Outcomes in adult ARDS

Question 28 from the first  paper of 2006 and  Question 2 from the first paper of 2003 asked the candidates who grew up in the post-ARMA trial era (2000 onwards) why the ARDS mortality data has failed to improve dramatically since the widespread acceptance of this strategy.

This 2006 question asked about the last two decades. Mortality in ARDS seems to have been declining steadily, at 1.1% per year (at least between 1994 and 2006). Since 2006, nine years have passed at the time of writing.

So, it is improving, but not very fast. These days (Villar et al, 2015) average in-hospital mortality seems to be somewhere just above 40%, ranging from 23% for mild ARDS to 60% for severe ARDS.

Why is ARDS mortality improving?

Specific widely-accepted therapies which have improved mortality are listed in the chapters on ventilation strategies for ARDS and supportive non-ventilation strategies for ARDS.

In short,

• Low tidal volume ventilation (that's probably the biggest one)
• Open-lung ventilation
• Improved supportive care and surveillance of complications (eg. early nutrition)
• Development of new strategies
  • Prone positioning
  • Earlier use of neuromuscular junction blockers
• ECMO is also more widely used, though it remains controversial

Why is ARDS mortality not improving enough?

Why has the current state-of-the-art in ARDS management has failed to produce significant improvements? Some points worth mentioning are listed below. There are lots of papers on this topic, but the single most useful one has been the review by Villar et al (2014)

Breakthroughs in ARDS management happen very infrequently.

Between the 1960s and now, there have been surprisingly few practice-changing management shifts. Since the ARMA trial (2000), there have been few studies which have demonstrated a mortality benefit for ARDS. Perhaps the exceptions to this are paralysis (which everybody has been doing all the time anyway) and low delta-P (which Amato et al only published in 2015). It would be unreasonable to expect major improvements in survival if there have been no major improvements in management.

Many widely-accepted therapies which were expected to improve mortality have failed to do so.

But people keep using them anyway. Some therapies which were expected to have a positive influence on mortality, and which have subsequently failed to show any benefit, are listed in the chapter on failed experimental therapies for ARDS. In addition to these, there are therapies which were long known to be unhelpful but which persisted in the repertoire of intesivists. These include inhaled pulmonary vasoodilators (nitric oxide and prostacycline), high frequency oscillatory ventilation (HFOV), ECMO (which remains a topic of debate) and several others. The course of such therapies follows a pattern  on initial excitement, experimentation, disappointment, denial, and gradual washout from mainstream practice. This course could run over many years. During these years, mortality for ARDS would not be expected to improve.

We expect an unrealistic improvement in mortality.

How fast do you want it to improve, seriously? The ARMA trial produced an 8.8% mortality benefit in their ARDS population. In spite of rapid and widespread adoption of lung-protective ventilation, ARDS mortality has not decreased by nearly this much in 2005. This may be because lung protective ventilation has had no realistic effect on crude mortality.  Massimo Zambon and JL Vincent (200) report on their analysis of 72 studies, which - if they are to be believed -  suggest that mortality from ARDS has been improving steadily at about 1.1% per year between 1994 and 2006. They hasten to point out that the greatest change in mortality since 1994 has been in-hospital mortality rather than ICU mortality.  We also have good data spanning the period before and after the 2000 ARMA study (1996 to 2005,  Erickson et al 2005). The crude 60-day mortality was 35% in 1996, and 26% in 2005. The trend towards improvement was already observed during the pre-ARMA 1996-2000 period. In fact, Milberg et al (1993) describes a pre-ARMA period (1983-1993) during which mortality also improved (from over 40% to 36%) in spite of violently lung-damaging ventilation strategies. Clearly, something other than lung protective ventilation was causing the improvement.  Thus, the mortality improvement we expected from lung-protective strategies may have simply failed to materialise outside of well-run clinical trials.

We are measuring mortality badly and inconsistently.

