Survey
* Your assessment is very important for improving the workof artificial intelligence, which forms the content of this project
* Your assessment is very important for improving the workof artificial intelligence, which forms the content of this project
Hygiene hypothesis wikipedia , lookup
Diseases of poverty wikipedia , lookup
Eradication of infectious diseases wikipedia , lookup
Public health genomics wikipedia , lookup
Epidemiology of metabolic syndrome wikipedia , lookup
Race and health wikipedia , lookup
Infection control wikipedia , lookup
Adherence (medicine) wikipedia , lookup
Epidemiology wikipedia , lookup
MAJOR ARTICLE Risk Factors for Mortality among Patients with Nonperinatal Listeriosis in Los Angeles County, 1992–2004 Ramon E. Guevara,1 Laurene Mascola,1 and Frank Sorvillo2 1 Acute Communicable Disease Control Program and 2Office of Vital Records, Los Angeles County Department of Public Health, Los Angeles, California Background. Listeriosis is a relatively rare foodborne disease with significant public health implications. The causative pathogen, Listeria monocytogenes, grows well in refrigeration, is associated with a case-fatality rate of 20%, and causes an estimated 28% of all foodborne disease–related deaths. Nevertheless, data on the risk factors for listeriosis mortality are limited. Methods. Using the passive surveillance listeriosis database of the County of Los Angeles Department of Public Health, we conducted a 13-year retrospective cohort study to describe nonperinatal listeriosis mortality in Los Angeles County during the period 1992–2004. A nonperinatal listeriosis case was defined as one occurring in a nonpregnant person 142 days of age who resided in Los Angeles County and had a culture positive for L. monocytogenes. Results. Unconditional multivariable logistic regression analysis of 281 nonperinatal listeriosis cases with 29 main effects variables resulted in finding nonhematological malignancy (odds ratio [OR], 5.92; 95% confidence interval [CI], 1.85–18.9), alcoholism (OR, 4.63; 95% CI, 1.36–15.8), age ⭓70 years (OR, 3.44; 95% CI, 1.50–7.87), steroid medication (OR, 3.34; 95% CI, 1.38–8.08), and kidney disease (OR, 2.94; 95% CI, 1.18–7.31) to be statistically significant risk factors for mortality. Other listeriosis mortality risk factors with adjusted odds ratios 11.5 included blood transfusion, asthma, black race, Asian race, use of antibiotics, hypertension, receipt of chemotherapy, and Hispanic race. Patients admitted to the hospital with a diagnosis of sepsis alone had the highest mortality (23.7%), whereas patients with cases of meningitis alone had the lowest mortality (3.13%). Conclusions. The findings of this study should be used to help researchers and clinicians focus on specific risk factors to prevent nonperinatal listeriosis–related deaths. Listeriosis is a foodborne disease caused by the bacterium Listeria monocytogenes. Although L. monocytogenes is a ubiquitous pathogen in the environment, it is a rare cause of human illness, compared with other foodborne pathogens. However, infection can lead to severe disease, including sepsis, meningitis, stillbirth, and death. L. monocytogenes is associated with a case-fatality rate of 20%, which is second only to that associated with Vibrio vulnificus (39%), and it causes an estimated Received 17 October 2008; accepted 3 February 2009; electronically published 27 April 2009. Reprints or correspondence: Dr. Ramon E. Guevara, Acute Communicable Disease Control, 313 N. Figueroa St., Rm. 212, Los Angeles, CA 90012 ([email protected]). Clinical Infectious Diseases 2009; 48:1507–15 2009 by the Infectious Diseases Society of America. All rights reserved. 1058-4838/2009/4811-0003$15.00 DOI: 10.1086/598935 28% of all foodborne disease–related deaths, which is second only to the amount attributable to nontyphoidal Salmonella (31%) [1]. Listeriosis mortality unrelated to pregnancy has been estimated at 35% in the United States [2] and as high as 44% in other countries [3– 7]. L. monocytogenes, which is an opportunistic and intracellular pathogen, usually causes disease among people with immature or compromised immune systems, particularly infants, pregnant women, elderly individuals, and persons receiving immunosuppressive medication [6, 8–13]. Previously identified risk factors for invasive listeriosis include advanced age [8, 14–16], cancer [3, 5, 12, 16, 17], receipt of immunosuppressive therapy (such as steroid medication) [8, 12, 15–20], alcoholism [3, 15, 16, 21], renal disease [3, 5, 12, 15– 17, 19, 20, 22], liver disease [3, 5, 10, 15, 16, 23–26], Nonperinatal Listeriosis Mortality • CID 2009:48 (1 June) • 1507 Table 1. Demographic and clinical characteristics of 285 patients with nonperinatal listeriosis and known survival outcome, Los Angeles County, 1992–2004. Variable No. (%) of patients Age, years !5 5–14 2 (0.70) 6 (2.11) 15–34 35–44 28 (9.82) 21 (7.37) 45–54 55–64 33 (11.6) 43 (15.1) 65–74 ⭓75 75 (26.3) 77 (27.0) Race and ethnicity Asian Black Hispanic White Sex Male Female 30 (10.5) 30 (10.5) 65 (22.8) 160 (56.1) 157 (55.1) 128 (44.9) Infection type Bacteremia (asymptomatic) Sepsis only Meningitis only Sepsis and meningitis Infection other than sepsis or meningitis 2 198 32 25 (0.70) (69.5) (11.2) (8.77) 28 (9.82) hemochromatosis [8, 16, 27], anemia [8, 16, 23, 28], diabetes mellitus [3, 5, 12, 15, 16, 21, 28], antacid use [15, 28–30], antibiotic use [28], and other factors that can decrease immunity in the intestine or throughout the body [2, 16, 31]. Few data exist on risk factors for mortality associated with L. monocytogenes infection. Using the Los Angeles County (LAC) surveillance database for the period 1992–2004, we assessed risk factors for listeriosis-related death among individuals with nonperinatal cases. PATIENTS, MATERIALS, AND METHODS Sources of data. The Acute Communicable Disease Control Program of the LAC Department of Pubic Health has conducted surveillance for listeriosis since the large Southern California outbreak in 1985 [32]. The Acute Communicable Disease Control Program passive surveillance system receives case reports from health care providers, hospitals, and laboratories throughout the county, because they are all required to report listeriosis by an LAC mandate. These reporting sources or their diagnostic laboratories send Listeria isolates to the LAC Public Health Laboratory (Los Angeles, California) for confirmation by conventional biochemical methods [33]. During case in1508 • CID 2009:48 (1 June) • Guevara et al. vestigations, medical professionals complete standardized epidemiology forms and case abstraction forms. With use of available paper documents archived at the Acute Communicable Disease Control Program, which included epidemiology forms, case abstraction forms, patient history and physical notes, laboratory reports, and death certificates, we confirmed that database case records met the Acute Communicable Disease Control Program surveillance criteria of laboratory-confirmed isolation of L. monocytogenes and patient residency in LAC, and we verified survival outcome, risk factors, and dates of disease onset, specimen collection, diagnosis, hospital admission, birth, and death. Definitions. Positive culture of L. monocytogenes from a normally sterile body site or from a stool culture obtained during a listeriosis outbreak defined a listeriosis case. A nonperinatal listeriosis case was defined as occurring in a nonpregnant case patient 142 days old. Listeriosis mortality was defined by a confirmed death without documented evidence of recovery or clearing of infection or by listeriosis being listed as a cause or contributory cause of death. All case survivors recovered or had culture results negative for L. monocytogenes. Infection type was categorized into mutually exclusive groups, as follows: bacteremia, sepsis only, meningitis only, sepsis and meningitis, and other infections. Bacteremia was defined as present in a patient with positive blood culture results but no symptoms, whereas sepsis was defined present in a patient with a positive blood culture result with any symptoms of illness (e.g., fever, chills, abdominal pain, diarrhea, vomiting, backache, and headache). Meningitis was defined as present in a patient with a positive cerebrospinal fluid (CSF) culture result. Other infection was defined as having symptoms and a positive culture result obtained for a site other than blood or CSF. Racial and ethnic groups were case-defined as mutually exclusive groups of Asian, black, Hispanic, and white. Preexisting risk factors that were routinely asked about on epidemiology forms included chronic alcoholism, gastrointestinal disease, autoimmune disorder, history of organ transplant, asthma, kidney disease, chronic lung disease, liver disease, iron overload, cancer, diabetes, human immunodeficiency virus (HIV) infection, AIDS, intravenous drug use, age ⭓65 years, receipt of antibiotics, receipt of chemotherapy, receipt of radiation therapy, receipt of chemoradiation, receipt of steroids, receipt of other immunosuppressive therapy, use of antacids, use of folk medicine, and history of gastric surgical procedures. Preexisting risk factors preceded onset of listeriosis. The presence of these factors was primarily defined by reporting physicians and infection-control professionals and was secondarily defined by medical chart review. Patients with AIDS, by definition, had HIV infection. The definition of lung disease did not include asthma. Receipt of chemoradiation was defined as receipt of both chemotherapy and radiation therapy. Figure 1. Survival outcome and mortality among 285 patients with nonperinatal listeriosis, Los Angeles County, 1992–2004. A total of 53 patients died (black bars), and 232 patients survived (gray bars). The black line with black diamond markers indicates mortality, expressed as a percentage. Analysis. In addition to calculating crude relative risks (RR) of mortality, we performed unconditional multivariable logistic regression to estimate adjusted odds ratios (OR) and 95% confidence intervals (CIs) for nonperinatal listeriosis mortality. Through causal modeling [34, 35] based on biological plausibility, prior knowledge, and available data, we established a 29-variable main effects model. Reference groups for sex, age, and race and ethnicity were female sex, age !70 years, and white, respectively. Age, sex, and race and ethnicity were assumed to be possible confounders for any preexisting risk factor. In addition, sex and race and ethnicity were considered to be confounders for age. Various interaction terms were explored, such as age and race and ethnicity; age and sex; sex and race and ethnicity; age, sex, and race and ethnicity; cancer and steroid use; cancer and receipt of nonsteroid cancer treatments; receipt of steroids and receipt of antibiotics; diabetes and heart disease or hypertension; kidney disease and receipt of transfusion; cancer and age; and cancer and race and ethnicity. For each risk factor, adjusted OR estimates were compared between the main effects model and a smaller multivariable logistic regression model that focused on the risk factor and possible confounders. These results will not be presented here, but most of the adjusted OR estimates were very similar between models. Data management and analysis were performed using Microsoft Access, Microsoft Excel, and SAS, version 9.1 (SAS). RESULTS Study population. More than one-half of the study population (53.3%) was ⭓65 years of age (table 1). The predominant race was white (56.1% of patients). There were slightly more male patients (55.1%) than female patients (44.9%). The most common infection type was sepsis only (69.5% of patients). With the exception of 7 patients with outbreak cases that occurred in 2001 (from the only outbreak that occurred during the study period) and a blood donor, all patients were hospitalized, and 90% of patients had invasive disease. Mortality. Of 285 cases of nonperinatal listeriosis with known survival outcome that occurred during the period 1992– 2004, 53 (18.6%) resulted in death. The annual number of deaths ranged from 2 to 9, and annual mortality ranged from 7.41% to 31.0% (figure 1). Fewer deaths were observed over time; the median number of annual deaths was 5 for the period 1992–1998 and 2 for the period 1999–2004. In addition, mean annual mortality was 22.0% for 1992–1998 and 13.3% for 1999–2004. Mortality was highest for patients with sepsis only (47 [23.7%] of 198 cases ended in death; median age, 68.5 years). Mortality was 12.0% for patients with both sepsis and meningitis (3 of 25 cases ended in death; median age, 62 years), 7.14% for patients with other infections (2 of 28 cases ended in death; median age, 62 years), and 3.13% for patients with meningitis only (1 of 32 cases ended in death; median age, 56.5 years). Neither of the 2 cases of bacteremia ended in death (median age, 67 years). Demographic risk factors. Nonperinatal listeriosis mortality for male patients was 19.8% (31 of 157 male patients died) and 17.2% for female patients (22 of 128 female patients died). Compared with female patients, male patients had a crude RR of death of 1.15 (95% CI, 0.70–1.88). In general, nonperinatal listeriosis mortality increased with age (table 2) and was highest for patients 65–74 years of age Nonperinatal Listeriosis Mortality • CID 2009:48 (1 June) • 1509 Table 2. Mortality among 285 patients with nonperinatal listeriosis, Los Angeles County, 1992–2004. Age, years !5 5–14 15–34 35–44 No. (%) of patients (n p 285) No. of deaths Mortality, % RR of mortality (95% confidence interval) 2 (0.70) 6 (2.11) 0 0 0.00 0.00 … … 28 (9.82) 21 (7.37) 4 1 14.3 4.76 1.00 0.33 (0.04–2.77) 45–54 55–64 33 (11.6) 43 (15.1) 5 7 15.2 16.3 1.06 (0.31–3.57) 1.14 (0.37–3.54) 65–74 ⭓75 75 (26.3) 77 (27.0) 15 21 20.0 27.3 1.35 (0.49–3.71) 1.84 (0.69–4.89) NOTE. RR, relative risk. (20.0%) and patients ⭓75 years of age (27.3%). Compared with patients !65 years of age, patients ⭓65 years of age had a 1.85 times higher risk of death (95% CI, 1.09–3.14). Similarly, compared with patients !65 years of age, patients ⭓75 years of age had a 2.13 times higher risk of death (95% CI, 1.20– 3.79). Analysis by race and ethnicity found mortality to be highest among Asians (26.7%; 8 deaths among 30 cases), followed by blacks (23.3%; 7 deaths among 30 cases), whites (17.5%; 28 deaths among 160 cases), and Hispanics (15.4%; 10 deaths among 65 cases). With Hispanics as the reference group, the crude RRs of nonperinatal listeriosis mortality were 1.73 (95% CI, 0.76–3.95) for Asians, 1.52 (95% CI, 0.64–3.60) for blacks, and 1.14 (95% CI, 0.59–2.21) for whites. Preexisting risk factors. In bivariable analyses, nonperinatal listeriosis mortality was highest for patients with blood transfusions (60.0%), patients with both hematological and nonhematological malignancies (40.0%), patients