Low anti-Factor Xa level predicts 90-day symptomatic venous thromboembolism in surgical patients receiving enoxaparin prophylaxis: A pooled analysis of eight clinical trials

Abstract

Objective:

To examine the relationship between enoxaparin dose adequacy, quantified with anti-Factor Xa (aFXa) levels, and 90-day symptomatic venous thromboembolism (VTE) and post-operative bleeding.

Summary Background Data:

Surgical patients often develop “breakthrough” VTE events—those which occur despite receiving chemical anticoagulation. We hypothesize that surgical patients with low aFXa levels will be more likely to develop 90-day VTE, and those with high aFXa will be more likely to bleed.

Methods:

Pooled analysis of eight clinical trials (N=985) from a single institution over a four year period. Patients had peak steady state aFXa levels in response to a known initial enoxaparin dose, and were followed for 90 days. Survival analysis log-rank test examined associations between aFXa level category and 90-day symptomatic VTE & bleeding.

Results:

Among 985 patients, 2.3% (n=23) had symptomatic 90-day VTE, 4.2% (n=41) had 90-day clinically relevant bleeding, and 2.1% (n=21) had major bleeding. Patients with initial low aFXa were significantly more likely to have 90-day VTE than patients with adequate or high aFXa (4.2% vs. 1.3%, p=0.007). In a stratified analysis, this relationship was significant for patients who received twice daily (6.2% vs. 1.5%, p=0.003), but not once daily (3.0% vs. 0.7%, p=0.10) enoxaparin. No association was seen between high aFXa and 90-day clinically relevant bleeding (4.8% vs. 2.9%, p=0.34) or major bleeding (3.6% vs. 1.6%, p=0.18).

Conclusions:

This manuscript establishes inadequate enoxaparin dosing as a plausible mechanism for breakthrough VTE in surgical patients, and identifies anticoagulant dose adequacy as a novel target for process improvement measures.

MINI ABSTRACT

This pooled analysis of eight clinical trials (N=985) examined the association between inadequate enoxaparin dosing, quantified by low anti-Factor Xa (aFXa) level, and 90-day symptomatic venous thromboembolism (VTE). The study showed that patients with initial low aFXa were significantly more likely to have 90-day VTE than patients with adequate or high aFXa (4.2% vs. 1.3%, p=0.007). High aFXa was not significantly associated with 90-day bleeding or major bleeding.

INTRODUCTION

Venous thromboembolism (VTE) is an important safety issue amongst hospitalized patients. VTE represents a substantial burden to patients, providers, and healthcare systems and causes 250,000 hospitalizations per year in the United States alone^1,2. Pulmonary embolus (PE) causes 33,000 of deaths per year in post-operative patients^3,4. Deep venous thrombosis (DVT) has lifelong ramifications, most notably the post-thrombotic syndrome. This lifelong condition reduces productivity, is an independent predictor of poor quality of life after DVT, and creates substantial burden for patients and healthcare systems^5,6. Patients with VTE have a risk of VTE recurrence of 3% per year for the rest of their life⁷. VTE prevention is the dominant initial strategy to minimize morbidity and mortality^1,2. Unfortunately, even aggressive VTE prevention strategies are not completely effective.

Surgery is a recognized risk factor for VTE, and surgical patients are at substantially higher risk for VTE than medically ill patients^8,50. Surgeons typically provide post-operative chemical prophylaxis to minimize risk for VTE. “Breakthrough” VTE events--VTE that occurs despite receipt of guideline-compliant chemical prophylaxis--are frustrating for clinicians, potentially life threatening for patients, and represent a substantial resource and economic burden to hospital systems. Prior clinical trials have shown that between 2% and 10% of highest risk surgical patients have a VTE event despite receiving standardized enoxaparin prophylaxis^9–11; importantly, available data cannot explain why “breakthrough” VTE occur. Current VTE prophylaxis guidelines fail to address the noteworthy proportion of patients known to have “breakthrough” VTE events in response to standard prophylaxis, and this gap in knowledge has important ramifications for patient safety in the peri-operative period.

