mortality, morbidity, and microbiology of endemic cholera among ...

Am. J. Trop. Med. Hyg., 63(1, 2), 2000, pp. 12–20 Copyright 䉷 2000 by The American Society of Tropical Medicine and Hygiene

MORTALITY, MORBIDITY, AND MICROBIOLOGY OF ENDEMIC CHOLERA AMONG HOSPITALIZED PATIENTS IN DHAKA, BANGLADESH EDWARD T. RYAN, UJJWAL DHAR, WASIF A. KHAN, MOHAMMED ABDUS SALAM, ABU S. G. FARUQUE, GEORGE J. FUCHS, STEPHEN B. CALDERWOOD, AND MICHAEL L. BENNISH Tropical and Geographic Medicine Center, Division of Infectious Diseases, Massachusetts General Hospital, Boston, Massachusetts; Department of Medicine, and Department of Microbiology and Molecular Genetics, Harvard Medical School, Boston, Massachusetts; Clinical Sciences Division, International Centre for Diarrhoeal Disease Research (ICDDR,B: Centre for Health and Population Research), Dhaka, Bangladesh; Department of Pediatrics, Louisiana State University School of Medicine, New Orleans, Louisiana; Division of Geographic Medicine and Infectious Diseases, Tupper Research Institute, New England Medical Center, Boston, Massachusetts

Abstract. Mortality and morbidity associated with cholera acquired in a modern endemic setting have not been well defined. In Dhaka, Bangladesh from 1986 to 1996, we found that causative agents of cholera shifted over time, varying by serogroup, biotype, and serotype. At the International Centre for Diarrhoeal Disease Research (ICDDR,B: Centre for Health and Population Research) in 1996, 19,100 cholera patients were treated, 887 (4.6%) were admitted, and 33 died (mortality rate ⫽ 3.7% of cholera inpatients, 0.14% of all cholera patients). When cholera inpatients who were discharged improved were compared with those who died, bacteremia (odds ratio [OR] ⫽ 10.5, 95% confidence interval [CI] ⫽ 2.9–37.9), radiographic evidence of pneumonia (OR ⫽ 3.1, 95% CI ⫽ 1.2–7.7), and acidosis as estimated by the serum bicarbonate value (OR ⫽ 0.893, 95% CI ⫽ 0.825–0.963) were independently associated with death by multivariate analysis. Pneumonia was the leading cause of death and accounted for two-thirds of all deaths among individuals with cholera in this study. Death in hospitalized patients with cholera acquired in a modern endemic setting is, therefore, extremely rare, and most frequently due to concomitant infection, especially pneumonia. INTRODUCTION

PATIENTS AND METHODS

Cholera is a watery diarrhea caused by the gram-negative bacillus Vibrio cholerae serogroup O1 or O139; if untreated, it is often lethal. Each year, five to seven million cases of cholera occur worldwide, resulting in approximately 100,000 deaths.1 Cholera epidemics can occur when V. cholerae is introduced into populations lacking protective immunity.2,3 The morbidity and mortality associated with epidemic cholera are primarily related to severe fluid and electrolyte depletion, and epidemic cholera can be associated with extremely high death rates. Of the approximately 500,000– 800,000 refugees in Goma, Zaire in 1994, for example, an estimated 50,000 died within a 21-day period due to explosive epidemics of cholera and shigellosis.4,5 Despite such dramatic manifestations of epidemic cholera, most cholera cases actually occur in areas of the world endemic for this infection, such as in Bangladesh.2,6–9 Endemic cholera results when V. cholerae organisms persist in the environment and when infection of a human population is ongoing and longterm.8,10,11 Due to the effects of varying degrees of pre-existing immunity, experience with the use of rehydration fluid therapy, and the influence of malnutrition and other co-morbid conditions common in cholera endemic areas, the clinical features of endemic cholera are distinct from those of epidemic cholera.6,8,10,12,13 Despite this, the morbidity and mortality currently associated with endemic cholera remain less well defined than those associated with epidemic cholera.11,14–16 In particular, causes of death among individuals hospitalized with cholera in a modern endemic setting remain largely unexamined. In this study, we examined data from patients presenting with cholera in an endemic setting in urban Bangladesh. We characterized clinical and microbiologic features of cholera, and identified features associated with hospital admission and death.

Study site and patients. The International Centre for Diarrhoeal Disease Research in Bangladesh (ICDDR,B: Centre for Health and Population Research) maintains a Clinical Research and Service Centre in Dhaka, Bangladesh. The Centre serves an urban population of approximately 8,000,000 individuals,17 and provides care for more than 100,000 individuals with diarrhea each year. Since 1979, when the population of Dhaka was estimated to be approximately three million,17 the ICDDR,B has maintained a surveillance system that obtains detailed information on every 25th patient presenting to the Centre (every 50th patient after 1995); this information includes a detailed history, physical examination, and microbiologic examination of stool or rectal swab sample for bacterial, viral, and parasitic enteric pathogens.18 Vibrio cholerae organisms are identified on taurocholate-tellurite-gelatin plates using standard microbiologic methods; serogrouping and serotyping are performed using specific antisera.18–23 The Centre maintains an outpatient treatment pavilion as well as an inpatient ward. Most individuals presenting with diarrhea to the Centre are treated in the outpatient pavilion where rehydration therapy (oral and intravenous) and antibiotics (oral amoxicillin, erythromycin, or doxycycline) can be administered; patient education programs also occur in the pavilion. Most outpatients are treated and released within 24 hr of presenting to the Centre. Individuals who require additional care are admitted to the inpatient unit. To determine the cause of death and complications occurring in individuals with cholera, we reviewed the hospital record of every second patient admitted to the inpatient unit with microbiologically confirmed cholera in 1996. In addition, we reviewed the hospital record of every patient with cholera in 1996 who 1) died in the hospital, 2) was referred to another hospital for additional care, or 3) left the hospital against medical advice. We also reviewed the records of all

12

13

ENDEMIC CHOLERA IN BANGLADESH

TABLE 1 Cholera presentations, admissions, and deaths, International Centre for Diarrhoeal Disease Research, Bangladesh, 1986–1996 Cholera cases Presenting patients* Year

