Open Access

Thrombocytopenia in critically ill surgical patients: a case-control study evaluating attributable mortality and transfusion requirements

  • François Stephan1,
  • Jacquesde Montblanc1,
  • Ali Cheffi1 and
  • Francis Bonnet1
Critical Care19993:151

DOI: 10.1186/cc369

Received: 6 July 1999

Accepted: 24 September 1999

Published: 25 October 1999

Abstract

Background

That thrombocytopenia results in increased mortality ortransfusion requirements has not been confirmed by previous studies. Weperformed a case-control study in which 36 patients who developed severethrombocytopenia of less than 50×109 platelets/l were carefullymatched for the severity of underlying disease and other importantvariables.

Results

Seventeen (47%) thrombocytopenic patients died, versus 10 (28%)matched control patients who were not thrombocytopenic.Nine pairs had adiscordant outcome, and in eight of these pairs the thrombocytopenic patientdied (exact binomial probability 0.037). The estimated attributable mortalitywas 19.5% (95% confidence interval 3.2–35.8), and the estimated odds ratio was2.7 (95% confidence interval 1.02–7.10). Thrombocytopenic patients hadcomparable values for severity of illness scores between day of admission andday of thrombocytopenia, in contrast with control patients who had astatistically significant decrease in severity of illness scores during thesame period. Thirty (83%) of the thrombocytopenic patients required transfusionof blood products, versus 21 (58%) control patients (paired χ2test 4.92, P < 0.04). The estimated attributable transfusionrequirement was 25% (95% confidence interval 5.4–44.6), and the estimated oddsratio was 1.52 (95 confidence interval 1.05–2.20).

Conclusion

The present study suggests that thrombocytopenia of less than50 × 109 platelets/l may be a marker for more severe illness andincreased risk of death, rather than causative, because a true causalrelationship is not established. Thrombocytopenia also leads to an excess ofblood product consumption.

Keywords

APACHE II score bleeding case-control study intensive care unit mortality prognosis sepsis transfusion thrombocytopenia underlying disease

Introduction

Thrombocytopenia is a well known complication in intensive care unit (ICU) patients. It has been associated with various risk factors, but mainly with sepsis [1,2,3]. The incidence of thrombocytopenia of less than 100 × 109platelets/l in ICU patients has been reported to vary from 23 to 41%, but lower frequencies (10–17%) have been reported for counts lower than 50 × 109 platelets/l [1,2,3]. Mortality rates as high as 38–54% have been observed, and have been reported to be proportional to the nadir of the platelet count [1,3,4]. Previous studies have not clearly demonstrated that thrombocytopenia results in increased mortality or increased transfusion requirements, however. Two independent factors have made this important and seemingly straightforward issue difficult to resolve.

First, mortality rates are high in such patients for many reasons. Numerous studies [1,3,5,6,7] have demonstrated that severe underlying illness predisposes to the development of thrombocytopenia in ICU patients, and the influence of thrombocytopenia on mortality is therefore difficult to assess. Although thrombocytopenia in the critically ill is more often a symptom than a disease process per se, it might increase mortality in several ways. Thrombocytopenia can result in a mild, moderate, or severe haemorrhagic disorder, which could enhance the risk of morbidity and mortality in critically ill surgical patients [8]. The adverse effects of anaemia in such patients have also recently been discussed [9,10]. Apart from its haemostatic effect, thrombocytopenia also increases the susceptibility to and severity of certain infections [11]. Although several studies have found a relationship between thrombocytopenia and the likelihood of death, especially in septic patients [1,3,4,7,12], thrombocytopenia has rarely been identified as an independent predictive factor of death using multiple logistic regression [5,6,13]. Nevertheless, these previous studies failed to reach any definite conclusions regarding whether thrombocytopenia per se was responsible for the poorer prognosis, or whether this higher mortality simply reflected more severe underlying illness.

Second, the threshold value for severe thrombocytopenia that is supposed to jeopardize the prognosis is difficult to determine. We and others [1,3,13] have suggested that a platelet count less than 50 × 109/l is associated with a poor outcome. Moreover, guidelines for platelet transfusion [8] have proposed that the threshold value of 50 × 109 platelets/l is indicative of platelet transfusion requirement in surgical patients.

One of the commonest methods to evaluate excess mortality is to perform a case-control study in which confounding variables (eg severity of underlying illness, reason for hospitalization, and so forth) are carefully matched in the two populations. To date, however, no case-control studies that have evaluated morbidity and mortality associated with thrombocytopenia in ICU patients have been published in which these important variables have been carefully matched.

