The present study was a prospective observational study which was conducted in ICU of Shri Krishna Hospital, between 1^st February 2015 to 1^st November 2016. Institutional Ethical clearance was obtained with approval letter number-HMPCMCE: HREC/2015/OUT. No. 169/15 (Human research Ethics Committee, HM Patel centre for Medical Care and Education, Karamsad, Gujarat, India).

Inclusion criteria: All individuals above 18 years of age having platelet count <1,50,000/μL or a fall in platelet count ≥500% from the ICU admission value were included.

Exclusion criteria: Patients less than 18 years of age and with platelet count >1,50,000/μL were excluded.

Sampling: Total number of admissions in critical care from 1^st February 2015 to 1^st November 2016 were 8055. Out of them there were 2403 patients who had thrombocytopenia (29.83%). Out of 2403 patients, 200 were those who had thrombocytopenia and sepsis. These 200 patients were taken in the study.

All patients with thrombocytopenia on admission or anytime during the stay in ICU were evaluated clinically and haematologically, after obtaining required consent. In case of patients developing thrombocytopenia after admission to ICU, the day of development of thrombocytopenia was considered as day one and was evaluated for the next days in ICU till the final outcome.

The data included demographic details of the patient, history of present illness, drug history, cause of admission to ICU, length of ICU stay as well as total hospitalisation (in number of days), complete blood count, kidney function tests, liver function tests, coagulation profile and general examination data.

Subsequent platelet counts were done on various days after admission to ICU and thrombocytopenia was considered as a platelet count of <150,000/μL. Thrombocytopenia is classified as very severe thrombocytopenia (platelet count ≤10,000/μL), severe thrombocytopenia (platelet count >10,000/μL but ≤50,000/μL) moderate thrombocytopenia (platelet count >50,000/μL but ≤1,00,000/μL) and mild (1,00,000-1,50,000/ μL) [8]. Sepsis is defined by the presence of infection and two or more signs suggesting reduced organ perfusion. In case of availability of more than one platelet count of a patient on a single day, the lower value was taken into consideration. Regarding drug-induced thrombocytopenia, the following general criteria was used for suspecting the presence of drug-induced thrombocytopenia [8].

The drug should bear a direct relationship with thrombocytopenia, usually within a week after beginning it for the first time [8].

After data collection and analysis, the rate of occurrence of thrombocytopenia in patients admitted to medical ICU was calculated. Mortality or recovery or prolonged hospitalisation was correlated with thrombocytopenia.

Descriptive statistics {Mean (SD), Frequency (%)} were used to depict characteristics of study population. Associations of different explanatory variables with thrombocytopenia at univariate level were evaluated using Chi-square/t-test depending on nature of the variables. The ANOVA was used to compare SOFA [9] and APACHE II [10] scoring systems. Epi InfoVersion 7 was used to collect data.

Mean age of the patients was 49 (SD 18.24) years and mean duration of hospital stay was 8.85 (SD 6.58) days. Maximum number of patients belonged to 18-30 and 51-60 years with 48 patients each. Out of 200, 115 (57.5%) were males and 85 (42.5%) were females [Table/Fig-1].

Total 99 (49.5%) patients presented with fever, 15 (7.5%) presented with bleeding from any site and 11 (5.5%) had presented with trauma. In the patients who had come with bleeding from any site as a presenting symptom, 4 (26.67%) patients had very severe thrombocytopenia (platelet count ≤10,000/μL), 8 (53.33%) patients had severe thrombocytopenia (platelet count >10,000/μL but ≤50,000/μL) and 3 (20%) patients had moderate thrombocytopenia (platelet count >50,000/μL but ≤1,00,000/μL). In patients who had major trauma (11 patients out of 200) had developed thrombocytopenia secondary to platelet consumption.

Majority of the patients had severe thrombocytopenia 72 (36%), 17 patients had very severe thrombocytopenia [Table/Fig-2].

There were 72 (36%) patients with deranged liver functions, 73 (36.5%) had deranged renal functions and 59 (29.5%) had deranged coagulation profile. Total 17 (8.5%) cases had chronic liver disease and 6 (3%) cases of patients who had thrombocytopenia had underlying chronic kidney disease. Total 89 (44.5%) of the 200 had sepsis and out of them 24 patients died, therefore, mortality rate among those with sepsis in thrombocytopenia was 27%. Mean SOFA score and APACHE II scores were 6.32 (SD 3.44) and 14 (SD 7.89), respectively.

[Table/Fig-3] shows a significant association between sepsis scoring systems and patient outcomes (SOFA p=0.0438, APACHEII p=0.0492).