1986 1987 1988 1989 1990 1991 1992 1993 1994 1995 1996 Total

Admitted (% of presenting patients)*

Died in-hospital

Admitted patients

Number

12,750 14,175 13,750 2,600 3,575 16,625 12,925 35,175 21,775 21,425 19,100 173,875

464 545 532 254 102 436 578 397 837 963 887 5,995

(3.6) (3.8) (3.7) (9.8) (2.9) (2.6) (4.5) (1.1) (3.8) (4.5) (4.6) (3.4)

Presenting patients*

Death rate per 100

19 21 14 10 3 20 26 24 51 27 33 248

4.1 3.9 2.6 3.9 2.9 4.6 4.5 6.0 6.1 2.8 3.7 4.1

0.15 0.15 0.10 0.38 0.08 0.12 0.20 0.07 0.23 0.13 0.17 0.14

* Extrapolated from surveillance data. From 1986–1995, a stool or rectal swab sample of every 25th patient was cultured for Vibrio cholerae. In 1996, a sample from every 50th patient was obtained.

persons, both inpatient and outpatients, infected with V. cholerae O1 or O139 who were identified as part of the surveillance system for 1996. To place the data for cholera in 1996 in perspective, we compared cholera presentations, cholera admissions to the inpatient unit, and cholera deaths for the years 1986–1996. The following information from the time of presentation was abstracted from the medical records of cholera inpatients cared for in 1996: age, gender, duration of diarrhea, stool consistency, recent illness, diet, use of rehydration therapy (oral and intravenous) before presentation, history of cough, and severity of dehydration. The following information from the time of admission for inpatients, or at the time of presentation for cholera outpatients identified as part of the surveillance system, was also abstracted: weight-forage as a percentage of the median weight for the age group;24 temperature; presence on physical examination of xerophthalmia, edema, labored/gasping respirations, lung field crepitations, abdominal distention, seizures, or unconsciousness; type of V. cholerae isolated in stool; and isolation of Salmonella or Shigella species in stool. Therapeutic interventions and outcome (categorized as discharged improved, discharged against medical advice, transferred, or died in-hospital) were also recorded. Statistical analysis. StatPac Gold version 3.2 (Walonick Associates, Minneapolis, MN) was used for database entry. Data analysis was performed in part using the Statistical Package for Social Sciences version 8.0 for Windows (SPSS, Chicago, IL). The significance of differences in proportions was calculated with the chi-square test or Fisher’s exact test if the size of any cell was ⱕ 5. For comparison of three or more groups, the chi-square test for m ⫻ 2 tables was used. The significance of differences in continuous variables involving three groups was tested using the Kruskal-Wallis test for K independent samples. For testing the significance of differences between two groups, the Student’s t-test for independent samples was used for normally distributed continuous data, and the Mann-Whitney U test was used for non-normally distributed data. Features independently associated with in-hospital death were determined by multiple logistic regression analysis;

variables found significantly associated with death in univariate analyses (P ⱕ 0.10) were included in the initial multivariate model. Variables were removed in a stepwise fashion if the probability associated with the likelihood ratio statistic for that variable exceeded 0.05. The final multiple logistic regression analysis included only those variables that were significant in the initial model. Multivariate odds ratios and confidence intervals were calculated from the coefficient of the multiple regression model. RESULTS

Cholera cases, 1986–1996. Based on extrapolations from the surveillance study sample of every 25th (1985–1995) or 50th (1996) outpatient, the estimated number of patients presenting with cholera to the ICDDR,B varied from 2,600 in 1989 to 35,175 in 1993 (mean ⫽ 12,834 per year) (Table 1). Based on phenotypic characteristics, V. cholerae serogroup O1 organisms can be classified as classical or El Tor biotypes; based on differences in antigenic determinants of the lipopolysaccharide O antigen, the organisms can be further subclassified into serotypes Inaba and Ogawa.20,25 Since 1987, V. cholerae O1 El Tor biotype has been the predominant cholera-causing organism isolated at the ICDDR,B with the exception of 1993, when V. cholerae O139 accounted for the majority of cholera cases (80.7%). When classical and El Tor biotypes were considered, Ogawa serotype accounted for the majority (89.0%) of all V. cholerae O1 isolates throughout the time period reviewed. From 1986 to 1996, the number of cholera cases caused by specific V. cholerae organisms peaked at different times; with V. cholerae O1 classical Ogawa peaking in 1987, V. cholerae O1 El Tor Inaba in 1988, V. cholerae O1 El Tor Ogawa in 1991, V. cholerae O139 in 1993, and V. cholerae O1 El Tor Ogawa again in 1995 (Figure 1). The number of individuals admitted to the inpatient unit yearly with cholera has generally been proportional to the number presenting with cholera each year and has ranged from a low of 102 in 1990 to a high of 963 in 1995 (mean ⫽ 470 per year). The admission rate of individuals present-

14

RYAN AND OTHERS

FIGURE 1. Number of cholera patients presenting to the International Centre for Diarrhoeal Disease Research, Bangladesh (ICDDR,B) in Dhaka, Bangladesh, by year, 1986–1996. Figures represent the number of cholera cases by type of Vibrio cholerae. Numbers of cases are estimated by multiplying the percentage of individuals infected with V. cholerae O1 or O139 identified as part of the surveillance system each year by the total number of individuals presenting that year. Throughout the period examined, cholera cases caused by specific V. cholerae organisms peaked at different times, varying by causative agent.

