- Email: [email protected]
Bloodstream infections in the elderly
Bloodstream Infections in the Elderly BURT R. MEYERS, M.D., EDWARD SHERMAN, M.D., MERYL H. MENDELSON, M.D., GLADYSVELASQUEZ, M.D., ELINA~RULEVITCH-CHIN, B.S., MICHAEL HUBBARD, Ph.D.. SHALOMZ. HIRSCHMAN, M.D. New York, New York
PURPOSE: Bacteremia in the elderly is associated with a different clinical course and a higher mortality rate when compared with that in younger age groups. In order to examine these issues in the aged, we reviewed the clinical course and factors involved in the outcome of 100 episodes of bloodstream infections in patients over 65 years of age. PATIENTSANDMETHODS: Thehospitalrecordsof all patients over 65 years of age at The Mount Sinai Hospital with a positive blood culture result during the period October 1984 to October 1986 were reviewed. Place of residence before hospital admission, site of acquisition of infection, source of bloodstream infection, and microorganism were analyzed. Antimicrobial therapy was defined as appropriate if initial therapy included one agent to which the isolate was sensitive, or inappropriate if the isolate was resistant. The following factors affecting survival were analyzed: age, sex, underlying diseases, clinical parameters on admission, white blood cell count, mental status, source of infection, microorganism isolated, antibiotic toxicity, and appropriate versus inappropriate antibiotic therapy. RESULTS: Most patients were female (63 percent), were febrile (90 percent), had an altered mental status (52 percent), and had a neutrophilic response (61 percent). Eighty-three percent of patients were admitted from the community (home), 14 percent were from long-term-care facilities, and 3 percent were transferred from other hospitals. Fifty percent of infections were nosocomial, and 44 percent were community (home and nursing home)-acquired. Gram-negative organisms accounted for 60 percent of isolates, with Esc~erichia coil (22 percent) and Klebsiella species (11 percent) predominating; 30 percent were gram-positive organisms, with Stk&1y1ococcus aureus (13 percent) and Streptococcus faecalis (10 percent) the most common. The overall survival was 60 percent; the survival rate was 65.8 percent for community-acquired (home) bacteremia, 75 percent for nursing home-acquired bacteremia, and 52.8 percent for hospital-acquired bacteremia. Survival for gramnegative isolates was 65 percent, versus 51.7 percent for gram-positive isolates. Survival was greatest in patients whose source of bacteremia was either the genitourinary tract (70 percent) or an intravascular device (78 percent) and poorest in patients with lower respiratory tract source (42 From the Division of Infectious Diseases, Department of Medicine, and the Department of Biomathematical Sciences, The Mount Sinai Hospital and The Mount Sinai School of Medicine of the City University of New York, New York, New York. Requests for reprints should be addressed to Burt R. Meyers, M.D., Box 1090, The Mount Sinai Hospital, One Gustave L. Levy Place, New York, New York 10029-6574. Manuscrtpt submitted October 11, 1988, and accepted in revised form January 16, 1989.
percent); all three patients with endocarditis died. Increased survival was observed in patients treated with appropriate antimicrobial agents regardless of age, source of infection, or bloodstream isolates. Inappropriate therapy, in conjunction with age greater than 85 years or with S. aureus infection, was associated with increased mortality. Toxicity was greatest with aminoglycosides (21 percent), followed by cephalosporins (19 percent), and was least with penicillins (12 percent); fewer cases of ototoxicity and nephrotoxicity were observed in aminoglycoside-treated patients older than 85 years compared with that in younger age groups. CONCLUSION: The data suggest that initial empiric antimicrobial therapy in the elderly be broad in scope and include coverage for S. aureus. Particular attention is necessary in selecting appropriate initial therapy in patients greater than 85 years of wife. acteremia is associated with a higher mortality B rate in the elderly as compared with younger-aged groups [l-6]. Factors contributing to this difference include senescence of both humoral [7] and cell-mediated [5,8] immune systems, and an increased incidence of underlying illnesses, including cardiovascular, pulmonary, and neoplastic diseases [9]. Increased mortality in the aged associated with bacteremia, pneumonia [lO,ll], or urosepsis [12] has been reported. Bacteremias with Streptococcus pneumoniae, Staphylococcus aureus, Escherichia coli, Bacteroides fragilis, and Pseudomonas aeruginosa are associated with decreased survival in the elderly [1,3,4,6]. In most instances, the clinical picture of bacteremia in the elderly differs markedly from that in younger populations [13,14]. Although most studies have shown increased mortality in the bacteremic elderly, there are reports of either no increase in mortality [15] or increased mortality limited to certain population subgroups with gram-negative bacteremia [2]. In order, therefore, to further delineate the problems of bacteremia and the factors associated with mortality in the elderly, we reviewed the clinical course and outcome of 100 episodes of bloodstream infections in 100 patients over the age of 65 at The Mount Sinai Hospital from October 1984 to October 1986.
PATIENTSAND METHODS We reviewed the hospital records of all patients 65 years of age and over with a positive blood culture result during the period October 1984 to October 1986. Bloodstream infection was defined as at least one positive blood culture from a patient with a consistent clinical setting. This applied to all pathogens except for Staphylococcus epidermidis, Bacillus species, diphtheroids, and Propionibacterium species, in April
1989 The American Journal of Medicine
Volume 86
379
BLOODSTREAM INFECTIONS IN THE ELDERLY / MEYERS ET AL
scan, magnetic resonance imaging), or both, in association with the bloodstream infection. (3) Endocurditis: presence of at least two positive blood culture results, clinical signs including embolic events, positive echocardiographic result, or autopsy findings. (4) Lower respiratory tract, urinary tract, skin or soft tissue1 wound, and graft: culture of the same organism from the primary site was required with clinical evidence of infection. (5) Indeterminate: none of the aforementioned criteria fulfilled. The place of residence prior to hospital admission was either the community (home), chronic-care facility (i.e., nursing home), or another hospital. Episodes of infection were determined to be either community(home), nursing home-, or nosocomially acquired: nosocomial infection was deemed present if infection occurred more than 24 hours after admission and was determined not to have been incubating at the time of admission. For additional analyses, the patients were stratified into three age groups: Group A, ages 65 to 74 (n = 41); Group B, ages 75 to 84 (n = 40); and Group C, greater than 85 years of age (n = 19). Antimicrobial therapy was defined as appropriate when initial therapy included at least one agent to which the isolate was susceptible. Toxicity was defined by the physician observer (e.g., skin rash, nausea, vomiting, diarrhea). Aminoglycoside nephrotoxicity was defined as a rise in serum creatinine greater than 0.5 mg/dl from the baseline level (prior to aminoglycoside therapy). Analysis of the contingency tables of variables was performed by chi-square test. The multiple logistic regression model was employed with survival as the outcome. The predictors of survival were as follows: ages 65 to 74,75 to 85, and over 85; number of underlying diseases; community, nursing home, or nosocomial acquisition; inappropriate therapy; white blood cell count less than 5,000/mm3; white blood cell count greater than 10,000/mm3; sources including genitourinary, lower respiratory tract, intra-abdominal, intravascular, skin or soft tissue/wound, graft, endocarditis, or indeterminate; organisms including E. coli, S. uureus, Klebsiella pneumoniae, Serratia murcescens, and P. ueruginosu.