with nonhematological malignancy (39.6%), patients who received radiation therapy (31.6%), patients with lung disease (30.0%), patients with asthma (30.0%), patients with kidney disease (29.6%), patients with hypertension (28.0%), patients with alcoholism (27.6%), patients with cancer (25.8%), and patients with liver disease (25.7%) (table 3). When comparing patients who had a specific preexisting risk factor with patients who did not have that risk factor, crude RRs of mortality were highest among patients with blood transfusions (3.36; 95% CI, 1.57–7.17), nonhematological malignancy (2.79; 95% CI, 1.74– 4.48), age ⭓70 years (1.99; 95% CI, 1.22–3.24), and kidney disease (1.85; 95% CI, 1.12–3.07). Of 20 patients with cases of nonperinatal listeriosis who had lung disease, 4 had lung cancer; 2 of these patients died. None of the 14 patients with nonperinatal listeriosis who did not have any recognized preexisting risk factor died. Multivariable modeling. With data from 281 patients with nonperinatal listeriosis (data from 4 patients could not be used, 1510 • CID 2009:48 (1 June) • Guevara et al. because their cancer could not be classified as hematological or nonhematological), the 29-variable main effects multivariable logistic regression model had statistically significant adjusted ORs for nonhematological malignancy (adjusted OR, 5.92; 95% CI, 1.85–18.9), alcoholism (adjusted OR, 4.63; 95% CI, 1.36–15.8), age ⭓70 years (adjusted OR, 3.44; 95% CI, 1.50– 7.87), receipt of steroid medication (adjusted OR, 3.34; 95% CI, 1.38–8.08), and kidney disease (adjusted OR, 2.94; 95% CI, 1.18–7.31) (table 4). Other risk factors with adjusted ORs 11.50 included undergoing blood transfusion (adjusted OR, 6.80), having asthma (adjusted OR, 4.60), black race (adjusted OR, 2.77), Asian race (adjusted OR, 2.70), receipt of antibiotics (adjusted OR, 1.70), having hypertension (adjusted OR, 1.69), receiving chemotherapy (adjusted OR, 1.66), and Hispanic race (adjusted OR, 1.55). None of the interaction terms explored were statistically significant; however, the effect estimate of the interaction of receipt of steroid medication and antibiotic use was strong (adjusted OR, 5.06; 95% CI, 0.86–29.7) and changed the effect estimates for Asian race (adjusted OR, 3.04; 95% CI, 0.90–10.2), receipt of blood transfusion (adjusted OR, 8.61; 95% CI, 0.70 -105.9), receipt of chemotherapy (adjusted OR, 1.95; 95% CI, 0.52–7.39), and having both hematological and nonhematological malignancies (adjusted OR, 2.20; 95% CI, 0.13–38.2). DISCUSSION With use of a multivariable logistic regression model to control confounding and estimate the direct effects of several possible causal risk factors, we found that nonhematological malignancy, alcoholism, age ⭓70 years, use of steroid medication, and kidney disease have strong statistically significant associations with nonperinatal listeriosis mortality. Other factors with strong but not statistically significant associations with listeriosis mortality were undergoing blood transfusion, having asthma, black race, Asian race, and use of both steroid medication and antibiotics. Table 3. Crude relative risks (RRs) of mortality among 285 patients with nonperinatal listeriosis, by preexisting risk factors, Los Angeles County, 1992–2004. No. of patients (n p 285) No. of deaths (n p 53) Risk, % RR (95% CI) 113 30 26.5 1.99 (1.22–3.24) 139 30 21.6 1.37 (0.84–2.24) Cancer Hematological malignancy Nonhematological malignancy Both hematological and nonhematological malignancies 93 46 48 5 24 6 19 2 25.8 13.0 39.6 40.0 1.71 0.67 2.79 2.21 Steroid use Receipt of radiation therapy 96 19 21 6 21.9 31.6 1.29 (0.79–2.11) 1.79 (0.88–3.64) Receipt of chemotherapy Receipt of chemoradiationb 50 12 24.0 1.38 (0.78–2.42) 9 2 22.2 1.20 (0.35–4.19) Receipt of other immunosuppressive drug History of organ transplant Autoimmune disorder Alcoholism 15 12 28 29 1 2 2 8 6.67 16.7 7.14 27.6 0.35 0.89 0.36 1.57 Asthma Lung diseasec 10 3 30.0 1.65 (0.62–4.39) 20 6 30.0 1.69 (0.83–3.47) Kidney disease Liver disease Diabetes Antibiotic use 54 35 68 76 16 9 14 17 29.6 25.7 20.6 22.4 1.85 1.46 1.15 1.30 Gastrointestinal disease Antacid use 25 24 2 3 8.00 12.5 0.41 (0.11–1.58) 0.65 (0.22–1.94) HIV infection AIDS Oral iron intake Otherd 13 7 9 1 0 1 7.69 0.00 11.1 0.40 (0.06–2.69) No deaths 0.59 (0.09–3.80) Hypertension 25 7 28.0 1.58 (0.80–3.13) 22 5 14 6 3 0 27.3 60.0 0.00 1.53 (0.74–3.17) 3.36 (1.57–7.17) No deaths Preexisting risk factor Age ⭓70 years Weakened immune systema Heart disease Receipt of blood transfusion None (unknown) (1.06–2.76) (0.30–1.47) (1.74–4.48) (0.73–6.65) (0.05–2.34) (0.25–3.24) (0.09–1.40) (0.82–3.00) (1.12–3.07) (0.78–2.73) (0.66–1.98) (0.78–2.17) NOTE. CI, confidence interval; HIV, human immunodeficiency virus. a Defined as having any of the following: autoimmune disorder, steroid use, receipt of chemotherapy, receipt of radiation, receipt of chemoradiation, immunosuppressive drug use, history of organ transplant, HIV infection, or AIDS. b Defined as receipt of both chemotherapy and radiation therapy. c The definition of lung disease does not include asthma. d Preexisting risk factor designated but not routinely determined. Our findings are supported by the 2 other published studies on listeriosis mortality. Bennion et al. [36] showed a decreasing trend in listeriosis mortality in the United States during the period 1990–2005, higher listeriosis mortality rates among Asians and blacks, increasing mortality rates with advancing age, and adjusted OR estimates reflective of the ones that we determined for kidney disease, diabetes, and liver disease. Our measure for liver disease, however, would correspond to the combination of their measures for liver and bile duct cancer, liver disease, and viral hepatitis. In addition, their statistically significant effect estimate for anemia suggests that the variable for blood transfusion might demonstrate statistical significance if data were available for a larger number of cases. The pop- ulation-based study from Denmark for the period 1994–2003 by Gerner-Smidt et al. [37] also presented similar mortality rates for all cases (21%) and for patients with bloodstream infection (21%), and they reported similar bivariable results for sex (RR, 1.0), alcoholism (RR, 1.5), hematological malignancy (RR, 0.59), nonhematological malignancy (RR, 2.0), and diabetes (RR, 1.35). The multivariable analysis of our study is unique, because it is based on a comprehensive causal model that describes the relationships among numerous risk factors for listeriosis. Gerner-Smidt et al. [37] published what is, to our knowledge, the only other cohort study thus far, and their multivariable models included serogroup and age, but they either aggregated most Nonperinatal Listeriosis Mortality • CID 2009:48 (1 June) • 1511 Table 4. Odds ratios (ORs) from a 29-variable logistic regression model based on causal diagrams for nonperinatal listeriosis mortality, Los Angeles County, 1992–2004. No. of patients (n p 281) No. of deaths (n p 52) OR (95% CI) 112 30 3.44 (1.50–7.87) Asian 28 7 2.70 (0.81–8.94) Black Hispanic 30 65 7 10 2.77 (0.81–9.47) 1.55 (0.58–4.12) 155 46 30 6 0.95 (0.44–2.05) 0.40 (0.08–1.98) Nonhematological malignancy Both hematological and nonhematological malignancies 48 5 19 2 5.92 (1.85–18.9) 1.36 (0.08–22.2) Receipt of steroid medication Receipt of chemotherapy 95 49 21 12 3.34 (1.38–8.08) 1.66 (0.46–5.92) Receipt of radiation therapy a Receipt of chemoradiation Receipt of other immunosuppressive drug Autoimmune disease 19 9 6 2 0.89 (0.19–4.17) 0.42 (0.05–3.70) 15 27 1 2 0.28 (0.03–3.10) 0.26 (0.04–1.56) HIV infection History of organ transplant 13 12 1 2 0.53 (0.05–6.25) 0.88 (0.12–6.35) Alcoholism Asthma Lung diseaseb Kidney disease 27 10 20 54 7 3 6 16 4.63 4.60 1.37 2.94 Liver disease Diabetes 34 66 8 14 1.45 (0.50–4.22) 1.01 (0.42–2.45) Gastrointestinal disease Antacid use Receipt of antibiotics Hypertension 25 24 76 25 2 3 17 7 0.50 0.29 1.70 1.69 Heart disease Receipt of blood transfusion Oral iron intake Receipt of steroid medication and antibioticsc 22 4 9 6 2 1 1.36 (0.40–4.66) 6.80 (0.61–76.3) 0.18 (0.01–3.39) 38 11 5.06 (0.86–29.7) Variable Age ⭓70 years Race and ethnicity Male sex Hematological malignancy (1.36–15.8) (0.81–25.9) (0.39–4.78) (1.18–7.31) (0.09–2.65) (0.06–1.28) (0.73–3.96) (0.47–6.02) NOTE.Of 285 patients with a known survival outcome, 4 did not have enough information in their records for their cancer to be classified as hematological or nonhematological. Therefore, the number of patients included in the multivariable analysis is 281. The reference groups for age, race and ethnicity, and sex were age !70 years, white race, and female sex, respectively. All variables except receipt of steroids and antibiotics were included in the main effects model. CI, confidence interval; HIV, human immunodeficiency virus. a Defined as receipt of both chemotherapy and radiation therapy. The definition of lung disease does not include asthma. Addition of this interaction term to the main effects model only changed the effect estimates for Asian race and ethnicity (OR, 3.04; 95% CI, 0.90–10.2), receipt of blood transfusion (OR, 8.61; 95% CI, 0.70 –105.9), receipt of chemotherapy (OR, 1.95; 95% CI, 0.52–7.39), and having both hematological and nonhematological malignancies (OR, 2.20; 95% CI, 0.13–38.2). b c of the preexisting risk factors into one variable to represent the presence of any preexisting risk factor or selected one preexisting risk factor of interest. Although this method of analysis increases statistical power, it may oversimplify the causal relationships among risk factors and listeriosis mortality. Because nonperinatal listeriosis cases can have a variety of different preexisting risk factors, we accounted for confounders and intermediate factors as distinct variables in a single multivariable model to estimate the direct effects of each specific risk factor on mortality. 