Abstract
Background:
The presence of a hypercoagulable disorder such as heparin-induced thrombocytopenia (HIT) may protect against anticoagulant-associated bleeding.
Objectives:
To determine the incidence of major bleeding in patients with suspected HIT.
Methods:
We performed a retrospective analysis of 310 patients suspected of having HIT from the Hospital of the University of Pennsylvania and an affiliated community hospital. We compared the cumulative incidence of major bleeding following suspicion for HIT by ultimate HIT status (HIT+ or HIT–) and exposure to an alternative anticoagulant (Tx+ or Tx–). Secondary outcomes included the incidence of new/progressive thrombosis and 30-day mortality.
Results:
The incidence of major bleeding was high in the HIT+Tx+, HIT –Tx+, and HIT–Tx– groups (35.7%, 44.0%, and 37.3%, respectively). The time to first major bleeding event did not differ between groups (p=0.24). Factors associated with increased risk of major bleeding included intensive care unit admission (HR 2.24, 95%CI 1.44–3.47), platelet count < 25 × 109/L (HR 2.13, 1.10–4.12), and renal dysfunction (HR 1.56, 1.06–2.27). 35.7% of HIT+Tx+, 13.8% HIT–Tx+, and 9.3% of HIT–Tx– patients experienced new or progressive thrombosis. Mortality was similar among the 3 groups (26.2% HIT+Tx+, 34.5% HIT–Tx+, and 26.7% of HIT–Tx– (p=0.34)).
Conclusions:
Among patients with suspected HIT, major bleeding was common regardless of HIT status. Contrary to our hypothesis, HIT+ patients were not protected from major bleeding. A better understanding of bleeding risk is needed to inform management decisions in patients with suspected HIT.
Keywords: Thrombocytopenia, Heparin/adverse event, Anticoagulants/adverse event, Anticoagulants/therapeutic use, Thrombosis
Background
Heparin-induced thrombocytopenia (HIT) is a prothrombotic complication of heparin exposure. Without appropriate treatment, patients with HIT develop thrombosis, which may be limb- or life-threatening, at an initial rate of approximately 6.1% per day [1]. HIT confirmatory laboratory tests are not available on a rapid basis at most institutions [2]. Thus, clinical prediction scores such as the 4Ts score or HIT Expert Probability (HEP) score are useful for estimating the pre-test likelihood of HIT and guiding initial management decisions [3]. With an intermediate or high score, clinical practice guidelines recommend stopping heparin and starting an alternative anticoagulant while awaiting confirmatory laboratory test results [4, 5].
The choice to treat patients empirically for HIT is a high-stakes decision. Treatment for HIT exposes patients to costly alternative anticoagulants, many of which have no reversal agent, and their attendant risk of bleeding [6–8]. The anticoagulants used for treatment of HIT include the parenteral direct thrombin inhibitors, argatroban and bivalirudin [9, 10], the indirect factor Xa inhibitors, fondaparinux and danaparoid [11], and the direct oral anticoagulants (DOACs). The incidence of major bleeding with parenteral alternative anticoagulants is reported to be ~1% per day [9, 11, 12]. This rate is derived from clinical trials of patients with HIT. The major bleeding rate in a real-world setting and in patients who are treated empirically for but ultimately found not to have HIT has not been well-studied [13]. Indeed, such patients may be at heightened risk for alternative anticoagulant-associated bleeding because their thrombocytopenia may be due to a prohemorrhagic diathesis (e.g., sepsis, drug-induced immune thrombocytopenia) rather than a prothrombotic diathesis such as HIT [14, 15]. It is crucial to better understand the bleeding risk associated with empiric alternative anticoagulant use in order to make evidence-based decisions regarding management of patients with suspected HIT.
Our primary aim was to determine the risk of major bleeding in patients with suspected HIT in a real-world setting. We hypothesized that among patients treated with an alternative anticoagulant, major bleeding would be more frequent in patients without HIT than in those in whom HIT was confirmed. Secondarily, we assessed for new/progressive thrombosis and 30-day mortality among patients with suspected HIT.
Methods
Study design
The study population was drawn from a prospective comparison of the HEP score and 4Ts score [16]. Details regarding eligibility and enrollment have been previously described [16]. In brief, 310 patients hospitalized at the Hospital of University of Pennsylvania or an affiliated community hospital were enrolled from June 2012 to January 2015. All patients in the study were suspected of having acute HIT, prompting the clinical care team to order laboratory testing for HIT. The decision to order testing for HIT was at the discretion of the clinical care team and use of a clinical prediction score was not required. Figure 1 describes the selection of comparator groups from the initial prospective cohort study. The protocol was approved by the University of Pennsylvania institutional review board. All subjects or their surrogates provided written informed consent.
Figure 1.
Study design. The patient population (n=310) was drawn from a prospective cohort study. Patients were classified based on whether they were treated with an alternative anticoagulant (Tx+) or not (Tx–) and whether they were classified as having HIT (HIT+) or not (HIT–). Bleeding and thrombotic events were collected retrospectively.
HIT, Heparin-induced thrombocytopenia; HEP score, Heparin-induced thrombocytopenia Expert Probability score; PF4/H ELISA, platelet factor 4/heparin enzyme-linked immunoassay; SRA, serotonin release assay
Procedures
At the time of study enrollment, patient demographics were obtained and the 4Ts and HEP scores were calculated by a member of the clinical care team. The day of suspicion for HIT (day 0) was defined as the day laboratory testing for HIT was ordered by the clinical care team or the day an alternative anticoagulant was started, whichever occurred first. Outcomes were assessed from day 0 until day 30, hospital discharge, or death. Events on day 0 were included in the analysis as occurring on day 0.5 if they occurred after the time of HIT laboratory test order or the initiation of an alternative anticoagulant. 30-day outcomes were obtained by telephone interview using standardized case report forms.
We extracted all laboratory results, transfusion records, medications and International Classification of Disease (ICD)-9 diagnosis codes from the hospital admission during which HIT was suspected using our health system’s clinical data warehouse (Penn Data Store). Baseline laboratory data were obtained at day 0 or 48 hours prior to day 0 (day −2). The admitting service of the patient was defined as medical or surgical, and the status of patients in an intensive care unit (ICU) status was obtained on day 0. The ICD-9 codes during the admission were used to calculate an Elixhauser Comorbidity index[17, 18]. A single reviewer (AP or DL) retrospectively reviewed the clinical progress notes and radiology reports from the index hospitalization to obtain detailed information regarding thrombosis and bleeding outcomes.
