Your browser is unsupported

We recommend using the latest version of IE11, Edge, Chrome, Firefox or Safari.

In this section

April 2015 FAQs


What drugs should be avoided in myasthenia gravis?

Is normal saline or heparin preferred as a catheter flush solution?

Should warfarin be resumed after gastrointestinal bleeding, and if so – when?


What drugs should be avoided in myasthenia gravis?


Myasthenia gravis is a rare autoimmune disease occurring in 1 to 2 people per 100,000 annually with a prevalence of 20 people per 100,000.1,2 Before 40 years of age, women are 3 times more likely to be affected by myasthenia gravis, whereas after age 50, males are predominately affected. 3 Overall, the prevalence has increased in the United States due to an increased occurrence in elderly patients and improved diagnostic strategies.

Normally, muscle contraction depends on the binding of acetylcholine released from motor nerve terminals to postsynaptic receptors on the muscle end-plate region.1,2 Muscle depolarization is terminated by acetylcholinesterase in the postsynaptic muscle membrane, which hydrolyzes the acetylcholine. However, in individuals with myasthenia gravis, acetylcholine receptor (AChR) antibodies bind to the AChR, cause internalization and degradation of AChR, block the binding of acetylcholine to AChR, which prevents muscles from contracting. Patients with seronegative myasthenia gravis do not have detectable AChR antibodies and may have anti-muscle-specific tyrosine kinase (MuSK) antibodies. Seronegative myasthenia gravis is typically more severe and requires immunosuppression, because cholinesterase inhibitors do not work.

In two-thirds of patients with myasthenia gravis, the presenting symptom is ptosis or diplopia.3Approximately 17% of patients present with chewing and swallowing problems, and 10% present with weakness of limbs. The symptoms typically become worse throughout the day. The disease may be limited to the external ocular muscles (a less severe form of the disease) or may be more generalized, involving muscles of the face, oropharyngeal areas, upper torso, and proximal extremities.4,5 Respiratory paralysis can also occur in very severe exacerbations. Although the disease is progressive, patients experience intermittent periods of very active disease and remission. Flares are triggered by surgery, infections, and certain medications. 6

Medications that worsen myasthenia gravis

Several medications are implicated in either inducing or worsening myasthenia gravis.4,7 Four mechanisms have been described to explain the interaction of these drugs and the disease: (1) neuronal transmission may be inhibited at the presynaptic terminal; (2) lack of acetylcholine release (possibly related to inhibition of calcium influx into the presynaptic terminal); (3) blockade of the postsynaptic AChRs, thereby preventing the binding of acetylcholine to the postsynaptic AChR; and (4) prevention of action potential transmission past the postsynaptic terminal due to changes in postsynaptic ion permeability. Another proposed mechanism is that the pyrimidine or pyridine moiety of certain drugs, such as voriconazole, interacts with AChR. 8

Many medications have been reported to have an effect on neuromuscular transmission. Because evidence of exacerbations or of first presentations of myasthenia gravis have mainly been published in case reports, it is difficult to determine a true incidence with each agent. In addition, questionable temporal relationships or other confounding factors sometimes make interpretation of the case reports difficult. Nonetheless, medications that have been implicated in myasthenia gravis (Table 1) should be used cautiously in this population. Additional details on some of these data are provided under the table.

Table 1. Medications that reportedly exacerbate myasthenia gravis.a,9


Aminoglycosides, ampicillin, azithromycin, clindamycin, colistimethate sodium, erythromycin, fluoroquinolones, imipenem, peramavir, telithromycin, tetracyclines, vancomycin, voriconazole

Anesthetics/anesthesia agents

Bupivacaine, cocaine, desflurane, diazepam, isoflurane, ketamine, lidocaine, prilocaine, procaine, procainamide, propofol, sevoflurane


Lidocaine, procainamide, quinidine


Cetirizine, ipratropium


Carbamazepine, gabapentin, phenytoin


Beta blockers, calcium channel blockers


Chloroquine, hydroxychloroquine


Chlorpromazine, haloperidol, lithium, phenothiazines




Alendronate, pamidronate, risedronate

Contrast agents

Diatrizoate meglumine, iothalamate


Methylprednisolone, prednisone


Codeine, hydromorphone, meperidine, morphine

Neuromuscular blockers

Cisatracurium, d-tubocurarine, pancuronium, rocuronium, succinylcholine, vecuronium

Ophthalmologic agents

Betaxolol, proparacaine, timolol, tropicamide


Atorvastatin, pravastatin, rosuvastatin, simvastatin

Miscellaneous agents

Carnitine, imiquimod, interferon alfa, levonorgestrel, magnesium sulfate, methocarbamol

a Not an all-inclusive list.

If myasthenia gravis symptoms such as muscle weakness or ocular symptoms occur, discontinuation of the causative medication is appropriate.4 Anticholinesterase drugs (eg, neostigmine methylsulfate, atropine, pyridostigmine) may be warranted, especially if the patient experiences respiratory distress. These agents may be used in combination with the offending medication if it cannot be discontinued abruptly (eg, antiepileptics). Ventilatory support may be temporarily required during respiratory failure in a myasthenic crisis. Summarized below in Table 2 are various medications that have been associated with exacerbations of myasthenia gravis and the temporal relationship of symptoms and the starting and discontinuing of the offending medication.

