Dying sooner or later? Predictors of pre-30 day mortality versus post-90 day

30v90tableHypothesizing that the determinants of death within 30 days of admission to ICU are different to those for death after 90 days, Garland et al looked at a Canadian 11 ICU registry. We’re familiar with survival curves levelling to population baseline rate after a few months (if you can make it 3 months it’s almost like it never happened!). Examining the curve between 1 and 90 days the section before 30 days is considerably steeper. So unsurprisingly early deaths relate to type and severity of the critical illness whereas by about day 90 co-morbidities have become the stronger predictor. Age was again a weak marker.

Perhaps sometime we forget that for patients with multiple concurrent medical problems their ‘pre-ICU’ Kaplan-Meier was already pretty steep. If we get them through 3 months, they rejoin their potentially ill-fated subgroup curve.

Does this suggest we should stop looking at 30 day mortality rates and make 90 day survival our default primary outcome when choosing our evidence?

The difficulty of ‘futility’

Futility tableUsing local consensus on the definition of futile, Huynh et al have returned to suggest and quantify an ‘opportunity cost’ (i.e. a detriment to others due non-futile care) incurred.

At a single centre, 5 unit hospital, delay in admission from A&E and delay in transfer to a tertiary ICU were examined and the impact of futile care on these delays assessed.

‘Futile’, ‘probably futile’ and non-futile treatment were not clearly defined and were subjective to the extent of depending on the specific coincidence of patient, doctor and hospital state of busyness. The same group has previously published on the same topic, finding that, in their single centre experience approximately 7% of hospital patient days were spent delivering costly futile care. 15% of those receiving futile care were still alive 6 months later but generally with a low quality of life. It is not clear in this week’s article how many patients were directly affected by delay in treatment, or to what detriment. In addition, figures from these studies will not be directly applicable to your institution, but the principles are intuitive and are likely to be transferable to some degree. It is surely no surprise that on less busy days there was a greater proportion of patients receiving futile treatment.

However, how often is it clear that treatment on the unit is futile in an absolute sense? Does the slippery term ‘futile’ really have a different definition for any given day/unit/doctor? Do we really operate a first-come-first-served system? The last chance to eke out what might be a tiny percentage chance to return to an acceptable life is often requested of intensivists. We know we are often a poor judge of the chance of success and also what an acceptable outcome is. Moreover, do some patients have a ‘better death’ on ICU, and if so is that of calculable value?

Immuno-nutrition worse than useless?

ImmunonutritionHeyland, SIGNET, REDOXS, ARDNSNET’s OMEGA, INTERSEPT, and now the Metaplus study. 

Approximately  300 patients, critically ill (APACHE >15) and ventilated were all given high protein (>1.2g/Kg/day), normal calorie (25Kcal/Kg/day) feed, either with or without Omega-3 acids, glutamine and antioxidants. All were started within 48 hours and stopped within 28 days.

No benefit in terms of length of stay or days on the ventilator. No difference in infection. No difference in subgroups except for an increase in adjusted (for age and disease severity) 6 month mortality amongst the medical patients.

Whilst most agree on high protein feeding, and initial permissive hypocaloric (~trophic) nutrition, evidence on the remaining contents is unclear. Regarding micronutrients Selenium recently performed well, but otherwise there have been no convincing and repeatable successes. And now it looks like the current batch of immuno-nutrition candidates have fallen short. Vitamin D is currently ‘research en-vogue’ but not yielding impressive results even at the basic science level, evident in another paper this week.

Moreover, the ‘ventilated patient’ is a heterogeneous set. Nutrients have been given enterally in some trials and parenterally in others. As always, larger studies in more defined subgroups are yet to happen.

The earlier the better? Maybe, maybe not for renal replacement therapy.

RRT timing

In this recent re-look at patients from a previous seminal study, a ‘nested’ cohort from the RENAL study with RIFLE ‘I’ acute kidney injury (GFR<50%, creatinine rise to twice baseline, UOP <0.5ml/kg for 12hr).were selected.