Outcome Definitions
The primary outcome was the cumulative incidence of major bleeding. Major bleeding was defined using modified International Society on Thrombosis and Haemostasis (ISTH) criteria to include any one of the following: (1) fatal bleeding, (2) bleeding into a critical area or organ, (3) fall in hemoglobin of ≥ 2 g/dL over 24 hours, (4) transfusion of two or more units of whole blood or packed red blood cells (pRBC) over 24 hours, and in surgical patients, (5) surgical site bleeding that required a second intervention [19, 20]. For all major bleeds meeting criteria by blood transfusion or hemoglobin fall alone, we assessed for documentation of bleeding in progress notes or radiology reports. All fatal bleeds were confirmed by consensus of two reviewers (AP and DL). Patients may have had multiple bleeding events, but only the date of the first bleeding event was used in the time to event analyses.
Secondary outcomes in the study were the cumulative incidence of progressive or new thrombosis on or following day 0 and mortality. New thrombosis was defined as a confirmed arterial or venous thrombosis on radiographic imaging (including ultrasound, computed tomography, magnetic resonance imaging, and angiography) [11] or clinical documentation of a pulseless extremity, gangrene, or necrotic/embolic skin lesions. Progressive thrombosis was defined as a radiographically confirmed increased clot burden on subsequent imaging. Mortality was assessed at 30 days by review of the medical record for patients who remained hospitalized and by telephone follow-up for discharged patients.
Definition of HIT
All patients underwent HIT laboratory testing with a polyspecific HIT antibody ELISA (PF4/H ELISA) (Immucor GTI Diagnostics Inc, Waukesha, WI, USA) and an in-house serotonin-release assay (SRA) [21]. In light of the high negative predictive value of the ELISA and SRA, all patients with a negative ELISA (<0.4 optical density units) and a negative SRA were classified as not having HIT [22]. Clinical case summaries of the remaining subjects were prepared using a standardized case report form to include detailed clinical information about the hospital course, the platelet count trend, exposure to heparin and other medications, thrombotic events, the results of the HIT antibody ELISA and SRA, and 30-day follow-up information. An independent adjudication panel of 3 experts in HIT (GA, MC, LR) reviewed the clinical case summaries and determined if patients were positive for HIT or negative for HIT as previously described [16]. HIT-positive status was defined as the consensus opinion of the adjudication panel.
Alternative anticoagulant exposure
Alternative anticoagulants included argatroban, lepirudin, bivalirudin, and fondaparinux (DOACs were not used to treat HIT and danaparoid was not available in the US during the study period). We assessed alternative anticoagulant treatment in two ways: first as a fixed exposure and then as an exposure varying over time.
The fixed exposure definition allowed us to assess the impact of the clinician’s initial treatment strategy on outcomes occurring at any time between day 0 and day 30. Using this fixed exposure definition, we classified patients into one of four groups by HIT status and treatment with an alternative anticoagulant on any day following suspicion for HIT as follows:
-
(1)
HIT+Tx+: HIT positive and treated with an alternative anticoagulant
-
(2)
HIT+Tx–: HIT positive and not treated with an alternative anticoagulant
-
(3)
HIT–Tx+: HIT negative and treated with an alternative anticoagulant
-
(4)
HIT–Tx–: HIT negative and not treated with an alternative anticoagulant
A limitation of this fixed exposure approach is that it does not take into account the temporal relationship between an outcome and alternative anticoagulant exposure. For example, a HIT-negative patient who was only treated with an alternative anticoagulant on days 0 and 1 and experienced a major bleed on day 10 would be classified in the HIT–Tx+ group, even though the bleed occurred 9 days after stopping the alternative anticoagulant and was therefore presumably unrelated to the alternative anticoagulant exposure. To overcome this limitation, we also evaluated alternative anticoagulant use as a time-varying exposure in which alternative anticoagulant exposure was reassessed each day. Subjects were classified as Tx+ for a given day only if they received an alternative anticoagulant on that day or on the previous calendar day. We included exposure on the previous day because we reasoned that, in light of the half-life of alternative anticoagulants, very recent exposure could potentially contribute to bleeding the following day.
Statistical Analysis
The characteristics of HIT+ and HIT– patients were compared with the Wilcoxon rank-sum test for continuous variables and the chi-square test or Fisher’s exact test for categorical variables. Categorical variables were modeled with a missing category if more than 5% of patients had missing data. Patients missing continuous variables were dropped from the analysis. We anticipated that patients the clinical team chose not to treat with an alternative anticoagulant might have a higher baseline bleeding risk and/or a lower risk of HIT than those who were treated. Thus, we performed a univariate logistic regression to assess the association of clinical factors, including factors potentially associated with bleeding risk (i.e., severity of thrombocytopenia, surgical status, renal and hepatic dysfunction) and thrombotic risk (i.e., 4Ts score, presence of thrombosis on day 0) with the physician’s choice to treat with an alternative anticoagulant.
Using the fixed definition of alternative anticoagulant exposure, we calculated cumulative incidence curves for major bleeding and thrombotic events using the Kaplan-Meier method. Log rank test statistics were performed to compare the HIT+Tx+, HIT–Tx+, and HIT–Tx– groups. Due to low numbers in the HIT+Tx– group, these patients were excluded from this portion of the analysis.
We performed a Cox proportional hazards regression analysis with HIT status as a covariate and with alternative anticoagulant exposure as a time-varying covariate. Univariate Cox regression was used to evaluate the association of other potential covariates including antiplatelet agent use, renal or hepatic dysfunction, intensive care unit (ICU) status, hospitalization on a surgical service, thrombosis, and degree of thrombocytopenia on day 0 with time to a first major bleeding event. Covariates with p<.01 were included in the full multivariable model with HIT status and alternative anticoagulant use. We also assessed any significant covariates from the univariate analyses for their interaction with HIT status, and included the interaction term in the full model if it was significant. Variables and/or interactions that were not significantly associated with time to first major bleeding event were sequentially removed to create a final reduced multivariable model. Missing data was not imputed. Analyses were performed using Stata 14.2 (College Station, TX). Statistical significance was defined as p≤0.05.
