A.B. Johan Groeneveld MD, PhD, FCCP, FCCM
, a, 
aVrije Universiteit Medical Centre and Institute for Cardiovascular Research at the Vrije Universiteit, Amsterdam, The Netherlands
Available online 24 January 2009.
Background
Risk factors for hospital-acquired infection and attributable mortality in surgical and critically ill patients are well-known. We sought to identify factors associated with increased mortality from hospital-acquired infections as compared with community-acquired infections in patients with new-onset fever and a presumed infectious focus (n = 212), in a department of internal medicine.
Methods
Demographic, clinical, and laboratory variables were studied for 2 days after inclusion. Septic shock and outcome were monitored for up to 7 and 28 days after inclusion, respectively.
Results
Of the 212 patients, 54 had hospital-acquired and 158 community-acquired infection, with septic shock rates of 15% and 4% and mortality rates of 24% and 6% (P = .001), respectively. Prior neurologic disease was associated with death. Patients with hospital-acquired infection had more often (intravascular) devices and underwent more often interventions, had a different distribution of infectious foci, and had more often bacteremia. Bacteremia-associated septic shock was associated with nonsurvival in both infection groups. The causative agents were not associated with outcome, and the clinical and laboratory host response associated with nonsurvival generally did not differ among infection groups.
Conclusion
Our data suggest that hospital-acquired infections carry a higher crude mortality rate than community-acquired infection in febrile medical patients, mainly because of more frequent use of devices and hospital interventions and resultant bacteremia and septic shock, rather than by differences in underlying diseases, causative agents, and clinical and laboratory host responses. The observations thus emphasize the continued importance of preventive measures on medical wards of our hospital and can be used for comparison with future studies.
Article Outline
Hospital-acquired (bloodstream) infections may increase over decades and may affect surgical or critically ill patients more often than medical patients so that the impact in the latter is relatively unknown.[1], [2], [3], [4], [5], [6], [7], [8], [9], [10] and [11] Indeed, nosocomial infection surveillance systems often mainly focus on surgical site and catheter-related infections.[11] and [12]
Risk factors for hospital-acquired infections, according to the literature, which is often limited to hospital-acquired (bloodstream) infection compared with noninfected controls, include old age, prior underlying diseases, changing patterns of causative agents and resistance patterns, invasive interventions and devices, immune depression, and others.[2], [3], [7], [9], [11], [13], [14], [15], [16], [17], [18], [19] and [20] Hospital-acquired infection is a risk factor for mortality, at least in some but not in all studies.[1], [2], [3], [5], [9], [13], [14], [16], [17], [20], [21] and [22] Predictors are controversial and may include more severe underlying disease and impaired host defense,[2], [5], [9], [17], [20] and [21] septic shock,[9], [16] and [21] increased virulence and antimicrobial resistance of hospital-acquired microbes, and inappropriate antimicrobial therapy.[5], [14], [16] and [21] More balanced comparisons of hospital- with community-acquired infections, which may obviate overestimation of risk factors and comorbidity in comparisons with noninfected controls, are rare,[9], [16], [18], [20] and [21] and only few studies address external versus patient-bound factors contributing to death in hospital-acquired infection.[5], [9], [13], [20] and [21] Understanding of the factors associated with mortality can have important preventive and therapeutic implications.
We therefore prospectively studied the demographic, clinical, and laboratory variables commonly used to evaluate patients with infection and 28-day mortality of hospitalized medical patients with fever and a clinically presumed focus of infection. The hypothesis was that increased mortality in hospital-acquired than in community-acquired infections is mainly caused by patient-bound rather than external factors and that these factors may help predicting such increase. Our results suggest the opposite, however.