Current guidelines for VTE chemoprophylaxis are largely based on studies that provided fixed-dose prophylaxis to the aggregate group of surgical patients. Fixed-dose enoxaparin prophylaxis has clearly been shown to provide VTE risk reduction amongst the aggregate surgical population, particularly among high-risk patients^2,8. However, recent data supports that surgical patients demonstrate variable pharmacodynamics in response to fixed doses of enoxaparin^12–15. This is clinically demonstrated by variable anti-Factor Xa levels (aFXa), a marker of extent of anticoagulation, in response to fixed enoxaparin dosing.

Existing published clinical trials are generally underpowered to associated adequacy of enoxaparin dosing, quantified by aFXa levels, and clinically relevant endpoints like 90-day bleeding or VTE events. An association between low aFXa levels and 90-day VTE events has been demonstrated in small, single center studies^16–19. Importantly, no study has definitively correlated aFXa with 90-day clinically relevant bleeding. The potential for type I and type II error exists in these smaller trials, especially because both VTE and bleeding are rare events. Although existing “one size fits all” enoxaparin dosing strategies benefit the aggregate group, it is plausible that patient-centric enoxaparin dosing could produce superior safety and effectiveness results.

We hypothesize that 1) surgical patients whose enoxaparin is inadequately dosed will be more likely to develop 90-day symptomatic VTE and that 2) surgical patients whose enoxaparin is over-dosed will be more likely to develop 90-day clinically relevant bleeding. This manuscript examines these hypotheses by pooling data from eight distinct clinical trials conducted in surgical patients over a four year period; all trials occurred at a single institution using identical inclusion and exclusion criteria, and all examined peak aFXa levels and followed patients for 90-day clinically relevant bleeding and symptomatic VTE.

METHODS

This study was designed as a pooled analysis of previously completed clinical trials from a single institution. All trials received IRB approval at the University of Utah and were registered on clinicaltrials.gov prior to recruitment.

The University of Utah has a longstanding interest in anticoagulant metabolism in hospitalized patients^14,20. Since 2014, that interest has been concentrated on surgical patients, with an emphasis on aFXa levels in response to prophylactic doses of enoxaparin.

Inclusion criteria were completed clinical trials from the University of Utah that were accepted for publication in the peer-reviewed literature and 1) enrolled patients after surgery under general anesthesia, 2) enrolled patients with no further operative procedures planned, 3) provided patients with prophylactic doses enoxaparin initiated after surgery, 4) initiated enoxaparin within 24 hours of surgery, 5) performed aFXa level monitoring with real time enoxaparin dose adjustment in response to aFXa levels.

Patients in all trials enrolled surgical patients who received prophylactic doses of enoxaparin according to the specific trial protocol. All patients had aFXa level monitoring and received real time enoxaparin dose adjustment, based on a clinical protocol (Figure 1), to target an optimal peak aFXa level. aFXa levels were obtained using a chromogenic assay from Diagnostica Stago (Parsippany NJ). Peak levels drawn at 3–5 hours after the third enoxaparin dose were included. For once daily dosing, low peak aFXa levels were defined as <0.3 IU/mL, in range peak aFXa levels as 0.3–0.5 IU/mL, and high peak aFXa levels as >0.5 IU/mL. For twice daily dosing, low peak aFXa levels were defined as <0.2 IU/mL, in range peak aFXa levels as 0.2–0.4 IU/mL, and high peak aFXa levels as >0.4 IU/mL. These ranges were chosen based on the institution’s extensive experience with real time enoxaparin dose adjustment guided by aFXa levels, and have previously been shown to optimize the balance between VTE risk reduction and post-operative bleeding^19,21,22.

Figure 1:

Enoxaparin administration and dose adjustment protocol for included trials

Outcome definitions

The primary independent variable for all trials was peak aFXa level in response to initial enoxaparin dosing. This was the aFXa level obtained prior to any dose adjustment. All patients were followed to 90 days after surgery using a combination of chart review and mandatory telephone or in-person contact. Patients unable to be reached at 90 days were considered lost to follow-up. Time series analysis allowed inclusion of these patients into the final analysis prior to censoring.