Results
Patient Characteristics
The demographic and clinical features of our cohort by HIT status are displayed in Table 1. Of the 310 patients included in the analysis, 44 (14.2%) patients were adjudicated as HIT+ and 266 (85.8%) patients were classified as HIT–.The groups were similar with respect to baseline demographic features. As expected, more HIT+ patients had a high probability 4Ts score (45.5% vs. 11.7%). HIT+ patients were also more likely to have thrombosis present on day 0 (63.6% vs. 32.3%, p<0.001). The majority of HIT+ patients had a HIT antibody ELISA ≥1.0 OD units (93.2%) and a positive SRA (68.2%).
Table 1.
Patient clinical and laboratory features by adjudicated HIT status
| HIT + N=44 |
HIT – N=266 |
p-value | |
|---|---|---|---|
| Demographic and clinical features | |||
| Age, median (25th, 75th percentile) years | 64 (53,74) | 67 (58,75) | 0.175 |
| Male, n (%) | 27 (61.4) | 137 (51.5) | 0.225 |
| Hospital, n (%) | |||
| Academic Center | 33 (75.0) | 160 (60.2) | 0.060 |
| Affiliated Community Center | 11 (25.0) | 106 (39.8) | — |
| ICU status, n (%) | 23 (52.3) | 159 (59.8) | 0.349 |
| Surgical service, n (%) | 28 (63.6) | 148 (55.6) | 0.321 |
| Elixhauser Comorbidity Index*, | 4 (2.5,5.5) | 4 (4,6) | 0.183 |
| median (25th, 75th percentile) | |||
| HIT clinical and laboratory features | |||
| Total 4Ts score, n (%) | |||
| Low (<4) | 1 (2.3) | 82 (30.8) | <0.001 |
| Intermediate (4–5) | 22 (50.0) | 136 (51.1) | — |
| High (>5) | 20 (45.5) | 31 (11.7) | — |
| Missing | 1 (2.3) | 17 (6.4) | — |
| Platelet Count†, median (25th, 75th percentile) (x 109/L) | 52 (34,92) | 61 (39,87) | 0.470 |
| Presence of thrombosis††, n (%) | 28 (63.6) | 86 (32.3) | <0.001 |
| Venous | 21 (47.7) | 46 (17.3) | — |
| Arterial | 6 (13.6) | 36 (13.5) | — |
| Both | 1 (2.3) | 4 (1.5) | — |
| Serotonin Release Assay, n (%) | <0.001 | ||
| Positive | 30 (68.2) | 1 (0.4) | |
| Negative | 13 (29.5) | 261 (98.1) | — |
| Indeterminate | 1 (2.3) | 2 (0.8) | — |
| Missing | 0 (0) | 2 (0.8) | — |
| Polyspecific PF4/H ELISA, n (%) | <0.001 | ||
| 0–0.39 optical density units | 0 (0) | 222 (83.5) | — |
| 0.4–0.99 optical density units | 3 (7.1) | 23 (8.6) | — |
| ≥1.00 optical density units | 41 (93.2) | 20 (7.5) | — |
| Missing | 0 (0) | 1 (0.4) | — |
HIT +, patients adjudicated to have HIT by consensus of expert panel; HIT–, patients adjudicated not to have HIT by consensus of expert panel; ICU, intensive care unit
Calculated based on ICD-9 codes during hospitalization
Platelet count on day 0 or up to 48 hours prior if no laboratory value available on day 0
Occuring on or before time of suspected HIT on day 0
Treatment for Suspected HIT
Treatment with an alternative anticoagulant was administered to 95.5% of HIT+ patients and 43.6% of HIT– patients (p<0.001). Two HIT+ patients did not receive an alternative anticoagulant because they died shortly after HIT laboratory testing was sent from intracranial hemorrhage. Among treated patients, the majority started alternative anticoagulant therapy on day 0 (76.2% of HIT+ patients vs. 84.5% of HIT– patients, p=0.228) (Table 2). The remainder of treated patients initiated alternative anticoagulant therapy by day 2. Argatroban was the most frequently prescribed alternative anticoagulant (88.1% of HIT+ and 76.7% of HIT– patients). The dose of argatroban prescribed was missing in 2 HIT+ and 12 HIT– patients. Among the patients with dosing information available, the median starting dose of argatroban was 1.2 mcg/kg/min in the HIT+Tx+ and 1 mcg/kg/min in the HIT–Tx+ population (p=0.101). The median total duration of alternative anticoagulant therapy was longer in HIT+ compared with HIT– patients (11.5 days vs. 3 days, p<0.001).
Table 2.
Treatment for suspicion of HIT by adjudicated HIT status.
| HIT+Tx+ | HIT –Tx+ | p-value | |
|---|---|---|---|
| N | 42 | 116 | — |
| Type of alternative anticoagulant, n (%) | 0.259 | ||
| Argatroban | 37 (88.1) | 89 (76.7) | — |
| Fondaparinux | 2 (4.8) | 13 (11.2) | — |
| Bivalirudin | 0 (0) | 6 (5.2) | — |
| Multiple agents | 3 (7.1) | 8 (6.9) | — |
| Day of initiation of alternative anticoagulant therapy, n (%) | 0.228 | ||
| Day 0 | 32 (76.2) | 98 (84.5) | — |
| Day 1 | 6 (14.3) | 14 (12.1) | — |
| Day 2 | 4 (9.5) | 4 (3.4) | — |
| Initial argatroban dosing†, median (25th,75th), mcg/kg/min | 1.2 (0.8,2) | 1 (0.3,2) | 0.101 |
| Duration of alternative anticoagulant therapy, median (25th,75th percentile), days | 11.5 (6,16) | 3 (2,6) | <0.001 |
HIT+Tx+, Patients who received treatment for HIT with an alternative anticoagulant and were adjudicated to be HIT positive by the consensus of the expert panel; HIT–Tx+, Patients who received treatment for HIT with an alternative anticoagulant and were adjudicated to not have HIT by the consensus of the expert panel
14 patients with missing dose of argatroban excluded.
Factors Predictive of Treatment with an Alternative Anticoagulant
Compared with patients with a low probability 4Ts score, physicians were more likely to treat with an alternative anticoagulant in patients with an intermediate (OR 1.96, 95%CI 1.13–3.39) or high probability 4Ts score (10.0, 95% CI 4.15–24.11). Presence of thrombosis on day 0 was also associated with increased odds of treatment with an alternative anticoagulant (OR 3.11, 95% CI 1.91–5.06). A higher Elixhauser comorbidity index was associated with decreased odds of receiving treatment with an alternative anticoagulant (OR 0.89, 95% CI 0.81–0.99). Severity of thrombocytopenia, serum creatinine, hemoglobin, and age were not associated with the decision to treat with an alternative anticoagulant (Table 3).
