Magnesium infusions for atrial fibrillation & torsade

November 1, 2015, 10:21 pm

≫ Next: Treatment of hemodynamically stable new-onset AF in critical illness

≪ Previous: 2015 ACLS Guidelines: What happened to VSE?

Introduction: Perpetual controversy

The use of magnesium for AF has been a controversial topic for decades. Magnesium is a normal electrolyte, so it is cheap and has an excellent safety profile. Ironically, this is also magnesium's Achilles heel, because this has caused the pharmaceutical industry to have no interest in it. This leaves us with relatively sparse clinical evidence.

Evidence for magnesium in atrial fibrillation

(1) Cardioversion

It is debatable whether magnesium alone causes cardioversion. Two meta-analyses published in 2007 reached opposite conclusions (Ho vs. Onalan). Overall Onalan et al. appears more reliable, as these authors were more discerning in the studies that they included (Ho et al. included many smaller studies and counted one study twice)(2). The results from Onalan et al. are shown below.

(2) Adjunctive agent for cardioversion

Even if magnesium alone fails to cause cardioversion, it may still improve the success of subsequent attempts at cardioversion, either electrical or medical. Adjunctive magnesium increases the likelihood of cardioversion due to class III anti-arrhythmic agents (e.g. ibutilide and dofetilide; Ganga 2013). Simultaneously, magnesium also reduces the risk of torsade de pointes, a significant problem with these agents. The combination of potassium and magnesium administration has been shown to improve the success rate of electrical cardioversion (Sultan 2012).

(3) Maintenance of sinus rhythm

Does magnesium help maintain sinus rhythm? Although indirect evidence, the highest quality data comes from studies regarding the prevention of postoperative atrial fibrillation. Magnesium has been shown in several RCTs and meta-analyses to reduce the incidence of postoperative atrial fibrillation by about 50% (Cochrane Review 2013).

(4) Improving rate control

Both the Onalan and Ho meta-analyses concur that magnesium improves rate control (e.g., figure from Onalan below). Although reductions in heart rate are statistically significant, the effect size is modest (e.g. average heart rate ~15 b/m lower than placebo group; Davey 2005, Walker 1996).

(5) Overall efficacy in AF

The likelihood of achieving either rate or rhythm control was higher with magnesium compared to placebo or calcium channel blocker (Onalan 2007):

(6) Safety

Compared to placebo or other treatments, magnesium showed a trend towards a reduced rate of major adverse effects (defined broadly as any event requiring additional intervention, treatment discontinuation, or considered significant by authors).

Magnesium pharmacology: The rationale for a continuous infusion

Less than 1% of the body's magnesium is located in the serum, with far greater amounts located within cells. Unfortunately, serum levels of magnesium may not reflect the intracellular magnesium stores. In particular, the serum magnesium level may be normal or slightly low despite substantial intracellular magnesium depletion. Magnesium deficiency is common in critical illness, where it is associated with mortality and arrhythmia (Tong 2005).

It is difficult to replete intracellular magnesium stores. This is commonly observed when trying to treat hypomagnesemia: a single intravenous dose may transiently increase the serum magnesium level, but the next day the magnesium level often hasn't improved much. There are two reasons for this. First, the total intracellular magnesium deficit may be much larger than the administered dose. Second, cells aren't great at absorbing magnesium, so much of the administered magnesium ends up getting excreted in the urine. For a patient with reasonable renal function, the only way to rapidly replete total body magnesium stores is with a continuous intravenous infusion (1).

Evidence: Magnesium infusions for critically ill patients with AF

There are two studies regarding magnesium infusions in critically ill AF patients.

Moran 1995

This was a prospective RCT comparing the use of magnesium vs. amiodarone among 42 ICU patients with atrial tachyarrhythmias (71% with AF or atrial flutter). Magnesium was provided as a bolus of 0.037 g/kg followed by an infusion of 0.025 g/kg/hr for 24 hours. The infusion was adjusted to target a therapeutic serum magnesium concentration of 1.4-2.0 mM (3.3-4.8 mg/dL). Patients with baseline creatinine > 3.4 mg/dL or urine output <400 ml/day were excluded, unless on continuous renal replacement therapy. For patients with moderate renal insufficiency (Cr 2.3-3.3 mg/dL and/or urine output <40 ml/hour), the infusion rate was decreased by a factor of two.