TABLE 2 Other organisms identified in stool of surveillance patients with Vibrio cholerae O1 or O139 Organism*

Rotavirus Other Vibrionaceae† Campylobacter jejuni Shigella spp. S. flexneri S. boydii S. dysenteriae non-1 Not speciated Salmonella spp. Giardia lamblia Entameoba histolytica / dispar Cryptosporidium parvum Multiple organisms Rotavirus & other Vibrionaceae† Rotavirus & S. boydii Rotavirus & Shigella spp. Rotavirus & Salmonella spp. & C. parvum E. histolytica/dispar & C. parvum E. histolytica/dispar & other Vibrionaceae† Total Salmonella spp. or Shigella spp. & Vibrio cholerae O1 or O139 Total other organism & Vibrio cholerae O1 or O139

No. identified (%) (All patients: n ⫽ 382)

22 14 10 9 4 2 2 1 5 4 2 2 7 2 1 1 1 1 1

(5.8) (3.7) (2.6) (2.4) (1.0) (0.5) (0.5) (0.3) (1.3) (1.0) (0.5) (0.5) (1.8) (0.5) (0.3) (0.3) (0.3) (0.3) (0.3)

17 (4.5) 84 (22)

* Rotavirus identified by enzyme linked immunosorbent assay; G. lamblia, E. histolytica / dispar, C. parvum identified by microscopy; all other organisms identified by culture. † V. cholerae non-O1/non-O139, other Vibrio spp., Aeromonas spp., Plesiomonas spp.

ing with cholera from 1986 to 1996 was 3.4% (range ⫽ 1.1– 9.8% per year of presenting cholera patients). Surveillance patients presenting with cholera, 1996. In 1996, 382 (17.6%) of the 2,171 individuals with diarrhea enrolled in the surveillance system were infected with V. cholerae O1 or O139. When extrapolated to the 108,550 individuals presenting to the ICDDR,B that year, an estimated 19,100 individuals presented with cholera to the ICDDR,B in 1996. Cholera patients had a median age of 10 years; 55.2% were ⱕ 15 years of age, 39.3 percent were ⱕ 5 years of age, and 12.0 percent were ⱕ 1 year of age. Fiftyfive percent were male. Watery stools were present in 97.9% of the cholera patients. Vibrio cholerae O1 El Tor Ogawa accounted for 74.6% of the isolates, 8.7% were V. cholerae O1 El Tor Inaba, and 16.7% were V. cholerae O139. Children and child-bearing age women with cholera were routinely treated with a three-day course of oral erythromycin, all other patients with cholera were routinely treated with either a one time oral dose of 300 mg of doxycycline or a three-day course of 100 mg (twice a day) of doxycycline. An additional enteric pathogen was identified from 84 (22%) of the 382 persons with cholera entered into the surveillance study (Table 2). Salmonella or Shigella species were isolated in 4.5% of the patients; none had bloody diarrhea suggestive of disease from these second infections. No patient identified with cholera as part of the surveillance system died or required referral to another hospital; 2.9% were admitted, and 3.2% left the Centre against medical advice. Comparison of cases with cholera treated as inpatients versus outpatients, 1996. In 1996, 5,720 patients were admitted to the ICDDR,B inpatient wards; 887 (15.5%) of

15


these individuals were infected with V. cholerae O1 or O139. Cholera inpatients represented 4.6% of the estimated 19,100 patients presenting with cholera in 1996. A number of differences were found between inpatients and outpatients with cholera (P ⱕ 0.05). When compared with cholera outpatients, inpatients were significantly younger (median age ⫽ 9 months versus 144 months), more malnourished (percentage of median weight-for-age, 65.4% versus 73.2%),24 more likely to have a longer duration of diarrhea (median ⫽ 72 hr versus 18 hr), co-infection with Salmonella or Shigella species (10.6% versus 3.8%), or blood in stool (2.9% versus 0.5%), more likely to have evidence of respiratory infection including cough (33.5% versus 23.7%), labored or gasping respirations (15.1% versus 0.8%), or lung field crepitations (29.4% versus 1.6%), and more likely to be febrile (temperature ⱖ 37.8⬚C, 33.6% versus 3.5%), and to have peripheral edema (6.4% versus 0%). Volume status and rehydration therapy. The majority (90.3%) of cholera outpatients and inpatients used oral rehydration therapy before presenting to the ICDDR,B. The majority of cholera outpatients (74.7%) and inpatients (78.3%) were treated with intravenous fluids within 24 hr of arriving at the Centre. Information on therapeutic complications of rehydration was available for cholera inpatients only. Of the 450 (50.7%) cholera inpatients whose serum sodium level was measured on admission, 12 (2.7%) had a serum sodium level ⱖ 146 mmol/L and 52 (11.6%) had a serum sodium level ⱕ 125 mmol/L. The contribution that rehydration therapy made to electrolyte disturbances in these individuals was unclear. Tetany or carpal pedal spasm developed in 31 (6.3%) cholera inpatients who were treated with intravenous fluids, and in none of the cholera inpatients who were treated solely with oral rehydration therapy. All patients who developed tetany or carpal pedal spasm were acidotic at admission, and all were receiving bicarbonate containing intravenous fluids; serum calcium levels were not checked. Two three-month-old malnourished children treated with intravenous fluid therapy developed overt congestive heart failure. One adult patient developed fatal nosocomial intravenous catheter/infusate related sepsis. Three adult patients had acute neurologic deficits consistent with hypovolemia related cerebral ischemia, and two adult patients had cardiac ischemic events. Of the 113 cholera patients whose serum creatinine levels were measured at admission, 31 (27.4%) had values ⬎ 200 ␮mol/L (normal ⫽ 60–140 ␮mol/ L). Reasons for referral. After being rehydrated and stabilized, 23 (2.6%) of the 887 cholera inpatients were referred to another hospital. Anuria, presumably secondary to cholera associated hypovolemia, was the most common reason for referral (12 of 23, 52.2%). Acute cerebral ischemia (3 of 23, 13%), neonatal sepsis (3 of 23, 13%), progressive pneumonia/respiratory failure (2 of 23, 8.7%), disseminated tuberculosis (1 of 23, 4.3%), surgical evaluation of the abdomen (1 of 23, 4.3%), and nutritional rehabilitation (1 of 23, 4.3%) accounted for the remainder. Information on the final illness outcome for these patients could not be obtained. Outcome and risk factors for death. Death rate, 1986 to 1996. From 1986 to 1996, the death rate of cholera inpatients was 4.1% (range ⫽ 2.6–6.1% per year); when chol-

era outpatients are included, the death rate was 0.14% (range ⫽ 0.07–0.38% per year) (Table 1). Admission characteristics by outcome, 1996. Data on admission from 377 patients with cholera in 1996 who were subsequently discharged improved were compared with admission data for the 33 patients admitted with cholera who subsequently died in-hospital (Table 3). In univariate analysis, inpatients who died in-hospital differed from those discharged improved in a number of important aspects, including level of dehydration, level of consciousness, nutritional status, degree of acidosis, incidence of hypoglycemia, evidence of respiratory infection, and likelihood of bacteremia. In multivariate analysis, factors independently predictive of death were bacteremia, radiographic evidence of pneumonia, and acidosis (Table 4). Causes of death. The hospital records of the 33 cholera patients who died in-hospital were reviewed to establish cause of death (Table 5). Six patients died of hypovolemic shock, four of whom were pulseless and unconscious upon arriving at the ICDDR,B. All six of these patients died within 12 hr of arriving; five died within 5 hr (Figure 2) (death rate due to hypovolemic shock ⫽ 0.7% of cholera inpatients, 0.03% of all patients presenting with cholera). Concomitant infection was the cause of death in the remaining 27 patients (death rate ⫽ 3.0% of cholera inpatients, 0.14% of all presenting cholera patients). Twenty of these patients had pneumonia, and two patients had pneumonia plus persistent Salmonella species serogroup B bacteremia/ paratyphoid fever. Overwhelming sepsis was the cause of death in the remaining five patients. The source of sepsis in one patient was felt to be intravenous catheter-infusate related. The source of sepsis in the remaining four patients was unclear. Three of the four individuals with sepsis of unclear origin were ⱕ 12 months of age. Blood was cultured at the time of admission in 29 of the patients dying in-hospital; seven (24.1%) had organisms isolated (Table 3). Bacteremia was eventually documented in 11 of the 26 patients (42.3%) who died of concomitant infections whose blood was cultured during their hospital stay. All 27 of the patients who died of concomitant infection received antibiotics. The initial antibiotic used in 18 of the patients was intravenous ampicillin with or without intravenous gentamicin. In 17 of these individuals, antibiotics were eventually changed to intravenous ceftriaxone with or without gentamicin. Twentyfive of the individuals who died of concomitant infection were treated with intravenous ceftriaxone before death, and 25 individuals were treated with intravenous gentamicin before death. Twelve individuals also received oral erythromycin, and six received intravenous metronidazole before death. DISCUSSION

The main finding of our study is that concomitant infection accounted for the majority of deaths in hospitalized patients with cholera cared for in this endemic area. Concomitant infection was the cause of four of every five deaths in cholera patients in this study, and accounted for almost all deaths in children twelve months of age or younger. Although concomitant infection has previously been recognized as contributing to the morbidity and mortality associ-

16

RYAN AND OTHERS

TABLE 3 Demographic and clinical characteristics of admitted cholera patients by outcome, International Centre for Diarrhoeal Disease Research in Bangladesh (ICDDR, B), 1996 Discharged improved (n ⫽ 377)

Died in-hospital (n ⫽ 33)

9 (4.5, 24) 211 (56.0) 72 (48, 168) (375)

8 (4.5, 29) 17 (51.5) 48 (24, 120)

0.879 0.756 0.223

0.8 (0.4–1.8)

361/374 (96.5) 10/376 (2.7) 345/375 (92.0) 109/372 (29.3) 132 (35.0) 7 (2.1) 44/319 (13.8) 333/371 (89.8) 5 (1.5) 116/376 (30.9)

30/32 (93.8) 0/32 29/33 (87.9) 10 (30.3) 13 (39.4) 1 (3.6) 5/30 (16.7) 30 (90.9) 2 (7.1) 18 (54.5)

0.334 1.000 0.341 0.938 0.753 0.471 0.591 1.000 0.092 0.010

0.5 0 0.6 1.1 1.2 1.8 1.3 1.1 5.2 2.7

66.1 ⫾ 14.3 (357/358) 8 (2.3) 22/373 (5.9) 7/334 (2.1) 124 ⫾ 18 (375) 121 (32.1) 41 ⫾ 11 (375) 28/244 (11.5) 23 (6.1) 7/373 (1.9) 106 (28.1) 65 (17.2) 21/356 (5.9) 3 (0.8) 4/375 (1.1)

59.2 ⫾ 12.3 (27/30) 1 (3.6) 5/32 (15.6) 8/33 (24.2) 124 ⫾ 31 (31) 13/32 (40.6) 44 ⫾ 10 (31) 9/20 (45) 7 (21.2) 6/32 (18.8) 15 (45.5) 4 (12.1) 7/32 (21.9) 6 (18.2) 1 (3.0)

133 ⫾ 6.7 (343) 3.4 ⫾ 1.3 (343) 16.8 ⫾ 7.1 (342) 12/213 (5.6) 34.3 ⫾ 6.6 (369) 15.1 ⫾ 6.7 (370)

134 ⫾ 8.5 (30) 3.4 ⫾ 1.6 (30) 13.2 ⫾ 6.9 (30) 6/29 (20.7) 34.4 ⫾ 7.3 (30) 16.9 ⫾ 7.2 (30)

Vibrio cholerae isolated in stool O1 Ogawa El Tor O1 Inaba El Tor O139 Salmonella/Shigella spp. in cholera stool Bacteremia**

294 (78.0) 62 (16.4) 21 (5.6) 38 (10.1) 10/272 (3.8)

24 (72.7) 5 (15.2) 4 (12.1) 3 (9.1) 7/29 (24.1)

0.634 0.958 0.131 1.000 ⬍0.001

0.8 0.9 2.3 0.9 8.3

Chest radiograph†† No infiltrate Perihilar infiltrate Peripheral ⫾ perihilar infiltrate

64/211 (30.3) 80/211 (37.9) 67/211 (31.8)

0/17 4/17 (23.5) 13/17 (76.5)

0.004 0.357 ⬍0.001

0 (0–0.6) 0.5 (0.1–1.7) 7.0 (2.0–30.2)

Characteristic

Age (mo)* Male gender Hours of diarrhea at presentation* At presentation, history of Watery stool Blood in stool Vomiting Fever Cough Measles in prior 3 mo (ⱕ5 yr of age)† Previous ICDDR,B visit ⬍ 2 weeks Use of oral rehydration solution/therapy (ORS/ORT) Adopted child (ⱕ5 years of age)† Severe dehydration at presentation Physical examination findings at admission Weight-for-age (of NCHS 50th percentile)§ (ⱕ15 years of age) Xerophthalmia (ⱕ5 years of age)†㛳 Edema Pulseless Heart rate (bpm)# Febrile (temperature ⬎37.8⬚C) Respirations/minute# Labored/gasping respirations Chest indrawing Cyanosis Lung field crepitations Abdominal distention Absent/decreased bowel sounds Unconscious Seizures Laboratory findings at admission Sodium (mmol/liter)# Potassium (nmol/liter)# Bicarbonate (mmol/liter)# Hypoglycemia (glucose ⱕ2 mmol/liter) Hematocrit (%)# Blood leukocytes (⫻ 1000 cells)/mm3#

