Anticoagulation (AC) in thrombocytopenic (TCP) cancer patients is not uncommon. Management is complex and informed mainly by inconsistent retrospective studies on venous thromboembolism. A few descriptive studies on factors influencing management show a variance in reported practice. Analytical data on patient characteristics influencing AC management are lacking.
Objectives: 1) Identify the patient characteristics associated with AC management strategies in TCP patients with hematological malignancy. 2) Evaluate whether a physician assessment of bleeding and thrombotic risk is associated with AC management.
We designed a decision-making analysis in the form of a case vignette study in a multinational and multicenter setting. First, in semi-structured interviews, 11 hematologists and thrombosis & hemostasis (T&H) specialists in Israel and the Netherlands were asked which patient variables influence their management of AC for any indication in this scenario. Then, the patient variables were refined based upon the number of interviewees selecting a given option. The 5 selected attributes with 2-5 levels each (Table 1) were then entered into an algorithm creating a balanced and reduced design out of a full factorial model. This generated 30 clinical cases of TCP patients with different combinations of hematological disease and treatment, AC indication and duration, and bleeding history. Each responder received 5 AC case vignettes and had to choose between the following options: 1) no change in AC; 2) no change in AC but transfuse platelets; 3) hold AC; 4) modify AC (i.e. AC class/dose). The survey was then re-piloted, designed as a website and distributed via mailing lists to national hematology and T&H societies in Israel, the Netherlands and Italy (N=886). In a multivariate analysis, mixed effects binomial logistic regression models were used to calculate the odds ratios of patient attributes in comparison to their respective reference levels, for each management strategy (vs. all others; i.e. one-vs-all coding). Random intercepts for each physician and country were incorporated where possible, and random slopes were calculated for thrombosis and bleeding risk perceived by each physician.
There were 168 responders, in Italy (59), Israel (52), the Netherlands (47) and other countries (10), answering 774 unique cases. In the 3 countries where the survey was distributed via professional societies, 18% (158/886) of population directly contacted, responded. The majority (80%) was from academic hospitals and 46% (77) from tertiary referral centers, and only 38% had institutional guidelines guiding management. Forty-one percent (69) specialized in T&H, 47% (79) in other malignant or benign hematology and 12% (20) in transfusion medicine. The responders reported a median professional experience of 15 years [IQR=17] and estimated seeing a median of 5 patients [IQR 8] with AC and TCP per month.
Management varied across countries. AC modification was the most common choice, made by 84% (141) of responders at least once. AC was held by 55% (93), while only 37% (62) and 42% (71) chose no change with or without platelet transfusion, respectively. Figure 1 depicts the odds ratios of utilizing different AC management strategies across patient and physician attributes. The physicians' thrombosis and bleeding risk assessment of individual cases seemed consistent with choice to hold or not change AC treatment, but not with transfusing platelets (Figure 1).
Lower platelet counts of 20,000/µL (vs. 40,000/µL), high risk AC indications and possibly recently indicated AC (2 weeks vs. 6 months) and leukemia with asparaginase-based treatment, are associated with platelet transfusion without change in AC. Holding therapy was more likely with lower platelets, major bleeding 4 months earlier (vs. never), while less likely with symptomatic pulmonary embolism (and aspirin for stroke). Thus, appropriate thrombotic and bleeding risk factors, representing all 5 assessed attributes, were taken into account when holding AC or transfusing platelets. While future evidence should inform us whether this decision making process is clinically justified, this analysis does suggest that management may be oversimplified by current practice guidelines.
Falanga: Advisory Board: Bayer, Janssen Speakers' bureau: Rovi, BMS: Membership on an entity's Board of Directors or advisory committees, Speakers Bureau.
Asterisk with author names denotes non-ASH members.