RESUMO
BACKGROUND: The storage duration of platelet (PLT) units is limited to 5 to 7 days. This study investigates whether PLT storage duration is associated with patient outcomes in critically ill patients. STUDY DESIGN AND METHODS: This study was a retrospective analysis of critically ill patients admitted to the intensive care unit (ICU) of two hospitals in Australia who received one or more PLT transfusions from 2008 to 2014. Storage duration was approached in several different ways. Outcome variables were hospital mortality and ICU-acquired infection. Associations between PLT storage duration and outcomes were evaluated using multiple logistic regression and also by Cox regression. RESULTS: Among 2250 patients who received one or more PLT transfusions while in the ICU, the storage duration of PLTs was available for 64% of patients (1430). In-hospital mortality was 22.1% and ICU infection rate 7.2%. When comparing patients who received PLTs of a maximum storage duration of not more than 3, 4, or 5 days, there were no significant differences in baseline characteristics. After confounders were adjusted for, the storage duration of PLTs was not independently associated with mortality (4 days vs. ≤3 days, odds ratio [OR] 0.88, 95% confidence interval [CI] 0.59-1.30; 5 days vs. ≤3 days, OR 0.97, 95% CI 0.68-1.37) or infection (4 days vs. ≤3 days, OR 0.71, 95% CI 0.39-1.29; 5 days vs. ≤3 days, OR 1.11, 95% CI 0.67-1.83). Similar results were obtained regardless of how storage duration of PLTs was approached. CONCLUSIONS: In this large observational study in a heterogeneous ICU population, storage duration of PLTs was not associated with an increased risk of mortality or infection.
Assuntos
Plaquetas , Preservação de Sangue , Mortalidade Hospitalar , Infecções , Unidades de Terapia Intensiva , Transfusão de Plaquetas , Adulto , Idoso , Austrália/epidemiologia , Estado Terminal , Humanos , Infecções/etiologia , Infecções/mortalidade , Pessoa de Meia-Idade , Fatores de Risco , Fatores de TempoRESUMO
BACKGROUND: Platelets are commonly transfused to critically ill patients. Reports suggest an association between platelet transfusion and infection. However, there is no large study to have determined whether platelet transfusion in critically ill patients is associated with hospital-acquired infection. METHODS: We conducted a multi-centre study using prospectively maintained databases of two large academic intensive care units (ICUs) in Australia. Characteristics of patients who received platelets in ICUs between 2008 and 2014 were compared to those of patients who did not receive platelets. Association between platelet administration and infection (bacteraemia and/or bacteriuria) was modelled using multiple logistic regression and Cox regression, with blood components as time-varying covariates. A propensity covariate adjustment was also performed to verify results. RESULTS: Of the 18,965 patients included, 2250 (11.9%) received platelets in ICU with a median number of 1 platelet unit (IQR 1-3) administered. Patients who received platelets were more severely ill at ICU admission (mean Acute Physiology and Chronic Health Evaluation III score 65 (SD 29) vs 52 (SD 25), p < 0.01) and had more comorbidities (31% vs 19%, p < 0.01) than patients without platelet transfusion. Invasive mechanical ventilation (87% vs 57%, p < 0.01) and renal replacement therapy (20% vs 4%, p < 0.01) were more frequently administered in patients receiving platelets than in patients without platelets. On univariate analysis, platelet transfusion was associated with hospital-acquired infection in the ICU (7.7% vs 1.4%, p < 0.01). After adjusting for confounders, including other blood components administered, patient severity, centre, year, and diagnosis category, platelet transfusions were independently associated with infection (adjusted OR 2.56 95% CI 1.98-3.31, p < 0.001). This association was also found in survival analysis with blood components as time-varying covariates (adjusted HR 1.85, 95% CI 1.41-2.41, p < 0.001) and when only bacteraemia was considered (adjusted OR 3.30, 95% CI 2.30-4.74, p <0.001). Platelet transfusions remained associated with infection after propensity covariate adjustment. CONCLUSIONS: After adjustment for confounders, including patient severity and other blood components, platelet transfusion was independently associated with ICU-acquired infection. Further research aiming to better understand this association and to prevent this complication is warranted.