P

0.009 0.512 0.051 ⬍0.001 0.983 0.429 0.096 ⬍0.001 0.006 ⬍0.001 0.058 0.609 0.005 ⬍0.001 0.345 0.528 0.963 0.010 0.012 0.929 0.195

OR (95% CI)

(0.1–5.2) (0–5.4) (0.2–2.6) (0.5–2.4) (0.6–2.6) (0.04–14.6) (0.4–3.6) (0.3–6.1) (0.5–33.2) (1.2–5.9)

1.5 (0.3–12.2) 3.0 (0.8–8.9) 15 (4.5–50.8) 1.5 (0.7–3.2) 6.3 4.1 12.1 2.1 0.7 4.5 27.7 2.9

(2.2–18.3) (1.5–11.4) (3.1–44.8) (1.0–4.6) (0.2–2.0) (1.6–12.5) (5.4–176.7) (0.06–30.3)

4.4 (1.2–14.0)

(0.3–1.8) (0.3–2.5) (0.6–7.6) (0.2–3.1) (2.6–27.0)

Values are no./total no. with data (%), unless otherwise noted. If data were available for all patients, no denominator is listed. * Values are median (25th, 75th centile) (no.). If data were available for all patients no (no.) is listed. † Values are no. (% age cohort). For the age cohort of ⱕ5 years of age, there were 341 cholera patients who were discharged improved and 28 cholera patients who died in-hospital. Data were available for all individuals in the age cohort. ‡ Values are no./total no. ⱕ1 year of age with data (%). For the age cohort of ⱕ1 year of age, there were 220 cholera patients who were discharged improved and 22 cholera patients who died in-hospital. No denominator is shown if data were available for all individuals in the age cohort. § Weight-for-age as a percentage of the median weight for the age group based on National Center for Health Statistics (NCHS) standards24. Values are mean ⫾ SD. (no./total no. ⱕ15 years of age). Based on rehydrated weight (defined as weight 72 hours after presentation, or the last recorded weight if patient was rehydrated, clinically stabilized, and discharged within 72 hours of presenting). 㛳 Conjunctival or corneal xerosis, Bitot’s spots, corneal erosion, or night blindness. # Values are mean ⫾ SD. (no.) ** For individuals discharged improved, the causative organisms were Acinetobacter spp. (4), Enterobacter spp. (2), Proteus sp. (1), Pseudomonas sp. (1), pneumococcus (1), and V. cholerae (1). For individuals who died in-hospital, the causative agents were pneumococcus (2), Salmonella serogroup B spp. (2), Klebsiella sp. (1), Acinetobacter sp. (1) and Enterobacter sp. (1). Individuals whose blood cultures grew only non-Staphylococcus aureus staphylococcal species were not considered to have bacteremia. †† Regarding radiographic severity of pneumonia, a perihilar infiltrate was defined as a pulmonary parenchymal infiltrate involving the perihilar one-third of at least one of the lung fields on chest radiography. A peripheral infiltrate was defined as a pulmonary parenchymal infiltrate involving the peripheral two-thirds of at least one of the lung fields on chest radiography. OR ⫽ odds ratio; CI ⫽ confidence interval.

17


TABLE 4 Admission characteristics associated with death of cholera patients in a multiple logistic regression analysis, International Centre for Diarrhoeal Disease Research in Bangladesh (ICDDR, B), 1996* Characteristic

OR

95% CI

Bacteremia† Peripheral ⫾ perihilar infiltrate on chest radiograph‡§ Serum bicarbonate (mmol/liter)

10.49 3.08 0.893㛳

2.91, 37.87 1.22, 7.71 0.825, 0.963

* Patients were included in the final multiple logistic regression analysis if data were available for all variables significant in initial models. Data on all characteristics were available for 320 of 410 (78%) admitted cholera patients who were discharged improved or who died in-hospital: 296 of 377 (78.5%) patients who were discharged improved; 24 of 33 (72.7%) patients who died in-hospital. † Individuals whose blood was never cultured were considered to have not been bacteremic. ‡ Individuals whose lung fields were never imaged were considered to not have a radiographic infiltrate. § A perihilar infiltrate was defined as a pulmonary parenchymal infiltrate involving the perihilar one-third of at least one of the lung fields on chest radiography. A peripheral infiltrate was defined as a pulmonary parenchymal infiltrate involving the peripheral two-thirds of at least one of the lung fields on chest radiography. 㛳 The odds ratio (OR) of less than 1 reflects the association of lower serum bicarbonate values (acidosis) with death. OR ⫽ odds ratio; CI ⫽ confidence interval.

ated with cholera,14,16,26–29 its role as a leading cause of death has not previously been reported. This finding is in large part due to the success of rehydration therapy programs and the virtual elimination of hypovolemia as a cause of death among hospitalized patients with cholera in this setting.30 The number of individuals with cholera who die before reaching medical attention in this modern urban area endemic for cholera, however, is not presently known.31 Two concomitant infections predominated in cholera patients in this study: respiratory infections and enteric coinfections. Mortality was primarily associated with the former. The pathogens responsible for the respiratory infections remain largely undefined at present,28,29,32 as is the possible contribution of cholera-induced vomiting and aspiration to the pulmonary pathology. In this study, one of every five surveillance system patients infected with V. cholerae O1 or O139 had a second enteric pathogen identified in stool. Asymptomatic, transient, intestinal passage of V. cholerae can occur in individuals immune to V. cholerae infection, and amelioration of the severity of the clinical features of cholera can be seen in individuals with partially protective immunity.8,33–39 Although it was not possible to discern the relative contributions of identified organisms to the clinical manifestations in any given patient, and although screening of stool for heatlabile enterotoxin-secreting Escherichia coli that can cause watery diarrhea was not routinely performed at the Centre, all patients reported in this study presented with diarrhea,