Table 3.
Factors associated with treatment with an alternative anticoagulant for suspected HIT
| Total N |
Tx + N (%) |
Tx – N (%) |
Odds Ratio | p-value | 95% CI | |
|---|---|---|---|---|---|---|
| 4Ts score | 310 | |||||
| Missing | 5 (3.16) | 13 (.8.55) | 0.72 | 0.561 | 1.13–3.39 | |
| Low (<4) | 29 (19.0) | 54 (38.9) | Reference | — | — | |
| Intermediate(4–5) | 81 (52.9) | 77 (55.4) | 1.96 | 0.016 | 1.13–3.39 | |
| High(>6) | 43 (28.1) | 8 (5.8) | 10.0 | <0.001 | 4.15–24.11 | |
| Platelet Count* | 309 | |||||
| ≥100 × 109/L | 36 (22.8) | 25 (16.6) | Reference | — | — | |
| 75–99 × 109/L | 26 (16.5) | 26 (17.2) | 0.69 | 0.338 | 0.33–1.46 | |
| 50–75 × 109/L | 38 (24.1) | 38 (25.2) | 0.69 | 0.293 | 0.35–1.37 | |
| 25–50 × 109/L | 40 (25.3) | 47 (31.1) | 0.59 | 0.119 | 0.30–1.15 | |
| ≤24 × 109/L | 18 (11.4) | 15 (9.9) | 0.83 | 0.676 | 0.35–1.96 | |
| Presence of Thrombosis* | 309 | 78 (49.4) | 36 (23.8) | 3.11 | <0.001 | 1.91–5.06 |
| ICU | 310 | 87 (55.1) | 95 (62.5) | 0.74 | 0.184 | 0.47–1.16 |
| Renal Dysfunction*† | 310 | 44 (27.9) | 40 (26.3) | 1.08 | 0.762 | 0.65–1.78 |
| Surgical Service | 310 | 84 (53.2) | 92 (60.5) | 0.66 | 0.191 | 0.47–1.16 |
| Hemoglobin* | 308 | |||||
| ≥9 g/dL | 90 (57.0 | 83 (54.6) | Reference | — | — | |
| <9 g/dL | 68 (43.0) | 69 (45.4) | 0.86 | 0.509 | 0.55–1.34 | |
| Age | 310 | |||||
| <49 years | 22 (13.9) | 17 (11.2) | Reference | — | — | |
| 50–75 years | 105 (66.5) | 95 (62.5) | 0.85 | 0.655 | 0.43–1.70 | |
| >75 years | 31 (19.6) | 40 (26.30 | 0.60 | 0.202 | 0.27–1.32 | |
| Elixhauser comorbidity Index | 310 | 0.89‡ | 0.025 | 0.81– 0.99 | ||
| <4 | 61 (38.6) | 45 (29.6) | — | — | — | |
| 4–6 | 51 (32.3) | 48 (31.6) | — | — | — | |
| >6 | 46 (29.1) | 59 (38.8) | — | — | — | |
Tx+, patient received treatment with alternative anticoagulant for suspected HIT; Tx–, patient did not receive treatment with alternative anticoagulant for suspected HIT
Occuring or or before time of suspected HIT on day 0
Defined as creatinine ≥2.0 mg/dL on day 0
Odds associated with increase of 1 point in Elixhauser comorbidity index (continuous variable)
Major Bleeding
The incidence of major bleeding events was high in all patient groups, occurring in 35.7% (15/42), 44.0% (51/116), and 37.3% (56/150) of patients in the HIT+Tx+ group, the HIT –Tx+ group, and the HIT –Tx – group, respectively (Table 4). The majority of bleeding events in all groups met criteria for major bleeding based on pRBC transfusion or hemoglobin fall (Table 4). There were 2 fatal bleeds each in the HIT+Tx+ and HIT+Tx– groups and none among HIT– patients. Critical organ bleeds occurred in none of the HIT+Tx+ patients, 6.9% of the HIT–Tx+ patients, and 4.6% of the HIT–Tx– patients. Descriptions of fatal and critical organ bleeds are provided in Supplemental Table 1.
Table 4. Description of major bleeding events following suspected HIT by treatment and HIT status.
There was no significant difference in the proportion of total bleeding events or the various types of major bleeding event by ISTH criteria among the HIT+Tx+, HIT+Tx-, and HIT-Tx- groups (p>0.05).
| HIT+Tx+ N=42 |
HIT−Tx+ N=116 |
HIT+Tx− N=2 |
HIT−Tx− N=150 |
|
|---|---|---|---|---|
| Total major bleeding events, N (%) | 15 (35.7) | 51 (44.0) | 2 (100) | 56 (37.3) |
| ISTH criteria fulfilled *: | ||||
| Fatal bleed, N (%) | 2 (4.8) | 0 (0) | 2 (100) | 0 (0) |
| Critical organ bleed, N (%) | 0 (0) | 8 (6.9) | 0 (0) | 7 (4.6) |
| 2 g/dL HgB fall and/or 2 unit pRBC transfusion, N (%) | 13 (30.9) | 43 (37.1) | 0 (0) | 49 (32.6) |
| + documented bleeding | 5 (11.9) | 13 (11.2) | 0 (0) | 18 (12.0) |
HIT+Tx+, patients who received treatment for HIT with an alternative anticoagulant and were adjudicated to be HIT positive by expert panel consensus; HIT–Tx+, patients who received treatment for HIT with an alternative anticoagulant and were adjudicated to not have HIT by expert panel consensus; HIT+Tx–, patients who did not receive treatment for HIT with an alternative anticoagulant and were adjudicated to have HIT by expert panel consensus; HIT–Tx–, patients who did not receive treatment for HIT with an alternative anticoagulant and were adjudicated to not have HIT by expert panel consensus; pRBC, packed red blood cell; HgB, hemoglobin
Patients may have met multiple ISTH criteria for major bleeding between days 0–30. Displayed in the table is the criterion of highest severity for each patient (i.e. fatal bleeding > critical organ bleeding > 2g/dL fall in HgB and/or transfusion of 2 or more units of pRBCs)
Contrary to our hypothesis, there was no significant difference in the cumulative incidence of major bleeding events among the HIT+Tx+, HIT–Tx+, and HIT–Tx– groups (p=0.24) (Supplemental Figure 1). Use of a more restricted definition of major bleeding that included only cases of clinically documented overt bleeding resulted in a lower cumulative incidence of major bleeding in all groups, but there was still no difference among groups (p=0.85).
