Off Label, On Target? FREE TO VIEW

In the lexicon of current medical practice, the term off–label is immediately understood through all levels of the profession. It refers to the use of a medication for a disease, or a dose, or method of administration that is not listed in the official labeling of its use. It is not only commonplace, but in some cases may represent the preferred therapy. On the other hand, the basis for off-label use may have the shakiest of support if using the evidence-based pyramid. Treatment is often based on anecdotes or small case series, although some therapies will have withstood the scrutiny of a randomized double-blind investigation. By its very nature though, off-label use represents extrapolation from prior experience and, more often than not, represents the early application of a yet-unproven therapy. It is important to note that this practice may not be inappropriate or ineffective, just unproven. The questions that naturally arise in this scenario span several areas, including concerns about safety, ethics, and responsibility.

On the other hand, the practice of medicine is replete with examples of treatment that strayed from usual or standard practice, only to subsequently be demonstrated as a viable or preferable option. Sildenafil in the treatment of pulmonary hypertension may be the latest example, but some may recall that at one time isoniazid also was used to treat depression. The phenomenon of off-label use is not restricted to medications. There are treatments and procedures that also fit this categorization, as the application of effective therapies is extended to other similar, and sometimes dissimilar, conditions. This is demonstrated by the increasing use of hyperbaric oxygen therapy or noninvasive ventilation in a host of conditions similar but disparate from those in the initial investigations. Of course, the evidence to support these variations in management may be difficult to accumulate or may occur several years after the procedure has been in place. It is unrealistic to expect the “gold standard,” randomized, placebo-controlled trial for every realm of medical care. There simply is not enough time, money, or investigators. It follows that success in one arena might lead to application in other similar, but distinct conditions.

In this issue of CHEST (see page 1281), Wongsurakiat and colleagues present data that highlights this very process. They document a change in ventilator settings over little more than a decade of observation (from 1990 to 2000). They investigated the ventilator management of cardiac arrest patients who were identified through a long-standing database and appropriately eliminated subgroups of patients that may have warranted special attention to ventilator settings (eg, COPD, ARDS, and other conditions). In that fashion, they were left with a group of patients whose management should be unencumbered by any special considerations and should reflect what would be considered the current best practice. Although this is a retrospective review, the authors were able to demonstrate definite changes in ventilator settings. Their major findings were a reduction in a 3-day average mean (± SD) set tidal volume (Vt) from 11.8 ± 1.5 to 8.4 ± 1.3 mL/kg (measured body weight) and an increase in the use of positive end-expiratory pressure (PEEP). In the first years of the study period, > 80%, and close to 90%, of patients had no PEEP added to their treatment, whereas by the end of the decade the converse was found, with < 20%, and probably closer to 10%, of patients receiving ventilation without PEEP. PEEP levels of ≥ 5 cm H₂O were used in about 10% of patients in the first years, and this had increased to about 90% of patients at the end of the decade. These changes were associated with an increase in radiographically evident atelectasis (58% vs 43%, respectively), with possibly higher oxygen requirements, but no changes in other outcome measures, including high plateau pressures, pneumonia, number of ventilator days, or mortality.

Based on the group profiled and with the elimination of confounding conditions, one can surmise that these changes in ventilator settings are an accurate reflection of the general approach to patients requiring mechanical ventilation during this time frame. Physicians at the end of the decade were using lower Vt levels and more PEEP than they were in previous years. This might not be that noteworthy, except that there exists little data to support this approach in patients following cardiac arrest with healthy, or presumably healthy, lungs. The use of Vt levels in the range of 10 to 15 mL/kg has been a long-standing recommendation, based in part on surgical and anesthesia practices directed at avoiding postoperative “microatelectasis” and the normalization of arterial blood gas values.¹ This in turn was influenced by experience in the management of polio patients with respiratory failure who experienced greater comfort with larger Vt levels.² This was certainly the prevailing practice during the late 1980s and early 1990s.

So then, what may have led to this change in management? There has been a plethora of literature addressing this very issue in the management of ARDS patients. It was recognized from animal studies that high peak airway pressures can be harmful to the lung, causing direct endothelial injury, inflammation, pulmonary vascular injury, and increased pulmonary vascular permeability.³ In some studies,⁴⁵ this occurred within an hour or less of exposure of healthy lungs to high inflation pressures. In an autopsy study of ARDS patients,⁶ there was greater airspace enlargement and alveolar overdistension in those patients subjected to higher mean peak airway pressures (56 ± 18 cm H₂O), Vt (12 ± 3 cm H₂O), and fraction of inspired oxygen (> 0.60). With as little as a third of the lung able to participate in gas exchange in ARDS patients,⁷ it is easy to appreciate the potential for further lung damage using high Vt levels that are transmitted to only a small portion of the lung. Some of this injury may be ameliorated with PEEP therapy, as more alveoli can be recruited to participate in gas exchange with volumes and pressures more evenly distributed and less injurious. Animal studies⁴⁸ have demonstrated less injury when the animal is subject to high inflation pressures with PEEP, as opposed to the high inflation pressures alone. The plain chest roentgenogram obtained in ARDs patients revealed diffuse, bilateral disease, but CT scans documented the marked inhomogeneity of airspace disease. The benefit of PEEP can be seen on CT scans that demonstrate alveolar recruitment and decreased airspace disease.⁹