Assuntos
Estado Terminal/terapia , Doença Iatrogênica/epidemiologia , Transfusão de Plaquetas/normas , Idoso , Austrália , Bacteriemia/epidemiologia , Bacteriemia/etiologia , Bacteriúria/epidemiologia , Bacteriúria/etiologia , Estado Terminal/epidemiologia , Feminino , Humanos , Unidades de Terapia Intensiva/estatística & dados numéricos , Modelos Logísticos , Masculino , Pessoa de Meia-Idade , Análise Multivariada , Transfusão de Plaquetas/estatística & dados numéricos , Estudos Prospectivos , Estudos Retrospectivos , Índice de Gravidade de DoençaRESUMO
CONTEXT: Administration of traditional chloride-liberal intravenous fluids may precipitate acute kidney injury (AKI). OBJECTIVE: To assess the association of a chloride-restrictive (vs chloride-liberal) intravenous fluid strategy with AKI in critically ill patients. DESIGN, SETTING, AND PATIENTS: Prospective, open-label, sequential period pilot study of 760 patients admitted consecutively to the intensive care unit (ICU) during the control period (February 18 to August 17, 2008) compared with 773 patients admitted consecutively during the intervention period (February 18 to August 17, 2009) at a university-affiliated hospital in Melbourne, Australia. INTERVENTIONS: During the control period, patients received standard intravenous fluids. After a 6-month phase-out period (August 18, 2008, to February 17, 2009), any use of chloride-rich intravenous fluids (0.9% saline, 4% succinylated gelatin solution, or 4% albumin solution) was restricted to attending specialist approval only during the intervention period; patients instead received a lactated solution (Hartmann solution), a balanced solution (Plasma-Lyte 148), and chloride-poor 20% albumin. MAIN OUTCOME MEASURES: The primary outcomes included increase from baseline to peak creatinine level in the ICU and incidence of AKI according to the risk, injury, failure, loss, end-stage (RIFLE) classification. Secondary post hoc analysis outcomes included the need for renal replacement therapy (RRT), length of stay in ICU and hospital, and survival. RESULTS Chloride administration decreased by 144 504 mmol (from 694 to 496 mmol/patient) from the control period to the intervention period. Comparing the control period with the intervention period, the mean serum creatinine level increase while in the ICU was 22.6 µmol/L (95% CI, 17.5-27.7 µmol/L) vs 14.8 µmol/L (95% CI, 9.8-19.9 µmol/L) (P = .03), the incidence of injury and failure class of RIFLE-defined AKI was 14% (95% CI, 11%-16%; n = 105) vs 8.4% (95% CI, 6.4%-10%; n = 65) (P <.001), and the use of RRT was 10% (95% CI, 8.1%-12%; n = 78) vs 6.3% (95% CI, 4.6%-8.1%; n = 49) (P = .005). After adjustment for covariates, this association remained for incidence of injury and failure class of RIFLE-defined AKI (odds ratio, 0.52 [95% CI, 0.37-0.75]; P <.001) and use of RRT (odds ratio, 0.52 [95% CI, 0.33-0.81]; P = .004). There were no differences in hospital mortality, hospital or ICU length of stay, or need for RRT after hospital discharge. CONCLUSION The implementation of a chloride-restrictive strategy in a tertiary ICU was associated with a significant decrease in the incidence of AKI and use of RRT. TRIAL REGISTRATION: Clinicaltrials.gov Identifier: NCT00885404.