TABLE I Characteristics
of Patient Groups
Age Group
(years)
Females*
A
65-74 75-84 85+
24 (59)
E Total
Males* i: {i:] 5 (26) 37
:i $ij 63
Admitted from Community (home) Nursing home Other hospital Acquisition of infection Community Nursing home Nosocomial
83% 14% 3% 39%
5%
Values in parenthesesare
percentages.
TABLE II Clinical Parameters Percent of Patients Te;grature
(OF)
2 8
97-99 99-101 >lOl Chills Altered mental status Tachypnea WB
:;
5,obo-10,000 10,000-20.000
34 28
>20,000 BC = white blood cell.
TABLE Ill Source of Bacteremia Source
Percent
RESULTS
Genitourinary Pneumonia Intra-abdominal Intravascular device Skin-soft tissue/wound Indeterminate Graft Endocarditis Multiple
which at least two positive results of blood cultures were required. The sources of infection were determined as follows: (1) Intravascular catheter: bloodstream infection in association with presence of purulent material or erythema/warmth/tenderness, or both, at the site of an intravascular device, or microbiologic clearance of bloodstream pathogen, or remission of signs/symptoms of infection only after removal of intravascular device. (2) Intru-abdominal source: culture of purulent material/tissue with the same organism or evidence of abscess collection, cholecystitis, biliary tract obstruction by noninvasive radiographic studies (i.e., songraphy, computed axial tomographic 380
April
1989 The American
Journal
of Medicine
Volume
86
The characteristics of patient groups are given in Table I. There were more patients in Groups A and B than in Group C; females predominated in all groups. Most patients were admitted from their homes (83 percent); 53 percent of infections were nosocomial (Table I). Underlying diseases were common: 33 patients had cardiovascular diseases, 33 had malignancy, and 11 had diabetes mellitus. The clinical parameters at the time bloodstream infection was initially diagnosed are given in Table II. A temperature of greater than 99“F was noted in 90 percent of patients. Altered mental status was the second most common presenting feature (52 percent). Most patients presented with white blood cell (WBC) counts between 5,000 and 20,000/mm3; 16 percent had WBC counts less than 5,000/mm3, and 38 percent had less than 10,000/mm3. The source of bloodstream infection is listed in Table III. Infections originating in the genitourinary tract were the most common (27 percent), followed by an intra-abdominal source (16 percent) and lower res-
BLOODSTREAM
INFECTIONS
IN THE ELDERLY
/ MEYERS
ET AL
TABLEIV Blood Culture isolates (N = 106) Organism Gram-negative bacilli Escherichia co/i Klebsiella sp. K. pneumoniae K. oxytoca
Community
Ni:kg
Hospital
25 12
3 2
36 10
:
: 1
-
-
Ei (60.4)’ :;
-
Klebsiella species Serra tia sp.
:
r
-
13
4
z 2
:
1
S. marcescens Serrafia species Proteus mirabilis Haemophilus influenzae Morganella morganii Acinetobacter calcoaceticus Pseudomonas aeruginosa Enterobacferspecies E. cloacae E. aerogenes
1 1 1 2 2 1
Gram-positive cocci Staphylococcus aureus Enterococcus faecalis Staphylococcus epidermidis Group A streptococci Strepotococcuspneumoniae
-i
Anaerobes b’acteroides fragilis Peptostrepfococcus Closfridium species
-
Fungi Candida species C. albicans C. tropicalis
-
Values in parentheses
5”
Total
A 5 :
-
: 2
i : -Y 6 5 1
; ;
15 8 6 -
7 -
!
;I
ifi (30.2)” 11 2
1
:
1 -
T -
i 1
: (6w i
i
if (2.8Y
:
:
are percentages.
piratory tract (12 percent). The source could not be determined in 21 patients. Gram-negative bacilli accounted for 60.4 percent of the blood culture isolates, with E. coli predominating (22 percent) (Table IV). Gram-positive cocci comprised 30.2 percent of the isolates; the predominant organisms were S. aureus and Enterococcus faecalis. B. fragilis was the most common anaerobic isolate (Table IV). Polymicrobial infections occurred in 6 percent of cases. Overall survival was 60 percent; survival was 65.8 percent for community (home)-acquired bacteremia, 75 percent for nursing home-acquired bacteremia, and 52.8 percent for hospital-acquired bacteremia. Patients with gram-negative isolates had a greater survival than those with gram-positive isolates (65 percent versus 51.7 percent, p = NS). Among all groups, survival was greatest in patients with one underlying disease (69 percent), less with two diseases (52 percent), and least with three diseases (45 percent) (p = NS). The relationship of survival to the source of bloodstream infection is examined in Table V. Survival was greatest when the source was determined to be an intravascular device (78 percent), followed by genitourinary (70 percent) and intra-abdominal sources (62 percent). Patients with a lower respiratory tract source had the poorest survival (42 percent). When all sources were compared with regard to survival, no differences were found (x2(10) = 12.62, p = 0.25). Survival was also examined in regard to source of blood-
stream infection and appropriate therapy. Among patients treated appropriately, there was no differry2;ypg the sources with respect to survival (x2(9) = 0.19). However, among those patients - , treated appropriately, a trend towards increased survival in those with genitourinary tract (x2(1) = 2.423, p = 0.12) and lower respiratory tract (x2(1) = 2.310, p = 0.13) infections and a significantly decreased survival with endocarditis (x2( 1) = 5.506, p = 0.025) were noted when each group was compared with the whole group. When survival was examined for all sources in patients treated inappropriately, no differences were noted (x2(6) = 5.33, p = 0.50) and survival was not associated with WBC count (x2(2) = 1.55, p = 0.46). , For all levels of WBC count, age was not related to survival. When survival from all sources was examined, a significant decrease was noted in patients with a WBC count less than 5,000/mm3 (x2(4) = 10.182, p = 0.037); mortality was increased in patients with a lower respiratory tract source compared to all other sources (x2(1) = 8.123, p = 0.004). Survival was not significantly related to age categories in patients treated appropriately (62 percent) (x2(2) = 0.656, p = 0.72) or those treated inappropriately (50 percent) (x2(2) = 3.09, p = 0.21). However, in those over 85 years of age, four of five (80 percent) patients treated inappropriately died compared with four of 14 (29 percent) treated appropriately (x2(1) = 4.00, p = 0,046) (Table VI). The relationship of isolate to survival in patients treated appropriately was also examined (Table VII). April 1989
The American
Journal
of Medicine
Volume
86
381
BLOODSTREAM INFECTIONS IN THE ELDERLY / MEYERS ET AL TABLE V Survival by Source of Infection
and Appropriate
Site
Number
Genitourinary Lower respiratory tract Intra-abdominal Intravascular device Skin and wound Indeterminate Graft Endocarditis Mixed * Values in parentheses
Antimicrobial
Therapy Survival*
Appropriate Therapy*
Number
27
21
fi 9
ii ‘Y $Ej 3 (50)
2;
7
1;
7 (‘E] 0 (0) l(33)
i 3
‘: (‘;:; 3 (60)
ll(61)
2 (100) 0 (0) l(50)
32 2
are percentages.
TABLE VI Survival
Related to Age and Therapy
Age
Group
(yea@
Appropriate Therapy*
A
$1;; 85+
s Total * Values in parentheses
Inappropriate Therapy*
Total*
p Value
22137 (59) 20/33 (61) 10/14 (71)
214 (50) :y: {:A;
24/41(59) 25/40 11/19 (62) (58)
0.715 0.591 0.046
52184 (62)
8116 (50)
60/100 (60)
0.373
are percentages.