Patients and methods
Patients
Three hundred consecutive patients with new-onset fever (body temperature >38.0°C axillary or 38.3°C rectally) admitted to the Department of Internal Medicine in a university hospital in a 1-year period (1995) were described elsewhere,23 but hospital- and community-acquired infection were not previously compared. We now report on the 212 patients in whom a clinical focus of infection was presumed and summarize the protocol. The study was approved by the local committee on ethics. All patients or their closest relatives gave informed consent before inclusion. The exclusion criteria were pregnancy, shock, and a life expectancy of less than 24 hours. Moreover, patients who had received cytokines (eg, interferon-γ, interleukin-2) or cytostatic drugs for treatment of solid tumors and malignant hematologic diseases were excluded. Patients were taken care of by physicians not involved in the study and who ordered blood tests and prescribed antibiotics and supportive measures. The clinical judgment by the treating physician was supplemented, if necessary, by imaging techniques, for assessing the clinically presumed infectious focus. At inclusion, demographic data were recorded. These included age, gender, and premorbidity, including active malignancies, neurologic disease, cardiovascular disease, respiratory disease, endocrine disease, genitourinary disease, autoimmune disease, skin and bone/joint disease, gastrointestinal disease, and surgery within 2 months prior to inclusion. The ICD-9 definitions were used to describe the disease states. We recorded factors that may predispose to infection, such as use of immunocompromising drugs (cytostatic drugs, corticosteroids), antimicrobial drugs, and devices and interventions done from 7 days prior up to study inclusion. Fever was considered hospital acquired if developing at least 72 hours after admission to the hospital. An estimation of the time interval between development of fever and inclusion in the study was made. The clinically presumed focus of infection and all local and blood cultures results during a follow up period of 7 days after inclusion were taken into study. On the day of admission, local specimens for microbiologic evaluation were collected, depending on the clinically presumed focus of infection, as judged by the treating physician. Local specimens were processed using standard procedures. Specific stains for mycobacteria (n = 1) or yeasts/fungi (n = 11) and cultures were performed on clinical grounds. Two blood cultures were obtained by venipuncture at inclusion as part of the protocol. Supplementary blood cultures were done thereafter as judged by the treating physician. Blood cultures were processed using delayed vial entry bottles for aerobic and anaerobic cultures and the Bactec 9120/9240 automatic analyzers (Becton Dickinson, Erembodegem, Belgium). Bottles were incubated for a maximum of 7 days. If the analyzers showed growth, Gram's stains were prepared, and identification and sensitivity cultures were processed. Blood cultures containing Staphylococcus epidermidis were considered contaminated if only 1 bottle revealed growth and there were no indwelling vascular catheters. All other positive blood cultures were thought to reflect bloodstream infection. The clinically presumed infection was considered proven if (local/blood) microbiologic results were positive and if the treating physician decided to prescribe or continue antimicrobial therapy based on these results. We classified infections into respiratory, including pneumonia and airway infections; skin/bone/joint infections; genitourinary tract infections; gastrointestinal; abdominal and hepatobiliary tract infections; intravascular infections, including catheter-related infection and endocarditis; and central nervous system infections. We recorded the performance status prior to admission as the World Health Organization score (0-4, with 4 indicating greatest disability).
We measured, at inclusion and both on the first and second day after the day of inclusion, clinical variables (temperature, respiratory rate, heart rate, mean arterial blood pressure, score on the Glasgow Coma Scale [GCS], Karnofsky performance score [0%-100%], body mass index [BMI]) and laboratory variables including leukocyte counts, platelet count, lactate, creatinine, albumin, bilirubin, alkaline phosphatase, alanine aminotransferase, aspartate aminotransferase, and lactate dehydrogenase levels in blood or plasma. All blood samples were collected in standard tubes (Becton and Dickinson, Erembodegem, Belgium). Laboratory variables were determined using a Symsec SE 9000 analyzer (Kolbe, Japan) and a Hitachi 747 automatic analyzer (Tokyo, Japan).
Patients were followed for 7 days for development of septic shock and for 28 days for outcome. A 28-day period is commonly used to estimate sepsis-related mortality. Septic shock was defined as a fall in arterial blood pressure below 90 mm Hg (or a systolic decrease >40 mm Hg in previous hypertension) in the presense of systemic inflammatory response syndrome criteria, with fever (or hypothermia), tachycardia, tachypnea, and leukocytosis (or leukopenia).24 In the presence of a presumed infection, patients meeting systemic inflammatory responses syndrome criteria were considered to suffer from sepsis.24 Patients discharged from the hospital within the follow-up period were classified as survivors.