We therefore designed a case-control study to determine to what extent severe thrombocytopenia (defined as < 50 × 109 platelets/l) increases mortality and blood product requirements in surgical ICU patients.

Methods

Study design

We performed a matched cohort study, with a matched control patient without thrombocytopenia for each thrombocytopenic patient (1:1 matching). The study was conducted in the Service de Réanimation Chirurgicale of Tenon University Hospital in Paris. This 8-bed ICU admits patients from all surgical departments and operating rooms of the hospital. The study period ran from January 1, 1996 to December 31, 1996, during which time 298 patients were admitted to the ICU and thrombocytopenic patients were prospectively identified.

Case identification

Platelet count was performed daily for all patients. Patients who experienced even a single episode of thrombocytopenia of less than 50 × 109platelets/l during the ICU stay were classified as 'cases'. Low platelet counts were confirmed by direct examination of the blood smear. Platelet transfusions were administered to actively bleeding patients and patients scheduled for emergency surgery if their platelet count fell below 50 × 109/l. Likewise, platelet transfusions were administered to patients at risk for bleeding complications (eg postoperative patients, or after gastrointestinal bleeding) when their platelet count fell below 20 × 109/l [8,14].

None of patients with a history of platelet disorders, haematologic malignancies or chemotherapy, splenectomy, mechanical heart valves, or patients undergoing cardiopulmonary bypass surgery were included in the patient population.

Matching and selection of control patients

Control patients had to have had no evidence of severe thrombocytopenia (< 50 × 109 platelets/l) at any time during hospitalization in the ICU. A computer-generated list of eligible control patients was obtained from a database that included 695 patients hospitalized between November, 1995 and March, 1998. Control patients were selected according to the following matching criteria: age (± 5 years), Acute Physiology and Chronic Health Evaluation (APACHE) II score calculated on the first day of ICU admission (± 5 points), primary diagnosis and duration of stay in the ICU (± 5 days). The list of potential control patients was reviewed for the best possible match, giving highest priority to primary diagnosis, duration of stay in the ICU, APACHE II score and age. In the case of multiple acceptable control patients, the one with the date of ICU admission closest to that of the patient was chosen.

Collection of data

The following information was recorded: primary diagnosis; age; sex; dates of admission and discharge from the ICU; previous health status; severity of underlying medical conditions stratified according to the criteria of McCabe and Jackson [15] as fatal, ultimately fatal and nonfatal; first-day organ dysfunction and/or infection (ODIN) score [16], based on the presence or absence of cardiac, respiratory, renal, hepatic, neurologic, haematologic dysfunctions and/or infection; red blood cells (RBCs); fresh frozen plasma and platelet transfusion requirements; and full blood count and coagulation studies (factor V assay and D-dimer detection were performed only when prothrombin time was abnormal). The severity of illness was evaluated with the first-day APACHE II score [17] and the first-day new Simplified Acute Physiology Score (SAPS) II [18].

The criteria of sepsis have previously been reported [19]. Shock was defined as a decrease in systolic blood pressure (< 90 mmHg) despite adequate vascular filling or the need for vasoactive drugs (dopamine > 5μ g/kg per min, dobutamine, epinephrine, or norepinephrine).

The definition of disseminated intravascular coagulation (DIC) required the following four criteria: a platelet count less than 150 × 109/l; a decrease in prothrombin level activity to less than 50%; a decrease in the level of factor V to less than 50%; and the presence of fibrin degradation products (D-dimers).

Statistical analysis

Analysis of prior data from our unit indicated that the odds ratio for death was 3.9 for cases and that the mortality rate for control patients was 23% [3]. On the basis of these results, we estimated that 28 patients in each group (control and case) would be required to test the null hypothesis at the 0.05 significance level with a power of 0.7 [20].

The attributable mortality due to severe thrombocytopenia was defined as the crude mortality rate of the control patients subtracted from that of the cases. The point estimate of the attributable mortality and 95% confidence intervals (CIs) were calculated as previously described [21]. McNemar's test with continuity correction and exact binomial probabilities were used to determine whether the crude mortality rates of the cases and control patients were significantly different [21]. Because the study involved a matched cohort, the odds ratio was used as a measure of relative risk. Because incidence of outcome is more than 10%, correction of the odds ratio was made according to the method of Zhang and Yu [22] in order to represent better the true relative risk. The 95% CI of the odds ratio was calculated using Miettinen's test-based approach [21].