The primary dependent variables for all trials were 90-day symptomatic VTE and 90-day clinically relevant bleeding and major bleeding. The VTE endpoint was defined as symptomatic VTE events that were confirmed with imaging. These included symptomatic deep venous thrombosis, pulmonary embolus, portal vein thrombosis, and inferior vena cava thrombosis. We chose to include central venous catheter associated DVT (CVC-DVT), when symptomatic and confirmed with imaging, in the 90-day VTE outcome. The most recent American College of Chest Physicians guidelines to comment on treatment of CVC-DVT (8^th edition) recommend the same duration of anticoagulation for CVC-DVT and non CVC-DVT²³, and patients who develop CVC-DVT are significantly more likely to develop the post-thrombotic syndrome on long term follow-up²⁴. As the recommended management strategy and complication profile are the same, CVC-DVT was included in the 90-day VTE outcome for all trials. The 2012 American College of Chest Physicians guidelines specifically advocate against screening duplex ultrasound for asymptomatic patients, including patients perceived to be at high risk for post-operative VTE². All included trials were designed as pragmatic trials that mirror clinical practice; none performed screening duplex ultrasound on asymptomatic patients. Patients were tested for VTE based on the presence of symptoms, and upon order from their physician.

90-day bleeding was defined as any bleeding requiring change in the course of care, including enoxaparin cessation, transfusion deemed unexpected by the surgeon, interventional or bedside procedure, unexpected return to the operating room, or unexpected visit to the emergency room. This intentionally broad definition was identical to the definition used in a seminal manuscript examining bleeding risk in patients who received prophylactic post-operative anticoagulation²⁵. In addition, we identified major bleeding events as rigorously defined by the International Society of Thrombosis and Hemostasis’ consensus definition of major bleeding for randomized controlled trials that examine antihemostatic drugs²⁶. Time to event was tracked for all bleeding events.

We limited our analysis to the impact of initial peak aFXa on 90-day outcomes. All trials performed real time, aFXa-level guided enoxaparin dose adjustment with repeated measures of aFXa at steady state (Figure 1), because out of range aFXa levels predispose patients to complications including VTE or bleeding^16–19. However, among all patients who received dose adjustment, 64.3% (281/437) received an adjusted dose but had no repeat aFXa level drawn due to hospital discharge. Thus, for the majority of patients whose initial aFXa level was out of range, the impact of real time dose adjustment was unknown and inclusion of repeat aFXa levels as a predictor variable introduced a serious confounder.

Analysis Plan

Data from individual trials were aggregated and stored using the Research Electronic Data Capture (REDCap) platform. Data were merged and analyzed using Stata15 (College Station, TX).

As a measure of enoxaparin effectiveness, we compared rates of VTE and bleeding between patients with low initial peak aFXa (<0.3 IU/mL for once daily and <0.2 IU/mL for twice daily) and those with in-range or high aFXa (≥0.3 IU/mL for once daily and ≥0.2 IU/mL for twice daily). As a measure of enoxaparin safety, we compared VTE and bleeding between patients with above-range peak aFXa (>0.5 IU/mL for once daily and >0.4 IU/mL for twice daily) and those with in-range or low aFXa (≤0.5 IU/mL for once daily and ≤0.4 IU/mL for twice daily).

The survival analysis log-rank test examined the association between initial peak aFXa level and 90-day symptomatic VTE. To examine the impact of initial adequate enoxaparin dosing, we compared 90-day VTE between patients with low versus in range/high aFXa based on initial enoxaparin dosing.

Similarly, survival analysis log-rank test examined associations between 90-day clinically relevant bleeding and peak aFXa level and major bleeding and peak aFXa level. We examined the impact of initial over-aggressive enoxaparin dosing by comparing 90-day bleeding between patients with low/in range versus high aFXa level.