Statistical analysis
Among the 3 determinations of each continuous variable in each patient of this study, the day 0, peak and nadir values were selected. Categorical variables were compared using the Fisher exact and χ2 tests, and odds ratio (OR) and 95% confidence intervals (CI) were calculated, where indicated. We determined with help of multiple logistic regression (forward method on the basis of likelihood ratio) whether there was a difference in categoric variables according to infection and outcome groups, respectively, and their respective first order interactions. A similar procedure was done for continuous variables, after logarithmic transformation to reach normal distributions (verified by the Kolmogorov-Smirnov test), using a general linear model and univariate analysis of variance (ANOVA), with time to inclusion as a covariate. Multiple backward (based on likelihood ratio) logistic regression analysis was done for prediction of nonsurvival in the infection groups by the smallest set of variables reaching statistical significance in univariate analyses. Goodness of fit was evaluated by the Hosmer-Lemeshow test. The OR and 95% CI are reported. Exact P values <.05 and >.001 are given, and P < .05 was considered as statistically significant. Data were expressed as median (range).
Results
Patient characteristics
Twenty-five percent of patients had hospital-acquired infection (Table 1). The overall rate of septic shock was 7%, and the 28-day mortality rate was 11%, and septic shock occurred more often in hospital-acquired infections and predisposed to mortality (Fisher exact test, P = .009 or less). When death was considered, on clinical grounds, to be caused, at least in part, by infection (n = 27), mortality was again higher in hospital-acquired infections (Fisher exact test, P < .001). The OR for septic shock in hospital-acquired versus community-acquired infection was 4.4 (95% CI: 1.5-13.3), whereas the OR for mortality with hospital-acquired (24%) versus community-acquired infection (6%) was 4.7 (95% CI: 1.9-11.5). Seventy-five percent (6/8) of nonsurviving septic shock patients had hospital-acquired infection and 25% (2/8) community-acquired infection.
Characteristics of patients and infections according to outcome
| Hospital-acquired | Community-acquired infection | |||
|---|---|---|---|---|
| Nonsurvivors n = 13 | Survivors n = 41 | Nonsurvivors n = 10 | Survivors n = 148 | |
| Age, yr | 62 (39-92) | 67 (27-92) | 74 (38-86) | 63 (17-97) |
| Gender, m/f | 7 (53) | 21 (51) | 4 (40) | 73 (49) |
| Origin, home | 12 (92) | 36 (89) | 6 (60) | 137 (93) |
| Nursing home | 1 (8) | 0 | 3 (30) | 3 (2) |
| Other hospital | 0 | 2 (5) | 0 | 1 (0) |
| Miscellaneous | 0 | 3 (7) | 1 (10) | 7 (5) |
| Premorbidity* | ||||
| Malignancies | 4 (31) | 10 (24) | 1 (10) | 27 (18) |
| Neurologic disease† | 4 (31) | 3 (7) | 4 (40) | 13 (9) |
| Cardiovascular disease | 5 (38) | 12 (29) | 1 (10) | 33 (22) |
