Hypoglycemia associated with bacteremicpneumococcal infections

I-Shiow Jan a, Tzu-Hsiu Tsai b, Jong-Min Chen a, Jih-Shuin Jerng b,Hsin-Fang Hsu b, Pei-Lun Hung a, Po-Ren Hsueh a,b, Li-Na Lee a,b,*

aDepartment of Laboratory Medicine, National Taiwan University College of Medicine and Hospital,7 Chung Shan South Road, Taipei 100, TaiwanbDepartment of Internal Medicine, National Taiwan University College of Medicine and Hospital, Taipei, Taiwan

Received 8 February 2008; received in revised form 22 August 2008; accepted 28 August 2008Corresponding Editor: Timothy Barkham, Tan Tock Seng, Singapore

International Journal of Infectious Diseases (2009) 13, 570—576

http://intl.elsevierhealth.com/journals/ijid

KEYWORDSHypoglycemia;Pneumococcalbacteremia;Serotype

Summary

Objectives: To evaluate the prevalence and associated presentations of hypoglycemia in bac-teremic pneumococcal infections, and serotypes of the isolates.Methods: This was a retrospective study of 70 episodes of pneumococcal bacteremia thatoccurred in 2004 and 2005.Results: We found hypoglycemia (plasma glucose <3.05 mmol/l)) in six (8.6%) episodes. Thepatients were three children (mean age 3 years 1month; range 1 year 5months—4 years 5months)and three adults (mean age 73.3 years; range 63—84 years). One child with asplenia and cyanoticheart disease had primary pneumococcal bacteremia. Of the other two children, one hadmeningitis and the other pneumonia. All the adults had cancer with previous chemotherapyand multilobar pneumonia, which progressed rapidly to respiratory failure. All patients devel-oped their first hypoglycemic episode within two hours after presentation. The average plasmaglucose during hypoglycemia was 1.78 � 0.78 mmol/l (range 0.33—2.94 mmol/l). One child andall of the adults died. Serotypes of isolates were those usually associated with severe pneumo-coccal infection: 6B and 19F in the children; 3, 14, and 23F in the adults. Only the asplenic childhad received pneumococcal vaccine.Conclusions: Hypoglycemia occurred in 8.6% of bacteremic pneumococcal infections and wasassociated with high mortality and serotypes that cause severe invasive disease. All patientssuspected of having septicemia should have their glucose checked to avoid missing hypoglycemialeading to a worsening of their already poor condition.# 2008 International Society for Infectious Diseases. Published by Elsevier Ltd. All rights reserved.

* Corresponding author. Tel.: +886 2 23123456x5359;fax: +886 2 23224263.

E-mail address: linalee@ntu.edu.tw (L.-N. Lee).

1201-9712/$36.00 # 2008 International Society for Infectious Diseases.doi:10.1016/j.ijid.2008.08.026

Introduction

Hypoglycemia has rarely been associated with severe infec-tion,1 except in cases ofmalaria2—4 and shigellosis.5,6 In 1934,

Published by Elsevier Ltd. All rights reserved.

Hypoglycemia in pneumococcal bacteremia 571

Magnusson7 reported the first three cases of bacteremia-associated hypoglycemia in children with fulminant menin-gococcemia and adrenal hemorrhage. In a few subsequentcase reports, hypoglycemia was found to occur in severesepsis caused by the encapsulated bacteria Streptococcuspneumoniae, Haemophilus influenzae, and Neisseria menin-gitidis,1,8—11 and to be an index of grave outcome. Thereported cases were young children,1,7 or adults who weresplenectomized,1,8,9,11 alcoholic,1 or on chronic dialysis.1,9,11

The prevalence of hypoglycemia in pneumococcal septice-mia, however, has never been studied. We conducted aretrospective study to evaluate the prevalence of hypogly-cemia in patients with bacteremic pneumococcal infections,and to present the clinical course of the disease and ser-otypes of the pneumococcal isolates.