TABLE VII Survival by Organism:
Appropriate
Therapy
Appropriate Therapy*
Organism E. co/i S. aureus K. pneumoniae S. marcescens P. aeruginosa S. faecalis Other Total * Values in parentheses
versus Inappropriate
Inappropriate Therapy*
14122 (64)
~&(%y
April
16/24 (67)
“0;; $0’
gp
6/14 (43) 215 (40) 5/12 (42)
4/8 (50) 17/27 (63) 52/84 (62)
: y: I::{
8/16 (50)
13/29 (45) 60/100 (60)
are percentages.
Survival was greatest with K. pneumoniae (80 percent) and least with P. aeruginosa (40 percent). The different rates of survival among the organisms in the subset were not significant (p = 0.9). The relationship between inappropriate therapy for the specific isolates and survival was examined; no other differences in outcome were noted (x2(5) = 10.67, p = 0.058). However, all three patients with S. aureus treated inappropriately died, whereas five of 13 with other organisms died (x2(1) = 3.692, p = 0.055). Multiple logistic regression analysis revealed that community-acquired infection (p = O.O2),intravenous source (p = 0.04), and S. marcescens infection (p = 0.05) were related to survival; inappropriate therapy (p = 0.05) was negatively associated with survival. Adverse effects were seen most frequently with aminoglycosides (21 percent), and least with penicillins (12 percent) (Table VIII). Patients over 85 years of age had the fewest adverse effects. Among the age groups, differences in toxicity were not significant. When the relationship of age (less than or greater than
382
212 (100) O/3 C-9
l/l (100)
;\i gj
Total*
1989 The American
Journal
of Medicine
Volume
86
85) and antibiotic toxicity in patients treated either appropriately or inappropriately was examined, no significant differences in toxicity were noted.
COMMENTS Bacteremic infections in the elderly have been described in relation to the sources [12,16], specific places of acquisition [2,14,15], hospital [2] or long-termcare facility [13,17], and specific pathogens such as Klebsiella species [18], E. coli [10,19], 5’. pneumoniae [3,15], or other gram-negative bacilli [2,11]. Our study examines data in regard to all these factors. The clinical picture of bloodstream infections in the elderly often differs from that in younger patients [13]. Gleckman and Hibert [20] noted that 13 percent of elderly patients with positive blood culture results were afebrile, whereas others noted that 69 percent of patients with bacteremic pneumococcal pneumonia had a temperature greater than 101°F; data were not given for the other 28 patients [16]. Only one of 100 patients with a community source had a temperature
BLOODSTREAM
less than 37.2% [14]. Fever occurred equally in elderly and younger patients with pneumococcal bacteremia [15]. Our data revealed only two patients with temperatures less than or equal to 97°F and 10 percent with temperatures less than 99°F. It appears, then, that most elderly bacteremic patients have fever as a presenting sign. An altered mental status was found in 52 percent of our patients; it has also been reported as a presenting sign in patients with bacteremic pneumonia [13,15]. Sixty percent of our patients had WBC counts greater than 10,000/mm3, compared with 70 percent in the series reported by Windsor [13]. The urinary tract is the most common source of bacteremia (34 percent) in elderly patients admitted from the community setting [ 141 and was the source in 27 percent of our patients; in long-term-care facilities, 24 to 56 percent of bacteremias originate in the urinary tract [13,17,21]. Infection was secondary to in-dwelling intravascular catheters in 7 percent of our patients. Comparable statistics are not available from other studies of the elderly. The incidence of bacteremia from an intra-abdominal source in our series was 14 percent, whereas this source constituted 20 percent of community-acquired bacteremias [14] and 8 percent in the bacteremias in a geriatric unit [13]. Gram-negative bacilli accounted for 60 percent of isolates in our study, compared with 42 to 76 percent in the long-term-care setting [13,17,21]. In our study, E. coli (23 percent of total isolates, 37.5 percent of gram-negative isolates) and K. pneumoniae (9.4 percent of total isolates, 15.6 percent of gram-negative isolates) were the predominant gram-negative isolates; P. aeruginosa (5.7 percent of total isolates, 9.4 percent of gram-negative isolates) and Enterobacter (6.6 percent of total isolates, 10.9 percent of gramnegative isolates) occurred less often. Although the frequencies of K. pneumoniae isolates were similar, we had fewer patients with E. coli infections and more with P. aeruginosa and Enterobacter isolates than the study of McCue [2]. Gram-positive cocci accounted for 30 percent of the total isolates, compared with 24 to 58 percent reported from long-term-care settings [13,17,21]. Of these isolates, S. aureus was the most common (13.2 percent of total isolates, 43.8 percent of gram-positive isolates), similar to that described by Setia et al [17]. Group D streptococci or enterococci (E. faecalis) accounted for 10.4 percent, compared with 7 to 16 percent in longterm-care facilities [17,21] and 5 percent from the community setting [14]. Enterococci have been reported with increasing frequency as nosocomial pathogens [22]. Only four isolates were S. pneumoniae (3.7 percent) compared with 12 percent from the community setting [14]. This is not unexpected since S. pneumoniae is not a common hospital-acquired pathogen [23]. Anaerobic organisms accounted for 6.6 percent of isolates, compared with 2 percent [14] and 3 percent [17] found by others in the elderly. This incidence is less than that reported for hospitalized patients of all ages with positive results of blood cultures [24]. The overall mortality in our study was 40 percent: 34.2 percent in community-acquired, 25 percent in nursing home-acquired, and 47.2 percent in nosocomially acquired bacteremias. This compares to reports of 26 percent mortality for community-acquired bac-
INFECTIONS
IN THE ELDERLY
/ MEYERS ET AL
TABLEVIII Antibiotic Toxicity Antibiotic
Toxicity
Aminoglycosidest (age 65-74) (age 75-84) (age 85+)
Group A
c”
5123 7125 o/10 12/58
Total Cephalosporins Group A
(age 65-74) (age 75-84) (age 85+)
F
(22) (28) (0) (21)*
3115 (20) 6125 (24)
018(0)
9/48 (19)*
Total Penicillins Group A
(age 65-74) (age 75-84) (age 85+)
F
;j;;
I;;.“’
l/l1 (8) 5/43 (12)Z
Total * Values in parentheses are percentages. t Aminoglycoslde toxicity is dependent on either increased en or creatinine. or physlcian observer recognition. !Not significant.
