Standard protocol for discharge for high risk csection anticoagulation

Am Fam Physician. 2019;100(7):426-434

Author disclosure: No relevant financial affiliations.

Anticoagulation therapy is recommended for preventing, treating, and reducing the recurrence of venous thromboembolism, and preventing stroke in persons with atrial fibrillation. Direct oral anticoagulants are first-line agents for eligible patients for treating venous thromboembolism and preventing stroke in those with nonvalvular atrial fibrillation. Vitamin K antagonists are recommended for patients with mechanical valves and valvular atrial fibrillation. Vitamin K antagonists inhibit the production of vitamin K-related factors and require a minimum of five days overlap with parenteral anticoagulants, whereas direct oral anticoagulants directly inhibit factor II or factor Xa, providing more immediate anticoagulation. The immediate effect of direct oral anticoagulants permits select patients at low risk to initiate treatment in the outpatient setting for venous thromboembolism, including pulmonary embolism. Low-molecular-weight heparin continues to be recommended as a first-line treatment for patients with venous thromboembolism and active cancer, although there is growing evidence of effectiveness for the use of direct oral anticoagulants in this patient population. Validated bleeding risk assessments such as HAS-BLED should be performed at each visit and modifiable factors should be addressed. Major bleeding should be treated with vitamin K and 4-factor prothrombin complex concentrate for patients already being treated with a vitamin K antagonist. Idarucizumab has been effective for reversing the anticoagulant effects of dabigatran, and andexanet alfa has been effective for reversing the effects of rivaroxaban and apixaban.

Vitamin K antagonists (e.g., warfarin [Coumadin]), unfractionated heparin, low-molecular-weight heparin (LMWH), and direct oral anticoagulants are commonly used for the prevention and treatment of systemic embolism associated with atrial fibrillation, stroke, and venous thromboembolism (VTE). LMWH and select direct oral anticoagulants can be used for anticoagulation therapy initiation on an outpatient basis.

WHAT'S NEW ON THIS TOPIC

Compared with vitamin K antagonists, direct oral anticoagulants have fewer overall drug-drug interactions, a comparable (if not lower) bleeding rate, a shorter half-life, and fixed dosing based on indication, drug interactions, and renal or hepatic function.

American Academy of Family Physicians guidelines recommend the use of oral anticoagulants in patients with a CHADS₂ score greater than 1 for the prevention of stroke in atrial fibrillation. The American Heart Association/American College of Cardiology/Heart Rhythm Society guidelines recommend a direct oral anticoagulant over a vitamin K antagonist, unless the patient has moderate-to-severe mitral stenosis or a mechanical heart valve.

In May 2018, andexanet alfa (Andexxa) was approved to reverse the anticoagulant effects of rivaroxaban (Xarelto) and apixaban (Eliquis) in patients with life-threatening or uncontrolled bleeding. Optimal dose, duration, need for repeat dosing, and mitigation of thromboembolic risk is yet to be delineated.

This article focuses on the indications for anticoagulation therapy, direct oral anticoagulant therapy, and recommendations from guidelines. Most of the recommendations are based on the 10th edition of the American College of Chest Physicians (ACCP) guidelines on antithrombotic therapy for VTE disease (Table 1). 1 – 5

Medication Overview

Indications for initiation and duration of therapy for vitamin K antagonists, direct oral anticoagulants, and LMWH are listed in Table 2. 1

• First episode of proximal DVT or PE attributed to reversible risk factor or surgery:

• Three months recommended over short-term use (grade 1B), longer use (grade 1B), or extended therapy (grade 1B)

• First episode of unprovoked proximal DVT or PE not attributed to a reversible risk factor:

• Low or moderate bleeding risk: extended use (lifelong) recommended over three months (grade 2B); high bleeding risk: three months recommended over extended use (grade 1B); recommend reassessing bleeding risk annually

• First episode of distal DVT attributed to a surgery or reversible risk factor:

• If without severe symptoms or risk factors of extension, suggest serial ultrasonography surveillance for two weeks instead of anticoagulation (grade 2C); if surveillance shows extension, recommend anticoagulation (grade 2C if it does not extend into proximal vessels; grade 1B if it extends into proximal vessels)
• If severe symptoms or risk factors of extension, recommend three months treatment over extended use (grade 1B)
• Risk factors for extension: unexplained D-dimer results; extensive DVT (> 5 cm) and/or involving multiple veins; close to proximal vein; unprovoked; cancer; previous VTE; inpatient