We reasoned that the unexpectedly high rate of major bleeding in the HIT+Tx+ group could be attributable to the greater duration of alternative anticoagulant exposure in this group compared with the HIT–Tx+ group (Table 2). To address this possibility, we assessed alternative anticoagulant exposure as a time varying covariate in a Cox proportional hazard model for major bleeding. Even when exposure time was accounted for in the model, alternative anticoagulant exposure was not significantly associated with an increased hazard of bleeding when adjusted for HIT status (HR 0.99, 95%CI 0.92–1.07, p=0.846) (Table 5A).
Table 5.
Cox Proportional Hazards Model of the association of major bleeding with HIT status and alternative anticoagulant exposure (A) and other clinical risk factors for bleeding (B).
| Bivariable Regression | |||
|---|---|---|---|
| Risk factor | Adjusted Hazard Ratio |
95% CI | p-value |
| HIT status | |||
| HIT − | Reference | — | — |
| HIT + | 0.82 | 0.45–1.46 | 0.501 |
| Alternative anticoagulant * | |||
| No | Reference | — | — |
| Yes | 0.99 | 0.92–1.07 | 0.846 |
| Multivariable Regression | |||
|---|---|---|---|
| Risk factor | Adjusted Hazard Ratio |
95% CI | p-value |
| HIT status | |||
| HIT − | Reference | — | — |
| HIT + | 0.79 | 0.43–1.43 | 0.431 |
| Alternative anticoagulant* | |||
| No | Reference | — | — |
| Yes | 1.00 | 0.93–1.07 | 0.998 |
| Intensive care unit | |||
| No | Reference | — | — |
| Yes | 2.24 | 1.44–3.47 | <0.001 |
| Renal dysfunction | |||
| Cr <2.0 mg/dL | Reference | — | — |
| Cr ≥2.0 mg/dL | 1.56 | 1.06–2.27 | 0.022 |
| Severity of thrombocytopenia | |||
| ≥100 × 109/L | Reference | — | — |
| 75–99 × 109/L | 0.76 | 0.36–1.59 | 0.462 |
| 50–75 × 109/L | 1.39 | 0.77–2.50 | 0.268 |
| 25–50 × 109/L | 1.41 | 1.11–2.53 | 0.243 |
| ≤24 × 109/L | 2.13 | 1.10–4.12 | 0.024 |
HIT +, patients adjudicated to have HIT by expert panel consensus; HIT–, patients adjudicated bit to have HIT by expert panel consensus;
Alternative anticoagulant received on that calendar day or day prior.
Other variables were evaluated for an association with hazard of major bleeding (Supplemental Table 2). In the final adjusted Cox model with HIT status and alternative anticoagulant use, admission to the ICU, serum creatinine ≥ 2.0 mg/dL, and a platelet count < 25 × 109/L, but not alternative anticoagulant exposure, were associated with a significantly increased adjusted hazard of major bleeding (Table 5B).
Because argatroban was the predominant alternative anticoagulant used in our study population, almost all major bleeding events in the HIT+Tx+ and HIT–Tx+ groups occurred in patients treated with argatroban (100% and 86.5%, respectively). The median initial dose of argatroban was lower in patients who experienced a major bleeding event than in those who did not (0.5 mcg/kg/min vs. 1.2 mcg/kg/min, p=0.0004). At least one activated partial thromboplastin time (APTT) value was supra-therapeutic (≥90 seconds) while receiving argatroban in 25% (3/12) of HIT+ patients and 12.5% (2/16) of HIT– patients with major bleeding events within 1 calendar day of receiving drug.
Thrombosis and mortality
There was a higher cumulative incidence of progressive/new thrombosis in HIT+Tx+ patients compared with either of the HIT– groups (p=0.001) (Figure 2); 35.7% (15/42) of HIT+Tx+, 13.8% (16/116) of HIT–Tx+, and 9.3% (14/150) of HIT–Tx– patients experienced new or progressive thrombosis during the study period (Supplemental Figure 2). Descriptions of thrombotic events are listed in Supplemental Table 3. Thirty-day mortality was high and similar among the 3 groups (26.2%, 34.5%, and 26.7%, respectively, p=0.34). Of note, 5.81% of patients were discharged from the hospital alive but were unable to be contacted at 30-days and were assumed alive in the analysis.
Discussion
We conducted a retrospective analysis of major bleeding events in a prospective cohort of patients with suspected HIT. We found that major bleeding was surprisingly common, occurring in 40.6% (126/310) of the overall cohort (Table 4). We did not find a significantly lower probability of major bleeding in HIT+ patients compared with HIT– patients, suggesting that although HIT is a prothrombotic disorder, it is not protective against bleeding.
In the pivotal trials leading to FDA-approval of argatroban for HIT (ARG-911 and ARG-915), major bleeding was reported in 11.1% and 6.1% of patients, respectively.3 A somewhat higher rate of major bleeding (17.6%) was reported in the lepirudin clinical trial program [12]. We observed a substantially higher frequency of major bleeding among HIT+ patients in our study (18/44, 40.9%) (Table 4). Reasons for this disparity may include differences in the way major bleeding was defined as well as differences in patients selected for clinical trials compared with the real-world population of patients enrolled in our study. Indeed, our findings are similar to those of another real-world study, in which bleeding events were observed in 48% of patients with HIT with thrombosis and 36% of patients with HIT without thrombosis [13].
To our surprise, major bleeding was no less common in HIT+ patients (18/44, 40.9%) than it was among HIT– patients (107/266, 40.2%) (Table 4). Although HIT+ patients were more likely to be treated with an alternative anticoagulant and to receive a longer duration of therapy than HIT– patients (Table 2), the hazard of major bleeding was no less in the HIT+ group when alternative anticoagulant exposure was analyzed as a time-varying covariate (Table 5). Differences in treatment intensity are also unlikely to account for our findings as the initial dose of argatroban was not different between the HIT+Tx+ and HIT–Tx+ groups (Table 2).