These findings became more intriguing with the association of lower mortality rates in ARDS patients who had been treated with lower set Vt levels and limits on peak inspiratory pressures, even to the point of hypoventilation and hypercapnia.¹⁰ First noted in a retrospective analysis,¹¹ this finding was duplicated in a prospective evaluation in which Vt levels were limited to 7 mL/kg, and peak inspiratory airway pressures to < 30 cm H₂O. In both analyses, the mortality rate was reduced by ≥ 50%. In a consensus conference on mechanical ventilation in 1993, and published later that year,¹² the potential for alveolar overdistension and lung injury was prominently noted. Recommendations were made to adjust ventilator settings to include a reduction in Vt levels in order to limit exposure to plateau pressures ≥ 35 cm H₂O. No specific range was endorsed, although Vt levels as low as 5 mL/kg were acceptable. This was also the sentiment expressed in recommendations on management in a major textbook published the following year.²

As the pulmonologists and others who were managing patients receiving mechanical ventilation were beginning to reevaluate their approach to these patients, several clinical trials were in progress during this decade. They addressed the concept of limiting alveolar distension by limiting Vt, peak inspiratory pressure, plateau pressures, or some combination of these variables.^13141516 Only the ARDS Network trial,¹⁶ comparing a Vt of 6 mL/kg (predicted body weight) with a Vt of 12 mL/kg, demonstrated a mortality benefit. Recruitment for that study started in March 1996 and was completed 3 years later. Coincidentally or not, the data presented by Wongsurakiat and colleagues reviewed patients from one of the participating institutions in that trial. Although not the focus of the trial, the cardiac arrest patients represent a cohort of patients who were treated alongside of the study patients. All medical staff, physicians and nonphysicians alike, must have had some knowledge of the study, its premise, and its purpose. It would follow that this may have influenced their management of other patients, in this case, cardiac arrest patients. Granted, the volumes set were not at the level used in the experimental arm of the study, but one cannot escape the possibility of that trial having an influence on decisions and management related to limits in plateau pressure and lower Vt levels.

But there are also data that argue against any influence of the findings of that trial on clinical practice. Following that report, two surveys¹⁷¹⁸ suggested that the move to lower set Vt levels for ARDS patients was lagging, even among teaching institutions. In fact, one report¹⁸ originated from the same hospital as this report. Nevertheless, there does appear to be a trend in the reduction of set Vt levels over several decades and investigations.¹⁷

How does one interpret these findings and what does this represent? Is this finding of lower set Vt and more use of PEEP limited to this single institution? The treatment of these patients at other facilities has not been closely examined. There is an acknowledged wide variation in the application of mechanical ventilation.¹⁹ However, in a 28-day survey with analysis of COPD and ARDS patients, the average set Vt levels were comparable to those reported by Wongsurakiat and colleagues. There is no breakdown by cardiac arrest patients. In an earlier 1-day corollary study,²⁰ it should be noted that the median set Vt of all patients categorized by country was in the range of 8 to 10 mL/kg (measured body weight), with the 75th percentile readings in the range of 9 to 12 mL/kg. Others have noted that ventilator management is frequently a reflection of local practices and have confirmed the impression of wide variation and incomplete adoption of recent clinical findings.²¹ The change over time in other institutions has not been similarly examined. Therefore, the decline of set Vt levels over time may not be limited to this single institution.

These changes over time in ventilator settings likely represent current practices at that institution. Does this in turn represent a “standard” of care? It is probably safe to say that it represents the most common practice at the time (“usual and customary” is another phrase that comes to mind). Having acknowledged this, it should also be recognized that current practice or standard of care are dynamic and elusive entities, subject to change as well as debate. This very issue has generated extensive debate with potentially profound implications on the conduct, ethics, and interpretation of this ARDs network trial as well as ongoing and planned investigations.²²²³²⁴ How this relates to other management cannot be known, but the data indicate that the average settings used in 1992 were different from those used in 1995, and that the changes have persisted through to 2000.

Although the basis for change represents the culmination of years of research, these changes are even more striking given the relatively short period of time in which they have occurred. Even with convincing clinical trials, the incorporation of treatment into practice may take years (estimated to be on the order of 5 to 10 years), and, in the case of thrombolytic therapy for myocardial infarction, was still lagging ≥ 6 years after the initial reports.²⁵²⁶ Several barriers have been identified, and they include issues with readiness to change, diffusion and dissemination of information, acceptance, mechanism, and costs of implementation.²⁷²⁸ Change does occur more readily in institutions that participated in the treatment-defining trials. Local practices and local leaders may be more important in affecting change than any other mechanism.²⁹ This is an intriguing aspect that is worth further investigation, since any process or mechanism that may have facilitated these changes would be invaluable in bringing other changes into clinical practice.

In summary, it is fair to surmise that changes over this decade were directed at ameliorating some of the risks of mechanical ventilation that affect both healthy and diseased lungs. The physicians in this report have extrapolated the experience with ARDS to other conditions, leading to a strategy that would decrease the exposure of the lung to the deleterious effects of mechanical ventilation. Other questions remain. Is this the right approach? Is it safe to treat one group of patients based on data from another group? What are the optimal settings? There remains some uncertainty about using very low Vt levels, but there is a definite trend in using lower Vt levels and PEEP. Is there any harm with settings that produce more atelectasis? Pneumonia and hypoxemia were not increased, but the study was not designed to adequately evaluate these outcomes. Do these changes make any difference in outcomes? The improvement in outcome in one group does not necessarily translate to improvement in other groups. These are all questions that will require more study and analysis. No answers are forthcoming. When clinicians are faced with this type of situation, they often resort to their instincts, based partly on science and partly on experience but squarely focused on the patient’s best interest. There has been much learned, with changes that seem headed in the right direction, but are these changes on target? Only time will tell.