Types and outcomes of cytopenia in critically ill patients

This cross-sectional study was conducted in the ICU of Imam Khomeini Teaching Hospital, affiliated to Tehran University of Medical Sciences, Tehran, Iran from 22 July 2016 to 15 May 2017. The study was approved by the Hospital Medical Ethics Committee and Institutional Review Board, and written informed consents were obtained from the patients or their family members. A sample size of 117 patients was calculated considering power of 80%, alpha error of 5% and the dropout rate of 10%.

ICU patients more than 18 years old presenting with any kind of cytopenia for more than 2 days were randomly selected and followed-up until being discharged from the ICU or being expired. The patients’ demographic and clinical data including, sex, age, the reason for admission, underlying diseases, medication history, drugs administered during the ICU stay and duration of ICU stay were all recorded. Prescribed drugs were categorized into 14 groups based on The Anatomical Therapeutic Chemical classification system [22]. The severity and the causes of each type of cytopenia were classified based on previous related studies [12,13,23–25].

The definitive diagnosis of cytopenia was made if at least two consecutive peripheral blood smears or complete blood counts showed blood cell counts below the lower limit of normal. The complete data regarding the patients’ cytopenia, including the onset (whether the patients had cytopenia on admission or they developed during the stay in ICU), the cause, the outcome (reversible or irreversible), the duration of cytopenia and the average value of hematological blood cells during the ICU stay were all collected. The outcome was considered as reversible if blood cell counts reached normal range or had an ascending pattern toward mild cytopenia during the ICU stay. White blood cell counts over 10,000 cells/μl for at least 2 consecutive days were reported as leukocytosis. Any abnormal findings in urine, cerebrospinal fluid, and blood, as well as trachea culture were noted during ICU admission. The mean value of serum creatinine, plasma magnesium level, plasma phosphorus level, C reactive protein, erythrocyte sedimentation rate, plasma iron level, plasma ferritin level and transferrin saturations were also documented. Any axillary body temperature above 38.3°C was considered as fever [26]. For patients with direct liver injury, including liver metastatic cancer, viral hepatitis, nonalcoholic fatty liver disease and autoimmune hepatitis, the Child–Pugh score was calculated based on the first reported value of albumin, total bilirubin, prothrombin ratio and international normalized ratio. Radiotherapy and drug-induced cytopenia were considered as causes of cytopenia when there were no confirmed reasons for blood cell counts below the lower limit of normal [13,27]. Disease-related causes of lymphopenia included Cushing syndrome and metastatic cancer [15]. The probability of causality assessment for cytopenia was recorded as probable, likely and definite. The case was considered as definite if it was supported by clinical condition and there were no other reasons for cytopenia, while if the cytopenia was induced by one or two causes and four or more causes other than the main cause, the case was identified as likely and probable, respectively [28].

The data were analyzed by SPSS version 24 (IL, USA). Categorical variables were presented as absolute and relative (percentage) frequencies. Continuous variables were expressed as mean ± standard deviation. Pearson test and Kruskal–Wallis test were used to show any possible correlation between variables. Fisher’s exact test and Chi-square test were used to statistically compare qualitative and quantitative data, respectively. The influence of different factors related to cytopenia on the overall outcome (expired or discharged) was evaluated by multivariate logistic regression analysis. p < 0.05 was considered as the significance level.

Totally, 117 cytopenic patients including 69 women (60%) and 48 men (40%) were assessed during the study period. The mean (SD) age of the patients was 53 (18) years and 58.97% of them were females. The demographic and clinical data of patients are shown in Table 1.

Totally, 13 patients were admitted to ICU due to elective surgery, 15 patients for trauma and 89 patients for other medical conditions, including pneumonia; sepsis; Guillain–Barré syndrome; pancytopenia; venous thrombosis; and hemolysis, elevated liver enzymes and low platelet count syndrome.