Table 2. Temporal relationship of drug administration, exacerbation, and resolution of myasthenia gravis exacerbation.a,7,10-16


Onset (from initiation)



12 hours to several days

48 hours


During infusion

10 to 90 minutes

Beta blockers (timolol, acebutolol, propranolol)

24 hours to several days

24 hours to 8 weeks

Bisphosphonates (alendronate, risedronate, pamidronate)

Same day of dose

1 day

Botulinum toxin


2 weeks


1 week to 3.5 years

5 days to 14 weeks


4 hours to 2 days

Several days


During infusion; 30 minutes after dose

2 to 3 days


2 days

48 hours

Interferon alfa

6 to 9 months

7 months

Iodinated contrast media (iothalamic acid, diatrizoate meglumine)


2 to 3 days in younger patients

4 to 7 weeks in older patients


3 weeks

resolved after edrophonium


10 days to 3 months

3 to 4 days


8 months

6 to 10 months


4 to 6 years

2 to 3 months


1 to 2 weeks

1 to 20 days


2 days to 2 weeks

12 hours to 10 days


72 hours

48 hours

Statins (atorvastatin, rosuvastatin, simvastatin, pravastatin)

8 weeks

2 to 7 months


15 minutes to 1.5 hours

<24 hours


4 days

2 weeks

a Not an all-inclusive list.


Antimicrobial agents may interact with voltage-gated calcium channels presynaptically or with AChR postsynaptically.6

Aminoglycosides are cited in numerous case reports involving their concomitant use with neuromuscular blockers.4,7,10 Postoperative respiratory depression was reported in nearly all cases. Limb or facial weakness has also been reported. Aminoglycosides have also exacerbated preexisting myasthenia gravis, and have led to worsening symptoms within 1 hour of administration.

Fluoroquinolones have consistently been associated with flares of myasthenia gravis. Recently, the US Food and Drug Administration Adverse Event Reporting System was queried for reports of myasthenia gravis exacerbations occurring in patients taking fluoroquinolones.17 Out of 27 reports, and an additional 10 reports found in the literature, 2 patients died and 11 patients required mechanical ventilation.

Telithromycin-induced myasthenia gravis is well-reported in medical literature.18,19

Antivirals such as voriconazole8 and peramavir20 have been recently reported to exacerbate myasthenia gravis.


Bisphosphonates (specifically alendronate, risedronate, and pamidronate) have been implicated in exacerbation of myasthenia gravis.16 In a case report with alendronate, the patient was switched to intravenous ibandronate, and no further exacerbations occurred on this therapy.

Cholinesterase inhibitor overdose

Overdoses of cholinesterase inhibitors may also exacerbate myasthenia gravis.7 Excessive doses can result in acetylcholine accumulation, which causes increased bronchial secretions leading to difficulty swallowing or breathing. Weakness 1 hour after administration of pyridostigmine could indicate overdose, while weakness 3 or more hours following a dose could indicate a suboptimal response to therapy.


Glucocorticoids, although a mainstay in the management of moderate to severe myasthenia gravis, can also cause muscle weakness.4,5,7,21,22 Patients are generally started on high doses of prednisone (60 to 100 mg/day) until the disease is in remission, then the dose is tapered to the lowest possible daily dose, and eventually switched to an every other day regimen. Approximately 25% to 75% of patients initiated on high-dose prednisone have an exacerbation of their disease in the first days to weeks of therapy, which is then followed by a period of remission.23 In one study, independent predictors of exacerbation caused by steroids included older age, bulbar symptoms, and severe neurologic presentation, especially in the initial phase of treatment. Proposed mechanisms include release of antibodies from degraded lymphocytes, increased cholinesterase activity in the neuromuscular junction, and increased immune-related reactions.

Neuromuscular blocking agents

The dose of neuromuscular blockers is typically reduced in individuals with myasthenia gravis, because this condition prolongs the duration of neuromuscular blockade, making them extremely sensitive to neuromuscular blocking agents. Exacerbations may occur when the dose of the neuromuscular blocking agent is too high, or unmasking of myasthenia gravis may happen in patients who have not yet been diagnosed.24,25


Approximately 1% to 7% of patients on penicillamine develop myasthenia gravis.1,3Penicillamine induces the formation of AChR antibodies in 90% of patients who develop myasthenia gravis while on this agent. While penicillamine is very well-documented to be a cause of myasthenia gravis, there are no reports of exacerbation in a patient already diagnosed with myasthenia gravis. Patients who develop myasthenia gravis while receiving penicillamine typically have a mild form of the disease, often limited to the extraocular muscles. Initial presentation of the disease varies from 2 to 12 months after therapy has begun. Most patients have resolution of the disease within 2 to 6 months following discontinuation.


Although statins are known to cause myotoxicities, myasthenia gravis exacerbations have not been well-reported in the literature.14,15,26,27 In several case reports, patients taking statins developed myasthenia-like symptoms; in many of these cases, AChR antibodies were present. There was variability in the timing of the presentation and resolution of the symptoms relative to statin therapy. Some authors suggest these symptoms could be due to several potential mechanisms, including underlying myasthenia gravis aggravated by the muscle toxicity of statins or antibody-mediated myasthenia gravis induced by statins. Statins can still be used in patients with myasthenia gravis with counseling on potential worsening of muscle weakness.26


Many other drugs have been found to worsen myasthenia gravis. The Myasthenia Gravis Foundation of America has a resource document for healthcare professionals that discusses medications that may exacerbate myasthenia gravis (
website_8%2010%2012.pdf).28 The document was last updated in August 2012.

In summary, many drugs have been implicated as a cause of myasthenia gravis or disease exacerbation. Although the literature is limited, caution and close monitoring when prescribing these agents is recommended, especially during an acute exacerbation.


1. Phillips LH. The epidemiology of myasthenia gravis. Semin Neurol. 2004;24(1):17-20.

2. Heldal AT, Owe JF, Gilhus NE, Romi F. Seropositive myasthenia gravis: a nationwide epidemiologic study. Neurology. 2009;73(2):150-151.