Differences in medians were tested using Wilcoxon's test for continuous variables and the χ2 test for categorical variables, with Yates' correction when appropriate. Quantitative data were expressed as means± standard deviation. P < 0.05 was considered statistically significant.

Results

Study population

During the study period, 36 of the 298 studied ICU patients developed severe thrombocytopenia of less than 50 × 109 platelets/l, producing a global incidence rate of 12 episodes of thrombocytopenia per 100 admissions. Thrombocytopenia occurred 3.2 ± 4.8 days (median 2 days, interquartile range 2.5 days, range 0–26 days) after ICU admission, for a mean duration of 3.4 ± 2.9 days (median 2 days, interquartile range 4 days, range 1–13 days). Thrombocytopenia was related to sepsis in eight patients; sepsis and DIC in 11 patients; bleeding in nine patients; bleeding and DIC in six patients; undetermined shock and DIC in one patient; and haemolysis-elevated liver enzymes-low platelets syndrome with DIC in one patient. Matching was performed for these 36 patients.

On admission to ICU, the mean platelet count was 127 ± 75 × 109/l (median 116) versus 199 ± 123 × 109/l (median 161) in cases and control patients, respectively (not significant). During ICU stay, mean nadir platelet count was 28 ± 14 × 109/l (median 29.5) versus 184 ± 142 × 109/l (median 114); mean nadir haemoglobin value was 6.9 ± 1.7g/dl (median 7.2) versus 8.7 ± 2.3g/dl (median 8; P = 0.009); and mean nadir leucocyte count was 6.2 ± 4.1 × 109/l versus 10.2 ± 4.9 × 109/l (P < 0.0004) in cases and control patients, respectively. Nineteen cases (52.7%) and eight control patients (22.2%) had evidence of DIC (P < 0.01).

Table 1

Results of matching of cases to control patients using four major criteria

 

Proportion of cases matched to control patients

Criterion

n/n

%

Same primary diagnosis

35/36

97

Same duration of stay in ICU (± 5 days)

31/36

86

Same APACHE II score (± 5)

29/36

81

Same age (± 5 years)

27/36

75

APACHE, Acute Physiology and Chronic Health Evaluation; ICA,intensive care unit.

Closeness of matching

The median age of the cases was 65 years (range 27–104 years, mean 60.7 years) versus 65 years (range 27–83 years, mean 60.6 years) for control patients. Twenty-one (58%) case-control pairs differed by no more than 3 years, and 27 (75%) by no more than 5 years (Table 1). The median APACHE II score of cases was 21 (range 5–41, mean 20.8) versus 21 for control patients (range 4–37, mean 22.7). The median duration of ICU stay of cases was 5 days (range 1–35 days, mean 8.3 days) versus 4 days for control patients (range 1–26 days, mean 6.4 days). Thirty-five (97%) of the pairs were matched for primary diagnosis. Overall, matching was successful for 243 out of 288 (84.3%) variables.

Because the study end point — mortality — is strongly related to severity of illness and/or initial respiratory status [17,18], we controlled matching by comparing another seven criteria that evaluated the severity of disease in the two groups of patients (Table 2). No significant differences in these indices were observed between cases and control patients. This was particularly the case when comparing the four variables that were individually predictive of survival for all patients: mean SAPS II score, mean number of organ-system failures, number of patients with cardiac failure on admission, and number of patients with renal failure on admission. The sex ratio of the two groups was also similar. Finally, the dates of admission of cases and control patients differed by less than 1 year in 27 pairs (75%).

Mortality

Seventeen (47%) cases died, versus 10 (28%) control patients (Table 3). Twenty-seven matched pairs had a concordant outcome (18 lived and nine died). Nine pairs had a discordant outcome, and in eight of these pairs the case died (exact binomial probabilities: 0.037). The estimated attributable mortality was 19.5% (95% CI 3.2–35.8), and the estimated odds ratio was 2.7 (95% CI 1.02–7.10).