The range of ICU stay was 2–79 days (with a median of 6 days). Totally, 29 patients (24.8%) died during the study period and 88 patients were discharged from ICU.

A total of 100 patients suffered from underlying diseases. The most prevalent underlying diseases were cardiovascular (48.69%), cancer (41%) and endocrine diseases (24.34%). The onset, outcome and mean value of duration of each type of cytopenia, as well as the mean value of blood cell counts are shown in Table 2. Most of patients did not have cytopenia on admission and became cytopenic during ICU stay. Cytopenia was reported irreversible for the majority of the patients. The results showed that 99.14, 32.17, 27.82, 19.13, 6.83, 5.21 and 5.21% of cytopenic patients suffered from anemia, lymphocytopenia, thrombocytopenia, leukopenia, neutropenia and pancytopenia, respectively.

The causes, frequencies and probability of causality assessment for each type of cytopenia are reported in Table 3. Blood loss due to surgery and trauma was the main cause of anemia in both expired and discharged groups by 61 cases (52.6% overall). One pancytopenic patient who had abortion with methotrexate due to ectopic pregnancy was the only case of drug-induced anemia. The reasons responsible for leucopenia in nearly 55% of patients were sepsis and viral infection. Six neutropenic patients distributed equally (33.3%) between three subclassifications (sepsis and viral infection, hematologic malignancy and drug induced and radiotherapy). Blood loss was the most common cause of lymphopenia (22 cases [59.5%]). Eight cases of thrombocytopenia were due to coagulopathy (overall 25%). Pancytopenia occurred in six patients, mainly due to the administration of chemotherapy drugs. All of these patients died.

The association between the overall outcome and demographic data including age, sex, cause of ICU admission, presence of underlying disease and the history of taking medications before ICU admission are presented in Table 4. The mean of age was significantly higher (p = 0.02) in expired patients in comparison to discharged ones (60.93 ± 17.4 vs 50.96 ± 18.6 years). However, no significant difference was observed between overall outcome and other mentioned factors.

Further analysis was conducted to show the impact of each type of cytopenia, the mean value of blood cell counts, reversible cytopenia and the cytopenia developed after ICU admission on the overall outcome. The results indicated that the overall outcome was significantly affected by Hgb concentration (p < 0.0001), leucopenia (p < 0.0001), neutropenia (p = 0.004), lymphopenia (p = 0.002), thrombocytopenia (p < 0.0001) and pancytopenia (p < 0.0001). The discharged group had a significantly higher rate of reversible lymphopenia than the expired group (p = 0.008). However, the reversibility of other types of cytopenia had no significant association with the outcome (Table 4).

Longer duration of anemia, lymphocytopenia and thrombocytopenia was associated with longer ICU LOS (p < 0.001 for all) and higher rate of mortality regarding overall outcome (p < 0.001; p < 0.001 and p = 0.004, respectively).

Discharged patients with lower amount of Hgb had a significantly longer ICU LOS (p = 0.022); however, this association was not significant in expired patients (p = 0.614). The outcome was not affected by anemia severity in both groups (discharged p = 0.472 and 0.385 in discharged and expired patients, respectively). Patients with mild thrombocytopenia had relatively higher ICU LOS than patients with normal range (p = 0.033). No significant relation was seen between the severity of neutropenia and ICU LOS in both discharged and expired patients.

According to multivariate logistic regression analysis and by adjusting for different variables (i.e., age, cause of ICU admission, hemoglobin amount, leukopenia, neutropenia, lymphopenia, thrombocytopenia and pancytopenia), only thrombocytopenia could independently predict mortality (OR: 0.485 and p = 0.016).