3. Meriggioli MN, Sanders DN. Disorders of neuromuscular transmission. In: Daroff RB, Fenichel GM, Jankovic J, Mazziotta JC, eds.Bradley's Neurology in Clinical Practice. 6th ed. Philadelphia, PA: Elsevier; 2012. Accessed March 24, 2015.

4. Barrons RW. Drug-induced neuromuscular blockade and myasthenia gravis.Pharmacotherapy. 1997;17(6):1220-1232.

5. Vincent A, Palace J, Hilton-Jones D. Myasthenia gravis. Lancet. 2001;357(9274):2122-2128.

6. Bhattacharyya S, Darby R, Berkowitz AL. Antibiotic-induced neurotoxicity. Curr Infect Dis Rep. 2014;16(12):448.

7. Wittbrodt ET. Drugs and myasthenia gravis - An update. Arch Intern Med. 1997;157(4):399-408.

8. Azzam R, Shaikh AG, Serra A, Katirji B. Exacerbation of myasthenia gravis with voriconazole. Muscle Nerve. 2013;47(6):928-930.

9. Aronson JK, ed Meyler's Side Effects of Drugs: The International Encyclopedia of Adverse Drug Reactions and Interactions. 15th ed. Amsterdam, The Netherlands: Elsevier Science; 2006.

10. Karcic AA. Drugs that can worsen myasthenia gravis. Postgrad Med. 2000;108(2):25.

11. Jones SC, Sorbello A, Boucher RM. Fluoroquinolone-associated myasthenia gravis exacerbation: evaluation of postmarketing reports from the US FDA adverse event reporting system and a literature review. Drug Saf. 2011;34(10):839-847.

12. Jennett AM, Bali D, Jasti P, Shah B, Browning LA. Telithromycin and myasthenic crisis.Clin Infect Dis. 2006;43(12):1621-1622.

13. Perrot X, Bernard N, Vial C, et al. Myasthenia gravis exacerbation or unmasking associated with telithromycin treatment. Neurology. 2006;67(12):2256-2258.

14. Hayashi K, Iwasa K, Morinaga A, Ono K, Yamada M. Exacerbation of myasthenia gravis by intravenous peramivir [published online ahead of print February 7, 2015]. Muscle Nerve.doi: 10.1002/mus.24594.

15. Kesikburun S, Guzelkucuk U, Alay S, Yavuz F, Tan AK. Exacerbation of myasthenia gravis by alendronate. Osteoporos Int. 2014;25(9):2319-2320.

16. Shanahan EM, Smith MD, Ahern MJ. Pulse methylprednisolone therapy for arthritis causing muscle weakness. Ann Rheum Dis. 1999;58(9):521-522.

17. Komiyama A, Arai H, Kijima M, Hirayama K. Extraocular muscle responses to high dose intravenous methylprednisolone in myasthenia gravis. J Neurol Neurosurg Psychiatry.2000;68(2):214-217.

18. Bae JS, Go SM, Kim BJ. Clinical predictors of steroid-induced exacerbation in myasthenia gravis. J Clin Neurosci. 2006;13(10):1006-1010.

19. Dunsire MF, Clarke SG, Stedmon JJ. Undiagnosed myasthenia gravis unmasked by neuromuscular blockade. Br J Anaesth. 2001;86(5):727-730.

20. Bowie RA. Myasthenia gravis unmasked by neuromuscular blockade. Br J Anaesth.2002;88(1):153-154.

21. Purvin V, Kawasaki A, Smith KH, Kesler A. Statin-associated myasthenia gravis: report of 4 cases and review of the literature. Medicine (Baltimore). 2006;85(2):82-85.

22. Gilhus NE. Is it safe to use statins in patients with myasthenia gravis? Nat Clin Pract Neurol. 2009;5(1):8-9.

23. Gale J, Danesh-Meyer HV. Statins can induce myasthenia gravis. J Clin Neurosci.2014;21(2):195-197.

24. Oh SJ, Dhall R, Young A, Morgan MB, Lu L, Claussen GC. Statins may aggravate myasthenia gravis. Muscle Nerve. 2008;38(3):1101-1107.

25. Mehrizi M, Fontem RF, Gearhart TR, Rascuzzi RM. Medications and myasthenia gravis: a reference for health care professionals. 2012.
website_8%2010%2012.pdf. Accessed March 25, 2015.

26. Yarom N, Barnea E, Nissan J, Gorsky M. Dental management of patients with myasthenia gravis: a literature review. Oral Surg Oral Med Oral Pathol Oral Radiol Endod.2005;100(2):158-163.

27. Kuczkowski KM. Labor analgesia for the parturient with neurological disease: what does an obstetrician need to know? Arch Gynecol Obstet. 2006;274(1):41-46.

28. Bedlack RS, Sanders DB. How to handle myasthenic crisis. Essential steps in patient care. Postgrad Med. 2000;107(4):211-214, 220-212.

Return to Top

Is normal saline or heparin preferred as a catheter flush solution?

Nearly all hospitalized patients require peripheral or central catheterization for administration of drugs and sampling of blood.1 A significant problem, however, is the potential for clot formation within the catheter, which may result in loss of patency.2 This may hinder medication administration and blood sampling and potentially necessitate catheter replacement. To maintain patency and prevent contact between incompatible medications, catheter lines are recommended to be flushed intermittently and before and after each administration of medication.3 Because of its anticoagulant activity, heparin is commonly incorporated in solutions used to flush and fill catheters.2

Risks of incorporating heparin in catheter flush solutions include the potential for dosing errors and rare adverse events, such as heparin-induced thrombocytopenia (HIT).3 To prevent these adverse outcomes, normal saline has been used as an alternative. Comparisons between normal saline and heparin catheter flush solutions have been of limited methodological quality and have provided some conflicting findings.1,4,5 Furthermore, recommendations and practice patterns vary.6,7 This document is a review of literature on the use of heparin and normal saline as flush solutions in various catheter types and patient-specific situations.