Out of 200 patients, 45 developed thrombocytopenia due to some medication/drug. Most commonly implicated drugs were chemotherapeutic agents [Table/Fig-4].

According to a study by Chakraverty R et al., medical and surgical Intensive Care Units (ICUs) had found platelet counts less than 100,000/μL in 20-40% of patients at some point during their stay in the unit, whereas severe thrombocytopenia (platelet counts <50,000/μL) occurred in 10-20% of patients [11]. The present study included all the critical care units in the hospital, and it revealed an overall incidence of thrombocytopenia of 29.83%. The mean age of the patients was 49 years with the minimum age being 18 years and maximum age being 86 years.

In this study, the mortality rate of patients with sepsis having thrombocytopenia was 27%. The association of chronic kidney disease and chronic liver disease in the thrombocytopenic patients was studied and concluded that 8.50% (n=17) cases had chronic liver disease and 30% (n=6) cases of patients who had thrombocytopenia had underlying chronic kidney disease, hence the association of thrombocytopenia and chronic liver disease was stronger than chronic kidney disease (p<0.001). This is comparable to the study by Afdhal N et al., which associates thrombocytopenia (platelet count, 50,000/μL-75,000/μL) with chronic liver disease (cirrhosis) occurring in approximately 13% of patients [12]. A 99v (49.5%) patients had presented with fever, 15 (7.5%) patients had presented with bleeding from any site and 11 (5.5%) patients had presented with trauma. In the patients who had come with bleeding from any site as a presenting symptom, four patients had very severe thrombocytopenia eight patients had severe thrombocytopenia and three patients had moderate thrombocytopenia. In patients who had major trauma (11 patients out of 200) and had developed thrombocytopenia secondary to platelet consumption, nadir of thrombocytopenia was day 3 or 4 (Mean day of nadir of thrombocytopenia-3.45) after which thrombocytopenia gradually recovered. Similar patterns of platelet count courses were described by Nijsten MW et al., in ICU patients after various surgeries [13].

Forty-five (22.5%) patients had a suspected DITP. Most commonly implicated drugs were chemotherapeutic agents, Linezolid, Valproate, Phenytoin, Beta-lactam and Heparin. Chemotherapy induced thrombocytopenia was seen in 24 (53.33%) patients out of 45. Heparin induced thrombocytopenia was seen in 2 (4.44%) patients. 89 (44.5%) patients had sepsis induced thrombocytopenia followed by viral haemorrhagic fever and consumptive coagulopathy 13.5% each. Similar study was done by Lee KH et al., which had 107 subjects in critical care out of which 53 patients were septic of which 31 patients developed thrombocytopenia. It was observed that majority of the patients in critical care had severe thrombocytopenia 72 (36%) patients and very severe thrombocytopenia was seen in only 17 patients (8.5%) among all the thrombocytopenic patients. Various studies found thrombocytopenia in 35%-45% of ICU patients, with a somewhat greater variability of 5%-20% for severe thrombocytopenia [7,14,15].

The study had used SOFA and APACHE II scores for assessment of severity of illness in thrombocytopenic patients in critical care, mean values (±SD) of both these scores were 6.32 (±3.44) and 14 (±7.89). The level of association between length of stay and outcome was not significant (p=0.1293) [Table/Fig-3]. In thrombocytopenic patients with sepsis, those patients who had poor outcome had the highest mean SOFA score of 9.04 (death) and 6.96 (DAMA) respectively. Whereas those with good outcomes had a mean SOFA score of 6.66 (discharge) and 6.20 (DOR) respectively (p=0.0438). Similarly, the association between APACHE II and outcome in sepsis with thrombocytopenia was also very significant (p=0.0492). A retrospective cohort study by Balci C et al., analysed data collected from patients admitted to a postoperative ICU. Patients were divided in two groups: Thrombocytopenic patients (P group) and non-thrombocytopenic (C group). The APACHE II score did not show statistically significant difference between groups in their study. The evaluation of the SOFA score, rather than APACHE II, correlates better with the outcomes in patients with sepsis and thrombocytopenia [16].

The limitation of this study was the small sample size. If the trend of platelet count could be obtained for all the subjects in the study it would have given a better understanding of trends and patterns of thrombocytopenia due to various causes.

Considering the results of this study, it was concluded that platelet counting is a simpler and more affordable method when treating septic patients. A daily assessment of thrombocytopenia should be considered as an important prognostic factor in sepsis, although authors have validated lengthy scores like SOFA and APACHE II to predict mortality. When treating thrombocytopenia in septic patients, interfering with platelet function, physician should treat the cause of thrombocytopenia to reduce the mortality.