Our results showed that 99.14% of cytopenic patients suffered from anemia and Hgb concentration below the lower limit of normal was the most common defect in blood cell counts among different types of cytopenia. Anemia is a common complication in ICU patients; almost 95% of our patients are anemic on day 3 of admission. As discussed in previous studies, the incidence of anemia and its management can affect the ICU LOS and mortality [29,30]. Culleton et al. followed up 17,030 older adults for 3 years and demonstrated that anemia is associated with a higher risk of mortality and hospitalization [20]. In patients who are predicted to have a longer duration of ICU stay, correction of nutritional deficiencies (iron, B12 and B9) must be considered to prevent anemia and consequent complications. Most of the critically ill patients demonstrate the pattern of anemia of inflammation and they are less sensitive to erythropoietin [31]. In our study, interventions and responses in anemic patients were not considered but the Hgb level was used as an indicator of prognosis and outcome. Anemia was caused mostly by blood loss (trauma and surgery) in our patients (52.6%). In this regard, phlebotomy is assumed as a major modifiable cause in critically ill patients [32]. Von Ahsen et al. examined 96 patients treated in the ICUs for more than 3 days. They found out that 77% of patients suffered from anemia. Also, it was shown that overt bleeding or diagnostic blood sampling were not the main causes of anemia in these patients, but, rather, anemia resulted mainly from various other reasons, such as occult gastrointestinal bleeding and renal replacement therapy [33]. It is reported that in severe traumatic brain injury, anemia (Hgb < 9 g/dl) along with compromised brain tissue oxygen tension is associated with inappropriate outcomes [34]. We could not detect the benefit of anemia reversibility in patients’ prognosis.

In our study, the patients were classified based on the severity of anemia. The results showed that a lower amount of Hgb was associated with a longer duration of hospitalization in discharged patients and just the incidence of anemia, independent of the severity, could affect the outcome in a statistically significant manner. Hgb concentration was statistically higher in discharged patients in comparison to expired individuals. Rasmussen et al. in their study showed that there is a relation between anemia (Hgb < 12 g/dl) and mortality in mechanically ventilated chronic obstructive pulmonary disease patients [35]. In the postoperative setting, the direct association between anemia and mortality was also observed [36–38]. There was a similar result regarding mortality and LOS with Hgb concentration of less than 9 g/dl [39].

It has been shown that more than 50% of critically ill patients experience thrombocytopenia during their ICU stay and 5–20% of these patients suffer from severe thrombocytopenia (platelet counts below 50 × 109/l) [40]. ICU patients usually have several comorbidities that affect platelet hemostasis and make this population prone to thrombocytopenia. Six major mechanisms have been introduced for thrombocytopenia induction, including hemodilution, increased platelet consumption, increased platelet destruction, decreased platelet production, improved platelet sequestration and pseudothrombocytopenia (laboratory artifacts) [41]. Defining the most common causes of cytopenia in each setting can help clinicians make treatment choices. More than a fourth (27.82%) of our cytopenic patients suffered from thrombocytopenia. In Levi et al. study [8], sepsis was the most common cause of thrombocytopenia; however, we found coagulopathy as the main cause (25.0%), followed by chronic diseases (18.75%), particular drugs or radiotherapy (15.62%) and massive blood loss (15.62%).

Lower platelet count and longer duration of thrombocytopenia lead to higher mortality in ICU. A similar study in ICU patients found out that more severe thrombocytopenia and lower platelets are associated with a higher mortality rate. Thus, it can be used as a valuable prognostic marker [11,24,42]. Our results showed that among all of the variables (incidence, duration and severity of each type of cytopenia), the only incidence of thrombocytopenia was significantly associated with mortality. Some studies demonstrated that both the incidence and severity of thrombocytopenia contribute to higher mechanical ventilation requirement and mortality in critically ill patients [42,43]. Akca et al. implicated that late onset thrombocytopenia in ICU leads to higher mortality rates in comparison to early thrombocytopenia. The patients were categorized into two groups; groups 1 and 2 developed thrombocytopenia at the first and second weeks of admission, respectively. It was reported that the second group had a higher mortality rate [9]. Also, another study showed that late onset thrombocytopenia leads to a higher mortality rate [11]. Based on our results, the onset of thrombocytopenia had no significant effect the on overall outcome.