Peripheral venous catheters

A 1998 meta-analysis by Randolph and colleagues compared heparin and normal saline in flushes of peripheral catheters.1 Meta-analysis of results for peripheral venous catheters showed intermittent flushing with heparin 10 units/mL every 8 hours did not reduce peripheral venous catheter clotting (relative risk [RR], 1.08; 95% confidence interval [CI], 0.55 to 2.1). In contrast, higher heparin concentrations (100 units/mL every 6 to 8 hours) significantly reduced clotting (RR, 0.52; 95% CI, 0.33 to 0.83). Continuous infusions of heparin 1 unit/mL significantly decreased the risk of phlebitis (RR, 0.55; 95% CI, 0.39 to 0.77), but not infusion failure rate (RR, 0.88; 95% CI, 0.72 to 1.07). Similarly, subsequent randomized controlled trials (RCTs) found benefit only in trials of 100-unit/mL heparin concentrations in post-orthopedic surgery and medical patients when compared to normal saline or no flushing.8,9The findings of benefit with higher heparin concentrations require further evaluation, as studies showing benefit were in limited populations and may not be widely generalizable.

A position statement published by the American Society of Health-Systems Pharmacists (ASHP) in 2012 addressed the practice of maintaining patency of peripheral indwelling intermittent infusion devices (ie, “locks”).2 The statement advocated the use of normal saline in preference to heparin-containing solutions in institutional settings in adults and children age 1 year and older. The recommendation was supported by several RCTs which found the solutions equally effective.10-12 Recommendations in the statement, however, do not apply to central venous catheters (CVCs) or arterial access devices.

Central venous catheters

The 2011 Infusion Nursing Society (INS) Standards of Practice state there is no established optimal flushing solution for central vascular access devices (CVADs).13 The INS recommends that preservative-free normal saline be used to flush CVADs, though non-valved, intermittently used CVADs can be locked with heparin solutions. These recommendations are generally supported by a recent review and meta-analysis that found no consistent benefit with either flush solution in CVCs.5,14 Most recently, a 2014 Cochrane meta-analysis found that although heparin was associated with reduced CVC occlusion rates in trials reporting outcomes at the catheter level (RR, 0.53; 95% CI, 0.29 to 0.94), studies reporting at the patient level found no difference (RR, 0.21; 95% CI, 0.03 to 1.7).5 In the meta-analysis, no significant differences existed in safety outcomes of mortality, hemorrhage, sepsis, or thrombosis.

Data from a recent survey of critical care nurses showed a variety of practices in flushing CVCs, but illustrate that normal saline may be preferred. 6 The 2012 survey found a majority of respondents (64.6%) used saline flushes exclusively, at a volume of 10 mL (63%) and a frequency of every 8 to 12 hours and after each use (72.5%). The most commonly reported heparin concentrations were 100 and 10 units/mL (37.5% and 29.7%, respectively). The authors proposed the use of normal saline could lead to cost savings and reduced adverse events.

Arterial catheters

Arterial catheters are often needed in critically ill patients for hemodynamic monitoring.1 The 2011 INS Standards of Practice state that existing studies are inconclusive to guide selection of an optimal arterial catheter flush solution.3 A recent review found evidence that suggested the use of heparin may be more beneficial if catheterization is maintained beyond 48 hours.15The Cochrane collaboration intended to perform a meta-analysis on trials comparing normal saline and heparin for maintaining arterial catheter patency, but could not do so because of the high degree of heterogeneity and bias in identified trials.4 While the meta-analysis by Randolph found the use of heparin in peripheral arterial catheters significantly reduced the risk of clot formation (RR, 0.51; 95% CI, 0.42 to 0.61), these results were mostly driven by a single trial. 1 Given limitations in the available literature, the INS recommends selecting an arterial flush solution based on the risk of catheter occlusion, the time the arterial catheter is required, and the risk for heparin sensitivity, bearing in mind that arterial catheters should not be routinely replaced.3

Special populations and considerations

The risk for catheter occlusion is hypothesized to be greater in specific patient groups. Groups in which the use of normal saline and heparin catheter flushes has been evaluated include pediatric and pregnant patients. The risk for thrombotic catheter occlusion is thought to be higher in children because of the use of smaller diameter catheters, the comparatively larger volume of blood vessels occupied by catheters, and lower levels of antithrombotic proteins in neonates.16 Despite this increased risk, data are limited. A 2010 Cochrane review found that limited data in neonatal patients mostly indicate no significant difference in the maintenance of peripheral intravenous catheter patency between normal saline and heparin, but study limitations precluded pooling results for the intended meta-analysis.17

Recommendations from nursing societies include the use of heparin in some pediatric situations, however. The National Association of Neonatal Nurses state that heparinized saline solutions should be used to periodically flush peripheral catheters.18 A 2011 position paper from the INS on the prevention of CVC occlusion in neonatal and pediatric patients recommended flushing CVCs with normal saline before and after accessing the port, and the use of heparin flushes at concentrations from 10 to 100 units/mL depending on the type of CVC.16 Peripherally inserted central catheters can be flushed with 10 units/mL of heparinized saline with dosages of 1 mL every 6 hours for 2-French catheters, or 2 to 3 mL every 12 hours for 2.6-French catheters and above. Nontunneled and tunneled CVCs can be flushed with 2 mL of heparinized saline at concentrations of 10 to 100 units/mL every 24 hours. If sodium chloride is used in neonates, bacteriostatic solutions should be avoided because of the risk for metabolic acidosis associated with the preservative benzyl alcohol.2,19