For the survival analysis, the rate of censoring was 3.5%, including 8 patients lost to followup, 21 patients dropped for protocol violations, 2 deaths, and 1 request to drop. These patients were treated as censored observations in the survival analysis

For pharmacokinetic optimization, the goal peak steady state aFXa level is different for once (0.3–0.5 IU/mL) versus twice (0.2–0.4 IU/mL) daily enoxaparin. Thus, survival analyses were initially pooled for all patients and then stratified by once versus twice daily drug administration.

The initial analysis plan included the potential to perform multivariable regression to examine the predictive role of aFXa level on VTE and bleeding when controlling for other factors. However, for the VTE outcome, only 23 events occurred. Similarly, only 28 patients with clinically relevant bleeding who also had aFXa levels were identified. Given the low number of total events, we abandoned use of multivariable regression due to high risk for model overfitting and the potential to generate invalid results²⁷.

RESULTS

The University of Utah conducted eight distinct trials meeting inclusion criteria between 2015 and 2019; collectively, 985 surgical patients were enrolled (Table 1). Patient populations included reconstructive surgery (3 trials; N=507)^18,28,29, thoracic surgery (2 trials; N=198)^30,31, colorectal surgery (1 trial; N=116)³², orthopedic surgery (1 trial; N=109)³³, and trauma surgery (1 trial, N=55)³⁴ (Table 2).

Table 1:

Demographics of included patients (N=985)

Characteristic	Value
Age in years, mean (range)	51.8 (18–94)
Ethnicity, % (n)
White	89.2% (879)
African American	0.6% (6)
Native American/Alaskan Native	1.0% (10)
Hispanic or Latino	7.6% (75)
Pacific Islander	1.6% (16)
Body mass index in kg/m2, mean (range)	28.6 (15.7–49.8)
Weight in kg, mean (range)	82.9 (30.4–185.0)
Female gender, % (n)	51.2% (504)
Diabetes receiving treatment, % (n)	14.3% (141)
Creatinine in mg/dl, mean (range)	0.83 (0.28–1.77)
Current smoker, % (n)	12.7% (125)
Primary surgical procedure type, % (n)
Reconstructive surgery	51.4% (507)
Thoracic surgery	20.1% (198)
Colorectal surgery	11.8% (116)
Orthopedic trauma surgery	11.1% (109)
Trauma surgery	5.6% (55)
Length of operation in minutes, mean (range)	230 (21 to 1064)
Length of hospital stay in days, mean (range)	7.1 (1 to 58)
Length of prophylaxis in days, mean (range)	11.3 (1 to 64)

Table 2:

Trials included in pooled analysis.

First author and year	Journal	Clinicaltrials.gov identifier	Patient population (N)	Trial design	Enoxaparin regimen	Primary and secondary endpoint(s)	Use of aFXa-guided enoxaparin dose adjustment
Pannucci 201718	Plastic and Reconstructive Surgery	NCT02411292	Plastic and reconstructive surgery (94)	Observational (interventional for out of range aFXa)	40mg once daily	Peak aFXa; 90-day clinically relevant bleeding and 90-day VTE	Yes
Wall 201834	American Journal of Surgery	NCT02704052	Trauma surgery (55)	Observational (interventional for out of range aFXa)	30mg twice daily	Peak aFXa; 90-day clinically relevant bleeding and 90-day VTE	Yes
Pannucci 201830	Annals of Thoracic Surgery	NCT02704052	Thoracic surgery (93)	Observational (interventional for out of range aFXa)	40mg once daily	Peak aFXa; 90-day clinically relevant bleeding and 90-day VTE	Yes
Pannucci 201828	Plastic and Reconstructive Surgery	NCT02687204	Plastic and reconstructive surgery (118)	Observational (interventional for out of range aFXa)	40mg twice daily	Peak aFXa; 90-day clinically relevant bleeding and 90-day VTE	Yes
Jones 201933	Journal of Orthopedic Trauma	NCT02704052	Orthopedic trauma (109)	Observational (interventional for out of range aFXa)	30mg twice daily	Peak aFXa; 90-day clinically relevant bleeding and 90-day VTE	Yes
Pannucci 201932	JAMA Surgery	NCT02704052	Colorectal surgery (116)	Observational (interventional for out of range aFXa)	40mg once daily	Peak aFXa; 90-day clinically relevant bleeding and 90-day VTE	Yes
Pannucci 202031	Annals of Thoracic Surgery	NCT03251963	Thoracic surgery (105)	Pre-post interventional trial	40mg once daily versus weight tiered regimen (30/40/50mg once daily)	Peak aFXa; 90-day clinically relevant bleeding and 90-day VTE	Yes
Pannucci in press29	Plastic and Reconstructive Surgery	NCT03212365	Plastic and Reconstructive surgery (295)	Randomized double blind	40mg twice daily versus 0.5mg/kg twice daily	Peak aFXa; 90-day clinically relevant bleeding and 90-day VTE	Yes