1512 • CID 2009:48 (1 June) • Guevara et al. Biological plausibility and prior knowledge were the driving factors of our analysis. Therefore, we did not seek to combine or exclude variables on the basis of statistical significance, especially when patient counts were small. Such methods would decrease the accuracy of effect measures and would add unnecessary assumptions to the analysis. Although our model distinguishes the effects of documented risk factors for listeriosis on mortality, additional studies involving a greater number of cases and a comprehensive causal model are required to provide more-exact estimates. Because the variables that were associated with statistically significant adjusted ORs in our analysis are fairly well recognized risk factors for listeriosis, we will not discuss their biological plausibility. However, these factors should be considered in future research on listeriosis mortality. Our results regarding the impact of blood transfusion, asthma, black race, Asian race, receipt of cancer treatment, and receipt of a combination of steroid medication and antibiotics are intriguing and require additional research. Because iron is a virulence factor in listeriosis and because iron overload and hemachromatosis are disease risk factors [13, 27, 38–40], future research should include measures on patient history of blood disorders, donation and receipt of blood components, and receipt of dialysis. Only additional studies of how asthma might cause listeriosis [41] will help to explain the strong adjusted OR that we found for asthma. Although Bennion et al. [36] had similar findings, researching social factors concerning health care access, cultural perceptions, and tendencies to seek care might elucidate why Asians and blacks have the highest risks for listeriosis mortality. The cancer treatment variables— receipt of steroid medication, chemotherapy, radiation, and chemoradiation—show a diminishing causal effect on listeriosis mortality. Regarding receipt of steroid medication and antibiotics, effect modification is possible, because antibiotics disrupt the intestinal flora and can allow L. monocytogenes to proliferate, resulting in an enhanced opportunity to infect the blood through the intestinal mucosa, whereas steroid medication suppresses cell-mediated immunity. Of interest, we found that mortality was lowest among patients with meningitis only, was higher among patients with both sepsis and meningitis, and was highest among patients with sepsis only. Possible explanations for this finding, which contrasts with that of the study by Gerner-Smidt et al. [37], include differences in definitions for infection types, in diagnostic testing practices, and in the treatment of patients with meningitis. Both studies lack data on antibiotic treatment. However, a possible explanation is that antibiotic treatment is more appropriate for patients with listeriosis who have meningitis only than it is for patients with other infection types, particularly sepsis. L. monocytogenes is a primary suspect in cases of bacterial meningitis, and treatment guidelines cite ampicillin for coverage [42]. However, such is not so in cases of sepsis, and initial antibiotic treatment might not cover L. monocytogenes. Differences in median patient ages might also explain differences in mortality by infection type. Although the 3 studies of listeriosis mortality had similar results, there are important differences regarding effect estimates of risk factors for mortality. The cohort studies support each other, regardless of differences in statistical significance, because differences stem from the different approaches in modeling the analysis. However, Bennion et al. [36] found lymphoid and hematopoeitic cancer, HIV infection, systemic lupus erythematosus, and history of organ transplant to be strongly associated with listeriosis in death certificates. Conclusions from their adjusted ORs require assumptions that the control subjects represent the base population (i.e., the control subjects represent the case patients, such that, had the control subjects been exposed to Listeria species, they would have been just as likely to become case patients and to have died), consistency of coding in death certificate data, high sensitivity and specificity of death certificates in identifying listeriosis deaths and predisposing conditions, and confounding only by age, race, sex, and HIV infection status. Nevertheless, their findings are important, because they are biologically plausible and, except for lymphoid and hematopoeitic cancer, the cohort studies might not have involved a sufficient number of cases to show a strong (or even positive) association in multivariable analyses. Thus, the variables that did not demonstrate strong or positive associations in our multivariable analysis should still be considered to be possible risk factors for death when assessing individual cases of listeriosis. Although these 3 