Materials and methods

Patients

This study was conducted at a 2000-bed university hospital innorthern Taiwan. We retrospectively searched the computerdatabase of our microbiology laboratory and retrieved allpatients with at least one blood culture positive for S.pneumoniae from January 2004 through December 2005.We then reviewed their charts and identified all patientswith hypoglycemia (plasma glucose level of<3.05 mmol/l).12

For the purposes of comparison, we also obtained a list ofpatients with bacteremia due to Staphylococcus aureus anddue to Escherichia coli for the year 2004, and studied theprevalence of hypoglycemia in these two groups.

The infection that led to pneumococcal bacteremia wasdetermined to be: (1) pneumonia, if there were clinical pre-sentationsofpneumonia and lobarpneumonic infiltrates at theinitial chest radiograph, with or without concomitant isolationof S. pneumoniae from blood and sputum; (2) meningitis, ifthere were clinical manifestations of meningitis and isolationof S. pneumoniae from the cerebrospinal fluid; (3) sponta-neous peritonitis, if there were clinical presentations of peri-tonitis, isolation of S. pneumoniae from ascites, and noevidenceof rupturedhollowviscus. Pneumococcal bacteremiawas definedasprimary, if therewerenoclinicalmanifestationsof local pneumococcal infection and S. pneumoniae was notisolated from sites other than blood. Septic shock was definedas previously stated.13 Acute respiratory distress syndrome(ARDS) was defined by bilateral pulmonary infiltrates on chest

Table 1 Pneumococcal infections causing bacteremia 2004—200

Pneumococcal infectioncausing bacteremia

Number of episodes

Adults

Community-acquired

Pneumonia 31Primary bacteremia 10Meningitis 4Spontaneous bacterial peritonitis 1

Total 46

a All were community-acquired.

radiograph, a ratio of arterial oxygen tension (PaO2) to thefraction of inspired oxygen (FiO2) of 200 or less, and theabsence of evidence of left atrial hypertension.14

Drug susceptibility

All S. pneumoniae isolates were tested for drug susceptibilityby the standardized disk diffusion methods.15 Minimum inhi-bitory concentration (MIC) tests were performed by themicrodilution method for the isolates from patients withhypoglycemia.15

Antibiotic treatment was considered concordant if at leastone of the antibiotics administered during the first 24 hoursshowed full in vitro sensitivity against the isolated strain.Therapies without this criterion were defined as discordant.

Serotyping and serogrouping of isolates

Serotyping was performed by slide agglutination with cap-sular typing sera and factor sera (Statens Serum Institut,Copenhagen, Denmark).

Data analysis

The Student’s t-test was used to analyze inter-group differ-ences for the age between patients with and those withouthypoglycemia. The Chi-square test was used to comparedifferences in the seasonal occurrence and mortalitybetween patients with and those without hypoglycemia.

Results

Case identification

During the years 2004 and 2005, there were 70 episodes (69patients) of bacteremia due to S. pneumoniae (Table 1). Fiftyepisodes (71%) occurred in 49 adults (38 male, 78%), theiraverage age being 59.8 � 20.0 years (range18—92 years). Fourwere nosocomial. Twenty episodes occurred in children (14male, 70%), with a mean age of 3 years 10 months (range 3months—14 years; 70% <4 years). Pneumonia was the mostcommon disease leading to bacteremia: 31/50 adults (62%)and 12/20 children (60%). Half of the children, and all of theadults except for one who was dead on arrival with unknownprevious health condition, had an underlying disease (Table 2).

5.

Childrena Total

Nosocomial

0 12 433 4 171 4 90 0 1

4 20 70

Table 2 Major underlying diseases of patients with bactere-mic pneumococcal infection.

No. (%) withknown underlyingdisease

Underlying disease Number(%)

Adults: 48 (98) Solid cancer 17 (34.7)Leukemia and myeloma 4 (8.2)Chronic obstructivepulmonary disease

7 (14.3)

Diabetes mellitus 5 (10.2)Alcoholic cirrhosis 4 (8.2)Cirrhosis, other causes 2 (4.1)Aortic aneurysm 3 (6.1)HIV infection 2 (4.1)Others 4 (8.2)

Children: 10 (50) Asthma 4 (20)Asplenism + CHD,a VSD,b

bronchiectasis, biliaryatresia post-liver-transplant,angiomatosis under cytoxan,atopic dermatitis

Each 1 (5)

a Congenital heart disease (right atrial isomerism, total anom-alous pulmonary return, endocardial cushion defect, and doubleoutlet right ventricle).b Ventricular septal defect.