The median 2005 Caprini score was 7 (range 1–20 and mean 7.3) (Supplemental Data File 1, http://links.lww.com/SLA/C708), meaning that the majority of patients were at high or highest risk for post-operative VTE^8,35. Patient demographics and individual risk factors for VTE are detailed in Tables 1 and 3, respectively.

Table 3:

Baseline risk factors for VTE, based on the 2005 Caprini Risk Assessment Model (N=985)

Risk Factor
ONE POINT FACTORS
Age 41–59	37.6% (370)
Minor surgery planned	2.7% (27)
Major surgery within 30 days	12.1% (119)
Varicose veins	7.8% (77)
History of IBD	10.2% (100)
Swollen legs (current)	7.1% (70)
BMI>25	64.0% (630)
Acute myocardial infarction <3 months	0.1% (1)
Congestive heart failure <1 month	0.3% (3)
Sepsis <1 month	2.7% (27)
Serious lung disease (including pneumonia) <1 month	0.9% (9)
Chronic obstructive pulmonary disease	3.6% (35)
Medical patient currently at bedrest	0.2% (2)
TWO POINT FACTORS
Age 60–74 years	28.7% (283)
Arthroscopic surgery	0 (0)
Malignancy (present or previous)	38.8% (382)
Major surgery >45 minutes	96.3% (949)
Laparoscopic surgery >45 minutes	8.0% (79)
Immobilizing plaster cast	0.6% (6)
Central venous access	23.7% (233)
THREE POINT FACTORS
Age ≥75	6.5% (64)
History of DVT/PE	11.5% (113)
Family history of DVT/PE	13.8% (136)
Any genetic hypercoagulable state	0.4% (4)
FIVE POINT FACTORS
Elective major lower extremity arthroplasty	0.5% (5)
Hip, pelvis, or leg fracture <1 month	12.6% (124)
Stroke <1 month	0.1% (1)
Multiple trauma <1 month	6.8% (67)
Acute spinal cord injury or paralysis <1 month	0 (0)
Females	n=504
ONE POINT FACTORS
Oral contraceptives	11.7% (59)
Pregnancy or postpartum (<1 month)	0 (0)
History of unexplained stillborn infant, recurrent spontaneous abortion (≥3), premature birth with toxemia or growth-restricted infant	7.5% (38)

Seven trials were conducted only at the University of Utah. One thoracic study included patients from the University of Utah and the University of Michigan. To avoid by-site confounding that might impact cohort homogeneity, all Michigan patients were excluded from this analysis.

Initial Peak aFXa and 90-day Venous Thromboembolism

Among 985 patients from the eight trials, 72 had unusable peak aFXa data and were excluded from the analysis for 90-day VTE and bleeding. Specific reasons for unusable peak aFXa data included peak aFXa drawn at incorrect time (n=26), patient discharged prior to lab draw (n=22), patient bled prior to aFXa (n=12), patient refused lab draw (n=3), patient switched to unfractionated heparin infusion (n=2), unplanned return to OR (n=2), medication distribution error (n=1), post-operative epidural placement (n=1), acute renal failure (n=1), patient requested study discontinuation (n=1), and reason unknown (n=1).