It is impossible to have an accurate representation of ARDS mortality during the period when there was no consistent definition for what ARDS actually is. For instance, Zambon et al (2008) quote studies form pre-1996 which claim that ARDS mortality was stable at around 40% since the 1960s. Since then apparently it has declined, but some studies still report mortality approaching 50%. Modern authors limit their meta-analysis to those studies which use the well-accepted ARDS/ALI definitions; however this may exclude studies which use other definitions, obscuring the "true" mortality rate. Not only that, but studies of mortality tend to vary in the time frame of data collection, which makes for difficult comparison: how do you compare a study of in-hospital mortality with a study of 30 day mortality or a study of ICU mortality? It is therefore impossible to say whether we have really improved mortality from ARDS, or whether it stayed the same, or whether there was an increase.

ARDS patients die of multi-organ system failure, not hypoxia.

What is killing these people is not susceptible to changes in ventilation strategy. ARDS-associated organ dysfunction is implicated in much of the mortality, and this may not be amenable to ventilation protocol changes. Thus, though we are ventilating our patients better, they are still developing multi-organ system failure and dying. Stapleton et al (2006) found that the most common cause of death in ARDS patients was actually sepsis and sepsis-associated organ damage, with respiratory failure accounting for only 13-19% of the deaths. In fact, the authors go on to say that "the increased survival over the past 2 decades is entirely accounted for by patients who present with trauma and other risk factors for their ARDS", and that the survival of ARDS patients with sepsis has not changed substantially over the preceding decade. This is consistent with the much older  findings and opinions of JL Vincent published in 1997 -  whose team attributed mortality to sepsis and MOSF in 49% of cases. Therefore, we cannot expect changes in ventilator management to produce miraculous cures in the ARDS population.

The ARDS patients today are sicker than they were 20 years ago.

Many of the patients who would have died of SIRS now survive and go on to develop ARDS in ICU. This population is sick: the SIRS also causes multi-organ system failure. All the other organ failures won't benefit from changes in ventilation protocols (see the points above). Hence, the ARDS we see today is likely to be associated with MOSF, and therefore have worse mortality. "Patients with ARDS are now older and more severely ill", complain Pierrakos et al (2012). These authors compared a 1993-1995 group to a 2006-2009 group. The modern ARDS population was older, more septic, staying in ICU for a shorter period, and featuring fewer patients with trauma and massive transfusion. The demographic change therefore reflects a trend towards increasing severity and complexity of co-existing illness in ARDS. It is in fact remarkable that any mortality improvement is observed between the two time periods.

There are fewer cases of  "mild" survivable ARDS.

Many patients who would have gone on to develop treatable "survivable" ARDS are now managed better and thus do not go on to develop ARDS, thereby disappearing from the data sets of mortality auditors.   Let us consider a population of mechanically ventilated patients who had "mild" ARDS and therefore had a reasonably low mortality. Let's say that the cause of ARDS in this population was primarily VILI resulting from inappropriate high-volume ventilation. Let's assume that in the post-ARMA era the incidence of such VILI has been greatly decreased. Thus, the incidence of "mild" ARDS should proportionally decrease, leaving only "moderate" and "severe" ARDS. If the population of ARDS patients is now composed of mainly moderate and severe cases, the overall mortality benefit from using lung-protective ventilation is buried in the demographic change.

Rapidly fatal causes of ARDS have become more prevalent.

The ARDS group is aetiologically heterogenous, and not all aetiologies have enjoyed an equal level of improvement in their management over the last few decades.   For instance, these days we see much less ARDS due to AIDS-related PJP, because the management of HIV infection has progressed to a point where florid AIDS is rare. The same might be said of various other previously prevalent conditions which have vanished into insignificance with a change in practice (eg. TRALI has disappeared with change to uniformly leukodepleted packed red cells). Instead, these days we might see more ARDS due to viruses and sepsis, and the population is older. Sepsis-induced inflammatory lung injury in the elderly is much harder to recover from than transfusion-associated lung injury in the fit young trauma patients.