Table

3Clinicalan

dlaboratory

data

ofsixpatients

withbac

teremic

pneumoco

ccal

infectionan

dhyp

oglycemia.

Case

No.

Age

/sex

Pneumoco

ccal

disease

Pre-existingdisease

Sourceof

isolate

Dateof

isolation

Body

temp.

(8C)

Shock/

DIC

Gluco

sein

initial

24h(m

mol/l)

Durationof

hyp

oglycemia

CRP

(mg/

l)WBC

(�10

9/l)

Outcome

(survival

time)

Cau

seof

death

11y5

m/M

Meningitis

-Bloodan

dCSF

2/26

/200

439

.6�/+

1.83

,2.39

,5.72

,7.44

2hours

248

10.44

Mentally

retarded

23y

6m/F

Pneumonia

-Blood

2/28

/200

539

.4�/n

ot

done

3.05

,2.61

,5.61

2hours

322

5.77

Reco

vered

34y

5m/F

Prim

ary

bacteremia

CHDan

dasplenism

Blood

3/24

/200

539

.7+/

+1.5,

1.06

,12

.67,

14.5

3hours

397.49

Died(28h)

Septicshock

484

y/M

Pneumonia

Prostateca

nce

runder

chemotherapy

Blood

12/1

7/20

0436

.1�/+

2.94

,9.39

,6.06

1.5hours

311

4.72

Died(52d)

ARDS

574

y/M

Pneumonia

Colonca

nce

runder

chemotherapy

Bloodan

dsputum

1/20

/200

536

.9+/

+3.72

,2.22

,6.72

4hours

173

3.55

Died(6

d)

ARDS

663y

/M

Pneumonia

Lungca

nce

runder

chemotherapy

Bloodan

dsputum

10/5

/200

536

.1�/n

ot

done

1.61

,16

.06,

1.33

,0.33

,4.56

,13

.33

20hours

255

26.30

Died(28h)

Septicshock

andARDS

DIC,disseminatedintrav

ascu

larco

agulation;CRP,C-reac

tive

protein

(norm

al:<10

mg/

l);WBC,whitebloodce

llco

unt;y,ye

ars;m,months;d,day

s;M,male;F,female;CSF,ce

rebrospinal

fluid;

CHD,cyan

oticco

nge

nital

heartdisease;ARDS,

acute

respiratory

distress

syndrome.

572 I.-S. Jan et al.

Prevalence and seasonality

Plasma glucose was analyzed in 55 episodes. For the rest, thepatients showed no signs of hypoglycemia. Hypoglycemia wasdocumented in six episodes (8.6%) (Table 3). The patientswere three children (one male; mean age 3 years 1 month)and three adults (all male; mean age 73.3 years). None haddiabetes, alcoholism, chronic renal failure, or history ofinsulinoma, hypothyroidism, or hypoadrenalism. Two chil-dren were previously healthy, while all three adults hadunderlying cancer. The previous plasma glucose levels ofall three adults were normal. All six episodes of pneumococ-cal bacteremia with hypoglycemia occurred in the coolerpart of the year (October through March), while in adultswithout hypoglycemia, 24 episodes (51%, p = 0.292), and inchildren without hypoglycemia, 10 episodes (59%, p = 0.455),occurred during cooler seasons.

For the comparison with patients with bacteremia due tootherpathogens,wefoundthat in theyear2004therewere601episodes of bacteremia due to S. aureus and 689 episodes ofbacteremia due to E. coli. In 227 (38%) of the 601 and 242 (35%)of the 689 episodes, plasma glucose levelswere checked.Afterexcluding those hypoglycemic episodes that were related toinsulin administration or liver cirrhosis, we found sepsis-asso-ciated hypoglycemia in three episodes (0.5%) of bacteremiadue to S. aureus and two episodes (0.3%) of bacteremia due toE. coli. These five hypoglycemic episodes occurred in January,March, April, July, and October (one in each).