For once daily dosing (n=376), 62.5% (n=235) of patients had low aFXa <0.3 IU/mL, 35.1% (n=132) had in-range aFXa 0.3–0.5 IU/mL , and 2.4% (n=9) had high aFXa >0.5 IU/mL. For twice daily dosing (n=537), 27.0% (n=145) of patients had low aFXa <0.2 IU/mL, 59.0% (n=317) had in range aFXa 0.2–0.4 IU/mL, and 14.0% (n=75) had high aFXa >0.4 IU/mL.

Twenty three patients (2.3%) had symptomatic 90-day VTE events. Symptomatic VTE was not significantly different between patients who received twice versus once daily enoxaparin (2.6% vs. 2.0%, p=0.67).

The primary analysis including all patients and enoxaparin regimens showed that patients with initial low aFXa were significantly more likely to have 90-day symptomatic VTE than patients with in range/high aFXa (4.2% vs. 1.3%, logrank p=0.007) (Figure 2). As low aFXa is defined differently for once daily (<0.3 IU/mL) versus twice daily (<0.2 IU/mL) enoxaparin, a stratified analysis was performed. The stratified analysis confirmed that initial low aFXa predicted symptomatic 90-day VTE in patients who receive twice daily enoxaparin (6.2% vs. 1.5%, logrank p=0.003), but not necessarily in patients who receive once daily enoxaparin (3.0% vs. 0.7%, logrank p=0.14) (Figure 3 and Supplemental Data File 2 , http://links.lww.com/SLA/C709).

Figure 2:

Impact of initial low aFXa on 90-day symptomatic VTE for all patients

Figure 3:

Impact of initial low aFXa on 90-day symptomatic VTE for patients who receive twice daily enoxaparin

Initial Peak aFXa and 90-day Clinically Relevant Bleeding

Forty one patients (4.2%) had clinically relevant, 90-day bleeding events. Twenty nine percent (12 of 41) of all bleeding events occurred prior to enoxaparin reaching steady state, and one additional bleeding event occurred in a patient whose labs were mistimed. Thus, 28 bleeding events could be correlated with initial peak steady state aFXa. Twenty one patients (2.1%) met ISTH criteria for major bleeding²⁶, and sixteen of these patients had peak aFXa levels drawn prior to bleeding.

Patients who received twice daily enoxaparin were significantly more likely to have clinically relevant bleeding (5.6% vs. 2.2%, p=0.009) or major bleeding (3.3% vs. 0.5%, p=0.003) than those who received once daily doses.

For all patients, there was no significant relationship between high aFXa and 90-day clinically relevant bleeding (4.8% vs. 2.9%, logrank p=0.34) (Supplemental Data File 3, http://links.lww.com/SLA/C710). High aFXa was defined differently for once (>0.5 IU/mL) versus twice (>0.4 IU/mL) daily enoxaparin dosing; to control for this confounder, an analysis stratified by dose frequency was performed. There was no significant association between initial high aFXa and 90-day bleeding in patients who received twice (5.3% vs. 4.6%, logrank p=0.78) or once (0% vs. 0.8%, logrank p=0.79) daily enoxaparin (Supplemental Data Files 4, http://links.lww.com/SLA/C711 and 5, http://links.lww.com/SLA/C712).

For all patients, major bleeding was not significantly different for patients with high vs. low or in range aFXa (3.6% vs. 1.6%, p=0.18) (Figure 4). A stratified analysis for once daily (0% vs. 0.5%, p=0.82) vs. twice daily (4.0% vs. 2.4%, p=0.43) showed no significant relationship between high vs. low or in range level based on dose frequency (Supplemental Data Files 6, http://links.lww.com/SLA/C713 and 7, http://links.lww.com/SLA/C714).