almost all had watery diarrhea, and all had V. cholerae O1 or O139 isolated from their stool, suggesting that V. cholerae was playing a pathogenic role in most patients. Death from cholera rarely resulted from hypovolemic shock in patients cared for in this endemic setting capable of appropriately managing fluid therapy. This is in sharp contrast to the situation in areas unable to implement appropriate rehydration therapy.5,30 Community education programs stressing the importance of rehydration therapy; the wide-spread availability and early home use of oral rehydration therapy; possibly the presence of pre-existing partial immunity protective against severe cholera; and, perhaps most importantly, the liberal and effective use of intravenous fluid therapy at the Centre probably accounted for this low mortality rate due to hypovolemia.5,8,34–38,40,41 With aggressive rehydration, hypovolemia-associated morbidities, including renal dysfunction and cardiac and neurologic ischemic events, were rare. Fluid replacement therapy was well tolerated by most patients and was rarely associated with fluid overload, intravenous access related nosocomial infections, possibly electrolyte imbalances, and tetany, the latter in individuals receiving intravenous solutions containing bicarbonate and possibly related to sudden alterations in blood pH and transient induction of hypocalcemia.42,43 When the causative agents of cholera were defined to the serogroup, biotype, and serotype level, we also found a pattern in which a sudden increase in cholera cases caused by a specific V. cholerae organism was followed within one to

TABLE 5 Causes of in-hospital death in patients admitted with cholera, International Centre for Diarrhoeal Disease Research in Bangladesh (ICDDR, B), 1996 ⱕ1 year of age (n ⫽ 22)* Diagnosis

Hypovolemic shock Concomitant infection Pneumonia Pneumonia and persistent Salmonella spp. bacteremia Sepsis Intravenous catheter/infusate related Unclear source Total concomitant infection

⬎1 year of age (n ⫽ 11)*

All patients (n ⫽ 33)*

Number with diagnosis (% of age cohort)

1 (4.5)

5 (45.5)

6 (18.2)

16 (72.7) 2 (9.1)†

4 (36.4) 0

20 (60.6) 2 (6.1)

0 3 (13.6)

1 (9.1)‡ 1 (9.1)§

1 (3.0) 4 (12.1)

21 (95.5)

6 (54.5)

27 (81.8)

* Number of patients dying in-hospital. † Patients with Salmonella spp. serogroup B isolated from more than one culture of blood sampled ⬎72 hours apart. ‡ Patient with Klebsiella sp. bacteremia. § Three year old patient with Enterobacter sp. bacteremia without clear source.

18

RYAN AND OTHERS

FIGURE 2.

Number of deaths in relation to duration of hospital stay, by cause, 1996.

two years by a decrease in frequency of that organism and a sudden increase in cases caused by a new V. cholerae organism. Such an occurrence was particularly evident when considering V. cholerae O139 infections. Vibrio cholerae O139 first emerged as a cause of cholera in 1992 in South Asia; immunity to V. cholerae O1 and O139 do not cross protect.44,45 As confirmed in this study, endemic cholera is primarily a pediatric disease.8,10,11 Early induction and, possibly, maintenance of partial or complete immunity by repetitive exposure to V. cholerae organisms usually protects older individuals who have resided long term in a cholera endemic area from developing clinically significant disease.37,46,47 In 1993, the population of Dhaka lacked immunity against V. cholerae O139, explaining, in part, the marked increase in cholera cases observed that year.23,45 Since 1993, the incidence of cholera caused by V. cholerae O139 has declined, possibly related to induction of immune responses in affected individuals, and/or to poorly understood environmental or bacterial factors.48–50 Reasons for the disappearance of classical V. cholerae O1 and the sudden increases and decreases in cholera cases caused by specific El Tor Ogawa and Inaba V. cholerae O1 organisms are less well understood, but may relate to varying degrees of immunologic cross-protection, environmental ecology of organisms, or clonal selection.8,48,50–55 In conclusion, causative agents of cholera vary over time in an endemic setting, perhaps related to immunologic crossprotection, clonal selection, and/or environmental ecology of organisms. Additionally, death in hospitalized patients with cholera acquired in a modern endemic setting capable of appropriately managing fluid replacement therapy is most

frequently due to concomitant infection, especially pneumonia. Further evaluation of such respiratory infections is warranted. Acknowledgments: We are grateful to Mohammad Mustafa, M. A. Malek, Ziaur Rahman, Md. Shafique Islam, Mohammad Ramzan Ali, and Humayun Kabir for assistance with data collection. Financial support: This work was supported in part by a WarrenWhitman-Richardson Traveling Fellowship, Harvard Medical School (Edward T. Ryan), an Aaron Diamond Fellowship in Medicine and Human Rights, Center for the Study of Society and Medicine, Columbia University College of Physicians & Surgeons (Edward T. Ryan), and by grants from the National Institute of Allergy and Infectious Diseases, AI/K0801332 (Edward T. Ryan), AI-40725 (Stephen B. Calderwood), and AI/HDO1671-01 (Michael L. Bennish), and an International Collaborations in Infectious Disease Research Award, HD39165 (Edward T. Ryan and Stephen B. Calderwood) from the National Institute of Child Health and Human Development. The ICDDR,B is supported by many donors including the United Nations Children’s Fund (UNICEF) and the Governments of Australia, Bangladesh, Belgium, Canada, Saudi Arabia, Sweden, Switzerland, the United Kingdom, and the United States. Authors’ addresses: Edward T. Ryan and Stephen B. Calderwood, Tropical and Geographic Medicine Center, Division of Infectious Diseases, Massachusetts General Hospital, Jackson 504, 55 Fruit Street, Boston, MA 02114. Ujjwal Dhar, Newton-Wellesley Hospital, Newton-Wellesley Hospital, 2014 Washington Street, Newton, MA 02462. Wasif A. Khan, Mohammed Abdus Salam, Abu S.G. Faruque, and George J. Fuchs, Clinical Sciences Division, International Centre for Diarrhoeal Disease Research (ICDDR,B: Centre for Health and Population Research), Mohakhali 1212, Dhaka, Bangladesh. Michael L. Bennish, Division of Geographic Medicine and Infectious Diseases, Tupper Research Institute, New England Medical Center, 750 Washington Street, Box 041, Boston, MA 02111. Reprint requests: Edward T. Ryan, Tropical and Geographic Medi-


cine Center, Division of Infectious Diseases, Jackson 504, Massachusetts General Hospital, 55 Fruit Street, Boston, MA 02114. 22. REFERENCES