Table 2

Matching criteria and clinical characteristics of the study population

 

Cases

Control patients

 

Parameter

(n = 36)

(n = 36)

P

Matching criteria

   

     Primary diagnosis

  

0.99

        Septic shock [n (%)]

12 (33)

12 (33)

 

        Postoperative bleeding, gastrointestinal hemorrhage, trauma [n (%)]

14 (39)

13 (36)

 

        Acute respiratory failure [n (%)]

3 (8)

3 (8)

 

        Complication during post partum [n (%)]

2 (6)

2 (6)

 

        Cardiac failure [n (%)]

1 (3)

2 (6)

 

        Miscellaneous [n (%)]*

4 (11)

4 (11)

 

    Duration of stay in ICU (days)

8.3± 8.4

6.5± 6.4

0.29

    APACHE II score (points)

22.7± 8.6

20.9± 8.3

0.35

    Age (years)

60.7± 19.1

60.6± 17.4

0.97

Characteristics

   

    Sex (male/female)

20/16

22/14

0.63

    Severity of underlying disease

  

0.81

        Fatal or ultimately fatal [n (%)]

19 (53)

18 (50)

 

        Nonfatal [n (%)]

17 (47)

18 (50)

 

    Previous health status

  

0.80

        No functional or moderate limitation [n (%)]

25 (69)

26 (72)

 

        Serious limitation or bedridden or institutionalized [n (%)]

11 (31)

10 (28)

 

    SAPS II score (points)

49.7 ± 18.9

46.7 ± 21.9

0.54

    Expected mortality (%)

46.0

39.1

    ODIN model on admission

  

        Cardiac failure [n (%)]

22 (61)

21 (58)

 

        Respiratory failure [n (%)]

30 (83)

30 (83)

 

        Haematologic failure [n (%)]

20 (55)

13 (36)

 

        Neurologic failure [n (%)]

6 (17)

10 (28)

 

        Hepatic failure [n (%)]

1 (3)

2 (5)

 

        Presence of infection [n (%)]

19 (53)

16 (44)

 

Total no. of organ system failure

2.8 ± 1.2

2.9 ± 1.2

0.60

*Miscellaneous included two cases of nosocomialpneumonia, one case of acute pancreatitis, and one case of undetermined shock.†Calculated from the equation proposed by La Gall et al. [18] APACHE, Acute Physiology and Chronic Health Evaluation;ODIN, organ dysfunstion and/or infection; ICU, intensive care unit; SAP,Simplified Acute Physiology Score.

Primary diagnosis in the nine discordant case-control pairs was septic shock in six pairs, acute pancreatitis in one pair, undetermined shock in one pair and haemorrhagic shock in one pair. Causes of mortalities included refractory septic shock in eight cases (47%) and four control patients (40%); multiple organ failure-related sepsis in five cases (29%) and three control patients (30%); uncontrolled bleeding in four cases (24%) and two control patients (20%); and one undetermined shock in one contol (10%).

Sensitivity analyses were done to explore whether mortality risk varies according to exposure. Thus, 17 cases experienced a thrombocytopenia below 50 × 109 platelets/l for more than 3 days. Mortality rate was 47%, versus 53 in patients without a prolonged thrombocytopenia (not significant). Likewise, among the 16 cases transfused with platelets, eight died (50%) compared with nine (45%) cases who were not transfused (not significant). A last sensitivity analysis was done on the timing variable. In 15 cases, thrombocytopenia occurred more than 2 days after ICU admission. Mortality rate was 65% in these patients compared with 35% in cases in whom thrombocytopenia occurred 2 days or less after ICU admission (P = 0.008).

Two time-point analyses of illness severity scores in the nine discordant pairs showed that the APACHE II, SAPS II and ODIN model scores were significantly higher for cases than for control patients on the day of thrombocytopenia (Table 4). Moreover, cases had comparable values of severity of illness scores between day of admission and day of thrombocytopenia; this was contrary to the situation with control patients, who had a statistically significant decrease in severity of illness scores during the same period (Table 4).

Blood product consumption

Thirty (83%) cases required transfusion of blood products, versus 21 (58%) control patients (Table 5). Thirteen pairs had discordant transfusion requirements, and in 11 of these pairs the cases were transfused (paired χ2 4.92, P < 0.04). The estimated attributable transfusion requirement was 25% (95% CI 5.4–44.6), and the estimated odds ratio was 1.52 (95% CI 1.05–2.20).

Units of RBCs transfused were 8.0 ± 7.0 (median 7.5) versus 2.6 ± 4.3 (median 1.5; P < 0.0001), and units of fresh frozen plasma transfused were 5.0 ± 7.0 (median 1.0) versus 1.0 ± 2.0 (median 0; P = 0.05) in cases and control patients, respectively. Sixteen cases (44.4%) received platelet transfusion (5.3 ± 7.7 units of platelets transfused), versus no control patients. In eight patients (one died), platelet transfusions resulted in a rise of platelet count to greater than 50 × 109/l, with a correction of thrombocytopenia a few days later. In eight patients (seven died), only a transient rise of platelet count was noted and thrombocytopenia persisted during the entire ICU stay, despite platelet transfusion. After onset of thrombocytopenia, 4.4 ± 7.2 units of RBCs and 2.5 ± 5.4 units of fresh frozen plasma were transfused. This represents about 50% of the total transfusion requirements.