Clinical presentation

Clinical presentation differed between adults and children(Table 2). All three children had high fever at presentation(average 39.4 8C). The two previously healthy children did

Figure 1 Chest radiograph of patient 6 at presentation (A),showing lobar consolidation of right lower, right middle, and partof the left lower lobe, and 24 hours later (B), showing rapidprogression of the pneumonic process.

Hypoglycemia in pneumococcal bacteremia 573

not have shock or respiratory failure, despite one experien-cing a severe meningoencephalitis and the other a pneumo-nia complicated with multiple lung abscesses. The only childwho died was congenitally asplenic. On arrival at the emer-gency room, she had shock, generalized petechiae, andsevere hypoxemia (oxygen saturation 37%, her usual satura-tion being 80% due to congenital heart disease with anintracardiac shunt). Among the six hypoglycemic patientsshe was the only one who had received a 23-valent pneumo-coccal vaccine (the conjugate vaccine was not availablelocally).

All three adults had pneumococcal pneumonia. On arriv-ing at the emergency service, none was febrile or hypother-mic, but all were in hypoxemia and impending respiratoryfailure (oxygen saturation 80%, 82%, and 74%, for cases 4, 5,and 6, respectively; Table 3). One was in shock. Two had livedin a nursing home, but all three had been ambulatory. Cases4, 5, and 6 were receiving anti-cancer chemotherapy 8, 16,and 8 days, respectively, before the onset of pneumonia. Theintermittently scheduled chemotherapy had been started 18,7, and 3 months, respectively, before bacteremia developed.The first chest radiograph of all three adult patients showedmultilobar pneumonia, which progressed rapidly to nearlycomplete bilateral opacification (Figure 1).

Hypoglycemia

Table 3 shows plasma glucose levels of the six patients duringthe initial 24 hours. In this period each patient had two to sixmeasurements of plasma glucose taken, and one to threedocumented episodes of hypoglycemia. A total of 10 hypo-glycemic episodes were documented in the six patients; nonepresented with cold clammy skin or sweating. The boy withmeningoencephalitis and the girl with asplenism were coma-tose at presentation. After glucose and antimicrobial ther-apy, the mentality of the boy improved slowly over severaldays; the girl never woke up. All three adults hadmild mentaldrowsiness at presentation that became clearer after glucoseinfusion. The average glucose level during hypoglycemia was1.78 � 0.78 mmol/l (range 0.33—2.94 mmol/l). All sixpatients documented their first hypoglycemic episode withintwo hours after arrival. For patients 1—5, hypoglycemia wasdetected only in the first four hours after arrival. Patient 6,however, had recurrent, severe hypoglycemia, despite con-stantly being treated with intravenous glucose. The lasthypoglycemic episode (0.33 mmol/l, or 6 mg/dl) was docu-mented 18 hours after arrival, 17 hours after antimicrobialsand glucose infusion were started. The two patients who hadglucose <1.67 mmol/l, or 30 mg/dl (patients 3 and 6) diedrapidly.

Other laboratory data

S. pneumoniae was isolated from sputum samples of twoadult patients (patients 5 and 6, Table 3). Both had a fulmi-nant course. White blood cell count ranged from 3.55 to26.3 � 109/l. Tests for disseminated intravascular coagula-tion (DIC) were performed in four, and all were positive. Theinitial level of C-reactive protein (CRP) was markedly ele-vated (173 to 322 mg/l) except in the child with cyanoticcongenital heart disease (39 mg/l). All three adults werehypoalbuminemic.

Table 4 Serotype, serogroup, drug susceptibility, and minimal inhibitory concentration of pneumococcal isolates.