Magnesium was more effective in conversion to sinus rhythm than amiodarone (figure above). Blood pressure was stable in both groups. Among patients who did not convert to sinus rhythm, heart rate reduction was similar in both groups (an average of ~19 b/m). However, this heart rate analysis excluded three patients in the magnesium group who needed to cross over to the amiodarone group due to uncontrolled ventricular rate.

This study has substantial limitations, most notably inclusion of various atrial tachyarrhythmias (e.g. acute AF, chronic AF, multifocal atrial tachycardia). Nonetheless, it does support the safety and efficacy of magnesium among sick ICU patients (average Apache II score of 22). It is notable that most of the patients in the magnesium group (17/21) responded to this therapy and did not require additional antiarrhythmic agents.

Sleeswijik 2008

This is a retrospective study of 29 critically ill patients with new-onset fast AF treated with a protocol involving magnesium with additional amiodarone if needed. First, patients were treated with a magnesium infusion as previously described by Moran (0.037 g/kg over 15 minutes followed by 0.025 g/kg/hr). After one hour, an amiodarone infusion was added if patients failed to respond (defined as ongoing AF with heart rate >110 b/m).

Following one hour of magnesium infusion, 16/29 patients responded (seven cardioverted to normal sinus rhythm and nine remained in AF with a heart rate <110 b/m). With ongoing magnesium infusion, all of these 16 patients eventually converted to sinus rhythm. Of the remaining 13 magnesium nonresponders, 11 cardioverted in response to amiodarone. The overall 24-hour cardioversion rate was 90%. AF did recur in 24% of patients (2 magnesium responders and 5 nonresponders), but in all cases this was successfully treated and none of the patients were discharged with AF.

There was absolutely no difference in baseline serum magnesium levels between patients who did and didn't respond to magnesium. This illustrates that magnesium may be effective even in patients with a normal baseline magnesium level (similar to torsade de pointes).

This study is limited by its retrospective observational design. Nonetheless, it provides support for the efficacy and safety of magnesium in critically ill patients (average Apache II score of 19). Another limitation of this study is that the magnesium infusion was stopped prematurely in patients who failed to respond to magnesium within one hour, which may have limited the efficacy of the magnesium-amiodarone combination (3).

Cardiac magnesium protocol: Putting evidence into practice

Despite safety and efficacy, magnesium infusions are rarely used for AF. The limitation on using magnesium is primarily logistic: it can be challenging to initiate and monitor a magnesium infusion. Publications on magnesium infusions used complex weight-based doses which are difficult to replicate at the bedside. In order to facilitate the safe use of magnesium infusions, the magnesium protocol below was developed. This is based on Moran and Sleeswijick et al., but it actually utilizes lower magnesium infusion rates to improve the margin of saftey (in our experience, this lower infusion rate remains adequate to achieve the target serum magnesium level).

In renal failure (i.e., GFR<30 ml/min), there is an increased risk of magnesium accumulation. Since these patients excrete less magnesium in their urine, using an infusion may be unnecessary. Instead, it might be safer and easier to replete magnesium with incremental doses while following serum levels.

Even if the patient cardioverts or improves prior to completion of the 24-hour infusion, it may be advisable to continue the full infusion. Magnesium has been shown to prevent AF, so completing a total-body magnesium load might reduce the risk of recurrence.

Safety

Magnesium infusions appear very safe in patients with sufficient renal function (e.g. GFR >30 ml/min) when protocolized and monitored adequately. Similar doses of magnesium have likewise been shown to be safe in critically ill patients with subarachnoid hemorrhage. Magnesium arguably has the best safety profile of any drug used for AF. It may cause minor flushing, tingling, or fatigue. Hypotension or bradycardia are rare complications, less common than with amiodarone or calcium channel blockers (Ho 2007). Unlike most anti-arrhythmic drugs, magnesium has no pro-arrhythmic effects, but instead decreasesthe risk of torsade de pointes (Ganga 2013). Magnesium infusions cause small decreases in serum calcium, of unclear clinical significance (Muroi 2008).

Where does magnesium fit in the greater context of AF treatment?

