to become septic when exposure was due to contaminated food. Patients with liver disease (aOR 4·1), a haematological disease (aOR 3·2), or malignancy (aOR 3·2) were more likely to become septic when infection was acquired via a non-foodborne exposure (Table 2). Having multiple pre-existing conditions was associated with severe illness. However, in patients with vibriosis who developed septicaemia, only liver disease had an attributable risk percentage >10% (24% of non-foodborne infections; 47% of foodborne infections).

Table 2. Pre-existing medical conditions associated with Vibrio vulnificus septicaemia

aOR, Adjusted odds ratio; CI, confidence interval; PAR, population attributable risk.

Vv infection results in high morbidity and mortality. Over 70% of infected patients become septic, of which significantly more patients become septic after acquiring infection via a foodborne exposure, and nearly half of septic patients die. Liver disease has already been well characterized as a host risk factor for serious infection, including sepsis and death [Reference Daniels4, Reference Horseman and Surani5, Reference Weis8, Reference Jones and Oliver10, Reference Dechet11]. In this analysis, we included other potential risk factors and found that only persons with liver disease, were at increased risk of developing septicaemia for both foodborne and non-foodborne exposure. Patients with a haematological disease, or an underlying malignancy, were at increased risk of developing septicaemia with a non-foodborne infection, but were not at increased risk if the exposure was foodborne. Although the reason for this discrepancy is unknown, both of these conditions may alter the immune system and impair wound healing, and are documented risk factors for other invasive bacterial infections [Reference Tilley and Roberts12].

Given that the results presented here are from a case-series analysis, we cannot determine the individual risk of infection for patients with specific conditions. Due to the relatively low percentage of patients with certain pre-existing conditions [e.g. malignancy (9%) or haematological disease (8%)], this analysis is unable to provide information on the importance of a specific type of condition (e.g. breast cancer). Moreover, this analysis is unable to characterize the severity of conditions (e.g. stage of malignancy, degree of neutropenia) associated with sepsis because such information is not currently collected in this surveillance system. Additionally, as we did not have molecular typing data, we were unable to evaluate the role of risk factors by particular Vv strain. This analysis covers data reported to COVIS from 1988 to 2006, which corresponds to the time period from the initiation of this surveillance system until the time that vibriosis became a nationally notifiable disease. While we do not believe the findings presented here will be altered with more recent data, continued analysis of surveillance data is necessary. To provide evidence for more refined recommendations, future iterations of this surveillance system should consider collecting molecular data as well as markers of disease severity.

While the general association between pre-existing medical conditions and Vv infection is well known, these results characterize the significance and relative importance of certain pre-existing medical conditions, by route of exposure, associated with Vv infection. As such, these results underscore the need for clinicians to advise patients with liver disease, malignancy, or a haematological condition to avoid exposing broken skin to warm seawater. Clinicians should further advise patients, especially patients with liver disease, of the risk of life-threatening illness after eating undercooked contaminated seafood. Clinicians should consider the diagnosis of Vv for seriously ill patients presenting with a recent history of raw shellfish consumption or wound exposure to the Gulf of Mexico or other warm marine environments which are likely to harbour Vv, culture specimens, and provide prompt treatment, including antibiotics.