Case No. Age Serogroup/serotype

Drug susceptibility by disk diffusion method MIC (mg/ml) Initial antibiotics

Pc Ery Clin Tet Cm Van Levo Pca Cro Van

1 1y5m 6B R R S R S S S 5 (R/R) 0.25 0.12 Van + cefotaxime2 3y6m 6B R R S R S S S 2 (R/S) 1 0.25 Cro3 4y5m 19F R R R R S S S 2 (R/S) 0.5 0.12 Van + ceftazidime4 84y 3 S R R R R S S 0.03 (S/S) 0.03 0.12 Flomoxefb

5 74y 14 R R R S S S S 0.25 (I/S) 0.06 0.25 Amoxicillin—clavulanate

6 63y 23F R R R R S S S 4 (R/R) 2 0.25 Cefepime

MIC, minimum inhibitory concentration; Pc, penicillin; Ery, erythromycin; Clin, clindamycin; Tet, tetracycline; Cm, chloramphenicol; Van,vancomycin; Levo, levofloxacin; Cro, ceftriaxone; R, resistant; S, susceptible; I, intermediate; y, years; m, months.a Based on the penicillin MIC interpretative criteria for Streptococcus pneumoniae (meningitis/non-meningitis) of the Clinical and

Laboratory Standards Institute (CLSI) 2008.15.b A cephamycin with a difluoromethylthioacetamido group at position 7.

574 I.-S. Jan et al.

Serotype, drug susceptibility, and antimicrobialtherapy

Table 4 shows the serotypes, drug susceptibilities, and MICs ofthe six pneumococcal isolates, and the initial antimicrobialtherapy the patients received. Although five of the six iso-lates were resistant to penicillin by disk diffusion method,MIC testing showed that only two isolates were resistant: onefrom the boy with meningitis with a MIC of 5 mg/ml, the otherfrom a pneumonia patient with a MIC of 4 mg/ml.15 Onlypatient 5 received a discordant initial antimicrobial. Otherswere treated with concordant initial antimicrobials includingthird- or fourth-generation cephalosporins, with or withoutvancomycin.

Course and fatality

Among the three children, only the one with asplenia, con-genital heart disease, and primary bacteremia died of ful-minant septicemia (mortality 33.3%). The boy withmeningitis recovered but had sequela of mental retardationand seizures. The mortality among 17 children with bactere-mic pneumococcal infection but no hypoglycemia was 5.9%(1/17, p = 0.677). All three adult patients progressed rapidlyto respiratory failure and required mechanical ventilation.They all died of multilobar pneumonia with respiratory fail-ure, with or without septic shock (Table 3). Compared withadult patients who had pneumonia but no hypoglycemia(n = 28), the average age was not different (73.7 � 10.5vs. 60.1 � 20.7, p = 0.277), but the mortality was higher(100% vs. 25%, p = 0.046).

Discussion

While hyperglycemia is often observed in the early phase ofsepsis,16 hypoglycemia due to sepsis has been described onlyrarely. Sepsis has been reported to be the cause of hypogly-cemia in 12% of hospitalized non-diabetic adults with hypo-glycemia, other than chronic renal failure (25%), alcoholintoxication (15%), and hepatic failure (12%).17 In rat anddog sepsis, hypoglycemia is caused by an increased glucoseutilization by macrophage-rich tissues (liver, spleen, lung,

ileum, skin),18,19 and decreased hepatic glucose produc-tion.19,20 The incidence of hypoglycemia in septicemia, how-ever, has never been reported. Our study shows thathypoglycemia occurred in 8.6% of bacteremic pneumococcalinfections. This is probably an underestimate, as not allpatients had plasma glucose tested. Compared with bacter-emia due to S. aureus or E. coli, the rate of hypoglycemia inpneumococcal septicemia was higher, probably reflecting themore invasive and fulminant nature of pneumococcal sepsis.That all six episodes of hypoglycemia associated with pneu-mococcal bacteremia occurred in the cooler part of the year(October to March) probably reflects the fact that invasivepneumococcal infections happen more often in coldweather.21 The observation that hypoglycemic episodesdue to S. aureus or E. coli do not show such seasonal variationprobably supports this speculation.

Our patients with hypoglycemia were young or old. Thethree hypoglycemic children had severe invasive disease,resulting in one death and one mental retardation. Infec-tion-associated hypoglycemia in children is well known inmalaria,2—4 and reported rarely in shigellosis,5,6 acute ful-minant hepatitis,22 pancreatitis/gastroenteritis due to rota-virus,23 and Reye’s syndrome.24 Septicemia-associatedhypoglycemia in children has been described in only fourcases,1,7 all due to Neisseria meningitidis. In our study itoccurred in three (15%) of the 20 pediatric patients, suggest-ing it is probably not very rare. A routine measurement ofblood glucose for any pediatric patient suspected of havingsepsis may be needed, to avoidmissing its presence, resultingin a worsening of an already severe disease.