Figure 4:

Impact of initial high aFXa on major bleeding for all patients

DISCUSSION

This pooled analysis of eight clinical trials, including 985 patients representing the broad spectrum of surgical patients, examined the association between initial peak aFXa level and 90-day VTE and bleeding in patients who received post-operative enoxaparin for VTE prevention. This study clearly demonstrates a significant association between low initial aFXa and 90-day symptomatic VTE. This analysis identifies enoxaparin dose adequacy as a novel and currently unconsidered paradigm for VTE risk reduction and prevention, and challenges current thought that the binary yes/no decision to provide chemical prophylaxis is sufficient. The pooled data did not demonstrate a significant relationship between initial peak aFXa and 90-day bleeding, and this finding is further discussed below.

This manuscript focuses on “breakthrough” VTE events, or events that happen despite anticoagulation being provided. However, Elliott Haut and the multidisciplinary Johns Hopkins Venous Thromboembolism Collaborative defined the broader scope of “potentially preventable VTE” as events that developed in patients who experienced a VTE prevention process failure³⁶. “Breakthrough” events as defined here would be a subset of Haut’s “potentially preventable VTE”. The Hopkins collaborative has demonstrated that mandatory clinical decision support can significantly increase provision of appropriate prophylaxis³⁷ and can concomitantly decrease racial disparities in prophylaxis provision³⁸. However, clinical decision support alone did not achieve their goal of “perfect VTE prophylaxis”, and new opportunities for quality improvement were identified³⁹.

The Hopkins collaborative has identified multiple factors contributing to breakthrough VTE, and has published evidence based-solutions. Their research demonstrates that over 12% of prescribed medication doses were not administered, that medications injected into the subcutaneous space were significantly more likely to be refused than oral medications, and medication non-administration is more likely in younger, male, and uninsured patients^40–42. The group has piloted patient-centric and bedside nurse-centric VTE education bundles⁴³, showing that both targeted interventions can significantly decrease medication non-administration. A similar process improvement project aimed at surgical resident prescribing patterns significantly increased appropriate provision of prophylaxis, but more importantly, significantly decreased rates of preventable VTE⁴⁴. This pooled analysis identifies anticoagulant dose adequacy as a novel target for future process improvement measures.

VTE risk varies at the level of the individual patient, not just by the type of procedure performed. Recent validation studies of the 2005 Caprini score and the Rogers score^45,46 have shown that a quantifiable, 15-fold variation in VTE risk exists among the overall surgical population. In addition, the effectiveness of chemical prophylaxis varies when provided to patients at different risk levels. Specialty-specific and meta-analysis data have only been able to definitively demonstrate a benefit for chemical prophylaxis in patients with 2005 Caprini scores of ≥7^8,11,47. As a result, national governing bodies advocate for individualized VTE risk assessment in surgical patients, and for the thoughtful provision of mechanical and/or chemical prophylaxis guided by patient-centric VTE risk^2,48. The paradigm of care for VTE prevention continues to shift toward an individualized, patient-centric approach.

VTE prevention strategies are becoming more focused on the individual patient, as evidenced by a rich body of surgical literature on individualized VTE risk assessment^35,45, and the variation in chemical prophylaxis effectiveness based on Caprini score^2,8,11,47. As VTE prevention becomes increasingly individualized, patient-centric enoxaparin dosing has received renewed focus in the surgical literature^13,16,20,49. Prior work from individual surgical populations has correlated patient-level factors with aFXa level in response to fixed doses. Specifically, lower aFXa levels in patients receiving fixed enoxaparin doses has been correlated with higher weight^{13–15,30,32}, increased total body surface area burned¹⁴, and increased total body surface area surgically injured¹⁸. Improved understanding of patient-level predictors of pharmacodynamics is relevant because prior small studies in isolated surgical populations have correlated low aFXa with all cause (symptomatic plus asymptomatic) VTE^16,17,19 or symptomatic VTE¹⁸, and one study in orthopedic surgery has associated high aFXa with increased risk for bleeding¹⁹. This study builds upon prior work, conducted in isolated surgical populations, through a pooled analysis of eight clinical trials conducted in a broad range of surgical patients.