VITAMIN K ANTAGONISTS

In the liver, vitamin K antagonists inhibit the cyclic interconversion of vitamin K, indirectly reducing clotting and synthesis for factors II, VII, IX, and X. Vitamin K antagonists also decrease levels of vitamin K–dependent anticoagulation proteins C and S; therefore, carboxylation inhibition can result in a paradoxical increased clotting risk when vitamin K antagonist therapy is initiated. Anticoagulant effects are delayed for five days after changes to dosing, including therapy initiation, because of the variable half-lives of previously formed circulating clotting factors. 4

Dosing. Patients receiving vitamin K antagonist therapy should be treated using a systematic process to optimize effectiveness and minimize adverse effects. Heparin or LMWH should be administered at initiation of the vitamin K antagonist and be continued for a minimum of five days and until the international normalized ratio (INR) is in the targeted therapeutic range for a minimum of 24 hours. The targeted INR range depends on indication for use and, at times, patient comorbidities.

For most patients, vitamin K antagonists should be initiated at a maintenance dosage of 5 mg per day. Older patients and persons with liver disease, poor nutritional status, or heart failure may require lower initiation dosages. 4 Diarrhea, fever, and hyperthyroidism can also potentiate the effect of vitamin K antagonists. Genetic factors can predispose patients to reduced vitamin K antagonist requirements, as well as resistance. Although there is a small subset of patients who may have unexpected responses to vitamin K antagonists, it is not currently recommended that patients undergo genetic testing. 4

After a baseline INR is determined, the next INR should be obtained after the patient has received two or three doses of the vitamin K antagonist. Monitoring should then be decreased to twice weekly until the INR is within the therapeutic range, then decreased to weekly, every other week, and finally monthly. 4 The ACCP guidelines recommend INR monitoring once every 12 weeks for patients who are stable (defined as at least three months of consistent results with no required adjustment of vitamin K antagonist dosing).

If the patient's INR becomes subtherapeutic or supratherapeutic, the frequency of monitoring should be increased until the INR stabilizes again. The ACCP provides recommendations for managing supratherapeutic INRs (Table 3). 4

Vitamin K antagonists should be taken at the same time every day. An advantage to evening administration is the ability to adjust or hold the dose the same day that the INR result becomes available. If the INR is not within the desired therapeutic range after excluding explanatory factors, a 5% to 20% increase or decrease in the total weekly dosage is recommended. 6 , 7 Patients should be provided with the simplest regimen to achieve the new total weekly dosage.

Drug and Food Interactions. Vitamin K antagonists are subject to many drug interactions. Select drug-drug interactions that are considered to potentiate or inhibit the effects of vitamin K antagonists are listed in Table 4. 8 When an interacting drug is initiated or discontinued, more frequent INR checks are recommended. The timing depends on the type of interaction, what the interacting drug is, and the clinical judgement of the physician.

Foods with high vitamin K concentrations, such as leafy green vegetables, have the potential to partially reverse anticoagulation effects of the vitamin K antagonist. 4 A consistent diet is more important than limiting dietary vitamin K.

LOW-MOLECULAR-WEIGHT HEPARIN

Considerations for parenteral medications are provided in eTable A. Dalteparin (Fragmin) and enoxaparin (Lovenox) are commonly used LMWHs in clinical practice. LMWH is derived from unfractionated heparin and has an increased affinity for factor Xa relative to thrombin. 4 LMWH's anticoagulant effect is primarily from factor Xa inhibition because of its smaller size and its lessened ability to inactivate thrombin compared with unfractionated heparin. Subcutaneous LMWH has a predictable absorption and degree of anticoagulation, so monitoring with anti–factor Xa levels is not routinely recommended. Although LMWH has a similar bleeding risk and lower heparin-induced thrombocytopenia risk compared with unfractionated heparin, a patient with a history of heparin-induced thrombocytopenia should not take LMWH. 1

DIRECT ORAL ANTICOAGULANTS

Warfarin was approved in 1954, and no other oral option existed for patients requiring long-term anticoagulation therapy until 2010 when the direct thrombin inhibitor dabigatran (Pradaxa) was approved. Since dabigatran's approval, four additional direct oral factor Xa inhibitors have been approved. Characteristics of these anticoagulants are provided in Table 5 9 – 13 and eTable B. Physicians should not automatically consider all patients taking vitamin K antagonists to be good candidates for direct oral anticoagulants because of the diversity in the characteristics of these medications.