The risk for catheter occlusion is also believed to be increased during hypercoagulable states like pregnancy and malignancy.20,21 However, results of RCTs in pregnant patients are conflicting and low sample sizes increase the risk for type II error.20,22,23 Similarly, guidelines from the American Society of Clinical Oncology on the care of CVCs in cancer patients state that conflicting and insufficient data preclude recommendations on the routine use of an optimal flush solution.21

Special consideration should be given to the compatibility of normal saline and heparin flushes with medications patients are receiving.2 In situations of incompatibility, the interacting medication may need to be “preflushed” with an alternative solution such as 5% dextrose. Common drugs that are incompatible with normal saline include amphotericin B conventional colloidal and diazepam.24,25 Heparin, in contrast, is known to be incompatible or has variable compatibility with a number of drugs (Table). 24,25

Table. Select drugs with known and variable incompatibility with heparin at Y-site.24,25

Known incompatibility

Variable incompatibility


Amphotericin B cholesteryl (Amphotec)




Daunorubicin hydrochloride




Doxycycline hyclate




Haloperidol lactate











Polymyxin B sulfate





Amphotericin B conventional colloidal



Antithymocyte globulin (rabbit)








Erythromycin lactobionate




Insulin (regular)



Methylprednisolone sodium succinate













Recommendations and practice patterns vary regarding the use of normal saline or heparin as a catheter flush solution.6 Given the lack of compelling data available to favor to the use of either, normal saline may be a cost- and complication-sparing option. Recommendations of the catheter manufacturer should be followed when available, institutional policies should be developed, and the potential for drug interactions with the flush solution should be considered.3 When heparin flush is utilized, the dose should be minimized and platelet monitoring should be considered to reduce the risk for HIT.


1. Randolph AG, Cook DJ, Gonzales CA, Andrew M. Benefit of heparin in peripheral venous and arterial catheters: systematic review and meta-analysis of randomised controlled trials.BMJ. 1998;316(7136):969-975.

2. Benner K, Lucas AJ. ASHP therapeutic position statement on the institutional use of 0.9% sodium chloride injection to maintain patency of peripheral indwelling intermittent infusion devices. Am J Health Syst Pharm. 2012;69(14):1252-1254.

3. Infusion Nurses Society. Infusion Nursing Standards of Practice. J Infus Nurs.2011;34(suppl 1):S1-110.

4. Robertson-Malt S, Malt GN, Farquhar V, Greer W. Heparin versus normal saline for patency of arterial lines. Cochrane Database Syst Rev. 2014;5:CD007364.

5. Lopez-Briz E, Ruiz Garcia V, Cabello JB, Bort-Marti S, Carbonell Sanchis R, Burls A. Heparin versus 0.9% sodium chloride intermittent flushing for prevention of occlusion in central venous catheters in adults. Cochrane Database Syst Rev. 2014;10:CD008462.

6. Sona C, Prentice D, Schallom L. National survey of central venous catheter flushing in the intensive care unit. Crit Care Nurse. 2012;32(1):e12-e19.

7. Mitchell MD, Anderson BJ, Williams K, Umscheid CA. Heparin flushing and other interventions to maintain patency of central venous catheters: a systematic review. J Adv Nurs. 2009;65(10):2007-2021.

8. Myrianthefs P, Sifaki M, Samara I, Baltopoulos G. The epidemiology of peripheral vein complications: evaluation of the efficiency of differing methods for the maintenance of catheter patency and thrombophlebitis prevention. J Eval Clin Pract. 2005;11(1):85-89.

9. Bertolino G, Pitassi A, Tinelli C, et al. Intermittent flushing with heparin versus saline for maintenance of peripheral intravenous catheters in a medical department: a pragmatic cluster-randomized controlled study. Worldviews Evid Based Nurs. 2012;9(4):221-226.

10. Epperson EL. Efficacy of 0.9% sodium chloride injection with and without heparin for maintaining indwelling intermittent injection sites. Clin Pharm. 1984;3(6):626-629.

11. Garrelts JC, LaRocca J, Ast D, Smith DF, Jr., Sweet DE. Comparison of heparin and 0.9% sodium chloride injection in the maintenance of indwelling intermittent i.v. devices. Clin Pharm. 1989;8(1):34-39.

12. Hamilton RA, Plis JM, Clay C, Sylvan L. Heparin sodium versus 0.9% sodium chloride injection for maintaining patency of indwelling intermittent infusion devices. Clin Pharm.1988;7(6):439-443.

13. Gorski L, Perucca R, Hunter MR. Central venous access devices: care, maintenance, and potential complications. In: Alexander M, Corrigan A, Gorski L, Hankins J, Perucca R, eds.Infusion Nursing: An Evidence-Based Approach. 3rd ed. St. Louis, MO: Saunders-Elsevier; 2010.

14. Dal Molin A, Allara E, Montani D, et al. Flushing the central venous catheter: is heparin necessary? J Vasc Access. 2014;15(4):241-248.

15. Kordzadeh A, Austin T, Panayiotopoulos Y. Efficacy of normal saline in the maintenance of the arterial lines in comparison to heparin flush: a comprehensive review of the literature. J Vasc Access. 2014;15(2):123-127.

16. Doellman D. Prevention, assessment, and treatment of central venous catheter occlusions in neonatal and young pediatric patients. J Infus Nurs. 2011;34(4):251-258.

17. Shah PS, Ng E, Sinha AK. Heparin for prolonging peripheral intravenous catheter use in neonates. Cochrane Database Syst Rev. 2005(4):CD002774.

