key: cord-0918169-e1tk8qvw
authors: Dua, Anahita; Thondapu, Vikas; Rosovsky, Rachel; Hunt, David; Latz, Christopher; Waller, H. David; Manchester, Scott; Patell, Rushad; Romero, Javier; Ghoshhajra, Brian; Eagleton, Matthew; Brink, James; Hedgire, Sandeep
title: DVT Protocol Optimization to Minimize Healthcare Worker Exposure in COVID-19
date: 2020-08-11
journal: J Vasc Surg Venous Lymphat Disord
DOI: 10.1016/j.jvsv.2020.08.005
sha: 84ee027246f6aa6594f749c30b7eac76d45b70ac
doc_id: 918169
cord_uid: e1tk8qvw

OBJECTIVES: There are no societal ultrasound guidelines detailing appropriate patient selection for deep vein thrombosis (DVT) imaging in COVID-19 patients nor are there protocol recommendations aimed at decreasing exposure time for ultrasound technologists. We aimed to provide COVID-19 specific protocol optimization recommendations limiting ultrasound technologist exposure while optimizing patient selection. METHODS: A novel two-pronged algorithm was implemented to limit the DVT ultrasound studies on COVID-19 patients prospectively which included direct physician communication with the care team and a COVID-19 specific imaging protocol was instated to reduce ultrasound technologist exposure. In order to assess pretest risk of DVT, sensitivity and specificity of serum D-Dimer in 500-unit increments from 500 to 8000 ng/mL and a receiver operating characteristic curve (ROC) to assess performance of serum D-Dimer in predicting DVT was generated. Rates of DVT, pulmonary embolism (PE) and scan times were compared using t-test and Fisher’s exact test (before and after implementation of the protocol). RESULTS: Direct physician communication resulted in cancellation or deferral of 72% of requested exams in COVID-19 positive patients. A serum D-Dimer > 4000ng/mL yielded a sensitivity of 80% and a specificity of 70% (CI: 0.54-0.86) for venous thromboembolism. Using the COVID-19 specific protocol, there was a significant (50%) reduction in scan time (p<0.0001) in comparison with conventional protocol. CONCLUSION: A direct physician communication policy between imaging physician and referring physician resulted in deferral or cancellation of a majority of requested DVT ultrasound exams. An abbreviated COVID-19 specific imaging protocol significantly decreased exposure time to the ultrasound technologist.

Given the increase in COVID-19 positive patients in the USA, there has been a sharp increase in 4 bedside imagining requests to evaluate for DVT in hospitals across the country. Imaging 5

technologists perform multiple bedside ultrasounds on both COVID-19 positive and non 6 COVID-19 patients across hospital systems. Hence, an increase in ultrasounds not only results 7 in significant COVID-19 exposure to vascular technologists but also creates a simultaneous risk 8 of transmission of COVID-19 to inpatients through the vascular technologists acting as potential 9 asymptomatic carriers. 5 

There are currently no COVID-19 specific guidelines providing recommendations for or against 11 use of DVT US nor is there a COVID-19 specific DVT scanning protocol that has been modified 12 to decrease the exposure time for vascular technologists. As the number of COVID-19 patients 13 increased, we aim to review our experience at a large, tertiary care hospital and provide COVID-14 19 specific protocol optimization recommendations focused on providing necessary care to 15 patients while limiting ultrasound technologist healthcare workers' exposure to identified. Each exam request was counted as a single data point regardless of whether the order 1 was for bilateral or unilateral US. Indications for the exam, DVT status, presence or absence of 2 pulmonary embolism (PE), serum D-Dimer level, and time from exam begin to exam completion 3 (per the Radiology Information System) were documented. At our institution the vascular laboratory and the radiology department both receive orders for 9

DVT US imaging and perform the studies. Hence, we implemented one protocol aimed at 10 decreasing ordering of studies in our vascular lab and another protocol aimed at decreasing 11 scanning time in the radiology department. This was structured so that each protocol could be 12 individually assessed to determine utility. 13

At our institution, we do not anticoagulation for muscular calf veins (soleal, gastrocnemial) After implementation of this algorithm by the vascular laboratory only, all DVT US on COVID-18 19 positive patients were reviewed to determine how many US orders were deemed unnecessary 19 by the ordering provider and canceled based on the algorithm. Statistical analysis included 20 descriptive statistics; both sensitivity and specificity of serum D-Dimer in 500 unit increments 21 from 500 to 8000 ng/mL were calculated and a receiver operating characteristic curve (ROC) 22 was generated to determine the serum D-Dimer value that was acceptably predictive of DVT. 1 From this, an area under the cure (AUC) was generated. Table I . 10