• 5 mg twice per day
• Decrease dose by 50% if on 5 mg or 10 mg twice per day and taking combined strong P-glycoprotein inhibitor and strong CYP3A4 inhibitor
• Avoid in patients taking combined P-glycoprotein and strong CYP3A4 inducers if on 2.5 mg twice per day
• Decrease dose to 2.5 mg twice per day if two of the following: age 80 years or older, body weight ≤ 60 kg (132 lb); serum creatinine ≥ 1.5 mg per dL (114 μmol per L)

• Avoid in patients taking combined P-glycoprotein and strong CYP3A4 inducers
• Reduce recurrence: 2.5 mg twice per day after six months or more of DVT or PE treatment
• Total hip or knee replacement surgery: 2.5 mg twice per day for 12 days after knee replacement surgery and 35 days after hip replacement surgery

• 10 mg twice per day for 7 days, then 5 mg twice per day
• Decrease dose by 50% if on 5 mg or 10 mg twice per day and taking combined strong P-glycoprotein inhibitor and strong CYP3A4 inhibitor
• Avoid in patients taking combined P-glycoprotein and strong CYP3A4 inducers if on 2.5 mg twice per day

• 80 mg with food for first dose, then 40 mg per day with food for 35 to 42 days
• Avoid use with P-glycoprotein inhibitors and CrCl < 30 mL per minute per 1.73 m 2 (0.50 mL per second per m 2 )

• 150 mg twice per day
• With P-glycoprotein inhibitors dronedarone (Multaq) or ketoconazole§ and CrCl 30 to 50 mL per minute per 1.73 m 2 (0.50 to 0.83 mL per second per m 2 ): 75 mg twice per day
• CrCl 15 to 30 mL per minute per 1.73 m 2 : 75 mg twice per day
• Avoid use with P-glycoprotein inhibitors and CrCl < 30 mL per minute per 1.73 m 2
• Avoid use with CrCl < 15 mL per minute per 1.73 m 2 with or without drug interaction
• Avoid with rifampin

• Total hip replacement surgery: 110 mg on first day, then 220 mg per day for 28 to 35 days
• Avoid with P-glycoprotein inhibitor and CrCl < 50 mL per minute per 1.73 m 2
• Avoid with CrCl < 30 mL per minute per 1.73 m 2 with or without drug interaction
• Avoid with rifampin

• 150 mg twice per day
• Avoid with P-glycoprotein inhibitor and CrCl < 50 mL per minute per 1.73 m 2
• Avoid with CrCl ≤ 30 mL per minute per 1.73 m 2 with or without drug interaction
• Avoid with rifampin

• 150 mg twice per day
• Avoid with P-glycoprotein inhibitor and CrCl < 50 mL per minute per 1.73 m 2
• Avoid with CrCl ≤ 30 mL per minute per 1.73 m 2 with or without drug interaction
• Avoid with rifampin

• 60 mg daily
• Avoid use in patients with CrCl > 95 mL per minute per 1.73 m 2 (1.59 mL per second per m 2 )
• CrCl 15 to 50 mL per minute per 1.73 m 2 : 30 mg per day
• Avoid use with rifampin

• 60 mg per day after 5 to 10 days of initial parenteral anticoagulant therapy (patients > 60 kg); 30 mg daily after 5 to 10 days of initial parenteral anticoagulant therapy (patients ≤ 60 kg)
• If taking certain P-glycoprotein inhibitors or CrCl 15 to 50 mL per minute per 1.73 m 2 : 30 mg per day
• Avoid use with rifampin

• 20 mg per day with evening meal
• Avoid with combined P-glycoprotein and moderate CYP3A4 inhibitors and CrCl 15 to 80 mL per minute per 1.73 m 2 (0.25 to 1.34 mL per second per m 2 )
• CrCl 15 to 50 mL per minute per 1.73 m 2 : 15 mg per day (patients with a CrCl < 30 mL per minute per 1.73 m 2 were not studied in trials)
• Avoid with combined P-glycoprotein and strong CYP3A4 inhibitors or inducers