| Endocrine disease | 4 (31) | 11 (27) | 3 (30) | 33 (22) |
| Genitourinary disease‡ | 3 (23) | 8 (20) | 1 (10) | 14 (9) |
| Autoimmune disease | 1 (8) | 4 (10) | 3 (30) | 12 (8) |
| Skin disease | 1 (8) | 5 (12) | 1 (10) | 9 (6) |
| Gastrointestinal disease | 0 | 2 (5) | 1 (10) | 1 (0) |
| Recent surgery | 2 (15) | 7 (17) | 0 | 19 (13) |
| Risk factors | ||||
| IC drugs | 4 (31) | 12 (29) | 1 (10) | 32 (22) |
| Prior antibiotics <7> | 4 (31) | 12 (29) | 2 (20) | 34 (23) |
| Devices/interventions§ | 6 (46) | 22 (54) | 1 (10) | 21 (14) |
| Biliary endoprosthesis | 0 | 2 (5) | 0 | 3 (2) |
| Thoracic drainage | 1 (8) | 3 (7) | 0 | 2 (1) |
| Urinary catheter | 2 (15) | 1 (2) | 0 | 3 (2) |
| Intravascular catheter | 0 | 7 (17) | 0 | 5 (3) |
| Hemodialysis | 0 | 2 (5) | 0 | 1 (0) |
| Peritoneal dialysis | 1 (8) | 0 | 0 | 2 (1) |
| Miscellaneous | 2 (15) | 7 (17) | 1 (10) | 5 (3) |
| Presumed focus of infection | ||||
| Respiratory | 6 (36) | 13 (32) | 4 (40) | 67 (45) |
| Skin/bone/joints | 4 (31) | 8 (20) | 3 (30) | 24 (16) |
GU tract | 2 (15) | 2 (5) | 1 (10) | 28 (19) |
| GI tract | 1 (8) | 12 (29) | 2 (20) | 21 (14) |
| Intravascular¶ | 0 | 6 (15) | 0 | 4 (3) |
| Central nervous system | 0 | 0 | 0 | 4 (3) |
| Proven infection# | 11 (85) | 18 (44) | 9 (90) | 68 (46) |
| Positive local culture/bacteremia** | 6 (46)/5 (38) | 6 (15)/12 (29) | 7 (70)/2 (20) | 41 (28)/27 (18) |
| Bacteria in local culture | ||||
| Enterobacteriaceae | 2 (15) | 3 (7) | 3 (30) | 20 (14) |
| Escherichia coli | 2 (15) | 2 (5) | 2 (20) | 10 (7) |
| Moraxella catarrhalis | 0 | 1 (2) | 0 | 1 (0) |
| Haemophilus species | 0 | 1 (2) | 0 | 4 (3) |
| Staphylococcus species | 2 (15) | 2 (5) | 1 (10) | 8 (5) |
| S aureus | 1 (8) | 1 (2) | 1 (10) | 7 (5) |
| Streptococcus species | 1 (8) | 0 | 0 | 2 (1) |
| S pneumoniae | 1 (8) | 0 | 0 | 2 (1) |
| Polymicrobial and miscellaneous | 5 (38) | 6 (15) | 4 (40) | 18 (12) |
| Bacteremia | ||||
| Enterobacteriaceae | 1 (8) | 3 (7) | 0 | 13 (9) |
| E coli | 1 (8) | 1 (2) | 0 | 8 (5) |
| Bacteroides species | 1 (8) | 0 | 1 (10) | 1 (0) |
| Staphylococcus species | 2 (15) | 4 (10) | 1 (10) | 8 (5) |
| S aureus | 1 (8) | 1 (2) | 1 (10) | 7 (5) |
| Streptococcus species | 1 (8) | 2 (5) | 0 | 5 (3) |
| S pneumoniae | 1 (8) | 0 | 0 | 3 (2) |
| Polymicrobial | 0 | 3 (7) | 0 | 0 |
| Sepsis | 13 (100) | 35 (85) | 10 (100) | 148 (94) |
| Septic shock†† | 6 (46) | 2 (5) | 2 (10) | 4 (3) |
| Time to inclusion, h‡‡ | 4 (1-28) | 2 (1-74) | 11 (1-124) | 28 (1-383) |
NOTE. Median (range) or number (percentage), where appropriate.
IC, immunocompromising; GU, genitourinary; GI, gastrointestinal.
Logistic regression: †P = 0.001 for interaction between infection and outcome groups: ‡P = .040, §P < .001, P = .050, and ¶P < .001 for infection groups, respectively; #P <>**P = .049 for infection groups; ††P < .001 for outcome groups; ANOVA: ‡‡P = .018 for infection groups.