The data on magnesium is sparse, so exactly where this may fit into a management scheme for AF remains unclear. The following are situations where magnesium may be particularly useful:

Other therapies are contraindicated (e.g. patients in whom hypotension limits the ability to use beta-blockers or calcium-channel blockers). Aside from baseline hypermagnesemia or neuromuscular disease (e.g. myasthenia gravis), there are few contraindications to magnesium.
Adjunctive agent for chemical cardioversion
Patients with hypomagnesemia
Critically ill patients (who may have a higher rate of subclinical magnesium deficiency than other populations)
AF refractory to conventional therapies

Treatment of hemodynamically stable new-onset AF in critical illness

November 9, 2015, 4:20 am

≫ Next: Investigation Bias: The freakonomics of when industry choses to sponsor a clinical trial

≪ Previous: Magnesium infusions for atrial fibrillation & torsade

Introduction with a clinical question

A 50-year-old woman with no prior medical problems was admitted to Genius General Hospital with severe influenza pneumonia and acute kidney injury. She was transferred to the ICU and treated with high-flow nasal cannula oxygen support. Over time she gradually improved with decreasing oxygen requirements and improving renal function. On hospital day two she developed new-onset atrial fibrillation (AF) with a ventricular rate of 130 b/m. She was symptomatic with palpitations but was hemodynamically stable. What is the best therapeutic approach for her?

Defining new-onset AF in critical illness (NAFCI)

NAFCI is used here to refer to patients with no prior history of AF who are admitted in sinus rhythm and subsequently develop new-onset AF while being treated for critical illness, generally in the ICU. The main differential diagnostic consideration is previously undiagnosed asymptomatic paroxysmal AF. Features that would support a diagnosis of NAFCI rather than chronic, paroxysmal AF may include symptomatic AF or severe physiologic stress triggering transition into AF (e.g. AF immediately following an epinephrine bolus).

NAFCI is common in critically ill patients with an incidence of ~10%, but little is known about its treatment (Yoshida 2015). Unfortunately, our understanding of AF treatment is based almost exclusively on studies of outpatients with chronic, recurrent AF occurring spontaneously - which is the polar opposite of NAFCI.

Arguments for attempting rhythm control

Available evidence regarding rate vs. rhythm control is derived from outpatients, where there is equipoise between these two strategies. Given that NAFCI patients often have transient AF due to acute stress, there may be a stronger rationale for attempting rhythm control in NAFCI compared to spontaneous AF. Additional arguments in favor of attempting rhythm control include:

Minimizing stroke & hemorrhage risks?

Perhaps the strongest argument for a rhythm-control strategy may be to avoid risks of stroke or anticoagulation. NAFCI is associated with a 2% risk of in-hospital ischemic stroke among patients with severe sepsis (Walkey 2011). High stroke rates may reflect hypercoagulability due to systemic inflammation.

Anticoagulation is usually recommended to reduce stroke risk for patients in AF >48 hours. Unfortunately, critically ill patients also have high rates of hemorrhage when anticoagulated for AF. Consequently, some authors have challenged the broad use of anticoagulation (Labbe 2015). There is no consensus about this: The rate of anticoagulation varies between 3% and 57% at different centers (Champion 2014, Koyfman 2015).

Realistically, the efficacy of anticoagulation among these patients is low. Most studies report that below one third of patients receive anticoagulation. Even when anticoagulation is provided, it may need to be interrupted and the level of anticoagulation may often be sub therapeutic. In this context, a prompt and effective rhythm control strategy might reduce the risk of stroke and other arterial thromboembolic events (e.g. mesenteric ischemia, limb ischemia).

Rhythm control has failed to prevent stroke among outpatients, possibly due to the occurrence of subclinical episodes of AF. NAFCI differs from outpatient AF because it presents a unique opportunity to pursue immediateand definitive rhythm control (e.g., critically ill patients on telemetry cannot have occult episodes of AF). Unfortunately, there is no evidence regarding the effect of a rhythm control strategy on stroke among NAFCI (1).

Improved mortality?

NAFCI is consistently associated with increased mortality, often as an independent risk factor (figure above; Yoshida 2015). Similarly, patients who are cardioverted to sinus rhythm have a lower mortality than patients in whom cardioversion fails (Meier enrich 2010). This may merely be a correlation, because NAFCI is associated with increased age, illness severity, and comorbidity. However, causality is also possible. Even if AF doesn't cause obvious hemodynamic instability, reduced cardiac function might impair the patient’s ability to cope with critical illness. For example, in the setting of kidney injury, reduced renal perfusion could be harmful. In the setting of ARDS, AF might increase pulmonary venous pressure, exacerbating pulmonary edema.

Reduced risk of future AF?

"Allowing patients to remain in AF for weeks to months will increase their risk for developing long-standing persistent AF"
- Bhave 2013

The longer a patient is in AF, the harder it is to transition back to normal sinus rhythm. Hence the clinical adage, "atrial fibrillation begets atrial fibrillation." The technical term for this process is electrical remodeling of the atria, which begins within minutes of transitioning into AF (Goette 1996).