blood urea nitro-
teremia [14], and 21 to 35 percent in the nursing home [17-221. It is not surprising that mortality was lower in patients with community-acquired gram-negative bacteremia, since the majority of these isolates were E. coli and Klebsiella species and originated in the urinary tract. In the long-term-care facility, 51 percent of all cases originated in the urinary tract [21], compared with 27 percent in our study, and probably accounts for the lower mortality rates noted in nursing homecompared with hospital-acquired bacteremias. Survival was examined with regard to source of bacteremia. It was greatest with a genitourinary source (70 percent), similar to that in reports from long-termcare facilities [17], but less than catheter-related urosepsis (80 to 92 percent) [2,12]. With a skin or soft tissue source, 50 percent survival was seen compared with 43 percent reported in the nursing home [17]. In both settings, bacteremic pneumonia was associated with a lower survival (42 percent). The relationship of survival to initial administration of appropriate therapy according to pathogen was examined. The highest rate of survival in patients given appropriate therapy was noted with K. pneumoniae (78 percent). Survival was 50 percent with E. fuecalis and P. aeruginosa; 55 percent mortality was noted with S. aureus. In a review of gram-negative bacteremia, McCue [2] noted that mortality was increased only in those patients receiving inappropriate therapy within the first 24 hours. In our study, none of the patients with S. aureus and one of four with E. faecalis who received inappropriate therapy survived. When survival was examined comparing appropriate and inappropriate therapy to the source of bacteremia, no significant differences were noted. This suggests that more practical methods of prevention of bacteremia, such as better attention to indwelling catheters, urinary drainage, and respiratory care, are warranted. The relationships of age, appropriate therapy, and survival revealed no differences in outcome when all groups were compared; however, in patients over 85 years of age, appropriate therapy yielded a higher sur-
April 1989
The American
Journal
of Medicine
Volume
86
383
BLOODSTREAM INFECTIONS IN THE ELDERLY / MEYERS ET AL
viva1 than inappropriate therapy. Patients 70 years and older who received appropriate single-drug antimicrobial therapy for gram-negative bacteremia had the lowest mortality [2]. P. aeruginosa, Enterobacter species, Serratia species, and enterococci accounted for 29.2 percent of our isolates. These pathogens are often antibiotic-resistant and more difficult to treat. The differences in reported mortality between our and previous studies may be due to the greater numbers of these isolates. The relationship of adverse effects to age and antibiotic class was examined. Toxicity was greatest with aminoglycosides (21 percent) and lower for cephalosporins (19 percent) and penicillins (12 percent); it was least in patients over 85 years of age receiving aminoglycosides (8 percent). For penicillins, no toxicity was seen in those over 85. Age has been associated with drug-induced nephrotoxicity and ototoxicity in patients treated with aminoglycosides [25]; in our study, only one of 13 aminoglycoside-treated patients greater than 85 years of age developed toxicity. This finding may relate to the caution used by the physicians treating this subgroup.
SUMMARY In a retrospective review of factors affecting outcome of bacteremia in the elderly, survival was best with isolates originating in the urinary tract and poorest with those from the respiratory tract. However, when all factors were considered simultaneously in a multiple regression analysis, appropriate therapy significantly increased survival. Factors associated with increased mortality in patients treated inappropriately were S. aureus infections, lower respiratory tract sources, and age greater than 85 years; with a WBC count less than 5,000/mm3, survival was decreased from all sources. Since the subgroup aged 85 years and older is the most rapidly growing segment of the elderly population, initial empiric therapy of bacteremia should be broad in scope and include coverage for S. aureus as well as gram-negative bacilli. Recent data from long-term-care facilities with respect to bacterial isolates reveal a trend toward more resistant gram-negative organisms and S. aureus, similar to those found in hospitalized patients [26-281. This suggests that empiric therapy should be approached in a similar manner for elderly patients in both the long-term-care facility and hospital settings. Use of new and more potent third-generation cephalosporins, monobactams, quinolones, and the carbapenems may increase future survival statistics; further studies are needed for confirmation of this premise.
384
April
1989
The American
Journal
of Medicine
Volume
86
REFERENCES 1. Chattopadhyay
9, Al-Zahawi M: Septicaemia and its unacceptably . high - mortality in the elderly. J Infect 1983; 7: 134-138. 2. McCue JD: Gram-negative bacillary bacteremia in the elderly: incidence, ecology, etiology, and mortality. J Am Geriatr Sot 1987; 35: 213-218. 3. Gruer LD, McKendrick MW, Geddes AM: Pneumococcal bacteremia-a continuing challenge. Q J Med 1984; 53: 259-270. 4. Vazquez F, Mendez FJ, Perez F, Mendoza MC: Anaerobic bacteremia in a general hospital: retrospective five-year analysis. Rev Infect Dis 1987; 9: 1038-1043. 5. Garibaldi RA, Nurse BA: Infections in the elderly. Am J Med 1986; 81(suppl 1A): 53-58. 6. Miller PJ, Wenzel RP: Etiologic organisms as independent predictors of death and morbidity associated with bloodstream infections. J Infect Dis 1987; 156:
471-477. 7. Powers DC, Nagel JE, Hoh J, Adler WH: Immune function in the elderly. Postgrad Med 1987; 81: 355-359. 8. Murasko DM, Nelson BJ, Silver R, Matour D, Kaye D: Immunologic response in an elderly population with a mean age of 85. Am J Med 1986; 81: 612-618. 9. Kovar M: Health of the elderly and use of health services. Public Health Rep
1977: 92: 9-19. 10. Jonas M. Cunha BA: Bacteremic EscherichiacoEpneumonia. Arch Intern Med 1982; 142: 2157-2159. 11. Phair JP, Bassaris HP, Williams JE, Metzger E: Bacteremic oneumonia due to gram-negative bacilli. Arch Intern Med 198% 143: 2147-2144. 12. Gleckman R, Blagg N, Hibert D. et al: Catheter-related urosepsis in the elderly: a prospective study of community derived infections. J Am Geriatr Sot 1982; 30:
255-257. 13. Windsor ICM: Bacteraemia in ageriatric
unit. Gerontology 1983; 291125-130. 14. Esposito AL, Gleckman RA, Cram S, Crowley M, McCabe F, Drapkin MS: Community-acquired bacteremia in the elderly: analysis of one hundred consecutive episodes. J Am Geriatr Sot 1980; 28: 315-319. 15. Esposito AL: Community-acquired bacteremic pneumococcal pneumonia. Arch Intern Med 1984; 144: 945-948. 16. Murphy TF, Fine BC: Bacteremic pneumococcal pneumonia in the elderly. Am J Med Sci 1984; 288: 114-118. 17. Setia U. Serventi I, Lorenz P: Bacteremia in a long-term facility. Arch Intern Med 1984; 144: 1633-1635. 18. Montgomerie JZ. Ota JK: Klebsiella bacteremia. Arch intern Med 1980; 140:
526-527. 19. Berk SL, Neumann P, Holtsclaw S, Smith JK: Escherichia cogpneumonia in the elderly. Am J Med 1982; 72: 899-902. 26. Gleckman R, Hibert D: Afebrile bacteremia. JAMA 1982; 248: 1478-1481. 21. Rudman D, Hontanosas A, Cohen Z, Mattson DE: Clinical correlates of bacteremia in a Veterans Administration extended care facility. J Am Geriatr Sot 1988;
36: 726-732. 22. Morrison Al Jr, Wenzel RP: Nosocomial coccus: ten years experience at a university
urinary tract infection due to Enter@ hospital. Arch Intern Med 1986; 146:
1549-1551. infection surveillance. MMWR 1986; 35: 17SS-25%. 24. Weinstein MP, Reller B, Murphy JR, Lichenstein KA The clinical significance of
23. Centers for Disease Control: Nosocomial
positive blood cultures: a comprehensive analysis of 500 episodes of and fungemia in adults. I. Laboratory and epidemiologic observations. Dis 1983; 5: 35-53. 25. Moore RD, Smith CR, Lipsky JJ. Mellitis ED, Lietman PS: Risk nephrotoxicity in patients treated with aminoglycosides. Ann Intern
bacteremia Rev Infect factors tor Med 1984;
loo: 352-357. 26. Gaynes RP, Weinstein RA, Chamberlin W, Kabins SA: Antibiotic-resistant flora in nursing home patients admitted to the hospital. Arch Intern Med 1985; 145: 1804-
1807. 27. Scheckler WE, Peterson PJ: Infections and infection control among residents of eight rural Wisconsin nursing homes. Arch Intern Med 1986; 146: 1981-1984. 28. Norman DC, Castle SC, Cantrell M: Infections in the nursing home. J Am Geriatr Sot 1987; 35: 796-805.