Table 3

Crude mortality, attributable mortality, and odds ratio of death due to severe thrombocytopenia in intensive care unit patients

 

Point estimate

 

Variable

n/n

%

95% CI

Crude mortality (cases)

17/36

47.2

 

Crude mortality (control patients)

10/36

27.8

 

Attributable mortality

7/36

19.4

3.2–35.8

Odds ratio

2.7

1.02–7.10

CI, confidence interval.

Table 4

Comparison of illness severity scores between cases and control patients on the day of admission and on the day of onset of thrombocytopenia in the nine discordant pairs

Score

Cases (n = 9)

Control patients (n = 9)

P

APACHE II (median)

   

   Day of admission

25.2 ± 4.2 (25)

24.0 ± 6.7 (26)

NS

   Day of thrombocytopenia

30.7 ± 8.3 (47)

12.9 ± 5.2 (11)*

0.04

SAPS II (median)

   

   Day of admission

46.6 ± 10.5 (47)

46.9 ± 16.9 (40)

NS

   Day of thrombocytopenia

58 ± 18.2 (51)

32.1 ± 10.4 (31)*

0.04

ODIN model (median)

   

   Day of admission

3.3 ± 1.4 (4)

3.0 ± 1.0 (3)

NS

   Day of thrombocytopenia

3.8 ± 0.67

0.7 ± 1 (0)*

0.008

Expected mortality (%)

   

   Day of admission

39.1

39.1

   Day of thrombocytopenia

64.0

12.8

*P < 0.05 by Wilcoxon's signed-ranktest (score on day of admission versus score on day of thrombocytopenia incontrol group). Calculated from the equation proposed by LeGall et al. [18].

Table 5

Crude transfusion requirements, attributable transfusion requirements, and odds ratio of transfusion requirements due to severe thrombocytopenia in intensive care unit patients

 

Point estimate

 

Variable

n/n

%

95% Cl

Crude transfusion requirements

   

   Cases

30/36

83.0

 

   Control patients

21/36

58.0

 

Attributable transfusion

9/36

25.0

5.4–44.6

requirements

   

Odds ratio

1.52

1.05–2.20

CI, confidence interval.

Discussion

The present results suggest that thrombocytopenia of less than 50×109platelets/l in ICU patients is associated with increased mortality, with a relative risk of 2.7 (95%CI 1.02-7.10), and with excess , with a relative risk of 1.52 (95%CI 1.05-2.20).

The incidence rate of thrombocytopenia reported in the present study (12% surgical ICU patients) is in the range reported in previous studies [1,3]. There are many explanations for the thrombocytopenia observed in the context of infection. DIC [23], immune mechanisms [23], and haemophagocytic histiocytes [24,25] are the commonest mechanisms of increased platelet destruction during bacterial infection. In critically ill surgical patients, blood loss with subsequent volume replacement using crystalloids or colloids, when sufficiently severe, can significantly decrease the platelet count [26]. Likewise, a fall in platelet count has been noted after blood transfusions, and the incidence of thrombocytopenia has been specifically reported to be directly proportional to the number of RBC transfusions [2].

Patients with more severe underlying illness are more likely to develop thrombocytopenia [1,2,3,5,6,7], and the severity of underlying illness has an important impact on the outcome of ICU patients with severe thrombocytopenia [1,3,5,6,7]. As previously suggested, mortality appears to be proportional to the nadir platelet count [1,3,4]. Relatively old studies [7,12,27] reported that thrombocytopenia was associated with an increased mortality rate in septic patients. Recent studies using multivariate analysis have reported conflicting results, however. Sprung et al [5], in a large prospective study, identified thrombocytopenia (< 100 × 109 platelets/l) as an independent predictor of poor prognosis in septic patients, corresponding to a relative risk of death of 1.66 (95%CI 1.06–2.60). In a recent multicenter prospective study, Brun-Buisson et al [13] also reported a relative risk of late mortality of 1.5 (95% CI 1.2–2.0) in severely ill septic thrombocytopenic patients (platelet count < 50 × 109/l). Another smaller study [6] showed comparable results in septic patients. On the other hand, Pittet et al [28] did not identify thrombocytopenia as an independent predictor of mortality at the onset of sepsis in ICU patients, like Brun-Buisson et al [13] did for early mortality. Finally, when the entire cohort of ICU patients was studied, thrombocytopenia was not identified as a variable that was independently associated with death in two studies [3,6]. These discrepencies may be attributable to differences in the variables and analytical methods used, and could also result from the complex interactions noted in multivariate analysis.