18. Altimier L, Brown B, Tedeschi L. National Association of Neonatal Nurses guidelines for neonatal nursing policies, procedures, competencies, and clinical pathways. 4th ed. Glenview, IL: National Association of Neonatal Nurses; 2006.

19. Brown WJ, Buist NR, Gipson HT, Huston RK, Kennaway NG. Fatal benzyl alcohol poisoning in a neonatal intensive care unit. Lancet. 1982;1(8283):1250.

20. Niesen KM, Harris DY, Parkin LS, Henn LT. The effects of heparin versus normal saline for maintenance of peripheral intravenous locks in pregnant women. J Obstet Gynecol Neonatal Nurs. 2003;32(4):503-508.

21. Schiffer CA, Mangu PB, Wade JC, et al. Central venous catheter care for the patient with cancer: American Society of Clinical Oncology clinical practice guideline. J Clin Oncol.2013;31(10):1357-1370.

22. Meyer BA, Little CJ, Thorp JA, Cohen GR, Yeast JD. Heparin versus normal saline as a peripheral line flush in maintenance of intermittent intravenous lines in obstetric patients.Obstet Gynecol. 1995;85(3):433-436.

23. Arnts IJ, Heijnen JA, Wilbers HT, van der Wilt GJ, Groenewoud JM, Liem KD. Effectiveness of heparin solution versus normal saline in maintaining patency of intravenous locks in neonates: a double blind randomized controlled study. J Adv Nurs. 2011;67(12):2677-2685.

24. Clinical Pharmacology [database online]. Tampa, FL: Gold Standard, Inc; 2015. Accessed January 20, 2015.

25. Micromedex Healthcare Series [database online]. Greenwood Village, CO: Truven Health Analytics; 2015. Accessed January 20, 2015.

Return to Top

Should warfarin be resumed after gastrointestinal bleeding, and if so – when?


The most concerning risk with use of systemic anticoagulants is bleeding events. Reports from randomized controlled trials of anticoagulants in patients with atrial fibrillation (AF) estimate an incidence of gastrointestinal bleeding (GIB) of approximately 2.1% annually with warfarin use.1 This statistic may be higher in an actual clinical setting. When GIB occurs in patients on anticoagulation, the American College of Chest Physicians make the recommendation to rapidly reverse anticoagulation, but there is no discussion in regards to if and when to resume anticoagulation.Patients are at increased risk for bleeding while taking warfarin when they are over the age of 75 years; have a history of uncontrolled hypertension; consume excessive amounts of alcohol; have poor compliance with their warfarin regimen; have unstable international normalized ratio (INR) control; have peptic ulcer disease, have other coagulation defects such as thrombocytopenia; and/or are taking concomitant non-steroidal anti-inflammatory drugs or antibiotics. Upon discontinuation or interruption of warfarin after GIB, patients are thought to be at an increased risk for thromboembolism due to a transient hypercoagulable state (and if they have any underlying diseases such as AF), but they are also thought to be at an increased risk for rebleeding if warfarin is resumed.4,5Therefore, the risk of thromboembolism with permanent anticoagulation discontinuation must be carefully weighed against the risk of a recurrent bleed with anticoagulation resumption, especially since both can result in fatalities. Currently there is a lack of consensus on the resumption of long-term anticoagulation after patients experience GIB. The limited data for anticoagulation resumption after GIB are reviewed here.

Review of the literature

Recurrent gastrointestinal bleed versus thromboembolism in patients with atrial fibrillation and warfarin interruption

A recent retrospective cohort study analyzed data from patients with AF who experienced a major GIB while taking warfarin, in order to test their hypothesis that restarting warfarin would result in benefits including prevention of thromboembolism and mortality.6 They also assessed what effect duration of warfarin interruption had on these outcomes, as well as on recurrent GIB. Data were obtained from the Henry Ford Health System in Detroit, Michigan, between 2005 and 2010. Patients with nonvalvular AF and documented GIB with resolution were included. They had to be taking warfarin for at least 1 year with recent prescription fills in order to be included in the analysis. Patient charts were further assessed to identify those had warfarin discontinued for ≥2 days due to GIB. They also had to have 2 years of follow-up data unless mortality was documented. Patients were excluded if they were under hospice care or had an indication for warfarin other than nonvalvular AF. Warfarin restart was defined as an entered prescription/order with documentation of INR ≥2.0. All analyses were adjusted for age, gender, race, number of blood product infusions, INR on admission, and bleeding and co-morbidity scores. Recurrent GIB was analyzed within 90 days of initial bleed since risk decreases after this timeframe, and thromboembolism was analyzed within 1 year of initial GIB.

The mean age of the entire cohort was 75 years and 47% of the patients were women.6Approximately half (49.1%) of the patients had been restarted on warfarin after a median duration of 50 days. Generally, there were more comorbidities in the group who had discontinued warfarin, although this was not statistically significant. Stroke risk and bleeding risk, as measured by CHADS2 and HAS-BLED, respectively, were both a score of 3 in both cohorts who had and had not restarted warfarin. Compared to those in whom warfarin had been discontinued, patients who had been restarted on warfarin had a significantly lower Framingham risk score, were more likely to be male, and less likely to have previous coronary artery disease, a history of falls, abnormal renal function, less likely to use a proton pump inhibitor, and less likely to have presented with GIB in the emergency room. The most common reasons for not restarting warfarin were physician preference and the patient’s inability to comply with follow-up appointments.