1. Institute of Medicine, 1986. New Vaccine Development: Establishing Priorities. Volume II. Diseases of Importance in Developing Countries. Washington, DC: National Academy Press. 2. Anonymous, 1998. Cholera in 1997. Wkly Epidemiol Rec 73: 201–208. 3. Swerdlow DL, Malenga G, Begkoyian G, Nyangulu D, Toole M, Waldman RJ, Puhr DN, Tauxe RV, 1997. Epidemic cholera among refugees in Malawi, Africa: treatment and transmission. Epidemiol Infect 118: 207–214. 4. Anonymous, 1995. Public health impact of Rwandan refugee crisis: what happened in Goma, Zaire, in July, 1994? Goma Epidemiology Group. Lancet 345: 339–344. 5. Siddique AK, Salam A, Islam MS, Akram K, Majumdar RN, Zaman K, Fronczak N, Laston S, 1995. Why treatment centres failed to prevent cholera deaths among Rwandan refugees in Goma, Zaire. Lancet 345: 359–361. 6. Feachem RG, 1981. Environmental aspects of cholera epidemiology. I. A review of selected reports of endemic and epidemic situations during 1961–1980. Trop Dis Bull 78: 675– 698. 7. Singh J, Sachdeva V, Bhatia R, Bora D, Jain DC, Sokhey J, 1998. Endemic cholera in Delhi, 1995: analysis of data from a sentinel centre. J Diarrhoeal Dis Res 16: 66–73. 8. Glass RI, Becker S, Huq MI, Stoll BJ, Khan MU, Merson MH, Lee JV, Black RE, 1982. Endemic cholera in rural Bangladesh, 1966–1980. Am J Epidemiol 116: 959–970. 9. Weil O, Berche P, 1992. The cholera epidemic in Ecuador: towards an endemic in Latin America. Rev Epidemiol Sante Publique 40: 145–155. 10. Martin AR, Mosley WH, Sau BB, Ahmed S, Huq I, 1969. Epidemiologic analysis of endemic cholera in urban East Pakistan, 1964–1966. Am J Epidemiol 89: 572–582. 11. McCormack WM, Mosley WH, Fahimuddin M, Benenson AS, 1969. Endemic cholera in rural East Pakistan. Am J Epidemiol 89: 393–404. 12. Clemens J, Albert MJ, Rao M, Qadri F, Huda S, Kay B, van Loon FP, Sack D, Pradhan BA, Sack RB, 1995. Impact of infection by Helicobacter pylori on the risk and severity of endemic cholera. J Infect Dis 171: 1653–1656. 13. Glass RI, Svennerholm AM, Stoll BJ, Khan MR, Hossain KM, Huq MI, Holmgren J, 1983. Protection against cholera in breast-fed children by antibodies in breast milk. N Engl J Med 308: 1389–1392. 14. Nalin DR, 1972. Mortality from cholera and other diarrhoeal diseases at a cholera hospital. Trop Geogr Med 24: 101–106. 15. Islam SS, Shahid NS, 1986. Morbidity and mortality in a diarrhoeal diseases hospital in Bangladesh. Trans R Soc Trop Med Hyg 80: 748–752. 16. Mitra AK, Khan MR, Alam AN, 1991. Complications and outcome of disease in patients admitted to the intensive care unit of a diarrhoeal diseases hospital in Bangladesh. Trans R Soc Trop Med Hyg 85: 685–687. 17. Bangladesh Bureau of Statistics, 1999. Statistical Year Book of Bangladesh. Dhaka, Bangladesh: Government of Bangladesh. 18. Stoll BJ, Glass RI, Huq MI, Khan MU, Holt JE, Banu H, 1982. Surveillance of patients attending a diarrhoeal disease hospital in Bangladesh. Br Med J (Clin Res Ed) 285: 1185–1188. 19. Monsur KA, 1961. A highly selective gelatin-taurocholate-tellurite medium for the isolation of Vibrio cholerae. Trans R Soc Trop Med Hyg 55: 440–442. 20. Pal SC, 1992. Laboratory diagnosis. Barua D, Greenough III WB, eds. Cholera. New York: Plenum Medical Book Company, 229–251. 21. Kay B, Bopp CA, Wells JG, 1994. Isolation and identification of Vibrio cholerae O1 from fecal specimens. Wachsmuth IK, Blake PA, Olsvik O, eds. Vibrio cholerae and Cholera: Mo-

23.

24.

25. 26. 27. 28.

29.

30.

31.

32.

33.

34.

35.

36.

37.

38.

39.

40.

41.

42.

43.