Near a third (32.17%) of our cohort were lymphopenic. Blood loss (e.g., trauma and surgery) was the main cause of lymphopenia (59.5%) in the present study. According to a retrospective review of a database of adult patients requiring ICU admission between 2002 and 2013 in the UK, 78% of patients had lymphopenia at admission and 91% of them developed at least one episode of lymphopenia during ICU stay [18].

Higher duration of lymphopenia can prolong the ICU LOS and negatively affect the outcome. Furthermore, the incidence of lymphopenia was associated with a higher mortality rate. A higher rate of reversible lymphopenia was observed among discharged patients. In contrast, prolonged and irreversible lymphopenia can lead to higher mortality. Lymphopenia developed during ICU-stay could not lead to higher mortality in comparison to that exists on ICU admission. Vulliamy et al. showed that the presence of lymphopenia on ICU admission and its persistence until day 3 could increase the risk of infection and just persistent lymphopenia, independent of onset, caused higher mortality [17]. Our current study also demonstrated that the longer length of lymphopenia was associated with increased mortality and ICU LOS. An increase in ICU LOS leads to an elevated overall hospitalization period. Our findings are consistent with those in which lymphopenia was considered as a predictor for poor outcome in terms of a longer hospital stay, in-hospital mortality and 1-year mortality after discharge [44].

Lymphopenia and neutrophil–lymphocyte count ratio can predict the possibility of bacteremia and somehow estimate the inflammation [45]. It has been shown that lymphopenia and an increase in the neutrophil count at the same time in critically ill patients with a prior history of diabetes is associated with poor prognosis [46]. However, based on our data, lymphocyte count did not affect mortality and length of stay.

Pancytopenia is related to poor prognosis, as all of pancytopenic patients expired in our study. However; one cannot make a definite statement in this regard due to the small population size (only 6 patients). All of these patients were anemic. On the other hand, the most common cause of pancytopenia was probably nutritional megaloblastic with concurrent iron deficiency [47]. Low doses of methotrexate can induce pancytopenia [48] and one of our cases had received this medication as an abortive therapy. Therefore, drug-induced pancytopenia is probable. We had three patients who had received induction therapy for acute myeloid leukemia and their pattern of cytopenia was compatible with chemotherapy-induced pancytopenia described by other studies [49].

The major strength of our investigation is the comprehensive study of the frequency of different types of cytopenia, related risk factors and its impact on clinical outcome in critically ill patients in the ICU.

Our study had some limitations which should be considered. We had no access to information before or after ICU admission. In addition, only random cytopenic patients were evaluated, and we had not the control group concurrently. Therefore, we cannot report the definite prevalence and also possible risk factors of cytopenia at our center. This negatively impacts the generalization aspect of the obtained results. The number of neutropenic and leukopenic patients was too small to detect any probable statistical differences. Thus, further studies with larger sample sizes are required.

Our results demonstrated that the rate of cytopenia, especially anemia, is relatively high in critically ill patients. Among all assessed variables, only thrombocytopenia could significantly and independently predict ICU mortality in our cohort. Knowing causes of cytopenia and its pattern, helps the clinicians to prevent cytopenia or choose the best treatment plan for cytopenic patients and reduce the cost imposed on patients and the healthcare system.

Assessing the effect of treatment and efficacy of managements of cytopenia along with determining relevant and possible risk factors of cytopenia in ICU patients are the era for further studies.

The authors have no relevant affiliations or financial involvement with any organization or entity with a financial interest in or financial conflict with the subject matter or materials discussed in the manuscript. This includes employment, consultancies, honoraria, stock ownership or options, expert testimony, grants or patents received or pending, or royalties.

No writing assistance was utilized in the production of this manuscript.

Research data are available per request.