Case–control studies require a considerably smaller sample size than the corresponding longitudinal studies, and is as effective as the cohort study as a basis for causal inference [29]. Critical to the validity of this study is the success in matching cases and control patients for important confounding variables, especially the severity of underlying illness. We used three variables that are strongly correlated with outcome in patients admitted to ICUs to match cases and control patients [17,18]: age, a severity of disease scoring system (APACHE II), and specific diagnostic categories according to the one main reason for admission. Matching was 84.3% successful for 288 possible variables. To verify the adequacy of matching for severity of underlying illness and primary diagnosis independently, we compared the cases and control patients with respect to another seven potentially confounding variables. No statistically significant differences in these indices were observed between cases and control patients. It may be argued that we matched patients at the time of admission. We realize that this does not reflect the severity of illness at the time of thrombocytopenia, and that matching on variables recorded at the time or just before the diagnosis of thrombocytopenia may have yielded different results. The reliability of systems such as APACHE II, SAPS II and ODIN is not sufficiently defined to study the daily probability of death, however. Moreover, thrombocytopenia is linked to admission characteristics and/or diagnosis, because nearly 60% of patients with thrombocytopenia occurred during the first 2 days after ICU admission.

We found that severe thrombocytopenia is associated with increased mortality, mainly in septic patients, thereby confirming the results of previous cohort studies [5,6,13]. In the ICU setting, however, thrombocytopenia is a symptom that has been associated with several acute diseases [1,2,3,23,24,25,26]. It is therefore difficult to distinguish causes from consequences of severe thrombocytopenia clearly in critically ill patients. For example, it is difficult to state that bleeding not represent the results of thrombocytopenia rather than the cause. Although matched case-control studies are attractive because matching assumes comparability for certain variables, the finding of a higher mortality in cases compared with in control patients only implies an association between mortality and thrombocytopenia. Thus, the present analysis identified an increased difference in severity of illness scores between cases and control patients between the day of admission and the day of thrombocytopenia. The finding that illness severity decreased over time in control patients, but remained constant in the cases, supports the notion that severe thrombocytopenia probably largely reflects the unfavourable course of the underlying disease and possibly the aggravating role of thrombocytopenia itself. Thrombocytopenia lasting for less than 3 days confers a similar risk to that of thrombocytopenia for many days. As the present results show, platelet transfusion failed to restore a normal platelet count, but normalization of platelet count, when it occured, was associated with better prognosis, as previously reported [3,24,27].

Patients who require intensive platelet support frequently have complex medical problems, including DIC, clinical bleeding and uncontrolled sepsis. Patients in this setting receive multiple drugs and frequent blood product transfusions. Patients with thrombocytopenia secondary to sepsis often require platelet transfusions, although their response is often suboptimal because of continued platelet destruction, reflecting the persistance of the underlying disease. Likewise, in patients receiving massive transfusion, thrombocytopenia is due both to consumption of platelets and to dilution by transfused blood and colloid fluid resuscitation. These patients may also have platelet function defects due to DIC, hypothermia and 'stunned' recently transfused platelets. Patients may remain midly thrombocytopenic for several days before having a rebound thrombocytosis [30]. Of interest, the mortality impact of thrombocytopenia occurred during the first 2 days after ICU admission, which is quite different from late-onset ICU-acquired thrombocytopenia. This result is in agreement with a previous analysis [13] that indicated that late mortality is essentially associated with the characteristics of the underlying disease.

The mechanism of increased mortality directly related to thrombocytopenia is difficult to assess, but several points should be considered. First, the present results show that severely thrombocytopenic patients required more blood product transfusions. In addition to the possible volume overload, higher postoperative infection rates have been reported in patients who received blood transfusions [31]. Another study [32], however, did not confirm this relationship. Second, platelets play key and multifactorial roles in antimicrobial host defences [11]. For example, in the absence of neutropenia, thrombocytopenia has been positively correlated with an increased incidence and severity of pneumonia in elderly patients [33]. Third, cases had a lower nadir haemoglobin value than control patients. Two recent studies have [9,10] shown that patients with a cardiovascular diagnosis and a high degree of acute illness appear to be at increased risk of death from a moderate degree of anaemia. Finally, severe thrombocytopenia is an important risk factor for spontaneous intracerebral haemorrhage associated with a high mortality in critically ill patients [34].