Assuntos
Injúria Renal Aguda/etiologia , Cloretos/administração & dosagem , Cloretos/efeitos adversos , Hidratação/efeitos adversos , Estado Terminal , Feminino , Mortalidade Hospitalar , Hospitais de Ensino , Humanos , Unidades de Terapia Intensiva , Soluções Isotônicas/administração & dosagem , Tempo de Internação , Masculino , Pessoa de Meia-Idade , Projetos Piloto , Estudos Prospectivos , Terapia de Substituição Renal , Lactato de Ringer , Solução Salina Hipertônica/administração & dosagem , Solução Salina Hipertônica/efeitos adversosRESUMO
OBJECTIVE: To assess the association of abnormalities of ionized calcium levels with mortality in a heterogeneous cohort of critically ill patients. DESIGN: Retrospective, combined clinical and biochemical study. SETTING: Four combined medical/surgical intensive care units. PATIENTS: Cohort of 7,024 adult critically ill patients. INTERVENTIONS: None. MEASUREMENTS AND MAIN RESULTS: We studied 177,578 ionized calcium measurements, from 7024 patients, with a mean value of 1.11 mmol/L (ionized calcium measured every 4.5 hrs on average). The unadjusted lowest and highest ionized calcium reported during intensive care unit stay were significantly different between intensive care unit survivors and nonsurvivors (p < .001). If hypocalcemia occurred at least once during the intensive care unit stay, the probability of intensive care unit mortality increased by 46%, 108%, and 150% for ionized calcium levels <1.15, 0.90, and 0.80 mmol/L, respectively. If hypercalcemia occurred at least once during the intensive care unit stay, the probability of intensive care unit mortality increased by 100%, 162%, and 190% for ionized calcium levels >1.25, 1.35, and 1.45 mmol/L, respectively. Similar trends were seen for hospital mortality. However, from multivariate logistic regression analysis, only an ionized calcium <0.8 mmol/L or an ionized calcium >1.4 mmol/L were independently associated with intensive care unit and hospital mortality. CONCLUSIONS: Within a broad range of values, ionized calcium concentration has no independent association with hospital or intensive care unit mortality. Only extreme abnormalities of ionized calcium concentrations are independent predictors of mortality.
Assuntos
Cálcio/sangue , Estado Terminal/mortalidade , Mortalidade Hospitalar , Hipercalcemia/sangue , Hipercalcemia/mortalidade , Hipocalcemia/sangue , Hipocalcemia/mortalidade , Adulto , Idoso , Biomarcadores/sangue , Análise Química do Sangue , Sinalização do Cálcio , Distribuição de Qui-Quadrado , Estudos de Coortes , Intervalos de Confiança , Cuidados Críticos/métodos , Feminino , Humanos , Concentração de Íons de Hidrogênio , Hipercalcemia/prevenção & controle , Hipocalcemia/prevenção & controle , Unidades de Terapia Intensiva , Masculino , Pessoa de Meia-Idade , Análise Multivariada , Razão de Chances , Estudos Retrospectivos , Sensibilidade e Especificidade , Análise de SobrevidaRESUMO
OBJECTIVES: The relationship between hyperglycemia and mortality is altered by the presence of diabetes mellitus. Biological adjustment to preexisting hyperglycemia might explain this phenomenon. We tested whether the degree of preexisting hyperglycemia would modulate the association between glycemia and outcome during critical illness in patients with diabetes mellitus. DESIGN: Retrospective observational study. SETTING: Two tertiary intensive care units. PATIENTS: Four hundred fifteen critically ill diabetic patients with HbA1c levels measured within 3 months of intensive care unit admission. INTERVENTIONS: None. MEASUREMENTS AND MAIN RESULTS: There were 9,946 blood glucose measurements in this study cohort (glucose measured 6.7 times per day; every 3.6 hrs on average). The median preadmission HbA1c level was 7.0%. There was no significant difference in HbA1c levels (p = .17) or time-weighted average of blood glucose concentrations (p = .49) between survivors and nonsurvivors. The time-weighted average of blood glucose concentrations during intensive care unit stay for nonsurvivors was lower than that of survivors when the HbA1c was >6.8%. In multivariate analysis, we found that there was a significant interaction between HbA1c and the time-weighted glucose level, indicating that the relationship between HbA1c and mortality changed according to the levels of time-weighted average of blood glucose concentrations (p = .008). As a consequence, in patients with higher (>7%) preadmission levels of HbA1c, the higher the time-weighted acute glucose concentration during intensive care unit stay (>10 mmol/L), the lower the hospital mortality compared with the lower HbA1c cohort (<7%). CONCLUSIONS: In patients with diabetes mellitus admitted to intensive care units, there was a significant interaction between preexisting hyperglycemia and the association between acute glycemia and mortality. These observations generate the hypothesis that glucose levels that are considered safe and desirable in other patients might be undesirable in diabetic patients with chronic hyperglycemia. Further studies are required to confirm or refute our findings.