Continuous outcomes were reported as means and standard deviations or median and 11 interquartile ranges (IQR), as appropriate. Dichotomous outcomes were reported as counts and 12 percentages. The protocol groups were compared using the student's T-test or Wilcoxon rank-13 sum test for continuous outcomes. Fisher's exact test was used for binary outcomes. All tests 14 were two-tailed with an alpha-level of 0.05 indicating statistical significance. Statistical 15 calculations were performed using R (version 3.6.2). 16 implementation for a total of 66 US requests. Application of the algorithm (including direct 1 conversations with the medical care team component) resulted in 18 of the 25 ordered studies 2 being canceled (72%). This was a joint decision between the referring faculty and vascular lab 3 physician to cancel (or defer) the exam. In all cases where the US was canceled, a physician to 4 physician conversation did occur; after implementation of the algorithm only 7 of the 25 (28%) 5 originally ordered ultrasounds were deemed necessary and performed. None of these 7 6 ultrasounds were positive for DVT. Overall, in the 48 patients who underwent DVT US 7 imaging during the 4-week study period (41 performed in the pre-protocol timeframe + 7 8 performed in the post protocol timeframe), the primary indication for the study was "swelling or 9

pain" in the limb (Table II) . 10 Implementation of the protocol decreased the DVT US volume by 72% as only 28% of COVID-11 19 positive patients had an indication for DVT US that would change medical management 12 based on the guidelines implemented by the vascular lab. This included patients who had renal 13 disease precluding the use of low molecular weight heparin (resulting in increased nursing 14 contact due to the need for heparin drip titration), patients where anticoagulation may be lethal if 15 a bleed ensued (i.e. patients who refuse blood products), or those with recent surgery or 16 hemorrhagic strokes. The patient scanning orders canceled occurred after an attending to 17 attending discussion regarding the utility of the scan in the face of a serum D-dimer > 4000 18 ng/mL and/or clinical suspicion of DVT. Serum D-Dimer levels were found to be accurate to 19 predict VTE (AUC 0.71, 95% CI 0.54-0.86). A serum D-Dimer cutoff of > 4000ng/mL yielded a 20 sensitivity of 80% and a specificity of 70% ( Figure 1) . 21

All 18 patients who had US requests canceled/deferred based on the algorithm were administered 22 therapeutic anticoagulation. There were no bleeding events associated with this patient group. 23 study period of which 53 (55%) were performed conventionally and 44 (45%) were performed 3 using the COVID-19 focused scanning protocol. Where there was no difference between groups 4 in demographics, serum D-dimer values, and DVT or PE positivity rates, the time to perform the 5 COVID-19 focused ultrasound was significantly lower (Table III) pneumonia, as well as particular patterns in hematologic testing including lymphopenia. 5, 6 In 13 addition, these patients have been noted to have abnormal coagulation testing including elevated 14 serum D-dimer levels, elevated fibrinogen, mildly prolonged prothrombin time and, rarely, 15 thrombocytopenia. 1, 7, 8 Markedly elevated serum D-dimer has been identified as a consistent 16 marker for mortality, moreover disseminated intravascular coagulation as per criteria 17 representing fulminant activation of coagulation and consumption of factors has been shown to 18 develop in as high as 71.4% of patients that succumbed to COVID-19 pneumonia. 1 19 In addition to laboratory derangements, COVID-19 patients have been shown to have variable 20 rates of thrombotic complications, as low as 7.7% to as high as 25% in critically ill patients. 9 ,10 21 Moreover in an autopsy series of 4 patients who died of COVID-19 from New Orleans revealed 22 that there was evidence of extensive micro-thrombi in pulmonary capillaries and venules, but 23 also evidence of diffuse alveolar hemorrhage, providing support that even in patients without 1 evidence of VTE on imaging studies, microthrombi may be prevalent. 11 The evidence for 2 anticoagulation is limited thus far although an area of active clinical investigation. It has been 3 more apparent through the COVID-19 experience globally, that even in the face of a negative 4 DVT US, full anticoagulation may be appropriate given that microthrombi form in these patients 5 causing significant end organ damage. 10 communication. While there are data to suggest that patients with COVID-19 may be at a higher 16 bleeding risk, we did not have any bleeding events in the patient cohort who were therapeutically 17 anticoagulated based on our algorithm. 12 We recognize that our study had a smaller number of 18 patients however given the current literature supporting the hypercoagulable state of COVID-19 19 patients and the associated D-Dimer increase even with no discernable DVT, we support full 20 anticoagulation in this patient group especially given the lack of bleeding events observed in our 21 study. 22 1 the technologist, but also in the spread of COVID through inpatients if technologists become 2 asymptomatic carriers. 3

