Decreasing Listeriosis Mortality in the United States, 1990–2005
- 1Los Angeles County Department of Public Health, Los Angeles, California
- 2Department of Epidemiology, School of Public Health, University of California at Los Angeles, Los Angeles, California
- Reprints or correspondence: Dr. Frank Sorvillo, 313 N. Figueroa St., Rm. 127, Los Angeles, CA 90012 (fsorvillo{at}ph.lacounty.gov).
Abstract
Background. Listeria monocytogenes is among the most virulent foodborne pathogens, with 20% of clinical infections resulting in death. To explore listeriosis-associated mortality in the United States and to evaluate prevention efforts, we reviewed vital records over a 16-year period to assess demographic, temporal, and seasonal trends.
Methods. Nonperinatal listeriosis-associated deaths from 1990 through 2005 were identified from multiple-cause–coded death records and were combined with US census data to calculate mortality rates. Poisson regression was used to model time trends, and logistic regression was used to identify comorbid conditions associated with listeriosis on the death record.
Results. Of the 37,267,946 deaths occurring in the United States during the 16-year study period, 1178 included listeriosis on the death record. Listeriosis-related mortality rates decreased annually by 10.74% from 1990 through 1996 and by 4.26% from 1996 through 2005. Seasonal trends show a distinct peak in mortality from July through October. After adjustment for age, sex, and race/ethnicity, listeriosis was positively associated with human immunodeficiency virus (HIV) infection (odds ratio, 4.19; 95% confidence interval, 3.06–5.73), lymphoid and hematopoietic cancers (odds ratio, 5.27; 95% confidence interval, 4.47–6.22), and liver disease (odds ratio, 2.05; 95% confidence interval, 1.54–2.73) on the death record.
Conclusions. Nonperinatal listeriosis-associated deaths in the United States have decreased, paralleling a decreasing trend in incidence. Strict monitoring of food manufacturing processes, as well as improved treatment for HIV infection, may have played influential roles in preventing human infections. Health care providers should be aware of seasonal listeriosis patterns, as well as conditions predisposing individuals to severe infection and death due to L. monocytogenes infection, to guide strategies for disease management and prevention.
For more than 2 decades, foodborne illness caused by Listeria monocytogenes has raised significant public health concern in the United States, Europe, and other areas of the world. Each year, there are an estimated 2500 cases in the United States, resulting in 2300 hospitalizations and 500 deaths [1]. This pathogen is more invasive than the pathogens associated with most types of foodborne illness, and conditions typically associated with Listeria infection include meningitis, encephalitis, sepsis, and febrile gastroenteritis [2]. Infections are most common among elderly individuals, newborns, pregnant women, and individuals with compromised immune systems [3].
Several reports have described patterns of listeriosis incidence in the United States and parts of Europe [4–7]. Although there is some indication that the number of Listeria–related infections are increasing in Germany, England, and Wales (particularly among the elderly population) [4, 5], data from the Foodborne Diseases Active Surveillance Network (FoodNet) show that the number of cases in the United States has decreased in recent years [2, 7]. The number of laboratory-confirmed cases in the United States decreased by 24% from 1996 through 2003, presumably because of improved testing and control of ready-to-eat foods for L. monocytogenes contamination, as well as other disease prevention efforts [7].
In spite of the availability of surveillance data on disease incidence, relatively little is known about listeriosis cases resulting in death. Information on mortality within the population is necessary, not only to more fully understand the burden of disease, but to evaluate the effectiveness of prevention efforts. Population-based mortality data have been used to investigate other infectious diseases [8, 9], but to our knowledge, these data have not yet been used in listeriosis research. In the present study, we examined vital records data to assess demographic, temporal, and seasonal trends in mortality due to listeriosis in the United States from 1990 through 2005.
Methods
Death records. Multiple-cause-of-death data from US death certificates have been collected, processed, and made available for public use by the National Center for Health Statistics for the years 1990–2005 [10]. These death certificates contain basic demographic information for each decedent, including age, sex, race/ethnicity, and place of residence (geographic data unavailable for 2005). In addition to designating an underlying cause, the physician or coroner filling out the death certificate may list up to 20 conditions that are believed to have contributed in some way to the death of the individual. Each of these conditions is coded on the basis of the International Classification of Diseases system for the year in which the death occurred (International Classification of Diseases, Ninth Revision [ICD-9] for the period 1990–1998 and International Classification of Diseases, Tenth Revision [ICD-10] for the period 1999–2005).