Assuntos
Glicemia/análise , Diabetes Mellitus/sangue , Diabetes Mellitus/mortalidade , Hemoglobinas Glicadas/análise , Mortalidade Hospitalar , Hiperglicemia/mortalidade , Doença Aguda , Idoso , Doença Crônica , Estudos de Coortes , Cuidados Críticos/métodos , Estado Terminal/mortalidade , Estado Terminal/terapia , Diabetes Mellitus/tratamento farmacológico , Feminino , Índice Glicêmico , Humanos , Hiperglicemia/complicações , Hiperglicemia/diagnóstico , Hipoglicemiantes/uso terapêutico , Unidades de Terapia Intensiva , Masculino , Pessoa de Meia-Idade , Análise Multivariada , Razão de Chances , Valor Preditivo dos Testes , Prognóstico , Estudos Retrospectivos , Medição de Risco , Análise de SobrevidaRESUMO
OBJECTIVE: To study the impact of diabetes mellitus on the relationship between glycemia and mortality in critically ill patients. DESIGN: Retrospective observational study. SETTING: Intensive care units of two university hospitals. PATIENTS: Cohort of 4946 critically ill patients including 728 patients with diabetes mellitus. INTERVENTION: None. MEASUREMENTS AND MAIN RESULTS: We assessed and compared the relationship between glycemia during intensive care unit stay and mortality in diabetic and nondiabetic patients. There were 125,036 blood glucose measurements (5.7 measurements/day on average). Intensive care unit mortality increased significantly with increasing mean blood glucose concentration in nondiabetes mellitus patients but not in diabetes mellitus patients. Nondiabetes mellitus patients with a time-weighted glucose concentration (Glu(Tw)) between 8.0 and 10.0 mmol/L were found to be 1.74 times more likely to die in intensive care unit as diabetes mellitus patients in the same range (odds ratio = 1.74 [1.13-2.68] p = 0.01). They were also more than three times more likely to die in the intensive care unit compared with diabetes mellitus patients when the Glut(w )was between 10.0 and 11.1 mmol/L (odds ratio = 3.34 [1.35-8.23] p = 0.009). Using multivariate logistic regression analysis, hyperglycemia was strongly and independently associated with outcome in nondiabetic patients (p < 0.001) but showed no significant association with outcome in diabetic patients. CONCLUSIONS: Unlike nondiabetic patients, diabetic patients show no clear association between hyperglycemia during intensive care unit stay and mortality and markedly lower odds ratios of death at all levels of hyperglycemia. These findings suggest that, in critically patients with diabetes mellitus, hyperglycemia may have different biological and/or clinical implications.