Patients with hospital-acquired infection had more often underlying genitourinary disease, but outcome groups did not differ in this respect. In contrast, nonsurvivors of both infection groups had more often neurologic disease than survivors (Fisher exact test, P = .001) but particularly in the community-acquired group. Foci of hospital-acquired infections were different from those in community-acquired infections (χ2 test, P = .001). Intravascular infections in particular were more frequent among hospital-acquired than community-acquired infections (Fisher exact test, P = .001). Hospital-acquired infections were more often associated with use of devices or interventions in the hospital, whereas the distribution differed among infection groups (χ2 test, P < .001) and the use related to mortality (Fisher exact test, P = .011).
Fifty-three percent of presumed infections were proven by positive cultures in hospital-acquired infections and 49% in community-acquired infections and more frequently in nonsurvivors than in survivors of either infection group. Hospital-acquired infections were more often associated with positive (blood) cultures than community-acquired infections, but the distribution among causative microorganisms did not differ. In both groups, bacteremia was more common in intravascular, soft tissue, and genitourinary than in respiratory and gastrointestinal infections (P = .006 or less). Fifty percent (4/8) of patients with hospital-acquired infection and septic shock had bacteremia and 67% (4/6) of community-acquired infections with septic shock so that bacteremia equally predisposed to septic shock in both infection groups (Fisher exact test, P = .003). The time interval between onset of fever and inclusion was higher for community-acquired infection but did not differ among outcome groups.
Clinical and laboratory variables
Some values differed between nonsurviving and surviving groups and hospital-acquired and community-acquired infection groups. Peak respiratory rate was particularly elevated in nonsurvivors of hospital-acquired infection, and the nadir GCS score was lower in nonsurvivors of community-acquired infection. In contrast to clinical variables, there were no interactions for infection groups and outcome for laboratory variables, except for lower day 0 platelet counts particularly in nonsurviving patients with hospital-acquired infection (Table 2 and Table 3). There was no effect of the time interval between fever onset and inclusion.
Clinical variables
| Hospital-acquired | Community-acquired infection | |||
|---|---|---|---|---|
| Nonsurvivors n = 13 | Survivors n = 41 | Nonsurvivors n = 10 | Survivors n = 148 | |
| Temperature, °C, day 0 | 39.1 (38.4-39.6) | 38.7 (38.2-40.0) | 38.9 (38.4-40.5) | 39.0 (38.0-41.3) |
| Peak | 39.2 (38.4-40.2) | 39.2 (38.3-40.4) | 39.5 (38.8-40.8) | 39.2 (38.3-41.3) |
| Respiratory rate, b/min, day 0* | 36 (10-50) | 24 (12-33) | 28 (15-48) | 24 (13-48) |
| Peak† | 36 (15-54) | 24 (12-40) | 30 (16-56) | 30 (14-48) |
| Heart rate, b/min, day 0 | 114 (60-140) | 103 (72-154) | 104 (84-120) | 100 (56-160) |
| Peak‡ | 120 (84-140) | 108 (80-170) | 116 (94-130) | 102 (76-160) |
| MAP, mm Hg, day 0§ | 90 (72-110) | 90 (73-123) | 79 (57-113) | 93 (57-157) |
| Nadir | 78 (50-100) | 83 (62-120) | 76 (57-93) | 83 (57-120) |
| GCS, day 0# | 15 (8-15) | 15 (10-15) | 11 (5-15) | 15 (9-15) |
| Nadir¶ | 15 (3-15) | 15 (3-15) | 11 (3-15) | 15 (8-15) |
| WHO score at home# | 2 (0-4) | 1 (0-3) | 2 (0-4) | 1 (0-4) |
| Karnofsky score, nadir** | 70 (50-100) | 80 (40-100) | 65 (40-100) | 90 (30-100) |
| BMI, kg/m2 | 24.6 (18.3-30.1) | 22.8 (17.1-34.7) | 25.6 (23.1-35.1) | 23.8 (15.9-45.3) |
NOTE. Median (range) or number (%), where appropriate.
MAP, mean arterial pressure; GCS, Glasgow Coma Scale; BMI, body mass index.