With the resolution of critical illness, most patients will spontaneously revert to sinus rhythm. However, some may not. It is possible that prompt cardioversion, prior to electrical remodeling, might reduce the likelihood of developing persistent atrial fibrillation or future paroxysms of AF (2).

Strategy for cardioversion and maintence of sinus rhythm

*Magnesium infusion protocol located here.

[1] Act quickly

The ideal timing of cardioversion is immediately after initiation of AF. AF causes electrical remodeling of the atrium which may make cardioversion more difficult over time. Thus, it is conceivable that a "golden hour" (or some time window) may exist following new-onset AF during which cardioversion would be most successful. Additionally, minimizing the duration of atrial fibrillation should reduce the risk of stroke (contrary to classic teaching, there is a tiny risk of forming thrombi before 48 hours; Airaksinen 2013).

[2] Investigate and remove triggers if possible

Any precipitating factors should be addressed (i.e., electrolyte abnormalities, catheter irritation of the atria, sympathomimetic drugs, hypoxemia, adrenergic states such undertreated pain/agitation or alcohol withdrawal). AF may be promoted by hypovolemia or volume overload, so volume status should be optimized. For patients who are on vasopressors, transitioning to agents with less beta-adrenergic activity may be considered.

[3] Cardiac magnesium infusion

This was explored in detail last week. Magnesium is a very safe therapy, which may improve rhythm control. For patients with intact renal function (e.g. GFR>30 ml/min), a magnesium infusion is needed to maintain high serum levels of magnesium and replete intracellular magnesium stores.

Sleeswijik 2008 described the use of magnesium alone initially, with amiodarone reserved for patients who fail to respond to magnesium. However, evidence supporting this strategy is limited. Currently it may be reasonable to start both amiodarone and magnesium immediately, with magnesium functioning as an adjunctive agent. Magnesium appears to work synergistically with amiodarone and other class-III antiarrhythmic agents (Cagli 2006, Ganga 2013).

[4] Amiodarone

Amiodarone is widely used among critically ill patients for several reasons. It promotes both cardioversion and maintenance of sinus rhythm. It rarely causes hypotension or secondary arrhythmias (e.g. torsade de pointes).

Pharmacology: Amiodarone for cardioversion

There are a wide range of doses reported for cardioversion of new AF, with little evidence comparing different doses. A typical regimen is loading with 300 mg followed by an infusion at 1 mg/minute for 24 hours.

A potential drawback to this regimen is that it will produce falling serum levels of amiodarone after the initial bolus. To counteract this, some authors have utilized a larger, tapering loading dose, which achieved a 92% conversion rate (e.g. 300 mg over one hour followed by 540 mg over the next three hours; Hou 1995). A simpler approach may be to administer an additional 150 mg amiodarone once or twice if the patient doesn’t cardiovert within 6-8 hours. Delayed re-loading occurs after the cardiac myocytes have absorbed magnesium, which might render them more responsive to amiodarone:

Pharmacology: Amiodarone for maintenance of sinus rhythm

The half-life of amiodarone depends on how long it has been administered. When used chronically, the half-life is extremely long. This fuels a common misconception that amiodarone can be stopped with no immediate clinical consequence. However, when it is first initiated, the half-life of a single intravenous dose is 18-36 hours (Hughes 2000). Therefore, premature discontinuation of amiodarone may lead to recurrent AF.

A multicenter Canadian study found that although 87% of patients could be cardioverted with amiodarone, 42% reverted back to AF during their ICU stay (Kanji 2012). To prevent recurrence, it seems reasonable to continue amiodarone for some period of time while the patient stabilizes (e.g., perhaps up to a week, depending on how rapidly the patient improves). This is similar to the accepted concept of using a short course of amiodarone for primary prevention of AF in patients undergoing cardiac surgery. When amiodarone is discontinued following several days of therapy it will have a longer half-life, providing some ongoing protection against recurrent AF.

Safety of amiodarone

Amiodarone is notorious for causing numerous complications (e.g. thyroid, lung, and eye disease). However, these complications only result from chronicaccumulation of amiodarone following months to years of treatment (Desai 1997).