In our study, cases are defined as US resident deaths having an ICD-9 code of 027.0 (“Listeriosis, Infection by Listeria monocytogenes, Meningitis by Listeria monocytogenes, Meningoencephalitis by Listeria monocytogenes, Septicaemia by Listeria monocytogenes; excludes: congenital listeriosis [771.2]”) or an ICD-10 code within the range A32.0-A32.9 (A32 = listeriosis; A32.0 = cutaneous listeriosis; A32.1 = listerial meningitis and meningoencephalitis; A32.7 = listerial septicemia; A32.8 = other forms of listeriosis; A32.9 = listeriosis, unspecified) listed as the underlying, immediate, contributing, or other cause listed in the cause of death section of the death record. Perinatal listeriosis cases (in which death occurred from 16 weeks after conception through 30 days postpartum) [11] have been excluded from our study, because the ICD-9 codes referencing these cases (655.4 and 771.2) encompass multiple congenital conditions that cannot be separated for analysis.
Mortality rates and trends. Mortality rates were calculated using bridged-race population estimates derived from US census data [12, 13] and were subsequently age-adjusted with weights from the 2000 US standard population data [14]. Bridged-race estimates of the population are necessary to control for changes in racial categorization across census years. To facilitate a review of temporal trends, we calculated age-adjusted mortality rates, cross-tabulated these by year, and then estimated corresponding temporal patterns using Poisson regression models. Mortality rates for race/ethnicity, sex, and geographic region were calculated with aggregated data from all years of our study to ensure stable rates (1990–2004 for geographic data). Race/ethnicity was divided into 5 categories: non-Hispanic white, non-Hispanic black, Hispanic, Asian/Pacific Islander, and Native American. For our geographic analysis, we divided the 50 states into 4 major regions and 9 subregions according to US census divisions [15].
Comorbidity analysis. Multiple-cause–coded mortality data facilitate a review of many factors influencing the death of an individual. To identify these factors, we searched listeriosis-related deaths for comorbidities frequently noted on the death certificate and compared these with data from a control group comprising all non–listeriosis-associated deaths. Statistical associations were estimated by creating unconditional logistic regression models with listeriosis as the outcome and with the selected comorbidity as the main exposure variable. We included age, race/ethnicity, and sex in each model to control for their potential confounding effects. HIV as a comorbidity was also tested as a potential confounder in each model. All calculations were performed with SAS, version 9.1 (SAS).
Results
Demographic and temporal trends. We identified 1178 listeriosis-related deaths during the period 1990–2005, representing 0.003% of the 37,267,946 total deaths among US residents. The overall crude mortality rate associated with listeriosis was 0.27 deaths per 1 million persons annually (95% CI, 0.25–0.28 deaths per 1 million persons annually), and after adjustment for age, this rate remained unchanged at 0.27 deaths per 1 million persons annually (95% CI, 0.26–0.29 deaths per 1 million persons annually) (table 1). Although >80% of listeriosis-related deaths occurred among non-Hispanic whites, race-specific mortality rates, adjusted for age, were similar for non-Hispanic blacks, Hispanics, Asians/Pacific Islanders, and non-Hispanic whites (range, 0.27–0.33 deaths per 1 million persons annually) (table 1). The most common underlying causes of mortality among listeriosis-related deaths included listeriosis (838 deaths; 71.1%), cancer (130 deaths; 11.0%), and diseases of the heart (59 deaths; 5.0%).
Figure 1 highlights the overall temporal trends in listeriosis-associated mortality from 1990 through 2005. According to a Poisson regression analysis, the data show a yearly decrease in age-adjusted mortality of 10.74% (95% CI, 7.22%–14.13%) from 1990 through 1996, followed by a decrease of 4.26% (95% CI, 1.51%–6.94%) from 1996 through 2005. Overall incidence decreased from 127 deaths in 1990 (age-adjusted mortality rate, 0.53 deaths per 1 million persons) to 55 deaths in 2005 (age-adjusted mortality rate, 0.18 deaths per 1 million persons). A summer through fall seasonal peak in listeriosis mortality (for the period from July through October) was observed, with the largest number of deaths occurring in September (148 deaths; 12.6%) and the fewest deaths occurring in February (68 deaths; 5.8%) (figure 2).