PURPOSE: Bacteremia in the elderly is associated with a different clinical course and a higher mortality rate when compared with that in younger age groups. In order to examine these issues in the aged, we reviewed the clinical course and factors involved in the outcome of 100 episodes of bloodstream infections in patients over 65 years of age. PATIENTSANDMETHODS: Thehospitalrecordsof all patients over 65 years of age at The Mount Sinai Hospital with a positive blood culture result during the period October 1984 to October 1986 were reviewed. Place of residence before hospital admission, site of acquisition of infection, source of bloodstream infection, and microorganism were analyzed. Antimicrobial therapy was defined as appropriate if initial therapy included one agent to which the isolate was sensitive, or inappropriate if the isolate was resistant. The following factors affecting survival were analyzed: age, sex, underlying diseases, clinical parameters on admission, white blood cell count, mental status, source of infection, microorganism isolated, antibiotic toxicity, and appropriate versus inappropriate antibiotic therapy. RESULTS: Most patients were female (63 percent), were febrile (90 percent), had an altered mental status (52 percent), and had a neutrophilic response (61 percent). Eighty-three percent of patients were admitted from the community (home), 14 percent were from long-term-care facilities, and 3 percent were transferred from other hospitals. Fifty percent of infections were nosocomial, and 44 percent were community (home and nursing home)-acquired. Gram-negative organisms accounted for 60 percent of isolates, with Esc~erichia coil (22 percent) and Klebsiella species (11 percent) predominating; 30 percent were gram-positive organisms, with Stk&1y1ococcus aureus (13 percent) and Streptococcus faecalis (10 percent) the most common. The overall survival was 60 percent; the survival rate was 65.8 percent for community-acquired (home) bacteremia, 75 percent for nursing home-acquired bacteremia, and 52.8 percent for hospital-acquired bacteremia. Survival for gramnegative isolates was 65 percent, versus 51.7 percent for gram-positive isolates. Survival was greatest in patients whose source of bacteremia was either the genitourinary tract (70 percent) or an intravascular device (78 percent) and poorest in patients with lower respiratory tract source (42 From the Division of Infectious Diseases, Department of Medicine, and the Department of Biomathematical Sciences, The Mount Sinai Hospital and The Mount Sinai School of Medicine of the City University of New York, New York, New York. Requests for reprints should be addressed to Burt R. Meyers, M.D., Box 1090, The Mount Sinai Hospital, One Gustave L. Levy Place, New York, New York 10029-6574. Manuscrtpt submitted October 11, 1988, and accepted in revised form January 16, 1989.
percent); all three patients with endocarditis died. Increased survival was observed in patients treated with appropriate antimicrobial agents regardless of age, source of infection, or bloodstream isolates. Inappropriate therapy, in conjunction with age greater than 85 years or with S. aureus infection, was associated with increased mortality. Toxicity was greatest with aminoglycosides (21 percent), followed by cephalosporins (19 percent), and was least with penicillins (12 percent); fewer cases of ototoxicity and nephrotoxicity were observed in aminoglycoside-treated patients older than 85 years compared with that in younger age groups. CONCLUSION: The data suggest that initial empiric antimicrobial therapy in the elderly be broad in scope and include coverage for S. aureus. Particular attention is necessary in selecting appropriate initial therapy in patients greater than 85 years of wife. acteremia is associated with a higher mortality B rate in the elderly as compared with younger-aged groups [l-6]. Factors contributing to this difference include senescence of both humoral [7] and cell-mediated [5,8] immune systems, and an increased incidence of underlying illnesses, including cardiovascular, pulmonary, and neoplastic diseases [9]. Increased mortality in the aged associated with bacteremia, pneumonia [lO,ll], or urosepsis [12] has been reported. Bacteremias with Streptococcus pneumoniae, Staphylococcus aureus, Escherichia coli, Bacteroides fragilis, and Pseudomonas aeruginosa are associated with decreased survival in the elderly [1,3,4,6]. In most instances, the clinical picture of bacteremia in the elderly differs markedly from that in younger populations [13,14]. Although most studies have shown increased mortality in the bacteremic elderly, there are reports of either no increase in mortality [15] or increased mortality limited to certain population subgroups with gram-negative bacteremia [2]. In order, therefore, to further delineate the problems of bacteremia and the factors associated with mortality in the elderly, we reviewed the clinical course and outcome of 100 episodes of bloodstream infections in 100 patients over the age of 65 at The Mount Sinai Hospital from October 1984 to October 1986.
PATIENTSAND METHODS We reviewed the hospital records of all patients 65 years of age and over with a positive blood culture result during the period October 1984 to October 1986. Bloodstream infection was defined as at least one positive blood culture from a patient with a consistent clinical setting. This applied to all pathogens except for Staphylococcus epidermidis, Bacillus species, diphtheroids, and Propionibacterium species, in April
1989 The American Journal of Medicine
Volume 86
379
BLOODSTREAM INFECTIONS IN THE ELDERLY / MEYERS ET AL
scan, magnetic resonance imaging), or both, in association with the bloodstream infection. (3) Endocurditis: presence of at least two positive blood culture results, clinical signs including embolic events, positive echocardiographic result, or autopsy findings. (4) Lower respiratory tract, urinary tract, skin or soft tissue1 wound, and graft: culture of the same organism from the primary site was required with clinical evidence of infection. (5) Indeterminate: none of the aforementioned criteria fulfilled. The place of residence prior to hospital admission was either the community (home), chronic-care facility (i.e., nursing home), or another hospital. Episodes of infection were determined to be either community(home), nursing home-, or nosocomially acquired: nosocomial infection was deemed present if infection occurred more than 24 hours after admission and was determined not to have been incubating at the time of admission. For additional analyses, the patients were stratified into three age groups: Group A, ages 65 to 74 (n = 41); Group B, ages 75 to 84 (n = 40); and Group C, greater than 85 years of age (n = 19). Antimicrobial therapy was defined as appropriate when initial therapy included at least one agent to which the isolate was susceptible. Toxicity was defined by the physician observer (e.g., skin rash, nausea, vomiting, diarrhea). Aminoglycoside nephrotoxicity was defined as a rise in serum creatinine greater than 0.5 mg/dl from the baseline level (prior to aminoglycoside therapy). Analysis of the contingency tables of variables was performed by chi-square test. The multiple logistic regression model was employed with survival as the outcome. The predictors of survival were as follows: ages 65 to 74,75 to 85, and over 85; number of underlying diseases; community, nursing home, or nosocomial acquisition; inappropriate therapy; white blood cell count less than 5,000/mm3; white blood cell count greater than 10,000/mm3; sources including genitourinary, lower respiratory tract, intra-abdominal, intravascular, skin or soft tissue/wound, graft, endocarditis, or indeterminate; organisms including E. coli, S. uureus, Klebsiella pneumoniae, Serratia murcescens, and P. ueruginosu.
TABLE I Characteristics
of Patient Groups
Age Group
(years)
Females*
A
65-74 75-84 85+
24 (59)
E Total
Males* i: {i:] 5 (26) 37
:i $ij 63
Admitted from Community (home) Nursing home Other hospital Acquisition of infection Community Nursing home Nosocomial
83% 14% 3% 39%
5%
Values in parenthesesare
percentages.