Breakthrough VTE are life or quality of life-threatening for patients^3–6, and events occur commonly-- between 2% and 10% of the time--amongst patients who receive guideline-compliant chemical prophylaxis^9–11. Prior clinical trials in surgical patients have shown that surgical patients metabolize a fixed enoxaparin dose differently, and that a “one size fits all” enoxaparin dose strategy produces widely variable peak aFXa levels. Prior findings are particularly relevant because this pooled analysis demonstrates a significant association between low aFXa and 90-day symptomatic VTE. This manuscript establishes inadequate enoxaparin dosing as a plausible mechanism for breakthrough VTE in surgical patients, and identifies anticoagulant dose adequacy as a novel target for process improvement measures.

LIMITATIONS

Pooled analyses are often limited by inclusion of dis-similar studies in order to augment the total number of patients. This manuscript is based on eight clinical trials performed in a short timeframe at a single institution. Thus, while included patients had a variety of surgical procedures, their baseline demographics and comorbid conditions would be more similar than a pooled analysis of national trials. All trials had identical inclusion and exclusion criteria, and similar dose adjustment algorithms. Utilizing trials data from only a single institution forces the reader to utilize the University of Utah’s pre-defined ranges for adequate peak aFXa levels (0.2–0.4 IU/mL for twice daily dosing and 0.3–0.5 IU/mL for once daily dosing)^21,22; we acknowledge that other institutions have considered different levels to represent adequate dosing.

Pooled data were acquired by enrolling patients into a rigorous clinical trials framework, which may not mirror day-to-day care. The Hopkins Collaborative has identified multiple process issues of day-to-day care, such as medication refusal or non-administration^39–42, which would represent barriers to implementation of patient-centric enoxaparin dosing outside of a clinical trials framework.

Under a research protocol at the University of Utah, a single aFXa lab can be completed for under nineteen dollars. In addition, there is a provider time requirement for review and real-time enoxaparin dose adjustment. The cost effectiveness of enoxaparin dose optimization and aFXa level monitoring is an important topic for future research.

This manuscript identifies initial inadequate enoxaparin dosing as a process measure which can impact downstream VTE events, but cannot comment on the impact of real time enoxaparin dose adjustment on symptomatic VTE or bleeding, as discussed in Methods. Similarly, we acknowledge that an ideal analysis would utilize multivariable regression to examine the predictive role of aFXa level on VTE and bleeding when controlling for other factors. Due to a paucity of outcome events, we could not use regression due to high risk of model overfitting²⁷.

This manuscript associates low initial aFXa level with development of 90-day VTE, but does not necessarily prove causation. It is plausible that low aFXa is a predictor for downstream VTE, as patients whose anticoagulation is initially inadequate may form a critical mass of clot in the immediate post-operative period. However, further research could examine whether low aFXa is merely a symptom of a larger disorder—such as a systemic inflammatory response causing both rapid enoxaparin clearance and increased likelihood of thrombosis.

Twenty nine percent (12 of 41) of all bleeding events occurred prior to enoxaparin reaching steady state, and one additional bleeding event occurred in a patient whose labs were mistimed. Thus, 32% of patients with bleeding had aFXa levels that were unknown. By definition, these patients were excluded from the time series analysis (Supplemental Data Files 3–5, http://links.lww.com/SLA/C710, http://links.lww.com/SLA/C711, http://links.lww.com/SLA/C712) because they did not have a peak aFXa level. A single center study has previously shown a logarithmic increase in hematoma risk with increasing aFXa level¹⁹, thus it is possible or even likely that the majority of patients who bled prior to steady state had high aFXa levels. This identified source of confounding for which we cannot control makes interpretation of our bleeding data more challenging.

CONCLUSION

Many surgical patients are at high risk for post-operative venous thromboembolism, and “breakthrough” VTE events are unfortunately common. This pooled analysis of eight prospective clinical trials demonstrates an association between inadequate initial enoxaparin dosing and 90-day symptomatic VTE. These data identify enoxaparin dose adequacy as a novel and important target for future process improvement studies to optimize VTE prevention after surgical procedures.