Figure 1
Age-adjusted mortality rate per 1 million persons, by sex and year, 1990–2005
Figure 2
Listeriosis-related deaths, by month, 1990–2005
The mean age at death was 67.9 years (median age, 71 years; range, <1–102 years), and individuals ⩾65 years of age accounted for nearly 70% of deaths (table 1). Individuals ⩾85 years of age had the highest mortality rate of any age group (2.12 deaths per 1 million persons annually). Male individuals accounted for slightly more deaths than did female individuals and had higher age-adjusted mortality rates (0.34 deaths per 1 million persons annually vs. 0.23 deaths per 1 million persons annually). During the study period, however, we observed a converging trend between the sexes (figure 1). From 1990 through 1996, mortality rates among the male population decreased by a mean of 10.45% per year (95% CI, 5.65%–15.01% per year), whereas rates among the female population decreased by 11.23% per year (95% CI, 5.96%–16.19% per year). During the period 1996–2005, mortality rates among the male population decreased by a mean of 4.98% per year (95% CI, 1.06%–8.74% per year), whereas the mortality rate among the female population decreased by 3.71% per year (95% CI, 7.47% decrease per year to 0.21% increase per year).
Table 1
Listeriosis-related deaths by sex, race, and age, with corresponding age-adjusted mortality rates in the United States, 1990–2005.
Geographic statistics. Listeria-related mortality varied to some degree across the 4 major geographic regions and 9 subregions of the United States. From 1990 through 2004, of the 4 geographic regions, the lowest mortality was found in the Midwest, which had a rate of 0.23 deaths per 1 million persons annually, whereas the West region had the highest rate (0.33 deaths per 1 million persons annually) (table 2). In the Pacific subregion (i.e., California, Oregon, Washington, Alaska, and Hawaii), there were 205 deaths, representing 18.3% of Listeria-related deaths and yielding a mortality rate of 0.36 deaths per 1 million persons annually. By contrast, there were 49 deaths related to listeriosis in the West North Central subregion (i.e., North Dakota, South Dakota, Minnesota, Nebraska, Iowa, Kansas, and Missouri), representing 4.4% of listeriosis-related deaths and an annual mortality rate of 0.17 deaths per 1 million persons.
Table 2
Listeriosis-related deaths, by geographic region and subregion, with corresponding age-adjusted mortality rates, United States, 1990–2004.
Disease associations. The frequency of HIV listing as a comorbidity among listeriosis-related deaths varied to some degree during the study period. From 1990 through 1995, 29 Listeria-related deaths (4.9%) also had HIV noted on the vital record, whereas only 20 such deaths (3.4%) were reported during the later years of the study. This difference in proportions, however, was not statistically significant (P=.22).
Listeriosis was positively associated with a number of infectious and chronic conditions recorded on the death certificate. After adjustment for age, sex, and race/ethnicity, associations were observed between listeriosis and HIV infection (OR, 4.19; 95% CI, 3.06–5.73), anemia (OR, 1.82; 95% CI, 1.32–2.52), and liver disease (OR, 2.05; 95% CI, 1.54–2.73) (table 3). Although no statistical association was observed between listeriosis and malignant neoplasms of all types (OR, 1.00; 95% CI, 0.88–1.14), lymphoid and hematopoietic cancers were >5 times as likely (OR, 5.27; 95% CI, 4.47–6.22) to be listed on death records with listeriosis than to be listed on death records without listeriosis. This association with lymphoid and hematopoietic cancers remained relatively unchanged (OR not altered by >5%) after controlling for the potential confounding effects of HIV infection, hepatitis, organ transplantation, and lupus.
Table 3
OR measuring the association between listeriosis and selected conditions on United States death certificates, 1990–2005.
Discussion
Nonperinatal listeriosis-related deaths decreased over the 16-year study period, highlighting an encouraging, although not entirely unexpected, mortality trend. Evidence shows that incidence rates of sporadic listeriosis in the United States similarly decreased during the years of our study, and there are several factors that may account for this downward trend [7, 16]. A zero-tolerance policy, prohibiting the sale of ready-to-eat meats contaminated with L. monocytogenes, was developed in 1987 by the Food Safety and Inspection Service of the US Department of Agriculture [17]. This policy has since been modified to increase on-site testing of meat and poultry processing plants [18, 19]. Results of a multistate case-control study published in 1992 prompted the Centers for Disease Control and Prevention to issue recommendations both to assist health care providers in the diagnosis and treatment of disease and to caution individuals who were at high risk against the consumption of certain foods [20]. Prevention efforts by the food industry included the development of a Hazard Analysis Critical Control Points strategy to minimize levels of L. monocytogenes at key points of the cooking and packaging processes prior to distribution [21].