TABLE II Clinical Parameters Percent of Patients Te;grature
(OF)
2 8
97-99 99-101 >lOl Chills Altered mental status Tachypnea WB
:;
5,obo-10,000 10,000-20.000
34 28
>20,000 BC = white blood cell.
TABLE Ill Source of Bacteremia Source
Percent
RESULTS
Genitourinary Pneumonia Intra-abdominal Intravascular device Skin-soft tissue/wound Indeterminate Graft Endocarditis Multiple
which at least two positive results of blood cultures were required. The sources of infection were determined as follows: (1) Intravascular catheter: bloodstream infection in association with presence of purulent material or erythema/warmth/tenderness, or both, at the site of an intravascular device, or microbiologic clearance of bloodstream pathogen, or remission of signs/symptoms of infection only after removal of intravascular device. (2) Intru-abdominal source: culture of purulent material/tissue with the same organism or evidence of abscess collection, cholecystitis, biliary tract obstruction by noninvasive radiographic studies (i.e., songraphy, computed axial tomographic 380
April
1989 The American
Journal
of Medicine
Volume
86
The characteristics of patient groups are given in Table I. There were more patients in Groups A and B than in Group C; females predominated in all groups. Most patients were admitted from their homes (83 percent); 53 percent of infections were nosocomial (Table I). Underlying diseases were common: 33 patients had cardiovascular diseases, 33 had malignancy, and 11 had diabetes mellitus. The clinical parameters at the time bloodstream infection was initially diagnosed are given in Table II. A temperature of greater than 99“F was noted in 90 percent of patients. Altered mental status was the second most common presenting feature (52 percent). Most patients presented with white blood cell (WBC) counts between 5,000 and 20,000/mm3; 16 percent had WBC counts less than 5,000/mm3, and 38 percent had less than 10,000/mm3. The source of bloodstream infection is listed in Table III. Infections originating in the genitourinary tract were the most common (27 percent), followed by an intra-abdominal source (16 percent) and lower res-
BLOODSTREAM
INFECTIONS
IN THE ELDERLY
/ MEYERS
ET AL
TABLEIV Blood Culture isolates (N = 106) Organism Gram-negative bacilli Escherichia co/i Klebsiella sp. K. pneumoniae K. oxytoca
Community
Ni:kg
Hospital
25 12
3 2
36 10
:
: 1
-
-
Ei (60.4)’ :;
-
Klebsiella species Serra tia sp.
:
r
-
13
4
z 2
:
1
S. marcescens Serrafia species Proteus mirabilis Haemophilus influenzae Morganella morganii Acinetobacter calcoaceticus Pseudomonas aeruginosa Enterobacferspecies E. cloacae E. aerogenes
1 1 1 2 2 1
Gram-positive cocci Staphylococcus aureus Enterococcus faecalis Staphylococcus epidermidis Group A streptococci Strepotococcuspneumoniae
-i
Anaerobes b’acteroides fragilis Peptostrepfococcus Closfridium species
-
Fungi Candida species C. albicans C. tropicalis
-
Values in parentheses
5”
Total
A 5 :
-
: 2
i : -Y 6 5 1
; ;
15 8 6 -
7 -
!
;I
ifi (30.2)” 11 2
1
:
1 -
T -
i 1
: (6w i
i
if (2.8Y
:
:
are percentages.
piratory tract (12 percent). The source could not be determined in 21 patients. Gram-negative bacilli accounted for 60.4 percent of the blood culture isolates, with E. coli predominating (22 percent) (Table IV). Gram-positive cocci comprised 30.2 percent of the isolates; the predominant organisms were S. aureus and Enterococcus faecalis. B. fragilis was the most common anaerobic isolate (Table IV). Polymicrobial infections occurred in 6 percent of cases. Overall survival was 60 percent; survival was 65.8 percent for community (home)-acquired bacteremia, 75 percent for nursing home-acquired bacteremia, and 52.8 percent for hospital-acquired bacteremia. Patients with gram-negative isolates had a greater survival than those with gram-positive isolates (65 percent versus 51.7 percent, p = NS). Among all groups, survival was greatest in patients with one underlying disease (69 percent), less with two diseases (52 percent), and least with three diseases (45 percent) (p = NS). The relationship of survival to the source of bloodstream infection is examined in Table V. Survival was greatest when the source was determined to be an intravascular device (78 percent), followed by genitourinary (70 percent) and intra-abdominal sources (62 percent). Patients with a lower respiratory tract source had the poorest survival (42 percent). When all sources were compared with regard to survival, no differences were found (x2(10) = 12.62, p = 0.25). Survival was also examined in regard to source of blood-
stream infection and appropriate therapy. Among patients treated appropriately, there was no differry2;ypg the sources with respect to survival (x2(9) = 0.19). However, among those patients - , treated appropriately, a trend towards increased survival in those with genitourinary tract (x2(1) = 2.423, p = 0.12) and lower respiratory tract (x2(1) = 2.310, p = 0.13) infections and a significantly decreased survival with endocarditis (x2( 1) = 5.506, p = 0.025) were noted when each group was compared with the whole group. When survival was examined for all sources in patients treated inappropriately, no differences were noted (x2(6) = 5.33, p = 0.50) and survival was not associated with WBC count (x2(2) = 1.55, p = 0.46). , For all levels of WBC count, age was not related to survival. When survival from all sources was examined, a significant decrease was noted in patients with a WBC count less than 5,000/mm3 (x2(4) = 10.182, p = 0.037); mortality was increased in patients with a lower respiratory tract source compared to all other sources (x2(1) = 8.123, p = 0.004). Survival was not significantly related to age categories in patients treated appropriately (62 percent) (x2(2) = 0.656, p = 0.72) or those treated inappropriately (50 percent) (x2(2) = 3.09, p = 0.21). However, in those over 85 years of age, four of five (80 percent) patients treated inappropriately died compared with four of 14 (29 percent) treated appropriately (x2(1) = 4.00, p = 0,046) (Table VI). The relationship of isolate to survival in patients treated appropriately was also examined (Table VII). April 1989
The American
Journal
of Medicine
Volume
86
381
BLOODSTREAM INFECTIONS IN THE ELDERLY / MEYERS ET AL TABLE V Survival by Source of Infection
and Appropriate
Site
Number
Genitourinary Lower respiratory tract Intra-abdominal Intravascular device Skin and wound Indeterminate Graft Endocarditis Mixed * Values in parentheses
Antimicrobial
Therapy Survival*
Appropriate Therapy*
Number
27
21
fi 9
ii ‘Y $Ej 3 (50)
2;
7
1;
7 (‘E] 0 (0) l(33)
i 3
‘: (‘;:; 3 (60)
ll(61)
2 (100) 0 (0) l(50)
32 2
are percentages.
TABLE VI Survival
Related to Age and Therapy
Age
Group
(yea@
Appropriate Therapy*
A
$1;; 85+
s Total * Values in parentheses
Inappropriate Therapy*
Total*
p Value
22137 (59) 20/33 (61) 10/14 (71)
214 (50) :y: {:A;
24/41(59) 25/40 11/19 (62) (58)
0.715 0.591 0.046
52184 (62)
8116 (50)
60/100 (60)
0.373
are percentages.