ANOVA: *P = .018 for outcome groups, P = .009 for interaction between infection and outcome groups; †P < .001 for outcome groups, P = .020 for interaction between infection and outcome groups; ‡P = .030, §P = 0.049 for outcome groups; #P < .001 for infection, outcome groups, and their interaction; ¶P < .001 for outcome groups and P = .019 for interaction with infection groups; #P = .016 for outcome groups; **P = .003 for outcome groups. No effect of time to inclusion.
Laboratory variables
| Hospital-acquired | Community-acquired infection | |||
|---|---|---|---|---|
| Nonsurvivors n = 13 | Survivors n = 41 | Nonsurvivors n = 10 | Survivor n = 148 | |
| Leukocytes, 109/L, day 0* | 17.0 (3.5-38.4) | 11.7 (2.6-41.4) | 16.6 (8.5-25.1) | 12.7 (0.3-42.3) |
| Peak† | 17.0 (3.6-41.9) | 12.0 (2.9-43.1) | 18.1 (9.9-32.6) | 12.8 (0.4-42.3) |
| Nadir‡ | 17.0 (3.4-33.2) | 8.3 (2.2-41.4) | 11.4 (3.1-24.1) | 8.9 (0.3-35.7) |
| Platelets, 109/L, day 0§ | 200 (30-321) | 256 (86-631) | 261 (114-361) | 201 (3-889) |
| Nadir | 182 (30-296) | 226 (70-631) | 140 (9-348) | 184 (3-825) |
| Lactate, mmol/L, day 0¶ | 1.5 (0.8-3.7) | 1.3 (0.6-3.2) | 1.7 (1.0-4.4) | 1.4 (0.6-5.4) |
| Peak# | 1.9 (1.2-5.2) | 1.6 (0.8-3.2) | 2.0 (1.0-4.4) | 1.7 (0-5.4) |
| Creatinine, μmol/L, day 0 | 131 (37-683) | 104 (50-609) | 111 (73-202) | 92 (45-1,885) |
| Peak | 141 (40-789) | 110 (51-987) | 142 (73-255) | 94 (45-1,885) |
| Albumin, g/L, day 0** | 22 (13-28) | 27 (9-41) | 23 (12-29) | 28 (15-43) |
| Nadir†† | 20 (13-28) | 26 (7-37) | 20 (12-24) | 24 (14-40) |
| Bilirubin, μmol/L, day 0 | 18 (6-290) | 9 (3-404) | 10 (7-56) | 11 (4-127) |
| Peak | 17 (6-302) | 10 (3-446) | 14 (7-68) | 13 (4-126) |
| AF, U/L, day 0‡‡ | 94 (25-444) | 77 (39-598) | 57 (38-163) | 67 (31-837) |
| Peak | 107 (29-444) | 79 (47-598) | 59 (38-163) | 75 (31-862) |
| ALAT, U/L, day 0¶¶ | 22 (11-362) | 17 (3-81) | 12 (1-129) | 12 (1-383) |
| Peak | 27 (12-769) | 19 (3-81) | 15 (3-151) | 17 (2-873) |
| ASAT, U/L, day 0 | 22 (12-304) | 17 (4-74) | 17 (8-86) | 16 (5-408) |
| Peak | 30 (12-1,219) | 18 (7-74) | 20 (9-86) | 19 (5-613) |
| LDH, U/L, day 0## | 300 (177-1022) | 230 (121-669) | 260 (158-817) | 248 (123-3,615) |
| Peak*** | 347 (177-3,650) | 235 (111-669) | 402 (163-859) | 261 (131-4,765) |
NOTE. Median (range) or number (%), where appropriate.
AF, alkaline phosphatase; ALAT, alanine aminotransferase; ASAT, aspartate aminotransferase; LDH, lactate dehydrogenase.
ANOVA: *P = .017, †P = .012, and ‡P = .003, for outcome groups, respectively; §P = .034 for interaction between infection and outcome groups; P = .048, ¶P = .030, #P = .048, **P < .001, and ††P = .001 for outcome groups; ‡‡P = .025, P = .018, and ¶¶P = .034 for infection groups, respectively; ##P = .031 and ***P = .002 for outcome groups. No effect of time to inclusion.