The three adult hypoglycemic patients in our study alldied of pneumococcal pneumonia. In the few previous casereports, hypoglycemia occurred in adult patients with septi-cemia due to S. pneumoniae, H. influenzae, Streptococcuspyogenes, S. aureus, Clostridium perfringens, and E. coli,with a mortality of 67%.1,8,9,25—27 The reported patients weremostly alcoholic,1 splenectomized,8,9,11 on chronic dialy-sis,1,8,11 or cirrhotic.27 Our patients, on the other hand, werereceiving antineoplastic therapy for solid tumors beforeacquiring pneumococcal pneumonia. Patients who arereceiving chemotherapy for cancer are immunosuppressedand prone to develop severe infection, including pneumo-coccal bacteremia.28 In cancer patients, pneumococcal

Hypoglycemia in pneumococcal bacteremia 575

bacteremia is often accompanied by pneumonia, as in ourpatients, but can be caused by infected catheters.28 Thelatter could be the reason that in our study, as many as 13(26%) of 50 adult episodes were primary bacteremia, for 43%of our adult patients had cancer. Hypoglycemia has beenassociated with non-islet cell tumors, especially mesothe-lioma,29 and rarely pulmonary squamous cell carcinoma.30

Our patients had carcinoma of the colon, prostate, and lung,but their tumors were not likely the cause of hypoglycemia,for their previous blood glucose levels were all normal. Noneof our patients had alcoholism, cirrhosis, or chronic renalfailure, which could cause hypoglycemia due to decreasedhepatic or renal gluconeogenesis.17 Thus in our patients,bacteremic pneumococcal pneumonia was the most impor-tant factor leading to hypoglycemia. We found that plasmaglucose could decline to as low as 0.33 mmol/l even after17 hours of concordant antimicrobial and glucose therapy,suggesting the necessity of monitoring blood glucose levelseven after its apparent normalization.

The isolates of our patients belong to those serotypes thatcommonly cause invasive disease, and are covered in the 7-or 23-valent vaccines.31 The hypoglycemia was not asso-ciated with any specific serotype, but rather a manifestationof severe pneumococcal infection. Previous studies haveshown that in invasive pneumococcal disease, serotype 3,older age, leukopenia, hypothermia, nursing home resi-dence, pre-existing lung disease, and need for mechanicalventilation are predictors of mortality.32,33 Our study showedthat hypoglycemia was also a marker of poor prognosis; thishas often been missed, even in a large prospective study.33

Due to the severity of such infections, the presence ofhypoglycemia can be masked by the devastating condition.Delay in its detection and treatment may worsen the alreadypoor prognosis. Other laboratory indices of poor prognosisinclude DIC, a known marker of sepsis.34 Levels of CRP in ourpatients were markedly high, except in the child with cya-notic heart disease. Lower CRP levels have been reported inchildren with right-to-left intracardiac shunt and chronichypoxemia than those with left-to-right shunt.35 The etiologyremains unknown.

Although five of the six patients received concordantantimicrobials, the mortality remained very high. Thus pre-vention of severe pneumococcal infection is more importantthan choosing appropriate chemotherapy. The asplenic childwas the only one who had previously received pneumococcalvaccine. The possible reason for vaccine failure may be thatthe 23-valent vaccine she received could not elicit a suffi-cient antibody response in asplenic patients.36 That all sixepisodes occurred in October through March probably sug-gests that high-risk populations should receive pneumococcalvaccine no later than early September.