• Total hip or knee replacement surgery: 10 mg daily for 12 days after knee replacement surgery or 35 days after hip replacement surgery
• Observe for blood loss in CrCl 30 to 50 mL per minute per 1.73 m 2
• Avoid use in CrCl < 30 mL per minute per 1.73 m 2
• Avoid with combined P-glycoprotein inhibitors and moderate CYP3A4 inhibitors and CrCl 15 to 80 mL per minute per 1.73 m 2
• Avoid with combined P-glycoprotein and strong CYP3A4 inhibitors or inducers

• 10 mg per day after six months or longer of standard anticoagulant therapy
• Avoid with combined P-glycoprotein inhibitors and moderate CYP3A4 inhibitors and CrCl 15 to 80 mL per minute per 1.73 m 2
• Avoid use in CrCl < 30 mL per minute per 1.73 m 2
• Avoid with combined P-glycoprotein and strong CYP3A4 inhibitors or inducers

• 15 mg twice per day with food for the first 21 days, then 20 mg per day with food for six months
• Avoid with combined P-glycoprotein inhibitors and moderate CYP3A4 inhibitors and CrCl 15 to 80 mL per minute per 1.73 m 2
• Avoid use in CrCl < 30 mL per minute per 1.73 m 2
• Avoid with combined P-glycoprotein and strong CYP3A4 inhibitors or inducers

Direct oral anticoagulants are first-line agents for eligible patients for the treatment of VTE and prevention of stroke in patients with nonvalvular atrial fibrillation. Compared with vitamin K antagonists, direct oral anticoagulants have the advantage of not requiring direct monitoring, having minimal drug-food interactions, and having a quicker onset of action to therapeutic effect. Direct oral anticoagulants have fewer overall drug-drug interactions (Table 6 5 , 9 – 13 ); a comparable (if not lower) bleeding rate; a shorter half-life; and fixed dosing based on indication, drug interactions, and renal or hepatic function. Among the direct oral anticoagulants, there are key differences including the need for parenteral anticoagulation lead-in, once or twice per day dosing, and degree of renal excretion. Of the direct oral anticoagulants, dabigatran and edoxaban (Savaysa) have the highest renal elimination (approximately 80% and 50%, respectively) and should be used with caution in patients with renal impairment. 14 Each of the direct oral anticoagulants has a risk of bleeding. Compared with vitamin K antagonists and with other direct oral anticoagulants, apixaban (Eliquis) has less major bleeding. In addition, rivaroxaban (Xarelto) and dabigatran have a higher risk of gastrointestinal bleeding compared with apixaban. 15 Real world experience with direct oral anticoagulants in terms of safety and effectiveness seems consistent with published trials. 15 , 16

Stroke Prevention in Atrial Fibrillation

In patients who have had acute ischemic stroke, the prevalence of comorbid atrial fibrillation is increasing, and atrial fibrillation–associated strokes have a higher mortality rate. 17 One study found that treatment decisions are often not guideline adherent. 18 The CHADS₂ (congestive heart failure; hypertension; age 75 years or older; diabetes mellitus; prior stroke, transient ischemic attack, or thromboembolism [doubled]) tool (https://www.mdcalc.com/chads2-score-atrial-fibrillation-stroke-risk) or CHA₂DS₂-VASc (congestive heart failure; hypertension; age 75 years or older [doubled]; diabetes mellitus; prior stroke, transient ischemic attack, or thromboembolism [doubled]; vascular disease; age 65 to 74 years; sex category) tool (https://www.mdcalc.com/cha2ds2-vasc-score-atrial-fibrillation-stroke-risk) can be used to estimate the risk of stroke. American Academy of Family Physicians (AAFP) guidelines recommend the use of oral anticoagulants with a CHADS₂ score higher than 1. 19 These guidelines also allow for anticoagulation in patients with a CHADS₂ score of 1 in certain circumstances. 19 The recently published American Heart Association (AHA) /American College of Cardiology (ACC)/Heart Rhythm Society (HRS) guidelines recommend a direct oral anticoagulant over vitamin K antagonists, unless the patient has moderate-to-severe mitral stenosis or a mechanical heart valve. 20 Anticoagulation is recommended for male patients with a CHA₂DS₂-VASc score of 2 or higher and female patients with a CHA₂DS₂-VASc score of 3 or higher. AHA/ACC/HRS guidelines, 20 and the recently published ACCP guidelines attempt to further specify recommendations for low-risk patients as defined by low CHA₂DS₂-VASc scores. 21