Assuntos
Glicemia , Estado Terminal/mortalidade , Diabetes Mellitus/fisiopatologia , Mortalidade Hospitalar , APACHE , Estado Terminal/classificação , Diabetes Mellitus/sangue , Diabetes Mellitus/tratamento farmacológico , Feminino , Índice Glicêmico , Humanos , Hiperglicemia/tratamento farmacológico , Hipoglicemiantes/uso terapêutico , Insulina/uso terapêutico , Unidades de Terapia Intensiva/estatística & dados numéricos , Modelos Logísticos , Masculino , Pessoa de Meia-Idade , Estudos Multicêntricos como Assunto , Estudos Retrospectivos , Análise de SobrevidaRESUMO
OBJECTIVE: Patients commonly receive i.v. fluids in the ED. It is still unclear whether the choice of i.v. fluids in this setting influences renal or patient outcomes. We aimed to assess the effects of restricting i.v. chloride administration in the ED on the incidence of acute kidney injury (AKI). METHODS: We conducted a before-and-after trial with 5008 consecutive ED-treated hospital admissions in the control period and 5146 consecutive admissions in the intervention period. During the control period (18 February 2008 to 17 August 2008), patients received standard i.v. fluids. During the intervention period (18 February 2009 to 17 August 2009), we restricted all chloride-rich fluids. We used the Kidney Disease: Improving Global Outcomes (KDIGO) staging to define AKI. RESULTS: Stage 3 of KDIGO-defined AKI decreased from 54 (1.1%; 95% confidence interval [CI] 0.8-1.4) to 30 (0.6%; 95% CI 0.4-0.8) (P = 0.006). The rate of renal replacement therapy did not change, from 13 (0.3%; 95% CI 0.2-0.4) to 8 (0.2%; 95% CI 0.1-0.3) (P = 0.25). After adjustment for relevant covariates, liberal chloride therapy remained associated with a greater risk of KDIGO stage 3 (hazard ratio 1.82; 95% CI 1.13-2.95; P = 0.01). On sensitivity assessment after removing repeat admissions, KDIGO stage 3 remained significantly lower in the intervention period compared with the control period (P = 0.01). CONCLUSION: In a before-and-after trial, a chloride-restrictive strategy in an ED was associated with a significant decrease in the incidence of stage 3 of KDIGO-defined AKI.
Assuntos
Hidratação/métodos , Hidratação/normas , Cloreto de Sódio/metabolismo , Injúria Renal Aguda/etiologia , Injúria Renal Aguda/prevenção & controle , Administração Intravenosa , Idoso , Serviço Hospitalar de Emergência/organização & administração , Feminino , Mortalidade Hospitalar , Humanos , Incidência , Modelos Logísticos , Masculino , Pessoa de Meia-Idade , Cloreto de Sódio/uso terapêuticoRESUMO
PURPOSE: There is conflicting evidence on the effect of red blood cells (RBC) storage duration and clinical outcomes. We aimed to investigate the association between RBC storage duration and clinical outcomes in patients admitted to the intensive care unit (ICU). MATERIALS AND METHODS: We retrospectively (2001-2011) studied adults admitted to the ICUs of 2 hospitals who received RBC. Using the mean, maximum and minimum age of RBC units transfused, we evaluated the association between RBC storage duration and mortality. We also analyzed the association between mean age of RBC units and length of stay (LOS) in survivors. We performed sensitivity analyses in patients who only received RBC in ICU and who only received leukodepleted RBC. RESULTS: We studied 8416 patients who received a median of 4 (interquartile range, 2-7) RBC units. After multivariate analysis, age of RBC was not independently associated with mortality, including in the subgroup analyses. Furthermore, there was no clinically relevant relationship between mean RBC age and LOS. CONCLUSIONS: RBC storage duration was not associated with increased mortality nor ICU and hospital LOS. These results support the view that the effect of RBC storage duration on outcomes in critically ill patients is uncertain.