studies have measures for specific risk factors, none of them have a measure for a person’s health status. We looked at a number of risk factors as predictors for listeriosis mortality, and we found no positive association. For this reason, a measure for how sick or weakened a person is should be established for more-accurate effect estimates of risk factors for listeriosis mortality. Our study has limitations. Laboratory-confirmed clearance of L. monocytogenes was not always obtained; serogrouping was not performed; antibiotic treatment data were not collected; the presence and absence of preexisting risk factors was not ascertained beyond the archived case records; and preexisting blood transfusion, hypertension, and heart disease data were not routinely collected, because they were not specified on the standardized forms. In addition, data on kidney dialysis were not collected. Given our findings regarding receipt of blood transfusion, dialysis might be an important intermediate factor in the causal pathway from kidney disease to death. Thus, the adjusted OR for kidney disease might be different when dialysis is accounted for. Types of cancer, diabetes, and lung disease, as well as stage of cancer, were not distinguished; however, with analysis of a large enough number of cases, they might show different effects on mortality. For a few risk factors, small patient counts caused wide 95% CIs and unstable effect estimates that could have been more easily affected by the number of covariates in multivariable analyses. Because of the number of risk factors that were considered in this study, another concern was finding a statistically significant association by chance as a result of multiple comparisons, despite the consideration of causal relationships and the results of bivariable analyses. Finally, causal modeling was based on current knowledge. ThereNonperinatal Listeriosis Mortality • CID 2009:48 (1 June) • 1513 fore, it is possible that not all possible confounders are represented in the models, and some factors that are designated as confounders might not actually be so. Invasive nonperinatal listeriosis is an important cause of preventable mortality. This study should help researchers and clinicians to focus on specific risk factors to prevent nonperinatal listeriosis–related deaths. Improvements in preventing mortality are strongly needed for patients who present with sepsis only. Additional population-based studies with multivariable analyses structured on causal models are required to confirm the findings of this study, specifically when more observations were needed and measures of risk factors were not routinely obtained. Furthermore, the mechanisms of how asthma, receipt of blood transfusion, kidney disease, and alcoholism increase the odds of mortality should be explored, because they might not be specific to listeriosis. Finally, public health and preventive medicine efforts should involve the education of medical communities, including pharmacists, regarding the use of steroids and antibiotics, and the general population should be educated on the associated risk groups and high-risk foods, so that individuals who are at high risk for infection, their health care providers, and their family caregivers remain aware of listeriosis and the increased risk of death. 22. Acknowledgments 23. We acknowledge Amy Gallagher for her help in the verification of case data. Financial support. County of Los Angeles, Department of Public Health. Potential conflicts of interest. L.M. is on the speakers’ bureaus for Merck and MedImmune. All other authors: no conflicts. 11. 12. 13. 14. 15. 16. 17. 18. 19. 20. 21. 24. 25. References 26. 1. Mead PS, Slutsker L, Dietz V, et al. Food-related illness and death in the United States. Emerg Infect Dis 1999; 5:607–25. 2. Gellin BG, Broome CV. Listeriosis. JAMA 1989; 261:1313–20. 3. Goulet V, Marchetti P. Listeriosis in 225 non-pregnant patients in 1992: clinical aspects and outcome in relation to predisposing conditions. Scand J Infect Dis 1996; 28:367–74. 4. Rocourt J, Bille J. Foodborne listeriosis. World Health Stat Q 1997; 50:67–73. 5. Siegman-Igra Y, Levin R, Weinberger M, et al. Listeria monocytogenes infection in Israel and review of cases worldwide. Emerg Infect Dis 2002; 8:305–10. 6. McLauchlin JS. Human listeriosis in Britain, 1967–85, a summary of 722 cases. 2. Listeriosis in non-pregnant individuals, a changing pattern of infection and seasonal incidence. Epidemiol Infect 1990; 104: 191–201. 7. Gillespie IA, McLauchlin J, Grant KA, et al. Changing pattern of human listeriosis, England and Wales, 2001–2004. Emerg Infect Dis 2006; 12: 1361–6. 8. Nieman RE, Lorber B. Listeriosis in adults: a changing pattern. Report of eight cases and review of the literature, 1968–1978. Rev Infect Dis 1980; 2:207–27. 9. Paul ML, Dwyer DE, Chow C, et al. Listeriosis: a review of eighty-four cases. Med J Aust 1994; 160:489–93. 