According to these recommendations, a CHA₂DS₂-VASc score as low as 1 for men and 2 for women warrants consideration for anticoagulation therapy. 21 Guidelines also recommend that if a therapeutic INR range of 2 to 3 cannot be attained more than 70% of the time, then consideration should be given to changing the treatment to a direct oral anticoagulant. Direct oral anticoagulants or vitamin K antagonists can also be used for the periods before and after cardioversion. Compared with vitamin K antagonists, direct oral anticoagulants are associated with a reduction in the incidence of stroke of 21% to 35% and a reduction in the incidence of intracranial hemorrhage of 33% to 60%. 22 – 25

One comparative effectiveness analysis looked at the treatment of patients with atrial fibrillation who may not have been well-represented in clinical trials because of multiple comorbidities. 26 This study used Medicare data to compare vitamin K antagonists with dabigatran and rivaroxaban in patients with atrial fibrillation and multiple chronic conditions. The study found that effectiveness was similar for all oral anticoagulants and risk of major hemorrhage was reduced among dabigatran users when compared with rivaroxaban (but not vitamin K antagonists) in patients with a moderate-to-high burden of chronic conditions. 26 Rates of gastrointestinal hemorrhage were highest in the rivaroxaban group. Finally, in this vulnerable population, the relative hazard of death rates was generally lower for rivaroxaban and dabigatran compared with vitamin K antagonists. 26

Bleeding Risk Assessment

Multiple risk assessment scoring systems have been developed, with inconsistent success in predicting major bleeding event incidence. 27 The ACCP guidelines recommend assessing bleeding risk for patients with VTE or atrial fibrillation as an essential step to guiding treatment decisions such as the duration of treatment. ACCP risk factors for VTE (e.g., advanced age, cancer, renal or hepatic failure) and an associated scoring system to categorize low (no risk factors), moderate (one risk factor), and high (two or more risk factors) risk should be used to determine treatment decisions. 1

The ACCP and AAFP recommend using the HAS-BLED (hypertension, abnormal renal function and liver function, stroke, bleeding, labile INR, elderly [older than 65 years], drugs and alcohol) scoring tool (https://www.mdcalc.com/has-bled-score-major-bleeding-risk) to assess risk of bleeding for patients with atrial fibrillation. 19 , 21 Because of the overlap in risk of ischemic stroke and bleeding, patients with the highest risk of ischemic stroke will commonly also have high bleeding risk. Bleeding risk in patients at high risk for ischemic stroke should rarely be used as a reason to withhold anticoagulation for patients with atrial fibrillation. 21 Risk should be evaluated at each visit and modifiable risk factors, such as alcohol consumption, anemia, anticoagulation control, and use of medications that increase risk of bleeding such as aspirin and nonsteroidal anti-inflammatory drugs, should be addressed. 21

Management of Bleeding

The ACC published an expert consensus decision pathway in 2017 on the management of bleeding for patients taking oral anticoagulants. 28 Management of bleeding for patients taking vitamin K antagonists depends on the severity of the bleed. For major bleeding (defined as all bleeds associated with hemodynamic compromise, occurring in an anatomically critical site, associated with a decrease of hemoglobin of 2 g per dL [20 g per L] or greater, or requiring transfusion of two units of packed red blood cells or more), the patient should receive supportive care, 4-factor prothrombin complex concentrate, and intravenous vitamin K. The only available 4-factor prothrombin complex concentrate is Kcentra, which is dosed based on INR and body weight. Kcentra contains the vitamin K dependent clotting factors (II, VII, IX, and X) as well as proteins C and S. The patient should also be given vitamin K intravenously to maintain vitamin K-dependent clotting factor levels once the effects of Kcentra have diminished. Kcentra is preferred over a fresh frozen plasma infusion because of its smaller volume, faster infusion rate, and superior effectiveness in INR reduction. For nonmajor bleeding (defined as any bleeding not classified as major) that does not require hospitalization or supportive care, holding vitamin K antagonists and possible administration of vitamin K is recommended depending on INR.