Although most bleeding events in HIT+ patients met major bleeding criteria by virtue of a fall in hemoglobin concentration and/or packed red cell transfusion, four HIT+ patients (9.1%) suffered fatal or critical site bleeds. By contrast, fatal or critical site bleeds occurred in 6.0% of HIT– patients (Table 4). Taken together, these findings suggest that HIT does not insulate patients from major or life-threatening bleeding.
We observed an unexpectedly high incidence of major bleeding in the HIT–Tx– group (56/150, 37.3%) (Table 4). There are a number of potential explanations for why bleeding was common in this group, even in the absence of exposure to an alternative anticoagulant. First, as expected in a cohort of patients suspected of having developed HIT, most HIT–Tx– patients were thrombocytopenic (median platelet count of 60 × 109/L). Second, clinicians may have deliberately chosen to avoid treatment with alternative anticoagulants in some patients precisely because they judged the patients to be at increased risk of bleeding. Indeed, patients who did not receive treatment with an alternative anticoagulant had a higher Elixhauser comorbidity index (Table 3), which clinicians may view as a marker of bleeding risk. Lastly, although by definition patients in the HIT–Tx– group did not receive alternative anticoagulants, they received other anticoagulants (e.g., heparin and low molecular weight heparin), which may have contributed to bleeding. In the HIT–Tx– group, 31.5% of bleeds occurred within 1 calendar day of receiving a prophylactic or therapeutic dose of unfractionated heparin, low molecular weight heparin, or warfarin.
Neither HIT status nor alternative anticoagulant exposure influenced bleeding risk, but we identified several variables associated with an increased risk of major bleeding including admission to the ICU, renal dysfunction, and severe thrombocytopenia (Table 5). Another study found recent surgery to be a risk factor for major bleeding on argatroban among critically ill patients, but our study did not find admission to a surgical service to be significant in multivariable analysis [23]. Most patients in the study received the hepatically cleared drug argatroban. Still, renal dysfunction correlated with increased bleeding risk. This could be attributable to uremic platelet dysfunction. Severity of thrombocytopenia, when adjusted for HIT status, was also predictive of major bleeding. While severity of thrombocytopenia is also a risk factor for thrombosis in patients with HIT [24], this observation challenges the prevailing belief that HIT-positive patients with severe thrombocytopenia are protected from bleeding.
In addition to evaluating bleeding outcomes, we found that progressive or new thrombosis occurred in 35.7% of HIT-positive patients on or after day 0 (Figure 2). This rate is somewhat higher than previous reports [13]. It was common practice at our institution to order four-extremity Doppler ultrasound in patients with confirmed acute HIT. Thus, some of the thrombotic events may have been present prior to day 0, but only recognized after the diagnosis of HIT was made.
Our study has several strengths and limitations. Whereas many previous studies have relied solely on laboratory testing or 4Ts score to define HIT, our use of a rigorous adjudication process to define HIT status was implemented to reduce the potential for misclassification. The study population was drawn from a prospective evaluation of clinical prediction scores for HIT[16]. All participants had a prospectively calculated 4Ts scores and were contacted to obtain 30-day outcomes. An important limitation in using this cohort was the fixed sample size of 310 patients. We were particularly limited by the number of HIT+ patients who received treatment and were only powered to detect large differences in the proportion of bleeding between groups. Nevertheless, due to the rarity of HIT, this study remains one of the largest to describe real-world outcomes in patients with suspected HIT. Another limitation is the retrospective collection of bleeding events. We were not able to ascertain whether all events that met ISTH criteria due to fall in hemoglobin or blood transfusion were truly related to major bleeding rather than other etiologies (e.g., hemolysis or dilution). Overt bleeding was only documented in the progress notes or imaging in 34.2% (36/105) of patients who experienced a fall in hemoglobin of at least 2 g/dL or who were transfused with 2 or more units of packed red blood cells. Our study was restricted to two hospitals within a single healthcare system. The predominant alternative anticoagulant used was argatroban. While this may reflect practice in some centers in the US, it does not reflect practice in other parts of the world where argatroban is used less frequently [25] nor does it reflect recent growth in the use of fondaparinux and DOACs to treat HIT [11]. Additional studies are needed to determine whether our findings are applicable to other centers and other alternative anticoagulants. Our study was limited by its observational nature. The decision regarding whether to treat with an alternative anticoagulant, with which agent, and at what dose was left to the discretion of the clinical team. It is likely that patients judged to be at higher risk of thrombosis were more likely to be treated with an alternative anticoagulant whereas those estimated to be at higher bleeding risk at the time of suspected HIT were more likely not to be treated.
Clinical practice guidelines recommend that patients with suspected HIT and an intermediate or high 4Ts score be treated empirically with an alternative anticoagulant while awaiting the results of HIT laboratory testing [5]. While such a recommendation incorporates an estimate of the risk of HIT (and therefore the risk of thrombosis) into clinical decision making, it does not include a formal assessment of bleeding risk. Our data provide information on the real-world incidence of major bleeding with argatroban as well as factors that may magnify or reduce an individual patient’s bleeding risk. Such information may be useful to clinicians for weighing not only the potential benefits but also the potential harms of treatment with an alternative anticoagulant in a patient with suspected HIT.
In summary, we observed a remarkably high rate of major bleeding in patients with suspected HIT, irrespective of HIT status or exposure to an alternative anticoagulant. Of greatest surprise, HIT-positivity was not protective from major or fatal bleeding. While thrombotic risk remains of paramount concern, our findings suggest that hemorrhagic risk also deserves strong consideration when weighing management options in patients with suspected HIT.
Supplementary Material
Cumulative incidence of major bleeding by HIT status and treatment with alternative anticoagulant. Bleeding events were assessed starting on day of suspicion for HIT (day 0) and defined using a modified ISTH definition for major bleeding
HIT+Tx+, Patients who received treatment for HIT with an alternative anticoagulant and were adjudicated to be HIT positive by expert panel consensus; HIT–Tx+, Patients who received treatment for HIT with an alternative anticoagulant and were adjudicated to not have HIT by expert panel consensus; HIT–Tx–, Patients who did not receive treatment for HIT with an alternative anticoagulant and were adjudicated to not have HIT by expert panel consensus
Cumulative Incidence of new or progressive thrombosis by HIT group and treatment. Thrombotic events were assessed starting on day of suspicion for HIT (day 0) and defined using a modified International Society on Thrombosis and Haemostasis definition of major bleeding.