19

lecular to Global Perspectives. Washington, DC: American Society for Microbiology, 3–25. Dhar U, Bennish ML, Khan WA, Seas C, Huq KE, Albert MJ, Salam MA, 1996. Clinical features, antimicrobial susceptibility and toxin production in Vibrio cholerae O139 infection: comparison with V. cholerae O1 infection. Trans R Soc Trop Med Hyg 90: 402–405. Mahalanabis D, Faruque AS, Albert MJ, Salam MA, Hoque SS, 1994. An epidemic of cholera due to Vibrio cholerae O139 in Dhaka, Bangladesh: clinical and epidemiological features. Epidemiol Infect 112: 463–471. World Health Organization, 1983. Measuring Change in Nutritional Status: Guidelines for Assessing the Nutritional Impact of Supplementary Feeding Programmes for Vulnerable Groups. Geneva: World Health Organization. Sakazaki R,Tamura K, 1971. Somatic antigen variation in Vibrio cholerae. Jpn J Med Sci Biol 24: 93–100. Struelens MJ, Bennish ML, Mondal G, Wojtyniak BJ, 1991. Bacteremia during diarrhea: incidence, etiology, risk factors, and outcome. Am J Epidemiol 133: 451–459. Samadi AR, Chowdhury AI, Huq MI, Shahid NS, 1985. Risk factors for death in complicated diarrhoea of children. Br Med J (Clin Res Ed) 290: 1615–1617. Islam SS, Khan MU, 1986. Risk factors for diarrhoeal deaths: a case-control study at a diarrhoeal disease hospital in Bangladesh. Int J Epidemiol 15: 116–121. Butler T, Islam M, Azad AK, Islam MR, Speelman P, 1987. Causes of death in diarrhoeal diseases after rehydration therapy: an autopsy study of 140 patients in Bangladesh. Bull World Health Organ 65: 317–323. Mahalanabis D, Molla AM, Sack DA, 1992. Clinical management of cholera. Barua D, Greenough III WB, eds. Cholera. New York: Plenum Medical Book Company, 253–283. Siddique AK, Akram K, Islam Q, 1988. Why cholera still takes lives in rural Bangladesh. Study of an epidemic. Trop Doct 18: 40–42. Tomashefski JFJ, Butler T, Islam M, 1989. Histopathology and etiology of childhood pneumonia: an autopsy study of 93 patients in Bangladesh. Pathology 21: 71–78. Levine MM, Black RE, Clements ML, Cisneros L, Nalin DR, Young CR, 1981. Duration of infection-derived immunity to cholera. J Infect Dis 143: 818–820. Cash RA, Music SI, Libonati JP, Craig JP, Pierce NF, Hornick RB, 1974. Response of man to infection with Vibrio cholerae. II. Protection from illness afforded by previous disease and vaccine. J Infect Dis 130: 325–333. McCormack WM, Islam MS, Fahimuddin M, Mosley WH, 1969. A community study of inapparent cholera infections. Am J Epidemiol 89: 658–664. McCormack WM, Chakraborty J, Rahman AS, Mosley WH, 1969. Vibriocidal antibody in clinical cholera. J Infect Dis 120: 192–201. Khan MU, Barua DK, Begum T, Shahidullah M, 1987. Vibriocidal titre in cholera cases and contacts: its value in assessing endemicity of or susceptibility to cholera. Trop Geogr Med 39: 271–275. Glass RI, Svennerholm AM, Khan MR, Huda S, Huq MI, Holmgren J, 1985. Seroepidemiological studies of El Tor cholera in Bangladesh: association of serum antibody levels with protection. J Infect Dis 151: 236–242. Kaper JB, Morris JGJ, Levine MM, 1995. Cholera [published erratum appears in Clin Microbiol Rev 1995 8(2): 316]. Clin Microbiol Rev 8: 48–86. Hirschhorn N, Kinzie JL, Sachar DB, Northrup RS, Taylor JO, Ahmad SZ, Phillips RA, 1968. Decrease in net stool output in cholera during intestinal perfusion with glucose-containing solutions. N Engl J Med 279: 176–181. Bhan MK, Mahalanabis D, Fontaine O, Pierce NF, 1994. Clinical trials of improved oral rehydration salt formulations: a review. Bull World Health Organ 72: 945–955. Lindenbaum J, Akbar R, Gordon RSJ, Greenough WB, Hirschorn N, Islam MR, 1966. Cholera in children. Lancet 1: 1066–1068. Wang F, Butler T, Rabbani GH, Jones PK, 1986. The acidosis

20

44.

45.

46.

47.

48.

49.

RYAN AND OTHERS

of cholera. Contributions of hyperproteinemia, lactic acidemia, and hyperphosphatemia to an increased serum anion gap. N Engl J Med 315: 1591–1595. Sarkar BL, De SP, Sircar BK, Garg S, Nair GB, Deb BC, 1993. Polymyxin B sensitive strains of Vibrio cholerae non-O1 from recent epidemic in India (letter). Lancet 341: 1090. Anonymous, 1993. Large epidemic of cholera-like disease in Bangladesh caused by Vibrio cholerae O139 synonym Bengal. Cholera Working Group, International Centre for Diarrhoeal Diseases Research, Bangladesh. Lancet 342: 387–390. Mosley WH, Ahmad S, Benenson AS, Ahmed A, 1968. The relationship of vibriocidal antibody titre to susceptibility to cholera in family contacts of cholera patients. Bull World Health Organ 38: 777–785. Mosley WH, Benenson AS, Barui R, 1968. A serological survey for cholera antibodies in rural east Pakistan. 1. The distribution of antibody in the control population of a cholera-vaccine field-trial area and the relation of antibody titre to the pattern of endemic cholera. Bull World Health Organ 38: 327–334. Faruque AS, Fuchs GJ, Albert MJ, 1996. Changing epidemiology of cholera due to Vibrio cholerae O1 and O139 Bengal in Dhaka, Bangladesh. Epidemiol Infect 116: 275–278. Morris JGJ, Losonsky GE, Johnson JA, Tacket CO, Nataro JP, Panigrahi P, Levin MM, 1995. Clinical and immunologic

50.

51.

52. 53. 54.

55.

characteristics of Vibrio cholerae O139 Bengal infection in North American volunteers. J Infect Dis 171: 903–908. Faruque SM, Ahmed KM, Abdul alim AR, Qadri F, Siddique AK, Albert MJ, 1997. Emergence of a new clone of toxigenic Vibrio cholerae O1 biotype El Tor displacing V. cholerae O139 Bengal in Bangladesh. J Clin Micro 35: 624–630. Sharma C, Nair GB, Mukhopadhyay AK, Bhattacharya SK, Ghosh RK, Ghosh A, 1997. Molecular characterization of Vibrio cholerae O1 biotype El Tor strains isolated between 1992 and 1995 in Calcutta, India: evidence for the emergence of a new clone of the El Tor biotype. J Infect Dis 175: 1134– 1141. Samadi AR, Chowdhury MK, Huq MI, Khan MU, 1983. Seasonality of classical and El Tor cholera in Dhaka, Bangladesh: 17-year trends. Trans R Soc Trop Med Hyg 77: 853–856. Huq MI, Glass RI, Stoll BJ, 1980. Epidemiology of cholera (letter). N Engl J Med 303: 643–644. Dalsgaard A, Skov MN, Serichantalergs O, Echeverria P, Meza R, Taylor DN, 1997. Molecular evolution of Vibrio cholerae O1 strains isolated in Lima, Peru, from 1991 to 1995. J Clin Microbiol 35: 1151–1156. Faruque SM, Abdul AA, Rahman MM, Siddique AK, Sack RB, Albert MJ, 1993. Clonal relationships among classical Vibrio cholerae O1 strains isolated between 1961 and 1992 in Bangladesh. J Clin Microbiol 31: 2513–2516.