As for possible causes of hypoglycemia in this population,we speculate that an overwhelming sepsis can cause hypo-glycemia by: (1) increased glucose utilization by macro-phage-rich tissues including liver, lung, spleen, ileum, andskin,19 and (2) depressed hepatic gluconeogenesis that hasresulted from decreased sensitivity to stress hormones, and/or adrenal failure.7,10,19

In summary, hypoglycemia developed in six (8.6%) of 70episodes of bacteremic pneumococcal infection. The threeadult patients had received anti-cancer chemotherapy, andall had multilobar pneumonia and hypoxemia at presenta-

tion. One of the three children and all three adults died,although five of them had received concordant antimicro-bials. All patients suspected of having septicemia should haveblood glucose checked, to avoid missing hypoglycemia,resulting in a worsening of an already poor condition. Theasplenic child who died was the only one who had received apneumococcal vaccine beforehand, although all the isolateswould have been covered by a 7- or 23-valent vaccine, andthe infection could have been prevented.

Conflict of interest: No conflict of interest to declare.

References

1. Miller SI, Wallace Jr RJ, Musher DM, Septimus EJ, Kohl S, BaughnRE. Hypoglycemia as a manifestation of sepsis. Am J Med1980;68:649—54.

2. Marsh K, Forster D, Waruiru C, Mwangi I, Winstanley M, Marsh V,et al. Indications of life-threatening malaria in African children.N Engl J Med 1995;332:1399—404.

3. Dzeing-Ella A, Nze Obiang PC, Tchoua R, Planche T, Mboza B,Mbounja M, et al. Severe falciparum malaria in Gabonese chil-dren: clinical and laboratory features. Malar J 2005;4:1.

4. Eholie SP, Ehui E, Adou-Bryn K, Kouame KE, Tanon A, Kakou A,et al. Severe malaria in native adults in Abidjan (Cote d’Ivoire).Bull Soc Pathol Exot 2004;97:340—4.

5. Shears P. Shigella infections. Ann Trop Med Parasitol 1996;90:105—14.

6. van den Broek JM, Roy SK, Khan WA, Ara G, Chakraborty B, IslamS, et al. Risk factors for mortality due to shigellosis: a case—control study among severely malnourished children in Bangla-desh. J Health Pop Nutr 2005;23:259—65.

7. Magnusson JH. Contribution to the knowledge of acute suprar-enal insufficiency in children. Acta Paediatr 1934;15:396—420.

8. Francioli P, Schaller MD, Glauser MP. Fulminant pneumococcalsepticemia in splenectomized patients. Schweiz MedWochenschr 1981;111:2014—6.

9. Latos DL, Stone WJ. Fulminant pneumococcal bacteremia in anasplenic chronic hemodialysis patient. Johns Hopkins Med J1978;143:165—8.

10. Rizzo M, Magro G, Castaldo P. OPSI (overwhelming postsplenect-omy infection) syndrome: a case report. Forensic Sci Int2004;146S:S55—6.

11. Igaki N, Matsuda T, Hirota Y, Kawaguchi T, Tamada F, Goto T.Streptococcal toxic shock syndrome presenting with sponta-neous hypoglycemia in a chronic hemodialysis patient: patho-physiological mechanisms. Intern Med 2003;42:421—3.

12. Virally ML, Guillauseau PJ. Hypoglycemia in adults. DiabetesMetab 1999;25:477—90.

13. LevyMM, Fink MP, Marshall JC, Abraham E, Angus D, Cook D, et al.2001 SCCM/ESICM/ACCP/ATS/SIS International Sepsis DefinitionsConference. Intensive Care Med 2003;29:530—8.

14. Bernard GR, Artigas A, Brigham KL, Carlet J, Falke K, Hudson L,et al. The American—European Consensus Conference on ARDS:definitions, mechanisms, relevant outcomes, and clinical trialcoordination. Am J Respir Crit Care Med 1994;149:818—24.

15. Clinical and Laboratory Standards Institute (2008). Performancestandards for antimicrobial susceptibility testing. 18th Informa-tional supplement. Document M100-S18. Wayne, PA: Clinical andLaboratory Standards Institute; 2008.

16. Brierre S, Kumari R, Deboisblanc BP. The endocrine system duringsepsis. Am J Med Sci 2004;328:238—47.

17. Mendoza A, Kim YN, Chernoff A. Hypoglycemia in hospitalizedadult patients without diabetes. Endocr Pract 2005;11:91—6.