HIT+Tx+, Patients who received treatment for HIT with an alternative anticoagulant and were adjudicated to be HIT positive by expert panel consensus; HIT–Tx+, Patients who received treatment for HIT with an alternative anticoagulant and were adjudicated to not have HIT by the consensus of expert panel; HIT–Tx–, Patients who did not receive treatment for HIT with an alternative anticoagulant and were adjudicated to not have HIT by expert panel consensus
Essentials.
-
–
Patients with HIT were not protected against major bleeding events compared to patients without HIT.
-
–
High risk features for bleeding were hospitalization in the ICU, renal dysfunction and severe thrombocytopenia.
Acknowledgments
This work was supported by the 2018 HTRS/Novo Nordisk Clinical Fellowship Award in Hemophilia and Rare Bleeding Disorders from the Hemostasis and Thrombosis Research Society (HTRS) to A. M. Pishko and the National Institutes of Health T32-HL007971-16A1 to A. M. Pishko.
Disclosures:
AP received salary support from an educational grant from Novo Nordisk, Inc. LR has served on an advisory board or speaker’s bureau for Pfizer and Novartis; has served on a data safety monitoring committee for Apellis; and has received research support from Baxalta and Alexion. AC has served as a consultant or advisory board member for Bioverativ, Genzyme, Kedrion, Stago, and Synergy and his institution has received research support on his behalf from Alexion, Bayer, Bioverativ, Novo Nordisk, Pfizer, Shire, and Spark. DBC has served as a consultant or advisory board member for Amgen, Rigel, Novartis, Ionis, Sanofi, and Bayer and has received research support from T2Biosystems, Syntimmune, and Momenta. MC has served on a data safety monitoring board for Bayer and Daiichi Sankyo; has served as a consultant or advisory board member for Shionogi, Octapharma, BMS Canada, Pfizer, Alexion and CSL Behring; and has received research funding from Bayer, Leo Pharma, and Heart and Stroke Foundation of Canada. GMA has served as a consultant for Apotex Pharmaceuticals. All other authors have no conflicts to disclose.
Footnotes
Addendum
A. M. Pishko, J. Guevara, P. Gimotty and A. Cuker designed the study. A. M. Pishko and D. S. Lefler collected the data. A. M. Pishko, P. Gimotty, and J. Guevara performed statistical analysis. A. M. Pishko, A. Cuker, P. Gimotty, and J. Guevara wrote the manuscript. All authors critically reviewed and approved the final version of the manuscript.
References
- 1.Greinacher A, Eichler P, Lubenow N, Kwasny H, Luz M. Heparin-induced thrombocytopenia with thromboembolic complications: meta-analysis of 2 prospective trials to assess the value of parenteral treatment with lepirudin and its therapeutic aPTT range. Blood. 2000; 96: 846–51. [PubMed] [Google Scholar]
- 2.Alving BM. How I treat heparin-induced thrombocytopenia and thrombosis. Blood. 2003; 101: 31–7. 10.1182/blood-2002-04-1089. [DOI] [PubMed] [Google Scholar]
- 3.Cuker A, Gimotty PA, Crowther MA, Warkentin TE. Predictive value of the 4Ts scoring system for heparin-induced thrombocytopenia: a systematic review and meta-analysis. Blood. 2012; 120: 4160–7. 10.1182/blood-2012-07-443051. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 4.Cuker A, Cines DB. How I treat heparin-induced thrombocytopenia. Blood. 2012; 119: 2209–18. 10.1182/blood-2011-11-376293. [DOI] [PubMed] [Google Scholar]
- 5.Cuker A, Arepally GM, Chong BH, Cines DB, Greinacher A, Gruel Y, Linkins LA, Rodner SB, Selleng S, Warkentin TE, Wex A, Mustafa RA, Morgan RL, Santesso N. American Society of Hematology 2018 guidelines for management of venous thromboembolism: heparin-induced thrombocytopenia. Blood Adv. 2018; 2: 3360–92. 10.1182/bloodadvances.2018024489. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 6.McMahon CM, Tanhehco YC, Cuker A. Inappropriate documentation of heparin allergy in the medical record because of misdiagnosis of heparin-induced thrombocytopenia: frequency and consequences. Journal of thrombosis and haemostasis : JTH. 2017; 15: 370–4. 10.1111/jth.13565. [DOI] [PubMed] [Google Scholar]
- 7.Cuker A. Heparin-induced thrombocytopenia (HIT) in 2011: an epidemic of overdiagnosis. Thromb Haemost. 2011; 106: 993–4. 10.1160/th11-09-0677. [DOI] [PubMed] [Google Scholar]
- 8.Patrick AR, Winkelmayer WC, Avorn J, Fischer MA. Strategies for the management of suspected heparin-induced thrombocytopenia: a cost-effectiveness analysis. Pharmacoeconomics. 2007; 25: 949–61. 10.2165/00019053-200725110-00005. [DOI] [PubMed] [Google Scholar]
- 9.Lewis BE, Wallis DE, Berkowitz SD, Matthai WH, Fareed J, Walenga JM, Bartholomew J, Sham R, Lerner RG, Zeigler ZR, Rustagi PK, Jang IK, Rifkin SD, Moran J, Hursting MJ, Kelton JG. Argatroban anticoagulant therapy in patients with heparin-induced thrombocytopenia. Circulation. 2001; 103: 1838–43. [DOI] [PubMed] [Google Scholar]
- 10.Joseph L, Casanegra AI, Dhariwal M, Smith MA, Raju MG, Militello MA, Gomes MP, Gornik HL, Bartholomew JR. Bivalirudin for the treatment of patients with confirmed or suspected heparin-induced thrombocytopenia. Journal of thrombosis and haemostasis : JTH. 2014; 12: 1044–53. 10.1111/jth.12592. [DOI] [PubMed] [Google Scholar]
- 11.Kang M, Alahmadi M, Sawh S, Kovacs MJ, Lazo-Langner A. Fondaparinux for the treatment of suspected heparin-induced thrombocytopenia: a propensity score-matched study. Blood. 2015; 125: 924–9. 10.1182/blood-2014-09-599498. [DOI] [PubMed] [Google Scholar]