Assuntos
Estado Terminal , Transfusão de Eritrócitos/estatística & dados numéricos , Eritrócitos , Mortalidade Hospitalar , Tempo de Internação , Manejo de Espécimes , Idoso , Estado Terminal/mortalidade , Feminino , Humanos , Unidades de Terapia Intensiva , Masculino , Pessoa de Meia-Idade , Análise Multivariada , Estudos Retrospectivos , Fatores de TempoRESUMO
PURPOSE: The aim of this study was to assess the association of phosphate concentration with key clinical outcomes in a heterogeneous cohort of critically ill patients. MATERIALS AND METHODS: This was a retrospective observational study at a general intensive care unit (ICU) of an Australian university teaching hospital enrolling 2730 adult critically ill patients. RESULTS: We studied 10504 phosphate measurements with a mean value of 1.17 mmol/L (measurements every 28.8 hours on average). Hyperphosphatemia (inorganic phosphate [iP] concentration > 1.4 mmol/L) occurred in 45% and hypophosphatemia (iP ≤ 0.6 mmol/L) in 20%. Among patients without any episodes of hyperphosphatemia, patients with at least 1 episode of hypophosphatemia had a higher ICU mortality than those without hypophosphatemia (P = .004). In addition, ICU nonsurvivors had lower minimum phosphate concentrations than did survivors (P = .009). Similar results were seen for hospital mortality. However, on multivariable logistic regression analysis, hypophosphatemia was not independently associated with ICU mortality (adjusted odds ratio, 0.86 [95% confidence interval, 0.66-1.10]; P = .24) and hospital mortality (odds ratio, 0.89 [0.73-1.07]; P = .21). Even when different cutoff points were used for hypophosphatemia (iP ≤ 0.5, 0.4, 0.3, or 0.2 mmol/L), hypophosphatemia was not an independent risk factor for ICU and hospital morality. In addition, timing of onset and duration of hypophosphatemia were not independent risk factor for ICU and hospital mortality. CONCLUSIONS: Hypophosphatemia behaves like a general marker of illness severity and not as an independent predictor of ICU or in-hospital mortality in critically ill patients.
Assuntos
Estado Terminal/mortalidade , Hipofosfatemia/mortalidade , Idoso , Distribuição de Qui-Quadrado , Feminino , Mortalidade Hospitalar , Humanos , Unidades de Terapia Intensiva , Modelos Logísticos , Masculino , Pessoa de Meia-Idade , Estudos Retrospectivos , Fatores de Risco , Estatísticas não Paramétricas , Taxa de SobrevidaRESUMO
OBJECTIVE: To estimate the ability of commonly measured laboratory variables to predict an imminent (within the same or next calendar day) death in ward patients. DESIGN: Retrospective observational study. SETTING: Two university affiliated hospitals. PATIENTS: Cohort of 42,701 patients admitted for more than 24 hours and external validation cohort of 13,137 patients admitted for more than 24 hours. INTERVENTION: We linked commonly measured laboratory tests with event databases and assessed the ability of each laboratory variable or combination of variables together with patient age to predict imminent death. MEASUREMENTS AND MAIN RESULTS: In the inception teaching hospital, we studied 418,897 batches of tests in 42,701 patients (males 55%; average age 65.8 ± 17.6 years), for a total of >2.5 million individual measurements. Among these patients, there were 1596 deaths. Multivariable logistic modelling achieved an AUC-ROC of 0.87 (95% CI: 0.85-0.89) for the prediction of imminent death. Using an additional 105,074 batches from a cohort of 13,137 patients from a second teaching hospital, the multivariate model achieved an AUC-ROC of 0.88 (95% CI: 0.85-0.90). CONCLUSIONS: Commonly performed laboratory tests can help predict imminent death in ward patients. Prospective investigations of the clinical utility of such predictions appear justified.