Changes in host factors likely played a role in further reducing Listeria-related mortality during the mid-1990s. The first protease inhibitor was approved by the Food and Drug Administration in late 1995 [22], and its use in combination with other antiretroviral therapy resulted in dramatic decreases in morbidity and mortality among patients with HIV infection and/or AIDS [23]. Cases of invasive disease caused by opportunistic organisms, such as L. monocytogenes, are now less frequent among those infected with HIV because of the considerable immunorestorative effects of this novel treatment [24]. Our data tend to support this trend, because we observed HIV infection and/or AIDS listed more frequently as a comorbidity among individuals with listeriosis-associated deaths from 1990 through 1995 (4.9% of cases) than from 1996 through 2005 (3.4%). Among men, who have a higher rate of HIV infection than women, we also observed a 30% drop in listeriosis-related mortality from 1995 to 1996 (figure 1).
Relatively little has been published on seasonal listeriosis patterns in the United States. Although our data show a clear increase in mortality from July through October, the reasons for this are unclear. Using data collected from the FoodNet surveillance system, Voetsch et al. [7] reported that listeriosis infections are more frequent among individuals ⩾50 years of age during these same months but that pregnancy-associated cases, as well as cases among younger individuals, do not necessarily follow this same trend. Given that ready-to-eat meats, such as sliced deli meats and hot dogs, are common vehicles of transmission, it is possible that these foods are served more frequently during the summer months (e.g., at picnics or barbecues, where they may be left unrefrigerated for long periods of time prior to consumption).
Our data also show variation in listeriosis-related mortality by geographic region, although the reasons for this are not evident. Pathogen exposure levels may vary by region according to local dietary preferences and habits. Underlying risk factors for infection may also be more prevalent in certain regional populations, leading to higher numbers of Listeria-related deaths in those areas.
We also observed an increase in Listeria-related deaths with age (table 1). Although this finding was expected, it has clinical relevance, because of the overall aging trend of the US population. Changes in food industry regulation have led to an overall decrease in listeriosis-related mortality, but clinicians should continue to offer dietary recommendations to patients ⩾55 years of age in an effort to further reduce mortality among this group.
L. monocytogenes has historically caused more reported infections among men than among women [2, 25], although the gap has narrowed in recent years. Our data show that men and women experienced similar numbers of deaths, but age-adjusted mortality rates for men were 48% higher than for women, suggesting that men tend to die from listeriosis at younger ages than women. A retrospective study of Vibrio vulnificus infection in Los Angeles County found that men are less likely than women to follow education measures warning against the consumption of contaminated foods [26]. Although dietary habits tend to differ between the sexes, the disparity in mortality may also be an indication of a larger population of immunocompromised men than immunocompromised women living in the United States. For example, in New York City, Sackoff et al. [27] recently found that men account for nearly 70% of persons with AIDS who are ⩾13 years of age.
Immunocompromised individuals have long been recognized as a group at high risk for opportunistic infections, such as listeriosis [7, 25]. T cell–mediated immune responses play a central role in maintaining the body's defenses against intracellular pathogenic organisms [28]. Cancer, organ transplantation, HIV infection, diabetes, renal disease, and liver disease have been recognized as specific conditions that predispose the individual to develop listeriosis [2, 25, 29, 30]. Whether this is due to the nature of the diseases themselves, the medications commonly prescribed to treat them, or a combination of the 2 is not clear and may vary by condition. In the case of chronic lymphocytic leukemia, hypogammaglobulinemia results from the cancer itself, whereas common treatment with purine analogs and monoclonal antibodies leads to T cell immunodeficiency, further increasing the risk of developing infection [31]. This malignancy comprises an important segment of the lymphoid and hematopoietic cancers, which have a notably strong association with listeriosis on the death certificate (table 2).
Our data also suggest disease associations that may not be widely recognized. Connective tissue disease, including systemic lupus erythematosus, has previously accompanied listeriosis as either a comorbid or preexisting condition in 2 case series and 1 outbreak [29, 30, 32]. Viral hepatitis and hemolytic anemia have also been linked to Listeria infection in case studies, although a statistical association had not been established prior to our study [33–36].