- 12.Lubenow N, Eichler P, Lietz T, Greinacher A, Hit Investigators G. Lepirudin in patients with heparin-induced thrombocytopenia - results of the third prospective study (HAT-3) and a combined analysis of HAT-1, HAT-2, and HAT-3. Journal of thrombosis and haemostasis : JTH. 2005; 3: 2428–36. 10.1111/j.1538-7836.2005.01623.x. [DOI] [PubMed] [Google Scholar]
- 13.Kuter DJ, Konkle BA, Hamza TH, Uhl L, Assmann SF, Kiss JE, Kaufman RM, Key NS, Sachais BS, Hess JR, Ness P, McCrae KR, Leissinger C, Strauss RG, McFarland JG, Neufeld E, Bussel JB, Ortel TL. Clinical outcomes in a cohort of patients with heparin-induced thrombocytopenia. Am J Hematol 2017; 92: 730–8. 10.1002/ajh.24759. [DOI] [PubMed] [Google Scholar]
- 14.Marler J, Unzaga J, Stelts S, Oliphant CS. Consequences of treating false positive heparin-induced thrombocytopenia. Journal of thrombosis and thrombolysis. 2015; 40: 512–4. 10.1007/s11239-015-1236-0. [DOI] [PubMed] [Google Scholar]
- 15.Akca S, Haji-Michael P, de Mendonca A, Suter P, Levi M, Vincent JL. Time course of platelet counts in critically ill patients. Critical care medicine. 2002; 30: 753–6. [DOI] [PubMed] [Google Scholar]
- 16.Pishko AM, Fardin S, Lefler DS, Paydary K, Vega R, Arepally GM, Crowther M, Rice L, Cines DB, Cuker A. Prospective comparison of the HEP score and 4Ts score for the diagnosis of heparin-induced thrombocytopenia. Blood Adv 2018; 2: 3155–62. 10.1182/bloodadvances.2018023077. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 17.Stagg V. ELIXHAUSER: Stata module to calculate Elixhauser index of comorbidity,” Statistical Software Components”. S458077, Boston College Department of Economics. [Google Scholar]
- 18.Moore BJ, White S, Washington R, Coenen N, Elixhauser A. Identifying Increased Risk of Readmission and In-hospital Mortality Using Hospital Administrative Data: The AHRQ Elixhauser Comorbidity Index. Med Care. 2017; 55: 698–705. 10.1097/mlr.0000000000000735. [DOI] [PubMed] [Google Scholar]
- 19.Schulman S, Kearon C, the SOCOAOTS, Standardization Committee Of The International Society On T, Haemostasis. Definition of major bleeding in clinical investigations of antihemostatic medicinal products in non-surgical patients. Journal of Thrombosis and Haemostasis. 2005; 3: 692–4. 10.1111/j.1538-7836.2005.01204.x. [DOI] [PubMed] [Google Scholar]
- 20.Schulman S, Angeras U, Bergqvist D, Eriksson B, Lassen MR, Fisher W. Definition of major bleeding in clinical investigations of antihemostatic medicinal products in surgical patients. Journal of thrombosis and haemostasis : JTH. 2010; 8: 202–4. 10.1111/j.1538-7836.2009.03678.x. [DOI] [PubMed] [Google Scholar]
- 21.Cines DB, Kaywin P, Bina M, Tomaski A, Schreiber AD. Heparin-associated thrombocytopenia. The New England journal of medicine. 1980; 303: 788–95. 10.1056/nejm198010023031404. [DOI] [PubMed] [Google Scholar]
- 22.Husseinzadeh HD, Gimotty PA, Pishko AM, Buckley M, Warkentin TE, Cuker A. Diagnostic accuracy of IgG-specific versus polyspecific enzyme-linked immunoassays in heparin-induced thrombocytopenia: a systematic review and meta-analysis. Journal of thrombosis and haemostasis : JTH. 2017; 15: 1203–12. 10.1111/jth.13692. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 23.Doepker B, Mount KL, Ryder LJ, Gerlach AT, Murphy CV, Philips GS. Bleeding risk factors associated with argatroban therapy in the critically ill. J Thromb Thrombolysis. 2012; 34: 491–8. 10.1007/s11239-012-0758-y. [DOI] [PubMed] [Google Scholar]
- 24.Greinacher A, Farner B, Kroll H, Kohlmann T, Warkentin TE, Eichler P. Clinical features of heparin-induced thrombocytopenia including risk factors for thrombosis. A retrospective analysis of 408 patients. Thromb Haemost. 2005; 94: 132–5. 10.1160/TH04-12-0825. [DOI] [PubMed] [Google Scholar]
- 25.Schindewolf M, Steindl J, Beyer-Westendorf J, Schellong S, Dohmen PM, Brachmann J, Madlener K, Potzsch B, Klamroth R, Hankowitz J, Banik N, Eberle S, Kropff S, Muller MM, Lindhoff-Last E. Frequent off-label use of fondaparinux in patients with suspected acute heparin-induced thrombocytopenia (HIT)--findings from the GerHIT multi-centre registry study. Thromb Res 2014; 134: 29–35. 10.1016/j.thromres.2014.03.029. [DOI] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.
Supplementary Materials
Cumulative incidence of major bleeding by HIT status and treatment with alternative anticoagulant. Bleeding events were assessed starting on day of suspicion for HIT (day 0) and defined using a modified ISTH definition for major bleeding
HIT+Tx+, Patients who received treatment for HIT with an alternative anticoagulant and were adjudicated to be HIT positive by expert panel consensus; HIT–Tx+, Patients who received treatment for HIT with an alternative anticoagulant and were adjudicated to not have HIT by expert panel consensus; HIT–Tx–, Patients who did not receive treatment for HIT with an alternative anticoagulant and were adjudicated to not have HIT by expert panel consensus
Cumulative Incidence of new or progressive thrombosis by HIT group and treatment. Thrombotic events were assessed starting on day of suspicion for HIT (day 0) and defined using a modified International Society on Thrombosis and Haemostasis definition of major bleeding.
HIT+Tx+, Patients who received treatment for HIT with an alternative anticoagulant and were adjudicated to be HIT positive by expert panel consensus; HIT–Tx+, Patients who received treatment for HIT with an alternative anticoagulant and were adjudicated to not have HIT by the consensus of expert panel; HIT–Tx–, Patients who did not receive treatment for HIT with an alternative anticoagulant and were adjudicated to not have HIT by expert panel consensus