Multivariate analysis
Death in hospital-acquired infection was predicted by a set of variables consisting of the presence of prior neurologic disease, septic shock, peak respiratory rate, and lactate dehydrogenase (P = .033 or less), independently of other clinical and laboratory variables that reached statistical significance in univariate analyses (Hosmer-Lemeshow test χ28 = 2.9, P = .94). For community-acquired infection, prior neurologic disease, nadir GCS score, low mean arterial blood pressure at day 0, septic shock, and low nadir albumin levels were predictive (P = .048 or less), independently of other clinical and laboratory variables (Hosmer-Lemeshow test χ28 = 1.3, P = .99). This indicates that prior neurologic disease and development of septic shock were major, independent, and common predictors of death.
For predicting death in the overall population, septic shock (OR, 11.7; 95% CI: 2.3-59.5), prior neurologic disease (OR, 10.2; 95% CI: 2.1-50.1), low nadir GCS score (OR, 1.42; 95% CI: 1.14-1.78), and high peak respiratory rate (OR, 1.11; 95% CI: 1.02-1.20) independently (P = .011 or less) contributed (Hosmer-Lemeshow test χ28 = 1.5, P = .99), whereas hospital-acquired infection did not (P = .074). This suggests that a higher crude mortality in hospital- than community-acquired infection can be largely explained by a higher rate of septic shock and respiratory distress.
Discussion
Our data suggest that higher rates of bacteremia, septic shock, and resultant mortality are not primarily caused by different underlying diseases, causative microorganisms, and clinical and laboratory host responses but by more frequent use of devices and hospital interventions and thus different foci in hospital-acquired than in community-acquired infections in febrile medical patients.
Most of the studies on hospital-acquired infections suggest attributable mortality (of at least 15%), but mechanisms are still poorly understood, and many studies address bloodstream infections or compare (causes of) mortality with noninfected hospitalized patients only.[1], [2], [5], [9], [15], [16], [18], [20], [21] and [22] In any case, reported mortality rates of hospital-acquired infections are consistent with ours, whereas independency of age and gender is not supported by some other studies.[2], [4], [5], [9], [13], [15], [16], [20] and [21] Our results also suggest low attributable mortality of hospital-acquired infection, independent of septic shock, in contrast to previous reports.[6], [9], [20] and [21] Hospital-acquired (bloodstream) infections in adults may be associated with malignancies and neurologic disease, which predipose to infections[2], [4], [5], [13], [15], [16], [17] and [20] but not in our study. In contrast, only a history of genitourinary disease was more frequent among patients with hospital-acquired infection in our cohort. In both types of infection, the presence of prior neurologic disease was nevertheless a poor prognostic sign, in accordance with the literature.[17] and [20] Indeed, premorbidity does not seem to have largely determined the difference in mortality between hospital- and community-acquired infections. Also, malnutrition may not have greatly predisposed to hospital-acquired infection, in contrast to the literature,25 as judged from similar BMI and albumin levels between hospital- and community-acquired infection groups, even though the latter was lower in nonsurvivors. Conversely, the characteristics of community-acquired infection in our hospitalized patients are in line with the literature.26
We studied hospital-acquired infection and not fever only because approximately 50% of hospital-onset fevers may be of noninfectious origin.[1], [2] and [3] Conversely, the foci of hospital-acquired infections agree with the literature on medical patients.[1], [2], [3] and [11] Although respiratory infections were predominant in both hospital-acquired and community-acquired infections, there was an altered distribution from respiratory to intravascular hospital-acquired infections, as reported previously.[9] and [13] In any case, the overall number of positive cultures and thus microbiologically proven infections did not differ among infection groups. In contrast, our data support the relatively high incidence of hospital-acquired bloodstream infections and associated dismal outcomes, reported by others, even when not compared with community-acquired infection.[2], [5], [13] and [18] The higher rate of septic shock predisposing to death, in hospital-acquired versus community-acquired infections, can be largely attributed to differences in frequency of devices/interventions, and thus infectious foci and bacteremia in the hospital, rather than to differences in underlying diseases and causative microorganisms, although there was a trend for more gram-positive bacterial involvement of hospital-acquired infections, in agreement with some previous observations.[1], [2], [3], [5], [9], [10], [11], [13], [14], [16], [18], [19], [20] and [21]
Surprisingly, body temperature did not predict mortality as it did in other[2], [9] and [26] but not all studies.[3] and [23] The only clinical variables showing interaction between infection groups on the one hand and outcome groups on the other were relatively higher respiratory rates (and GCS scores) in hospital-acquired infections leading to death, confirming other reports.9 Even though some laboratory values differed between infection groups, the prognostic significance of most of them was similar among the groups. The clinical and laboratory associations with outcome argue against a reduced clinical host response in hospital-acquired than in community-acquired infection, in contrast to prior suggestions.9 Overall, laboratory values hardly contributed to prediction of outcome of both infection groups.