Conclusion

The present report shows that thrombocytopenia of less than 50×109platelets/l is associated with excess mortality that is independent of the patient's age and initial severity of illness, and leads to excess blood product consumption, thus imposing a significant economic burden. Thrombocytopenia appears to be mostly a marker of severity of underlying processes, rather than causally related to death. Thus, the exact relationship between thrombocytopenia and mortality has yet to be elucidated, especially in septic patients. Further studies of the specific role of thrombocytopenia in shock and infections are necessary. Moreover, the risks and benefits of the various strategies for the management of thrombocytopenic ICU patients should be reevaluated in a variety of clinical settings.

Declarations

Acknowledgments

The authors thank Prof Jean-Yves Fagon for his critical review of the manuscript.

Authors’ Affiliations

(1)
Service d'Anesthésie-Réanimation Chirurgicale.Hôpital Tenon

References

  1. Baughman RP, Lower EE, Flessa HC, Tollerud DJ: Thrombocytopenia in the intensive care unit. Chest 1993, 104: 1243-1247.View ArticlePubMedGoogle Scholar
  2. Hanes SD, Quarles DA, Boucher BA: Incidence and risk factors of thrombocytopenia in critically ill trauma patients. Ann Pharmacother 1997, 31: 285-289.PubMedGoogle Scholar
  3. Stéphan F, Hollande J, Richard O, et al.: Thrombocytopenia in a surgical intensive care unit: incidence, risk factors and outcome. Chest 1999, 115: 1363-1370. 10.1378/chest.115.5.1363View ArticlePubMedGoogle Scholar
  4. Marshall JC, Cook DJ, Christou NV, et al.: Multiple organ dysfunction score: a reliable descriptor of a complex clinical outcome. Crit Care Med 1995, 23: 1638-1652. 10.1097/00003246-199510000-00007View ArticlePubMedGoogle Scholar
  5. Sprung CL, Peduzzi PN, Shatney CH, et al.: Impact of encephalopathy on mortality in the sepsis syndrome. Crit Care Med 1990, 18: 801-806.View ArticlePubMedGoogle Scholar
  6. Lee KH, Hui KP, Tan WC: Thrombocytopenia in sepsis: a predictor of mortality in the intensive care unit. Singapore Med J 1993, 34: 245-246.PubMedGoogle Scholar
  7. Kreger BE, Craven DE, McCabe WR: Gram-negative bacteria. Re-evaluation of clinical features and treatment in 612 patients. Am J Med 1980, 68: 344-355.View ArticlePubMedGoogle Scholar
  8. Practice guidelines for blood component therapy. A report by the American Society of Anesthesiologists task force on blood component therapy. Anesthesiology 1996, 84: 732-747. 10.1097/00000542-199603000-00032
  9. Carson JL, Duff A, Poses RM, et al.: Effect of anemia and cardiovasular disease on surgical mortality and morbidity. Lancet 1996, 348: 1055-1060. 10.1016/S0140-6736(96)04330-9View ArticlePubMedGoogle Scholar
  10. Hébert PC, Wells G, Tweeddale M, et al.: Does transfusion practice affect mortality in critically ill patients? Am J Respir Crit Care Med 1997, 155: 1618-1623.View ArticlePubMedGoogle Scholar
  11. Yeaman MR: The role of platelets in antimicrobial host defense. Clin Infect Dis 1997, 25: 951-970.View ArticlePubMedGoogle Scholar
  12. Svanbom M: A prospective study on septicemia. Clinical manifestations and complications, results of antimicrobial treatment and report of a follow-up study. Scand J Infect Dis 1980, 12: 189-206.View ArticlePubMedGoogle Scholar
  13. Brun-Buisson C, Doyon F, Carlet J, et al.: Incidence, risk factors, and outcome of severe sepsis and septic shock in adults. A multi-center prospective study in intensive care units. JAMA 1995, 274: 968-274. 10.1001/jama.274.12.968View ArticlePubMedGoogle Scholar
  14. Kitchens CS: Disseminated intravascular coagulation. Curr Opin Hematol 1995, 2: 402-406.View ArticlePubMedGoogle Scholar
  15. McCabe WR, Jackson GC: Gram-negative bacteria. I. Etiology and ecology. Arch Intern Med 1962, 110: 847-855.View ArticleGoogle Scholar
