Background: Platelet transfusions play an essential supportive role in treatment of hematologic cancer patients who are predisposed to thrombocytopenia. However, the optimal platelet count threshold for prophylactic transfusion is unknown in sub-Saharan Africa. We sequentially lowered thresholds for prophylactic platelet transfusion in Ugandan cancer patients and measured bleeding and adherence to these thresholds.

Methods: We prospectively followed patients admitted to the Uganda Cancer Institute (UCI) with a hematological malignancy from October 2014 to October 2015 in 3 sequential 4-month time-periods using incrementally lower thresholds for prophylactic platelet transfusion: platelet counts ≤ 30 x 109/L in period 1, ≤ 20 x 109/L in period 2, and ≤ 10 x 109/L in period 3. Clinically significant bleeding was defined as WHO grade ≥ 2 bleeding. The number of platelet units that UCI requested and received during the study period was abstracted from the Uganda Blood Transfusion Service records. We used GEE to compare the frequency of clinically significant bleeding, platelet transfusions, and log10-transformed platelet counts by study period, adjusting for age, sex, cancer type, chemotherapy, baseline platelet count, and baseline hemoglobin. Cox regression was used to compare survival by period.

Results: Overall, 188 patients were enrolled. The median age was 18 years (range 1-78), 26 (range 1-80) and 22 (range 2-75) in each period; 56-64% were male. Previous or current chemotherapy use was similar among the 3 periods and ranged from 39-45% of participants. Acute lymphoblastic leukemia, acute myeloid leukemia, and non-Hodgkin lymphoma accounted for most malignancies. Sixty-two patients in period 1 accrued 998 days of follow up, 69 in period 2 accrued 1300 days, and 57 in period 3 accrued 1047 days. The geometric mean platelet count x109/L was 36 (range 0-730) in period 1, 26 (range 0-723) in period 2, and to 21 (range 1-924) in period 3 (adjusted p = 0.55). Platelet transfusions were given to 42%, 55% and 45% of patients and on 65/842 (8%), 154/1268 (12%) and 84/1058 (8%) days for the 3 time periods. In adjusted models, period 3 had significantly fewer transfusions than period 1 (RR = 0.6, 95% CI 0.4-0.9; p = 0.01) and period 2 (RR=0.5, 95% CI 0.4-0.7; p<0.001). Most (54-60%) platelet transfusions were given as prophylaxis for low platelets only. The median pre-transfusion platelet counts x109/L were 10 (range 0-62), 7 (range 0-73), and 5 (range 1-35) for the 3 periods, respectively. The mean number of 60-80mL platelet concentrate units requested for the entire UCI per day was 17 for period 1, 21 for period 2, and 25 for period 3 but the mean platelet units received was only 5.8, 6.4 and 7.7, respectively.

Among all days with platelet counts below the trigger threshold, platelet transfusions were given within 1 day of the low platelet counts on 29 (32%) days in period 1, 76 (46%) days in period 2, and 40 (34%) in period 3. Platelet transfusions were rarely given for prophylaxis when platelet counts were above the threshold, and the overall number of days that transfusions were given according to our defined thresholds was 132 days (68%) in period 1, 241 days (72%) in period 2 and 224 (74%) in period 3.

Eighteen patients (30%), 23 (30%) and 15 (23%) had clinically significant bleeding on at least one day in period 1, 2 and 3, respectively. Of the 559 total days with bleeding assessed, clinically significant bleeding was noted on 45 days (8%) of 559 patient-days in time period 1, 72/809 (9%) days in period 2 and 38/755 (5%) in period 3 (p = 0.41 in adjusted analyses). Thirteen (21%) patients died in period 1, 15 (22%) in period 2, and 11 (19%) in period 3 (p = 0.72 in adjusted analyses).

Conclusion:Platelet counts were low among UCI inpatients with hematological malignancies and platelet transfusions were given to nearly half of patients. Low availability of blood products likely hindered clinicians' ability to provide transfusions when platelet counts dropped below our defined trigger thresholds, but overall adherence was good. Lowering the threshold for platelet transfusion led to fewer transfusions and did not change the incidence of clinically significant bleeding or mortality, suggesting that a threshold of 10 x 109/L platelets used in resource-rich countries may be a safe level for transfusions in Uganda.


No relevant conflicts of interest to declare.

Author notes


Asterisk with author names denotes non-ASH members.