Overall, the short-term use of intravenous amiodarone is well tolerated. The side-effect profile includes bradycardia, hypotension, and infusion-site phlebitis. Concerns have been raised about the possibility of acute-onset pulmonary toxicity, but this appears to be a myth (3). The risks of short-term intravenous amiodarone are likely lower than the risks of heparin anticoagulation or stroke from AF.

[5] Patients failing to cardiovert

Magnesium and amiodarone are both effective for rate control as well as rhythm control. Thus, even if patients fail to cardiovert, it is very likely that they will achieve adequate rate control (Karth 2001). Whether further efforts to achieve cardioversion (e.g. electrical cardioversion) are worthwhile is unclear, and may be judged on a case-by-case basis.

Resolution of the case

Our patient was started on magnesium and amiodarone infusions, with conversion to normal sinus rhythm a few hours later. Amiodarone was continued until she improved further and left the ICU a few days later. She was discharged from the hospital with normal renal function, normal sinus rhythm, and ongoing improvement in her pulmonary status.

New-onset AF is common among critically ill patients, but very little is known about its treatment.
New-onset AF correlates with increased stroke rate, ICU length of stay, and mortality. However, it is unknown whether AF causes increased mortality.
Theoretical arguments favoring an attempt at rhythm control include improved cardiac function, reduced stroke risk, and reduced risk of persistent AF.
Combining magnesium and amiodarone yields a high rate of cardioversion among new-onset AF.
Although amiodarone has substantial long-term toxicity, short courses of intravenous amiodarone are well tolerated.

Notes

(1) There are three theoretical reasons for anticoagulation in AF:

(a) Prevention of thrombus formation in a patient with ongoing AF.
(b) Resolution of existing thrombus prior to cardioversion (for a patient who has been in AF >48 hours, anticoagulation may be performed for 4 weeks prior to cardioversion)
(c) Prevention of new thrombus formation aftercardioversion. AF causes an atrial tachymyopathy, so even after conversion to sinus rhythm the atria may not contract effectively for 1-2 weeks. Thus, for a patient who has been in AF >48 hours, anticoagulation is advisable following conversion to sinus rhythm.

For a patient with NAFCI, prompt cardioversion (ideally <24 hours after the onset of AF) with sustained rhythm control should avoid the need for anticoagulation for any of these three reasons.

(2) Even among patients who spontaneously convert back to sinus rhythm, there is a significant risk of future episodes of AF following discharge. These episodes of AF may be subclinical, presenting initially with stroke (Walkey 2014). Although some authors have proposed the use of post-discharge holter monitoring, this remains a challenging clinical problem. If prompt control of AF within the hospital could reduce the risk of recurrent episodes outside the hospital, this would be of great value.

(3) Some concern has been raised regarding the possibility of acute-onset pulmonary toxicity within days of initiating amiodarone. The strongest support for this concept is a RCT evaluating the role of amiodarone in preventing AF among patients undergoing lung resection (Mieghem 1994). The study was stopped prematurely due to three patients who developed respiratory failure following amiodarone and right pneumonectomy, even though this difference was not statistically significant. Respiratory failure following right pneumonectomy is not uncommon, so it is entirely possible that this occurrence may have been due to chance.

Subsequently, there have been a handful of case reports describing acute-onset respiratory failure following amiodarone initiation, typically in the setting of complex patients undergoing multiple interventions. This evidence doesn't allow us to either prove or disprove whether acute-onset amiodarone lung toxicity is a real entity. Amiodarone pulmonary toxicity is a diagnosis of exclusion. The pathologic pattern of amiodarone exposure (foamy macrophages) may be found in patients taking amiodarone without clinical lung disease, so even a complete autopsy cannot determine whether the patient had amiodarone pulmonary toxicity.

Upon reviewing the literature, it appears that acute-onset amiodarone pulmonary toxicity may be a "trashcan diagnosis:" a diagnostic label which has been affixed to patients with unexplained respiratory failure and a variety of radiographic and pathologic features. It is possible that these patients may have been suffering from other causes of respiratory failure including pneumonia, aspiration pneumonitis, and various idiopathic interstitial pneumonias (Lee 2012, Boriani 2012, Kharabsheh 2002).

Perhaps most notably, recent RCTs involving amiodarone have failed to detect acute pulmonary toxicity. This led the American Association of Thoracic Surgery 2014 guidelines for postoperative AF to conclude that "In the nonsurgical population, it is commonly accepted that amiodarone-related pulmonary toxicity does not occur with short term (<1 month) exposure."