18. Lang CH, Dobrescu C. Sepsis-induced increases in glucose uptakeby macrophage-rich tissues persist during hypoglycemia. Meta-bolism 1991;40:585—93.

576 I.-S. Jan et al.

19. Maitra SR, Wojnar MM, Lang CH. Alterations in tissue glucoseuptake during the hyperglycemic and hypoglycemic phases ofsepsis. Shock 2000;13:379—85.

20. Clemens MG, Chaudry IH, McDermott PH, Baue AE. Regulation ofglucose production from lactate in experimental sepsis. Am JPhysiol 1983;244:R794—800.

21. Spanjaard L, van der Ende A, Rumke H, Dankert J, van Alphen L.Epidemiology of meningitis and bacteremia due to Streptococcuspneumoniae in the Netherlands. Acta Paediatr Suppl 2000;89:22—6.

22. Srivastava KL, Mittal A, Kumar A, Gupta S, Natu SM, Kumar R,et al. Predictors of outcome in fulminant hepatic failure inchildren. Indian J Gastroenterol 1998;17:43—5.

23. Nigro G. Pancreatitis with hypoglycemia-associated convulsionsfollowing rotavirus gastroenteritis. J Pediatr Gastroenterol Nutr1991;12:280—2.

24. Maheady DC. Reye’s syndrome: review and update. J PaediatrHealth Care 1989;3:246—50.

25. Walsh TD. Hypoglycaemia and septicaemia. Postgrad Med J1984;60:431—2.

26. Gradon JD, Lutwick LI. Fatal hypoglycemia associated with Clos-tridium perfringens bacteremia. J Infect Dis 1990;161:1038.

27. Nouel O, Bernuau J, Rueff B, Benhamou JP. Hypoglycemia: acommon complication of septicemia in cirrhosis.Arch Intern Med1981;141:1477—8.

28. Kumashi P, Girgawy E, Tarrand JJ, Rolston KV, Raad II, Safdar A.Streptococcus pneumoniae bacteremia in patients with cancer.Disease characteristic and outcomes in the era of escalating drugresistance (1998—2002). Medicine (Baltimore) 2005;84:303—12.

29. Lloyd RV, Erickson LA, Nascimento AG, Kloppel G. Neoplasmscausing nonhyperinsulinemic hypoglycemia. Endocr Pathol1999;10:291—7.

30. Nakamura H, Imamura T, Kimura N, Niho Y, Okeda T, Yanase T.Hypoglycemia, hypopotassemia and hyperleukocytosis asso-ciated with squamous cell carcinoma of the lung. Jpn J Med1982;21:22—8.

31. Scott JA, Hall AJ, Dagan R, Dixon JM, Eykyn SJ, Fenoll A, et al.Serogroup-specific epidemiology of Streptococcus pneumoniae:associations with age, sex, and geography in 7000 episodes ofinvasive disease. Clin Infect Dis 1996;22:973—81.

32. Martens P, Worm SW, Lundgren B, Konradsen HB, Benfield T.Serotype-specific mortality from invasive Streptococcus pneu-moniae disease revisited. BMC Infect Dis 2004;4:21.

33. Kalin M, Ortqvist A, Almela M, Aufwerber E, Dwyer R, HenriquesB, et al. Prospective study of prognostic factors in community-acquired bacteremic pneumococcal disease in 5 countries. JInfect Dis 2000;182:840—7.

34. Bick RL. Disseminated intravascular coagulation current con-cepts of etiology, pathophysiology, diagnosis, and treatment.Hematol Oncol Clin North Am 2003;17:149—76.

35. Hovels-Gurich HH, Schumacher K, Vazquez-Jemenez JF, Qing M,Huffmeier U, Budding B, et al. Cytokine balance in infantsundergoing cardiac operation. Ann Thorac Surg 2002;73:601—8.

36. Petrasch S, Kuhnemund O, Reinacher A, Uppenkamp M, ReinertR, Schmiegel W, et al. Antibody responses of splenectomizedpatients with non-Hodgkin’s lymphoma to immunization withpolyvalent pneumococcal vaccines. Clin Diagn Lab Immunol1997;4:635—8.