10. Rettally CA, Speeg KV. Infection with Listeria monocytogenes following 1514 • CID 2009:48 (1 June) • Guevara et al. 27. 28. 29. 30. 31. 32. 33. 34. 35. orthotopic liver transplantation: case report and review of the literature. Transplant Proc 2003; 35:1485–7. Newton L, Hall SM, Pelerin M, McLauchlin J. Listeriosis surveillance: 1991. Commun Dis Rep CDR Rev 1992; 2:R142–4. Skogberg K, Syrjanen J, Jahkola M, Renkonen OV, et al. Clinical presentation and outcome of listeriosis in patients with and without immunosuppressive therapy. Clin Infect Dis 1992; 14:815–21. Vazquez-Boland JA, Kuhn M, Berche P, et al. Listeria pathogenesis and molecular virulence determinants. Clin Microbiol Rev 2001; 14: 584–640. Ciesielski CA, Hightower AW, Parsons SK, Broome CV. Listeriosis in the United States: 1980–1982. Arch Intern Med 1988; 148:1416–9. Buchholz U, Mascola L. Transmission, pathogenesis, and epidemiology of Listeria monocytogenes. Infect Dis Clin Prac 2001; 10:34–41. Slutsker L, Schuchat A. Listeriosis in humans. In: Ryser E, Harth EH, eds. Listeria, listeriosis and food safety. New York: Marcel Dekker, 1999: 75–95. Cherubin CE, Appleman MD, Heseltine PN, Khayr W, Stratton CW. Epidemiological spectrum and current treatment of listeriosis. Rev Infect Dis 1991; 13:1108–14. Mascola L, Sorvillo F, Neal J, Iwakoshi K, Weaver R. Surveillance of listeriosis in Los Angeles County, 1985–1986: a first year’s report. Arch Intern Med 1989; 149:1569–72. Harisdangkul V, Songcharoen S, Lin AC. Listerial infections in patients with systemic lupus erythematosus. South Med J 1992; 85:957–60. Kraus A, Cabral AR, Sifuentes-Osornio J, Alarcon-Segovia D. Listeriosis in patients with connective tissue diseases. J Rheumatol 1994; 21:635–8. Ortel S. Listeria-meningitis and -septicaemia in immunocompromised patients. Acta Microbiol Hung 1989; 36:153–7. Stamm AM, Dismukes WE, Simmons BP, et al. Listeriosis in renal transplant recipients: report of an outbreak and review of 102 cases. Rev Infect Dis 1982; 4:665–82. Chadwick RG, Graham JM. Chronic active hepatitis, haemolytic anaemia and Listeria monocytogenes bacteraemia. Postgrad Med J 1978; 54: 55–7. Peetermans WE, Endtz HP, Janssens AR, van den Broek PJ. Recurrent Listeria monocytogenes bacteraemia in a liver transplant patient. Infection 1990; 18:107–8. Limaye AP, Perkins JD, Kowdley KV. Listeria infection after liver transplantation: report of a case and review of the literature. Am J Gastroenterol 1998; 93:1942–4. De Vega T, Echevarria S, Crespo J, Artinano E, San Miguel G, Pons Romero F. Acute hepatitis by Listeria monocytogenes in an HIV patient with chronic HBV hepatitis. J Clin Gastroenterol 1992; 15:251–5. Mossey RT, Sondheimer J. Listeriosis in patients with long-term hemodialysis and transfusional iron overload. Am J Med 1985; 79: 397–400. Nakajima T, Kodaira M, Masuda Y, et al. Five cases of listeriosis in the elderly. Kansenshogaku Zasshi 1990; 64:1468–73. Ho JL, Shands KN, Friedland G, Eckind P, Fraser DW. An outbreak of type 4b Listeria monocytogenes infection involving patients from eight Boston hospitals. Arch Intern Med 1986; 146:520–4. Schlech WF, Chase DP, Badley A. A model of food-borne Listeria monocytogenes infection in the Sprague-Dawley rat using gastric inoculation: development and effect of gastric acidity on infective dose. Int J Food Microbiol 1993; 18:15–24. Poropatich R, Phillips YY. Listerial brain abscess in long-standing sarcoidosis. South Med J 1992; 85:554–6. Linnan JM, Mascola L, Lou XD, et al. Epidemic listeriosis associated with Mexican-style cheese. N Engl J Med 1988; 319:823–8. Schuchat A, Swaminathan B, Broom CV. Epidemiology of human listeriosis. Clin Microbiol Rev 1991; 4:169–83. Greenland S. Modeling and variable selection in epidemiologic analysis. Am J Public Health 1989; 79:340–9. Greenland S, Brumback B. An overview of relations among causal modelling methods. Int J Epidemiol 2002; 31:1030–7. 36. Bennion JR, Sorvillo F, Wise ME, Krishna S, Mascola L. Decreasing listeriosis mortality in the United States, 1990–2005. Clin Infect Dis 2008; 47:867–74. 37. Gerner-Smidt P, Ethelberg S, Schiellerup P, Christensen JJ, et al. Invasive listeriosis in Denmark 1994–2003: a review of 299 cases with special emphasis on risk factors for mortality. Clin Microbiol Infect 2005; 11:618–24. 38. Lorber B. Listeriosis. Clin Infect Dis 1997; 24:1–11. 39. Van Asbeck BS, Verbrugh HA, van Oost BA, et al. Listeria monocytogenes meningitis and decreased phagocytosis associated with iron overload. Br Med J (Clin Res Ed) 1982; 284:542–4. 40. Guevara RE, Tormey MP, Nguyen DM, Mascola L. Listeriosis monocytogenes in platelets: a case report. Transfusion 2006; 46:305–9. 41. Van Loveren H, Steerenberg PA, Garssen J, Van Bree L. Interaction of environmental chemicals with respiratory sensitization. Toxicol Lett 1996; 86:163–7. 42. van de Beek D, de Gans J, Tunkel AR, Wijdicks EFM. Communityacquired bacterial meningitis in adults. N Engl J Med 2006; 354:44–53. Nonperinatal Listeriosis Mortality • CID 2009:48 (1 June) • 1515