There is similar concern regarding the possibility of acute amiodarone-induced hepatic failure. To date there are perhaps eleven credible case reports in the literature describing this complication (Stratton 2015). If this is a real phenomenon, it seems to be extremely rare.

Image credits:

http://www.freeimages.com/photo/rough-sea-1-1245311

http://www.freeimages.com/photo/waves-101-1222750

↧

Investigation Bias: The freakonomics of when industry choses to sponsor a clinical trial

November 16, 2015, 12:51 am

≫ Next: Brief rant: Still no evidence that azithromycin increases mortality

≪ Previous: Treatment of hemodynamically stable new-onset AF in critical illness

Background: Publication bias

Over the last several years, publication bias has received a considerable amount of attention. In its most blatant form, this is when a drug company sponsors several trials, but only publishes the trials which yeild positive results. Growing awareness of this problem has led to the development of trial registries, with increasing pressure on industry to publish all trial results (positive or negative). Investigation bias is a similar problem.

Investigation bias

Investigation bias refers to the fact that companies will only sponsor studies which they believe are likely to improve sales of their product. That seems reasonable enough. However, it can lead to an unfortunate situation wherein therapies are only partially evaluated.

When an intervention is first developed, it must undergo at least some testing to be accepted and FDA approved. Once the intervention is approved and enters clinical use, things get more complicated. As the intervention becomes more popular, the company has more to lose from additional clinical trials (if the trials fail to show benefit, or reveal an unexpected complication). Additionally, as the intervention grows more popular, the company has less to gain from a positive trial (the intervention is already being used, so a positive trial may not improve sales substantially).

Game theory: When is it beneficial to run an additional clinical trial?

To illustrate how this may work, let’s imagine a very simple situation. Suppose that a drug is currently being used for x% of the market share (the maximal number of patients for whom it could be prescribed). If a clinical trial is run, the probability that it will yield a positive result is p_success. For the sake of simplicity, let us imagine that this will be a definitive trial which will either drive the market share up to 100% (if the trial is successful) or down to zero (if the trial is negative). If this scenario were played out a thousand times, the average gain and average loss in market share would be:

If the likelihood of having a positive study (p_success) is 50%, this yields the following graph:

In this scenario, it only makes sense to run a clinical trial if the current market share is <50%. If the drug is already occupying >50% of the market, the risk of running a trial outweighs the benefit.

Now let’s suppose that we have a very promising drug, which we truly believe is going to work well.We estimate the likelihood of obtaining a positive trial with this wonderdrug (p_success) to be 90%. In this case, the average gain and loss will be as shown below. Even if the drug is already quite popular (occupying up to 90% of the market share), it would still make sense to run another clinical trial.

Finally, let’s suppose instead that we have a rather dodgy drug which we don’t really expect to work well. We estimate the likelihood of obtaining a positive trial (p_success) with this drug to be 25%. As shown below, for this drug it is advantageous to run another clinical trial only if the market share is very low (<25%).

As these examples illustrate, it is unwise to run a clinical trial if the current market share is greater than the likelihood of obtaining a positive trial. This is an oversimplification, but it illustrates a basic point: when the drug’s popularity exceeds its probable effectiveness, further investigation of the drug is a poor investement for the company.

What does this imply about investigation bias?

After an intervention is approved, its market share will improve over time. The company will continue to study it, until a stopping point is reached when the company believes that the risk of a negative trial outweighs the possible benefit of a positive trial.

For a drug which the company believes is going to be extremely successful (e.g., p_successguessed to be 90%), investigation bias is not a big problem. The company will continue running clinical trials until the market share is very high. By investigating the drug thoroughly, the company has much to gain and little to lose. This is good business and good medicine.

The problem arises for a drug which the company doesn’t think works very well. Or, perhaps, the drug is initially promising but further investigation suggests that it may not work well. For example, suppose that the company estimates the likelihood of a successful clinical trial is 15%. As soon as the drug becomes somewhat popular, the company will stop further testing on the drug. This is good business but poor medicine.

Herein lies the real danger of investigation bias: a drug which becomes prematurely popular, before rigorous research has proven its efficacy. If the company believes that the drug’s popularity has exceeded its effectiveness, it will immediately halt further investigation into the drug. This may cause the drug to be used for years or decades, without any firm evidence basis.

Example #1: Activated protein C (APC), a.k.a. Drotecogin alpha (XIGRIS)

In 2001, the PROWESS trialfound that APC caused a stunning 6% absolute mortality reduction in septic shock. However, there were some concerns about this study including premature termination and a change in the recruitment protocol during the study (Finfer 2008). Additionally, subgroup analysis suggested that the drug was effective only in the sickest patients.