TABLE VII Survival by Organism:
Appropriate
Therapy
Appropriate Therapy*
Organism E. co/i S. aureus K. pneumoniae S. marcescens P. aeruginosa S. faecalis Other Total * Values in parentheses
versus Inappropriate
Inappropriate Therapy*
14122 (64)
~&(%y
April
16/24 (67)
“0;; $0’
gp
6/14 (43) 215 (40) 5/12 (42)
4/8 (50) 17/27 (63) 52/84 (62)
: y: I::{
8/16 (50)
13/29 (45) 60/100 (60)
are percentages.
Survival was greatest with K. pneumoniae (80 percent) and least with P. aeruginosa (40 percent). The different rates of survival among the organisms in the subset were not significant (p = 0.9). The relationship between inappropriate therapy for the specific isolates and survival was examined; no other differences in outcome were noted (x2(5) = 10.67, p = 0.058). However, all three patients with S. aureus treated inappropriately died, whereas five of 13 with other organisms died (x2(1) = 3.692, p = 0.055). Multiple logistic regression analysis revealed that community-acquired infection (p = O.O2),intravenous source (p = 0.04), and S. marcescens infection (p = 0.05) were related to survival; inappropriate therapy (p = 0.05) was negatively associated with survival. Adverse effects were seen most frequently with aminoglycosides (21 percent), and least with penicillins (12 percent) (Table VIII). Patients over 85 years of age had the fewest adverse effects. Among the age groups, differences in toxicity were not significant. When the relationship of age (less than or greater than
382
212 (100) O/3 C-9
l/l (100)
;\i gj
Total*
1989 The American
Journal
of Medicine
Volume
86
85) and antibiotic toxicity in patients treated either appropriately or inappropriately was examined, no significant differences in toxicity were noted.
COMMENTS Bacteremic infections in the elderly have been described in relation to the sources [12,16], specific places of acquisition [2,14,15], hospital [2] or long-termcare facility [13,17], and specific pathogens such as Klebsiella species [18], E. coli [10,19], 5’. pneumoniae [3,15], or other gram-negative bacilli [2,11]. Our study examines data in regard to all these factors. The clinical picture of bloodstream infections in the elderly often differs from that in younger patients [13]. Gleckman and Hibert [20] noted that 13 percent of elderly patients with positive blood culture results were afebrile, whereas others noted that 69 percent of patients with bacteremic pneumococcal pneumonia had a temperature greater than 101°F; data were not given for the other 28 patients [16]. Only one of 100 patients with a community source had a temperature
BLOODSTREAM
less than 37.2% [14]. Fever occurred equally in elderly and younger patients with pneumococcal bacteremia [15]. Our data revealed only two patients with temperatures less than or equal to 97°F and 10 percent with temperatures less than 99°F. It appears, then, that most elderly bacteremic patients have fever as a presenting sign. An altered mental status was found in 52 percent of our patients; it has also been reported as a presenting sign in patients with bacteremic pneumonia [13,15]. Sixty percent of our patients had WBC counts greater than 10,000/mm3, compared with 70 percent in the series reported by Windsor [13]. The urinary tract is the most common source of bacteremia (34 percent) in elderly patients admitted from the community setting [ 141 and was the source in 27 percent of our patients; in long-term-care facilities, 24 to 56 percent of bacteremias originate in the urinary tract [13,17,21]. Infection was secondary to in-dwelling intravascular catheters in 7 percent of our patients. Comparable statistics are not available from other studies of the elderly. The incidence of bacteremia from an intra-abdominal source in our series was 14 percent, whereas this source constituted 20 percent of community-acquired bacteremias [14] and 8 percent in the bacteremias in a geriatric unit [13]. Gram-negative bacilli accounted for 60 percent of isolates in our study, compared with 42 to 76 percent in the long-term-care setting [13,17,21]. In our study, E. coli (23 percent of total isolates, 37.5 percent of gram-negative isolates) and K. pneumoniae (9.4 percent of total isolates, 15.6 percent of gram-negative isolates) were the predominant gram-negative isolates; P. aeruginosa (5.7 percent of total isolates, 9.4 percent of gram-negative isolates) and Enterobacter (6.6 percent of total isolates, 10.9 percent of gramnegative isolates) occurred less often. Although the frequencies of K. pneumoniae isolates were similar, we had fewer patients with E. coli infections and more with P. aeruginosa and Enterobacter isolates than the study of McCue [2]. Gram-positive cocci accounted for 30 percent of the total isolates, compared with 24 to 58 percent reported from long-term-care settings [13,17,21]. Of these isolates, S. aureus was the most common (13.2 percent of total isolates, 43.8 percent of gram-positive isolates), similar to that described by Setia et al [17]. Group D streptococci or enterococci (E. faecalis) accounted for 10.4 percent, compared with 7 to 16 percent in longterm-care facilities [17,21] and 5 percent from the community setting [14]. Enterococci have been reported with increasing frequency as nosocomial pathogens [22]. Only four isolates were S. pneumoniae (3.7 percent) compared with 12 percent from the community setting [14]. This is not unexpected since S. pneumoniae is not a common hospital-acquired pathogen [23]. Anaerobic organisms accounted for 6.6 percent of isolates, compared with 2 percent [14] and 3 percent [17] found by others in the elderly. This incidence is less than that reported for hospitalized patients of all ages with positive results of blood cultures [24]. The overall mortality in our study was 40 percent: 34.2 percent in community-acquired, 25 percent in nursing home-acquired, and 47.2 percent in nosocomially acquired bacteremias. This compares to reports of 26 percent mortality for community-acquired bac-
INFECTIONS
IN THE ELDERLY
/ MEYERS ET AL
TABLEVIII Antibiotic Toxicity Antibiotic
Toxicity
Aminoglycosidest (age 65-74) (age 75-84) (age 85+)
Group A
c”
5123 7125 o/10 12/58
Total Cephalosporins Group A
(age 65-74) (age 75-84) (age 85+)
F
(22) (28) (0) (21)*
3115 (20) 6125 (24)
018(0)
9/48 (19)*
Total Penicillins Group A
(age 65-74) (age 75-84) (age 85+)
F
;j;;
I;;.“’
l/l1 (8) 5/43 (12)Z
Total * Values in parentheses are percentages. t Aminoglycoslde toxicity is dependent on either increased en or creatinine. or physlcian observer recognition. !Not significant.