The management of nonmajor bleeding with direct oral anticoagulant therapy typically involves holding the anticoagulant and implementing control measures. Direct oral anticoagulants have half-lives of approximately 12 hours; therefore, holding a dose results in a relatively fast decline in anticoagulant effect. Idarucizumab (Praxbind) is a monoclonal antibody fragment that binds directly to dabigatran, leading to 88% to 98% of patients having concentrations of unbound dabigatran in safe levels within 15 minutes of administration and hemostasis restoration at a median of 11.4 hours. 29 For patients taking dabigatran, idarucizumab is recommended for life-threatening or ongoing bleeding, as well as the need to reverse for emergent procedures.

Andexanet alfa (Andexxa) is a genetically modified variant of factor Xa that binds and sequesters factor Xa inhibitors. Andexanet alfa has been approved to reverse the anticoagulant effects of rivaroxaban and apixaban in patients with life-threatening or uncontrolled bleeding. The ANNEXA-4 study showed that 79% of patients with acute major bleeding treated with andexanet alfa achieved effective hemostasis. 30 Key selected exclusion criteria from this study included impending surgery, a major thrombotic event within two weeks before enrollment, intracranial hemorrhage with a Glasgow Coma Scale score of less than 7, or receipt of one of the following agents within seven days before screening: vitamin K antagonist, dabigatran, 4-factor prothrombin complex concentrate, blood, or plasma. Of note, 18% of patients had a thrombotic event during 30 days of follow-up, highlighting the potential prothrombotic risk that this agent carries. Optimal dose, duration, and the need for repeat dosing and mitigation of thromboembolic risk have yet to be delineated. Cost can limit availability and use of andexanet alfa.

VTE Treatment and Cancer

VTE treatment in patients with active cancer is challenging because of the increased risk of VTE recurrence and bleeding related to therapy. 31 – 33 Guidelines have recommended LMWH as the anticoagulant of choice for patients with cancer and VTE. 1 Evidence is emerging for the increased use of direct oral anticoagulants for certain patients with cancer. The 2019 National Comprehensive Cancer Network guidelines on cancer-associated VTE include rivaroxaban as a monotherapy option. 34 Apixaban is included as an acceptable alternative for patients who refuse LMWH or who have compelling reasons to avoid LMWH. 35

The SELECT-D trial evaluated dalteparin with rivaroxaban in patients with active cancer. 2 The study found an absolute VTE recurrence reduction of 7% at six months in favor of rivaroxaban (11% vs. 4%; hazard ratio [HR] = 0.43; 95% CI, 0.19 to 0.99). 2 The major bleeding rate at six months was 4% for dalteparin and 6% for rivaroxaban (HR = 1.83; 95% CI, 0.68 to 4.96). 2 Most of the major bleeding was gastrointestinal, primarily in patients with esophageal or gastroesophageal cancer. Clinically relevant nonmajor bleeding was higher in the rivaroxaban group (13% vs. 4%; HR = 3.76; 95% CI, 1.63 to 8.69). The Hokusai VTE cancer trial evaluated dalteparin with edoxaban in patients with active cancer. 3 The primary outcome (recurrent VTE and/or major bleeding) did not differ between treatment groups (P = .006 for noninferiority). 3 There was a decrease in recurrent deep venous thrombosis in favor of the edoxaban group (3.6% vs. 6.7%; HR = 0.56; CI, 0.32 to 0.97) but an increase in major bleeding in that group (6.9% vs. 4.0%; HR = 1.77; 95% CI, 1.03 to 3.04). 3 Gastrointestinal malignancy was also found to be a risk factor of increased gastrointestinal bleeding when using a direct oral anticoagulant vs. LMWH. 3 Therefore, direct oral anticoagulants should be used with caution in patients with cancer who have a history of gastrointestinal malignancy or bleeding.

This article updates previous articles on this topic by Wigle, et al., 36 and du Breuil and Umland. 37

Data Sources: A PubMed search was completed in Clinical Queries using the key terms outpatient, anticoagulation, warfarin, dabigatran, rivaroxaban, apixaban, edoxaban, betrixaban, heparin, low-molecular-weight heparin, dalteparin, enoxaparin, patient self-monitor, and INR. The search included meta-analyses, randomized controlled trials, clinical trials, clinical guidelines, and reviews. Also searched were the National Guideline Clearinghouse, Essential Evidence Plus, UpToDate, the Cochrane database, and the Agency for Healthcare Research and Quality Evidence Reports. Search dates: June 28, 2018, and May 2, 2019.