Our vascular and radiology group has now collaborated to develop an algorithm to curtail the 4 ordering of DVT US. This algorithm included a serum D-Dimer level > 4000 which correlated 5 with an acceptable sensitivity of 80% in our cohort. Chinese investigators early in the pandemic 6 reviewed the prevalence of venous thromboembolism (VTE) in patients with COVID-19 and 7

concluded that serum D-Dimer was indeed a good index for predicting VTE in patients with 8 severe disease. In their study of 88 patients with COVID 19, they identified 1.5 µg/mL as a good 9 cut-off parameter with a sensitivity of 85% and specificity of 88.5% with a positive predictive 10 value of 70.8% and negative predictive value of 94.7% of VTE identification. 9 Hence, if clinical 11 suspicion with supported laboratory values can yield a high likelihood of thrombus and 12 anticoagulation is not contraindicated, treatment should be initiated with therapeutic 13 anticoagulation after a physician to physician discussion about patient risk with therapeutic 14 anticoagulation. It is important to acknowledge that even with a negative ultrasound finding, a 15 patient with a high D-dimer may have microthrombi which is best managed with therapeutic 16 anticoagulation. 17

We have now implemented an algorithm throughout our institution that should decrease the 18 number of DVT US ordered by 72% and decreases technologists' time per exam by 50% based 19 on our data. 20

The scanning protocol modifications are detailed (Table I) There is no uniform consensus on anticoagulation in patients with isolated calf DVT. 16 The 12 CACTUS study randomized 259 patients (non-COVID) into treatment and placebo arms and 13 showed no significant difference in the primary outcomes such as proximal propagation of the 14 thrombus or pulmonary embolism. 17 patients in such a way that to decrease unnecessary exposure to US technologists. Another 7 limitation may be that the modifications of venous ultrasound recommended in this study were 8 evaluated in isolation so we cannot state that combining the two modalities would be more, less, 9 or equally efficacious. 10

Across the country an increase in COVID-19 patients is resulting in a rise of DVT US testing 12 requests which in turn results in significant exposure for vascular technologists. No protocol 13 currently exists to triage patients to ensure those who would benefit from US would receive it 14 and to minimize health care worker exposure in those who may not need it. We have 15 implemented a novel algorithm throughout our institution that decreases the number of DVT US 16 by 72% as well as decreases technologist time per exam by 50%. We have implemented these 17 protocols in both the vascular lab and the radiology department. We recommend implementing 18 both protocols as a joint effort to decrease technologist exposure time and volume of DVT US at 19

institutions caring for patients with COVID-19. 20 pulmonary embolism and scan times with conventional vs. COVID-19 specific US protocol *. for Thrombosis (n=48)

Abnormal coagulation parameters are associated with 3 poor prognosis in patients with novel coronavirus pneumonia

Anticoagulant treatment is associated with 6 decreased mortality in severe coronavirus disease 2019 patients with coagulopathy

Plasminogen Activator (tPA) Treatment for COVID-19 Associated Acute Respiratory 10

Distress Syndrome (ARDS): A Case Series

Upper Extremity Venous Duplex Evaluation

COVID-19: the case for health-care worker screening 17 to prevent hospital transmission

Therapeutic and triage strategies 20 for 2019 novel coronavirus disease in fever clinics

Anticoagulation in patients with isolated 3 distal deep vein thrombosis: a meta-analysis

Acute Pulmonary Embolism

Associated with COVID-19 Pneumonia Detected by Pulmonary CT Angiography

Do isolated calf deep vein thrombosis need anticoagulant treatment?

Anticoagulant 11 therapy for symptomatic calf deep vein thrombosis (CACTUS): a randomised, double-12 blind, placebo-controlled trial

A STROBE cohort 15 study of 755 deep and superficial upper-extremity vein thrombosis

We would like to thank Melanie L Orlowski and Janice Write for their 22 support.