When the FDA considered approval of APC, the initial vote was evenly tied. Ultimately the FDA approved APC in 2001 with a request for further evaluation in patients with less severe septic shock.This led Lilly to sponsor the ADDRESS trial in 2005, which confirmed that APC was indeed ineffective in such patients.

Unlike the FDA, the European Medicines Agency (EMA) approved APC in 2002 with a requirement for undergoing annual review. In 2007, this annual review found that the evidence supporting APC was weak and called for further studies. To satisfy the EMA, Lilly performed the PROWESS-SHOCK trial, which was an attempt to replicate the initial PROWESS trial. This was a negative study, which led to the immediate withdrawal of APC from the market.

This illustrates how investigation bias may discourage replication of a positive study. Ideally, a dangerous and expensive drug shouldn’t be prescribed to thousands of patients over eleven years on the basis of a single positive study. Ideally, the PROWESS trial should have been replicated earlier. However, from Lilly’s standpoint, a smarter strategy was instead to study APC in populations where the drug wasn’t already being used (e.g. septic children in the RESOLVE trial). If this trial had been positive, it would have expanded the use of APC. Alternatively, if it were negative (as was the case), it wasn’t a major loss to the company. Thus, studying the drug in a new patient population is a low-risk, high-benefit strategy.

Alternatively, replicating the PROWESS trial was a high-risk undertaking, because it threatened the entire APC market. Although the PROWESS trial was very positive, subsequent data didn’t look so encouraging for APC. Lilly may have realized that a replication of PROWESS probably wouldn’t be as positive as the initial trial. Consequently, Lilly waited until it was pressured by the EMA to undertake a replication study. The rest is history.

Example #2: Inferior Vena Cava Filters

The use of IVC filters currently is based on a single RCT in 1998 (PREPIC-1). This was not an overwhelmingly positive study: IVC filters reduced the risk of PE, increased the risk of DVT, and yielded no mortality benefit. The results may have been biased because patients who were randomized not to receive an IVC filter knew that they didn’t receive a filter, which may have increased anxiety about recurrent PE, leading to an increased intensity of subsequent scans.

Nearly two decades have gone by, with IVC filters being permanently implanted in thousands of patients. Meanwhile, there has been no replication of PREPIC-1. Why not? IVC filters have been broadly accepted by the medical community, with a consensus to use them in patients with pulmonary emboli who can’t receive anticoagulation. Their popularity probably exceeds their efficacy. Industry has little to gain and much to lose from attempting to replicate PREPIC-1.

The PREPIC-2study was performed to investigate whether temporary IVC filters improve outcomes in patients who could receive anticoagulation, a controversial indication for which IVC filters were only occasionally used. The study was sponsored by the French Department of Health but supported indirectly with a free supply of IVC filters from industry. This study is similar to the RESOLVE trial of APC in septic children: it tested IVC filters in a situation where they were rarely used. The study was negative, but only caused limited damage to the IVC filter market (more discussion of these studies here).

Detecting investigation bias: Hearing the dog that doesn’t bark

Investigation bias can be very subtle, to the point of being nearly invisible. It is impossible to be critical of a study which doesn’t exist. Especially in critical care, we are used to encountering topics about which little evidence exists. Thus, a lack of evidence supporting various interventions often goes unnoticed.

A famous Sherlock Holmes story turns around Sherlock’s noticing the absence of a dog barking as evidence that the perpetrator was the dog’s owner. In this same sense, it may sometimes be possible to sense the presence of investigation bias based on the absence of studies which really ought to be done. When a drug is intensively investigated for a few years and then, amid ongoing controversy, investigations abruptly stop… what happened? Why were no further trials done? Perhaps the company lost faith in the drug and felt that ongoing studies wouldn’t pay off. It is impossible for us to know.

Conclusions

The scientific method is based upon ongoing study until a topic is understood, driven by a fundamental quest to understand the topic. Unfortunately, industry-funded research is a perversion of this process, driven instead with an endpoint of improving sales. Investigation bias is created because industry will only pursue studies to the extent that they represent a good financial investment for the company. In some cases, this may halt further research long before a topic is well understood, leading to the ongoing use of a harmful or ineffective therapy. Unfortunately, this form of bias is difficult to detect or contend with.