Together with different foci of infections associated with more frequent use of in-hospital interventions and devices, a higher frequency of bacteremia-related septic shock and resultant mortality could have been caused by less appropriate antibiotic treatment, a difference in study timing, or combinations. The limitations of our study thus include absence of evaluation of bacterial resistance and appropriate antibiotic therapy. Because drug resistance may also occur in community-acquired infections and all patients were treated in the same medical wards, we may assume equally appropriate antibiotic coverage. In Denmark, similar Escherichia coli strains causing hospital- and community-acquired bacteremia suggest that hospital-acquired bacterial strains belong to the patients normal flora.27 We may inadvertently have classified health care-related as community-acquired infections15 but used the 72 (rather than the more common 48) hours after admission to the hospital as the criterion to separate hospital- from community-acquired infection because we were primarily interested in infections truly acquired in the medical wards of our hospital. Indeed, national nosocomial infection surveillance programs often mainly focus on surgical infections, including the “PREZIES” in The Netherlands.12 In any case, the number of patients coming from other health care settings did not differ among our infection groups. We cannot exclude that the low attributable mortality in hospital-acquired infection and the predominance of external over patient-bound factors therein relates in part to study design, ie, comparison with community-acquired infection rather than with noninfected controls, and thereby explains discrepancy with some literature suggesting predominance of patient-bound factors and comorbidity in acquiring and dying from hospital-acquired infections.[4], [6], [9], [20] and [21] In view of small numbers, we also cannot definitively conclude that septic shock acquired in the hospital carries a similar mortality than that following community-acquired infection. Our results may appear outdated, at first glance, but may nevertheless serve benchmarking purposes and usable for comparison with future studies on medical patients. Moreover, the study suggests the importance of the nature of the control group for comparison with hospital-acquired infection, and there is no reason to believe that this is time dependent. Finally, our results may not be translated to other hospitals because of geographical differences.[6], [8], [9], [10], [11], [12], [13], [19], [20] and [21]
In conclusion, our data suggest that hospital-acquired infections carry a higher crude mortality rate than community-acquired infections because of in-hospital devices/interventions and increased bacteremia-induced septic shock, rather than by differences in underlying diseases, causative agents, and host responses, in febrile patients hospitalized in medical wards. Indeed, prior neurologic disease and septic shock are major common predictors of death. Predominance of external over patient-bound factors in mortality emphasizes the continued importance of preventive measures, if feasible, which may include improved asepsis techniques and others to prevent catheter-related infection.
References
PDF (835 K) | View Record in Scopus | Cited By in Scopus (16) PDF (955 K) | View Record in Scopus | Cited By in Scopus (25) PDF (118 K) | View Record in Scopus | Cited By in Scopus (17) PDF (241 K) | View Record in Scopus | Cited By in Scopus (40) PDF (149 K) | View Record in Scopus | Cited By in Scopus (2) PDF (121 K) | View Record in Scopus | Cited By in Scopus (4) PDF (287 K) | View Record in Scopus | Cited By in Scopus (29) PDF (598 K) | View Record in Scopus | Cited By in Scopus (31)Conflicts of interest: The author reports no conflicts of interest.
0 comments:
Post a Comment