Thromboembolism was less likely to occur in patients who had been restarted on warfarin compared to those who had discontinued (adjusted hazard ratio [HR], 0.71; 95% confidence interval [CI], 0.54 to 0.93; p=0.010).6 Recurrent GIB was not significantly different between groups overall (adjusted HR, 1.18; 95% CI, 0.94 to 1.10; p=0.47). However, when results were stratified based on time, GIB was significantly more likely to occur when patients had been restarted within 7 days of interruption as compared to not being restarted (adjusted HR, 3.27; 95% CI, 1.82 to 5.91; p=0.002). Mortality was significantly lower in patients who had been restarted on warfarin, with an adjusted hazard ratio of 0.67 (95% CI, 0.56 to 0.81; p<0.0001). There were 4 deaths due to thromboembolism in patients who had discontinued warfarin, and 1 death due to GIB in a patient who had resumed warfarin. The benefits of decreased thromboembolism and mortality seemed to be lost when restart of warfarin occurred >30 days after interruption. This study showed that resumption of anticoagulation may be associated with benefits of decreased mortality and thromboembolism, without an increase in risk for GIB when restarting warfarin after 1 week and within 1 month of initial GIB.

Recurrent gastrointestinal bleed versus thromboembolism in patients with atrial fibrillation, renal disease and warfarin interruption

Patients with chronic kidney disease (CKD) are at an even greater risk for upper GIB.7 An analysis of patients with renal impairment was performed within the cohort of patients in the previous study performed by Qureshi et al.6 Twelve percent of the 1329 patients had CKD (defined as an estimated glomerular filtration rate of ≤60 mL/min) and 7.2% had end stage renal disease (ESRD).7 Warfarin was restarted in 56% and 35.4% of these patients, respectively. There was a trend toward increased risk of GIB in patients with CKD restarted on warfarin as compared to those with CKD not restarted on warfarin, but it was not significant (HR, 2.50; 95% CI, 0.85 to 7.28; p=0.095). Patients with CKD restarted on warfarin had a significant decrease in thromboembolism as compared to those with CKD not restarted (HR, 0.06; 95% CI, 0.02 to 0.16; p<0.0001), and there was a non-significant trend toward decreased mortality (HR, 0.70; 95% CI, 0.46 to 1.07; p=0.10). In the cohort of patients with ESRD, compared to those with ESRD who had not resumed warfarin, those who did had a significantly higher risk of GIB (HR, 1.72; 95% CI, 1.29 to 2.30; p<0.0001), a significantly lower risk of thromboembolism (HR, 0.44; 95% CI, 0.27 to 0.73; p=0.002), and a significantly lower mortality risk (HR, 0.22; 95% CI, 0.13 to 0.40; p<0.0001). Overall, it was concluded that although thromboembolism and mortality risks may decrease with resumption of warfarin in patients with AF and CKD, there is an increased risk of GIB.

Recurrent gastrointestinal bleed versus thromboembolism in a general population with anticoagulation interruption

Another recent observational study utilized a prospective timeframe to assess the risk of thromboembolism versus GIB in hospitalized patients who experienced anticoagulation interruption.8 One hundred ninety-seven patients with acute GIB on systemic anticoagulation were identified through review of medical records from April 2013 to March 2014 at Beth Israel Deaconess Medical Center in Boston, Massachusetts. Patients were categorized as having resumed anticoagulation if it was resumed within 72 hours of hospital discharge; otherwise, patients were categorized as having interrupted anticoagulation. Patients were followed for 90 days with assessment for use of anticoagulants, emergency department visits or hospital readmissions related to GIB, and thromboembolism. Analyses were adjusted for age, gender, comorbidities, transfusion requirements, and active malignancy. Patients were censored at 90 days or after having thromboembolism, GIB, or death, whichever came first.

The median age of the entire cohort was 75 years and 58% were male.8 Warfarin was the most commonly used anticoagulant (74%), followed by enoxaparin (8%), dabigatran (6%), rivaroxaban (6%), unfractionated heparin (6%), and apixaban (1%). The indications for anticoagulation were AF (58%), history of venous thromboembolism (VTE) (29%), prosthetic valve (9%), portal vein thrombosis (3%), and post-surgical procedure (2%). Stroke risk and bleeding risk, as measured by CHADS2 and HAS-BLED, respectively, were the same in both cohorts. Sixty-one percent of patients were restarted on anticoagulation at hospital discharge. Patients were more likely to have anticoagulation resumed if they had a previous stroke or transient ischemic attack, and if they had a previous GIB. They were more likely to have anticoagulation discontinued if they had an active malignancy.

Follow-up rates were similar between groups after 90 days.Thromboembolism was more common in patients who had discontinued anticoagulation as compared to those who had resumed it (8.0% vs. 0.8%, respectively; p=0.003). Three of the events were stroke. Fourteen percent of patients experienced a recurrent GIB, with a median time of 13 days after discharge. Although there were more GIBs in the cohort who had resumed anticoagulation compared to those who had not, it did not reach statistical significance (HR, 2.17; 95% CI, 0.861 to 6.67; p=0.10. Death occurred in 8% of the cohort, and there was no difference in this outcome between those who had resumed or discontinued anticoagulation. None of the deaths were attributed to GIB or thromboembolism. The investigators concluded that anticoagulation continuation within 72 hours of discharge after initial GIB was associated with reduced risk of thromboembolism within 90 days, without an increased risk for recurrent GIB.