Related posts

Image credits:

https://en.wikipedia.org/wiki/Poker

https://simple.wikipedia.org/wiki/Roulette#/media/File:13-02-27-spielbank-wiesbaden-by-RalfR-093.jpg

https://en.wikipedia.org/wiki/Sherlock_Holmes#/media/File:Strand_paget.jpg

↧

Brief rant: Still no evidence that azithromycin increases mortality

November 18, 2015, 7:52 pm

≪ Previous: Investigation Bias: The freakonomics of when industry choses to sponsor a clinical trial

Introduction

It has long been known that some macrolides (e.g. erythromycin) cause torsade de pointes. However, azithromycin has a much lower affinity for cardiac potassium channels than erythromycin, so it has less effect on the heart. For many years it was believed that azithromycin lacked cardiac toxicity.

Controversy was sparked in 2012 when a retrospective study by Ray et al found that azithromycin use correlated with an increased risk of cardiovascular death. However, this correlational study may have been confounded because patients treated with azithromycin were more likely to have COPD or pneumonia.

Subsequent correlational studies revealed conflicting findings, with one large study failing to replicate Ray's findings (Svanstrom 2013). More importantly, meta-analyses of prospective RCTs of azithromycin have found it to be safe. This topic was previously explored in detail here.

New meta-analysis by Cheng 2015, J Am Coll Cardiol

This article is a meta-analysis of all studies relating macrolide use to cardiac events, with an impressive total sample size of over twenty million patients. Unfortunately, this includes several large retrospective studies (including conflicting studies by Ray 2012 and Svanstrom 2013). These large retrospective studies will over-power and drown out any signal from smaller, higher quality RCTs included in the meta-analysis.

The study found that azithromycin was associated with greater rates both of cardiovascular death and of the composite of sudden cardiac death or ventricular tachyarrhythmia:

It is accepted that erythromycin is considerably more arrhythmogenic than azithromycin. However, this study found these two drugs to have nearly identical effects on the risk of arrhythmia (green boxes above). This should cast doubt on the external validity of these results.

Heterogeneity was found between different types of studies and between different macrolides (blue boxes above; note that these p-values test homogeneity between strata). Such heterogeneity challenges the validity combining these studies together into a single meta-analysis.

For a moment, let's suspend skepticism and suppose that this data is entirely accurate. In that case, the most important finding might be that azithromycin is associated with a trend toward reduced all-cause mortality (red box above). The only way to explain this would be that azithromycin exerts some life-saving effect (e.g. anti-inflammatory or anti-infective) which offsets the increase in cardiac death that it causes. Thus, even if we were to accept the validity of this study, it may still support the use of azithromycin.

Ultimately, this meta-analysis is driven largely by retrospective, correlational studies. Thus, it constitutes a lower level of evidence than meta-analyses of RCTs, which have found azithromycin to be safe (e.g. Baker 2007, Almalki 2014). Ironically, this paper concludes with the statement that "large RCTs are warranted to confirm these findings" -- which unfortunately neglects the fact that numerous RCTs and meta-analyses of these RCTs have already been done.

Accompanying editorial by Viskin et al. , J Am Coll Cardiol

This is a disquieting editorial, which raises the possibility that litigation of the pharmaceutical industry could lead to the discontinuation of the entire class of macrolide antibiotics. Really? Azithromycin is notable for its excellent safety profile. Eliminating azithromycin would increase the use of fluoroquinolones, which probably have more side effects (e.g. delirium and the selection of MRSA and clostridium difficile).

The editorial suggests that correlative data can be used to establish causation for two reasons (see highlighted text below). These arguments are based on data from erythromycin, but do no apply to azithromycin. As previously discussed, neither electrophysiologic data nor RCTs suggest that azithromycin causes arrhythmia.

It is accepted that azithromycin is significantly safer than erythromycin. Thus, it is not valid to combine different types of macrolides within the same study.
This meta-analysis is primarily driven by retrospective observational studies. As such it constitutes a lower level of evidence than pre-existing meta-analyses of RCTs (which have found azithromycin to be safe).
If we are worried that azithromycin is killing patients, the most robust and patient-centered outcome to evaluate is all-cause mortality. This study and meta-analyses of RCTs have all found a trend toward reduced all-cause mortality with azithromycin.
Available evidence suggests that azithromycin does not cause torsade de pointes or increased mortality.

Related posts

Azithromycin doesn't cause torsade de pointes or increase mortality

Conflicts of Interest: Never.

↧