Several important limitations are associated with the use of multiple-cause-of-death data that require consideration. Although these data are population based and contain large numbers of observations, death certificates may contain errors, which have been attributed to a variety of factors [37]. Redelings et al. [38] point out that, instead of reporting all prevalent diseases on the death certificate, physicians more commonly list only those that are believed to play an instrumental role in the death of the individual. Therefore, the death certificate of an individual who dies of a serious infection, such as listeriosis, may be less likely to list chronic conditions—such as cardiovascular disease, stroke, or hypertension—even if these are present at the time of death.
A second limitation is the chronological ambiguity of diseases listed on an individual death record [39]. For instance, although we note a strong positive association between HIV infection and listeriosis (OR, 4.19; 95% CI, 3.06–5.73), the data do not explicitly indicate which developed first. They merely indicate that both conditions were present at the time of death.
Underreporting and misclassification of listeriosis may also limit our study, because some discrepancy exists between estimates of listeriosis-related deaths from prior studies [1, 7] and the number of cases in which listeriosis was recorded on the death certificate. Data from FoodNet have been used to estimate that 500 individuals die each year of listeriosis in the United States [1], yet actual death records include only 127 nonperinatal Listeria-associated deaths for 1990 and just 55 such deaths for 2005. The exclusion of perinatal Listeria-associated deaths in our study accounts for some, but likely not all, of this discrepancy. Although it is unclear which data source most closely reflects the true disease burden in the population, the possibility of misclassification error exists in our data. Listeriosis-associated deaths might be recorded on death certificates as “unspecified septicemia” (ICD-9 code 038.9 or ICD-10 code A41.9) in the absence of laboratory test results positive for L. monocytogenes.
Finally, mortality rates may be distorted because of errors in population estimates, particularly for race/ethnicity. Because estimates of the at-risk population factor into the denominator for rate calculations, such errors can lead to biased estimates of mortality rates.
Because of the limitations of multiple-cause-of-death data and the potential for error in estimating population size, we urge caution in interpreting our results. However, when the direction of incidence trends, the biologic plausibility of associations identified, and a general agreement with prior literature are considered, these findings likely reflect true patterns of listeriosis-associated mortality in the United States. The decreasing trend in listeriosis-associated mortality over a 16-year period demonstrates the largely preventable nature of Listeria infection, as well as the success of collaborative efforts by the US Department of Agriculture, US Food and Drug Administration, Centers for Disease Control and Prevention, and food industry in limiting the spread of this opportunistic pathogen. Changes in host factors among the population, particularly with respect to HIV infection status and treatment, have played a minor (although important) role in influencing listeriosis-associated mortality.
Despite these recent successes, listeriosis remains a cause of preventable mortality in the United States. Prevention efforts in educating high-risk groups, such as the elderly population and those with decreased immune function, should emphasize extra caution during the summer and early fall. Health care providers should be aware of conditions that may predispose individuals to developing severe infection and death due to L. monocytogenes and incorporate this knowledge into strategies for disease management.
Acknowledgments
We thank Matthew Redelings, Loren Lieb, and Margaret Shih for their contributions.
Financial support. Los Angeles County Department of Public Health.
Potential conflicts of interest. L.M. is on the speakers' bureaus for Merck and MedImmune. All other authors: no conflicts.
- Received January 16, 2008.
- Accepted May 16, 2008.
- © 2008 by the Infectious Diseases Society of America
References
This Article
- AbstractFree
- » Full Text (HTML)Free
- Full Text (PDF)Free
Classifications
-
- Articles and Commentaries
Services
Citing Articles
Google Scholar
PubMed
Related Content
Share
- Email this article
-
CiteULike
Connotea
Delicious
Facebook
Google+
Mendeley
Twitter
Published on behalf of
IDSA Members: For your free access to this journal, log in via the IDSA members area.
The Journal
Impact Factor: 8.186
5-Yr impact factor: 7.898
For The Media
For Authors
- Services for authors
- Self-archiving Policy
- IDSA Conflict of Interest policy
- Instructions for Authors
- ICMJE Form
- Submit Now!
Open access options for authors visit Oxford Open
This journal enables compliance with the NIH Public Access Policy
Corporate Services
Alerting Services
Related IDSA Publications