Similarly, a retrospective cohort study using data from the Kaiser Permanente Colorado database found warfarin resumption post-GIB in a general population to be associated with a reduction in thromboembolism.9 Data from 442 patients with a warfarin-associated GIB were analyzed, and warfarin had been resumed in 58.8% of these patients. The median time to resumption was 4 days. Approximately half of the patients included were receiving warfarin for AF, one-fourth were receiving it for treatment or secondary prevention of VTE, and 10% were receiving it for prevention of thromboembolism related to a prosthetic heart valve. The mean age of the patients was 74 years, and half were male. Older patients and those in whom GIB source was unidentified were less likely to resume warfarin. After a 90-day follow-up period, 0.4% of patients who had resumed warfarin had a thromboembolism compared to 5.5% who did not resume warfarin (p<0.001). The hazard ratio was adjusted for propensity scores, comorbidities, age, and sex (adjusted HR, 0.05; 95% CI, 0.01 to 0.58). In patients who resumed warfarin, thromboembolism rates did not differ based on duration of interruption. Numerically more patients who resumed warfarin experienced a recurrent GIB within 90 days compared to those who did not resume warfarin (10% versus 5.5%, p=0.09; adjusted HR, 1.32; 95% CI, 0.50 to 3.57). Rates of GIB increased when patients were restarted on warfarin between 1 and 7 days of the initial GIB. None of the recurrent GIBs resulted in death. Risk for death was also significantly decreased in patients who restarted warfarin (adjusted HR, 0.31; 95% CI, 0.15 to 0.62), and risk of death was lowest when warfarin was resumed between 15 and 90 days after initial GIB. The investigators concluded that resumption of warfarin within 90 days of initial GIB was associated with decreased risk for thrombosis and death.


There are quite a few limitations that have to be considered given the observational design of these studies.6-9 Investigators are limited to the quality, accuracy, and completeness of the data that have already been recorded. Statistical analyses can minimize confounders but not eliminate them, and it is impossible to account for all factors that contribute to these events. Loss to follow-up can overestimate or underestimate effects of resuming anticoagulation after GIB. A major limitation to consider is selection bias. It may be that patients in whom clinicians had decided to not resume anticoagulation had greater comorbidities that increased their risk for thromboembolism, recurrent GIB, or death, regardless of warfarin resumption or not. Detection bias may have been present since patients who did not resume anticoagulation may have had less contact with their clinicians, which could have led to an underestimation of events. Diversity in all of the studies is limited, so it is difficult to extrapolate these data to a general US population. The data for events in patients with CKD are especially limited by the small sample size.7

Information on this topic with use of the novel oral anticoagulants (NOACs) is minimal. It is important that resumption of NOACs after GIBs be assessed since the risk for GIBs is even higher with novel oral anticoagulants.1 A meta-analysis assessed the RE-LY, ROCKET AF, ARISTOTLE, and ENGAGE AF-TIMI 48 studies, and found that collectively, dabigatran, rivaroxaban, apixaban, and edoxaban increased the risk for GIBs by approximately 23% in comparison to warfarin (relative risk, 1.23; 95% CI, 1.03 to 1.46, p=0.01), with an absolute risk increase of 0.3%.


Data from 3 US observational studies have shown that resumption of warfarin within 3 months of initial GIB can reduce incidence of thromboembolism. Risk for recurrent GIB does not seem to be significantly increased when warfarin resumption occurs after 1 week of initial GIB. Some of the data suggest that warfarin resumption may also reduce risk for death. Patients with CKD also receive the benefit of reduced thromboembolism with warfarin resumption, but may be at a greater risk for recurrent GIB. Observational trials on risk and benefit of resumption of NOACs after GIB have yet to be conducted.


1. Loffredo L, Perri L, Violi F. Impact of new oral anticoagulants on gastrointestinal bleeding in atrial fibrillation: A meta-analysis of interventional trials. [published online ahead of print February 7, 2015]. Dig Liver Dis. doi: 10.1016/j.dld.2015.01.159.

2. Holbrook A, Schulman S, Witt DM, et al; American College of Chest Physicians. Evidence-based management of anticoagulant therapy: antithrombotic therapy and prevention of thrombosis, 9th ed: American College of Chest Physicians Evidence-Based Clinical Practice Guidelines. Chest. 2012;141(suppl 2):e152S-e184S.

3. Fitzmaurice DA, Blann AD, Lip GYH. Bleeding risks of antithrombotic therapy. BMJ. 2002;325(7368):828-831.

4. Lee JK, Kang HW, Kim SG, Kim JS, Jung HC. Risks related with withholding and resuming anticoagulation in patients with non-variceal upper gastrointestinal bleeding while on warfarin therapy. Int J Clin Pract. 2012;66(1):64-68.

5. Ananthasubramaniam K, Beattie JN, Rosman HS, Jayam V, Borzak S. How safely and for how long can warfarin therapy be withheld in prosthetic heart valve patients hospitalized with a major hemorrhage? Chest. 2001;119(2):478-484.

6. Qureshi W, Mittal C, Patsias I, et al. Restarting anticoagulation and outcomes after major gastrointestinal bleeding in atrial fibrillation. Am J Cardiol. 2014;113(4):662-668.

7. Khalid F, Qureshi W, Qureshi S, Alirhayim Z, Garikapati K, Patsias I. Impact of restarting warfarin therapy in renal disease anticoagulated patients with gastrointestinal hemorrhage.Ren Fail. 2013;35(9):1228-1235.

8. Sengupta N, Feuerstein JD, Patwardhan VR, et al. The risks of thromboembolism vs. recurrent gastrointestinal bleeding after interruption of systemic anticoagulation in hospitalized inpatients with gastrointestinal bleeding: a prospective study. Am J Gastroenterol. 2015;110(2):328-335.

9. Witt DM, Delate T, Garcia DA, et al. Risk of thromboembolism, recurrent hemorrhage, and death after warfarin therapy interruption for gastrointestinal tract bleeding. Arch Intern Med. 2012;172(19):1484-1491.

Return to Top