  16. Fagon J-Y, Chastre J, Novara A, Medioni P, Gibert C: Characterization of intensive care unit patients using a model based on the presence or absence of organ dysfunctions and/or infection: the ODIN model. Intens Care Med 1993, 19: 137-144.View ArticleGoogle Scholar
  17. Knaus WA, Draper EA, Wagner DP, Zimmerman JE: APACHE II: a severity of disease classification system. Crit Care Med 1985, 18: 818-829.View ArticleGoogle Scholar
  18. Le Gall J-R, Lemeshow S, Saulnier F: A new simplified acute physiology score (SAPS II) based on a European/North American multi-center study. JAMA 1993, 270: 2957-2963.View ArticleGoogle Scholar
  19. American College of Chest Physicians/Society of Critical Care Medicine Consensus Committee: Definitions for sepsis and organ failures and guidelines for the use of innovative therapies in sepsis. Crit Care Med 1992, 20: 864-874.View ArticleGoogle Scholar
  20. Towsend TR: New aspects of case-control studies and clinical trials. In: Prevention and Control of Nosocomial Infections. Edited by Wenzel RP. Baltimore: Williams and Wilkins 1987, 578-590.Google Scholar
  21. Kirkwood BR: . Essential of Medical Statistics. Cambridge, MA: Blackwell Scientific Publications; 1988.Google Scholar
  22. Zhang J, Yu KF: What's the relative risk? A method of correcting the odds ratio in cohort studies of common outcomes. JAMA 1998, 280: 1690-1691. 10.1001/jama.280.19.1690View ArticlePubMedGoogle Scholar
  23. Wilson JJ, Neame PB, Kelton JG: Infection-induced thrombocytopenia. Semin Thromb Hemostas 1982, 8: 217-233.View ArticleGoogle Scholar
  24. François B, Trimoreau F, Vignon P, et al.: Thrombocytopenia in the sepsis syndrome: role of hemophagocytosis and macrophage colony-stimulating factor. Am J Med 1997, 103: 114-120. 10.1016/S0002-9343(97)00136-8View ArticlePubMedGoogle Scholar
  25. Stéphan F, Thiolière B, Verdy E, Tulliez M: Role of hemophagocytic histiocytosis in the etiology of thrombocytopenia in patients with sepsis syndrome or septic shock. Clin Infect Dis 1997, 25: 1159-1164.View ArticlePubMedGoogle Scholar
  26. Spiess BD: Hemorrhagic problems during the immediate postoperative period. In: Post anesthesia Care. Edited by Vender JS, Spiess BD. Philadelphia: WB Saunders; 1992, 131-140.Google Scholar
  27. Oppenheimer L, Hryniuk WM, Bishop AJ: Thrombocytopenia in severe bacterial infections. J Surg Res 1976, 20: 211-214.View ArticlePubMedGoogle Scholar
  28. Pittet D, Thiévent B, Wenzel RP, et al.: Bedside prediction of mortality from bacteremic sepsis. A dynamic analysis of ICU patients. Am J Respir Crit Care Med 1996, 153: 684-693.View ArticlePubMedGoogle Scholar
  29. Austin H, Hill HA, Flanders D, Greenberg RS: Limitations in the application of case-control methodology. Epidemiol Rev 1994, 16: 65-76.PubMedGoogle Scholar
  30. Schroeder ML: Principles and practice of transfusion medicine. In: Wintrobe's Clinical Hematology. Edited by Lee GR, Foerster J, Lukens J, et al. Baltimore: Williams and Wilkin; 1998, 817-874.Google Scholar
  31. Edna TH, Bjerkeset T: Association between blood transfusion and infection in injured patients. J Trauma 1992, 33: 659-661.View ArticlePubMedGoogle Scholar
  32. Vamvakas EC, Moore SB, Cabanela M: Blood transfusion and septic complications after hip replacement surgery. Transfusion 1995, 35: 150-156. 10.1046/j.1537-2995.1995.35295125738.xView ArticlePubMedGoogle Scholar
  33. Feldman C, Kallenbach JM, Levy H, et al.: Comparison of bacteraemic community-acquired lobar pneumonia due to Streptococcus pneumoniae and Klebsiella pneumoniae in an intensive care unit. Respiration 1991, 58: 265-270.View ArticlePubMedGoogle Scholar
  34. Oppenheim-Eden A, Glantz L, Eidelman LA, Sprung CL: Spontaneous intracerebral hemorrhage in critically ill patients: incidence over six years and associated factors. Intens Care Med 1999, 25: 63-67. 10.1007/s001340050788View ArticleGoogle Scholar

Copyright

© Current Science Ltd 1999

Advertisement