blood urea nitro-
teremia [14], and 21 to 35 percent in the nursing home [17-221. It is not surprising that mortality was lower in patients with community-acquired gram-negative bacteremia, since the majority of these isolates were E. coli and Klebsiella species and originated in the urinary tract. In the long-term-care facility, 51 percent of all cases originated in the urinary tract [21], compared with 27 percent in our study, and probably accounts for the lower mortality rates noted in nursing homecompared with hospital-acquired bacteremias. Survival was examined with regard to source of bacteremia. It was greatest with a genitourinary source (70 percent), similar to that in reports from long-termcare facilities [17], but less than catheter-related urosepsis (80 to 92 percent) [2,12]. With a skin or soft tissue source, 50 percent survival was seen compared with 43 percent reported in the nursing home [17]. In both settings, bacteremic pneumonia was associated with a lower survival (42 percent). The relationship of survival to initial administration of appropriate therapy according to pathogen was examined. The highest rate of survival in patients given appropriate therapy was noted with K. pneumoniae (78 percent). Survival was 50 percent with E. fuecalis and P. aeruginosa; 55 percent mortality was noted with S. aureus. In a review of gram-negative bacteremia, McCue [2] noted that mortality was increased only in those patients receiving inappropriate therapy within the first 24 hours. In our study, none of the patients with S. aureus and one of four with E. faecalis who received inappropriate therapy survived. When survival was examined comparing appropriate and inappropriate therapy to the source of bacteremia, no significant differences were noted. This suggests that more practical methods of prevention of bacteremia, such as better attention to indwelling catheters, urinary drainage, and respiratory care, are warranted. The relationships of age, appropriate therapy, and survival revealed no differences in outcome when all groups were compared; however, in patients over 85 years of age, appropriate therapy yielded a higher sur-
April 1989
The American
Journal
of Medicine
Volume
86
383
BLOODSTREAM INFECTIONS IN THE ELDERLY / MEYERS ET AL
viva1 than inappropriate therapy. Patients 70 years and older who received appropriate single-drug antimicrobial therapy for gram-negative bacteremia had the lowest mortality [2]. P. aeruginosa, Enterobacter species, Serratia species, and enterococci accounted for 29.2 percent of our isolates. These pathogens are often antibiotic-resistant and more difficult to treat. The differences in reported mortality between our and previous studies may be due to the greater numbers of these isolates. The relationship of adverse effects to age and antibiotic class was examined. Toxicity was greatest with aminoglycosides (21 percent) and lower for cephalosporins (19 percent) and penicillins (12 percent); it was least in patients over 85 years of age receiving aminoglycosides (8 percent). For penicillins, no toxicity was seen in those over 85. Age has been associated with drug-induced nephrotoxicity and ototoxicity in patients treated with aminoglycosides [25]; in our study, only one of 13 aminoglycoside-treated patients greater than 85 years of age developed toxicity. This finding may relate to the caution used by the physicians treating this subgroup.
SUMMARY In a retrospective review of factors affecting outcome of bacteremia in the elderly, survival was best with isolates originating in the urinary tract and poorest with those from the respiratory tract. However, when all factors were considered simultaneously in a multiple regression analysis, appropriate therapy significantly increased survival. Factors associated with increased mortality in patients treated inappropriately were S. aureus infections, lower respiratory tract sources, and age greater than 85 years; with a WBC count less than 5,000/mm3, survival was decreased from all sources. Since the subgroup aged 85 years and older is the most rapidly growing segment of the elderly population, initial empiric therapy of bacteremia should be broad in scope and include coverage for S. aureus as well as gram-negative bacilli. Recent data from long-term-care facilities with respect to bacterial isolates reveal a trend toward more resistant gram-negative organisms and S. aureus, similar to those found in hospitalized patients [26-281. This suggests that empiric therapy should be approached in a similar manner for elderly patients in both the long-term-care facility and hospital settings. Use of new and more potent third-generation cephalosporins, monobactams, quinolones, and the carbapenems may increase future survival statistics; further studies are needed for confirmation of this premise.
384
April
1989
The American
Journal
of Medicine
Volume
86
REFERENCES 1. Chattopadhyay
9, Al-Zahawi M: Septicaemia and its unacceptably . high - mortality in the elderly. J Infect 1983; 7: 134-138. 2. McCue JD: Gram-negative bacillary bacteremia in the elderly: incidence, ecology, etiology, and mortality. J Am Geriatr Sot 1987; 35: 213-218. 3. Gruer LD, McKendrick MW, Geddes AM: Pneumococcal bacteremia-a continuing challenge. Q J Med 1984; 53: 259-270. 4. Vazquez F, Mendez FJ, Perez F, Mendoza MC: Anaerobic bacteremia in a general hospital: retrospective five-year analysis. Rev Infect Dis 1987; 9: 1038-1043. 5. Garibaldi RA, Nurse BA: Infections in the elderly. Am J Med 1986; 81(suppl 1A): 53-58. 6. Miller PJ, Wenzel RP: Etiologic organisms as independent predictors of death and morbidity associated with bloodstream infections. J Infect Dis 1987; 156:
471-477. 7. Powers DC, Nagel JE, Hoh J, Adler WH: Immune function in the elderly. Postgrad Med 1987; 81: 355-359. 8. Murasko DM, Nelson BJ, Silver R, Matour D, Kaye D: Immunologic response in an elderly population with a mean age of 85. Am J Med 1986; 81: 612-618. 9. Kovar M: Health of the elderly and use of health services. Public Health Rep
1977: 92: 9-19. 10. Jonas M. Cunha BA: Bacteremic EscherichiacoEpneumonia. Arch Intern Med 1982; 142: 2157-2159. 11. Phair JP, Bassaris HP, Williams JE, Metzger E: Bacteremic oneumonia due to gram-negative bacilli. Arch Intern Med 198% 143: 2147-2144. 12. Gleckman R, Blagg N, Hibert D. et al: Catheter-related urosepsis in the elderly: a prospective study of community derived infections. J Am Geriatr Sot 1982; 30:
255-257. 13. Windsor ICM: Bacteraemia in ageriatric
unit. Gerontology 1983; 291125-130. 14. Esposito AL, Gleckman RA, Cram S, Crowley M, McCabe F, Drapkin MS: Community-acquired bacteremia in the elderly: analysis of one hundred consecutive episodes. J Am Geriatr Sot 1980; 28: 315-319. 15. Esposito AL: Community-acquired bacteremic pneumococcal pneumonia. Arch Intern Med 1984; 144: 945-948. 16. Murphy TF, Fine BC: Bacteremic pneumococcal pneumonia in the elderly. Am J Med Sci 1984; 288: 114-118. 17. Setia U. Serventi I, Lorenz P: Bacteremia in a long-term facility. Arch Intern Med 1984; 144: 1633-1635. 18. Montgomerie JZ. Ota JK: Klebsiella bacteremia. Arch intern Med 1980; 140:
526-527. 19. Berk SL, Neumann P, Holtsclaw S, Smith JK: Escherichia cogpneumonia in the elderly. Am J Med 1982; 72: 899-902. 26. Gleckman R, Hibert D: Afebrile bacteremia. JAMA 1982; 248: 1478-1481. 21. Rudman D, Hontanosas A, Cohen Z, Mattson DE: Clinical correlates of bacteremia in a Veterans Administration extended care facility. J Am Geriatr Sot 1988;
36: 726-732. 22. Morrison Al Jr, Wenzel RP: Nosocomial coccus: ten years experience at a university
urinary tract infection due to Enter@ hospital. Arch Intern Med 1986; 146:
1549-1551. infection surveillance. MMWR 1986; 35: 17SS-25%. 24. Weinstein MP, Reller B, Murphy JR, Lichenstein KA The clinical significance of
23. Centers for Disease Control: Nosocomial
positive blood cultures: a comprehensive analysis of 500 episodes of and fungemia in adults. I. Laboratory and epidemiologic observations. Dis 1983; 5: 35-53. 25. Moore RD, Smith CR, Lipsky JJ. Mellitis ED, Lietman PS: Risk nephrotoxicity in patients treated with aminoglycosides. Ann Intern
bacteremia Rev Infect factors tor Med 1984;
loo: 352-357. 26. Gaynes RP, Weinstein RA, Chamberlin W, Kabins SA: Antibiotic-resistant flora in nursing home patients admitted to the hospital. Arch Intern Med 1985; 145: 1804-
1807. 27. Scheckler WE, Peterson PJ: Infections and infection control among residents of eight rural Wisconsin nursing homes. Arch Intern Med 1986; 146: 1981-1984. 28. Norman DC, Castle SC, Cantrell M: Infections in the nursing home. J Am Geriatr Sot 1987; 35: 796-805.