Assuntos
Estado Terminal/mortalidade , Testes Diagnósticos de Rotina/estatística & dados numéricos , Modelos Teóricos , Idoso , Causas de Morte/tendências , Feminino , Mortalidade Hospitalar/tendências , Equipe de Respostas Rápidas de Hospitais , Humanos , Masculino , Valor Preditivo dos Testes , Estudos Retrospectivos , Índice de Gravidade de Doença , Vitória/epidemiologiaRESUMO
OBJECTIVE: To estimate the ability of commonly measured laboratory variables to predict imminent (within the same or next calendar day) medical emergency team (MET) calls, ICU admission or death. METHODS: We performed a retrospective observational study of ED patients. We estimated the ability of each laboratory variable or combination of variables together with patient age to predict imminent MET calls, ICU admission or death. We externally validated our findings in patients from a different hospital. RESULTS: We studied 160 341 batches in 71 453 ED patients (average age: 59.9 ± 22.1 years) for a total of 1 million individual measurements. There were 341 MET calls, 160 ICU admissions from the wards and 858 deaths. Multivariable modelling achieved a receiver operating characteristic area under the curve (ROC-AUC) of 0.69 (95% CI 0.63-0.74) for imminent MET call with prediction occurring a mean of 11.9 h before the call. Additionally, it achieved a ROC-AUC of 0.82 (95% CI 0.73-0.87) for imminent ICU admission. Finally, it achieved a ROC-AUC of 0.90 (95% CI 0.87-0.91) for imminent death. When tested using an additional 37 367 batches from a cohort of 21 430 ED patients from a second teaching hospital, the multivariate model achieved a ROC-AUC of 0.70 (95% CI 0.66-0.73) for imminent MET call, a ROC-AUC of 0.84 (95% CI 0.78-0.90) for imminent ICU admission. Finally, it achieved a ROC-AUC of 0.89 (95% CI 0.86-0.91) for imminent death. CONCLUSIONS: Commonly performed laboratory tests can help predict imminentâ MET calls, ICU admission or death in ED patients. Prospective investigations of the clinical utility of such predictions appear desirable.
Assuntos
Técnicas de Laboratório Clínico , Serviço Hospitalar de Emergência , Mortalidade Hospitalar , Unidades de Terapia Intensiva/estatística & dados numéricos , Equipe de Assistência ao Paciente/estatística & dados numéricos , Adulto , Idoso , Idoso de 80 Anos ou mais , Austrália , Feminino , Humanos , Masculino , Pessoa de Meia-Idade , Análise Multivariada , Admissão do Paciente/estatística & dados numéricos , Valor Preditivo dos Testes , Curva ROC , Estudos RetrospectivosRESUMO
OBJECTIVE: To determine whether mild or moderate hypoglycemia that occurs in critically ill patients is independently associated with an increased risk of death. PATIENTS AND METHODS: Of patients admitted to 2 hospital intensive care units (ICUs) in Melbourne and Sydney, Australia, from January 1, 2000, to October 14, 2004, we analyzed all those who had at least 1 episode of hypoglycemia (glucose concentration, <81 mg/dL). The independent association between hypoglycemia and outcome was statistically assessed. RESULTS: Of 4946 patients admitted to the ICUs, a cohort of 1109 had at least 1 episode of hypoglycemia (blood glucose level, <81 mg/dL). Of these 1109 patients (22.4% of all admissions to the intensive care unit), hospital mortality was 36.6% compared with 19.7% in the 3837 nonhypoglycemic control patients (P<.001). Even patients with a minimum blood glucose concentration between 72 and 81 mg/dL had a greater unadjusted mortality rate than did control patients (25.9% vs 19.7%; unadjusted odds ratio, 1.42; 95% confidence interval, 1.12-1.80; P=.004.) Mortality increased significantly with increasing severity of hypoglycemia (P<.001). After adjustment for insulin therapy, hypoglycemia was independently associated with increased risk of death, cardiovascular death, and death due to infectious disease. CONCLUSION: In critically ill patients, an association exists between even mild or moderate hypoglycemia and mortality. Even after adjustment for insulin therapy or timing of hypoglycemic episode, the more severe the hypoglycemia, the greater the risk of death.