The APACHE III score for those supported late was significantly lower suggesting there may have been something different (a covariate, so not propensity matched) that added to confidence in delaying RRT.

Moreover only about 40% of those supported early had sepsis, against 60% of those receiving RRT after 48hr. This possibly reflects an element of faith in source control and circulatory support – give resuscitation a chance before firing up the filter. Presumably they were also less likely to be fluid overloaded, an indication for RRT know to carry poor outcome.

Urea level does not correlate well with time from start of AKI using RIFLE criteria. This is an issue that needs to be considered when looking at other studies on the same subject.

Acknowledging the dangers in concluding causality in observational studies, earlier RRT was not significantly associated with improved outcome but there was a suggestion of benefit (a trend towards mortality if delayed) that needs more exploring; large prospective studies in groups eg severe sepsis.

Do the dosing studies then need to be redone as well? An interesting single unit experience in saving money by switching to low volume (20ml/kg/hr) CVVHDF without changing clinical outcome was published this week.

Staffing the unit – more questions than answers?

UK staffing survey tableA number of articles in the last few months have focused on staffing structure in the ICU.

Previously we’ve seen that ‘high intensity staffing‘  (a dedicated ICU team, or at least the obligatory consultation by one) reduces mortality, in the USA. However night-time intensivists don’t seem to do the same unless there’s a ‘low intensity’ daytime staffing model. Cross-covering fellows may even improve outcome, perhaps through a ‘second opinion’ effect, as suggested in a single centre study.

Now in England, after the recent core standards (ICS and ESICM), a national survey suggests neither consultants’ working pattern nor clinical experience affect mortality (adjusted). Grade of covering night doctor again appeared not to correlate with outcome. The suggestion that intensivist involvement in handover is associated with mortality can almost certainly be ignored as chance (given the small proportion where there is no involvement). Particularly as cardiac arrest in the preceding 24 hours seemed not to make a difference!

Telemedicine  has also shown promising results across the Atlantic in a recent meta-analysis.

Perhaps a bigger issue is staff at the bedside. Isn’t it likely that nursing staffing ratios and continuity affect clinical measures? And how do critical care practitioners and trainees fit into these results?

In any case, aren’t mortality statistics a blunt tool for this analysis? Clinically, longer term outcome measures, readmission rates and other quality indicators may be better. The organisational and meta-outcomes such as staffing moral, staff retention, patient and family satisfaction, and the trainees experience may be better still.


The golden hour for antibiotics in sepsis, reiterated. Shoot first, ask questions later?

Antibiotic timingSSC international data from more than 28,000 patients with severe sepsis or septic shock emphasises early source control is paramount  (closely followed by BP management). Get antibiotics in within 1 hour (2 at the very most) – the clock is ticking! Here, this was timed from the moment of triage or, on the wards, the moment the observations met criteria – differing slightly from previous work.

Possible signs in the data suggest it’s particularly those with severe sepsis but no hypotension that action was delayed. Also maybe under-recognition of liver dysfunction as organ failure.

Some concern that having a hair trigger for sepsis may keep antibiotic use high? Early de-escalation needs to remain high on the check-list for post-resuscitation management.



Recruitment manoeuvres – looking for an evidence base

recruitment manoeuvreThe PHARLAP open lung approach showed promising short term benefit earlier on in the year in ARDS, but now the evidence for recruitment manoeuvres alone has been meta-analysed by Suzumura. Nearly 1600 patients. No standardized strategy, illness severity or timing. They at least seem safe but outcome gain is less clear. An attempt at isolating the studies with least potential for bias suggests a mortality improvement. An NNT of 17 will encourage enthusiasts but sceptics might pull at the statistics. Evidence strongest in the moderate-severe group. Unclear if previous hints of more help in extra-pulmonary ARDS still stand. Staircase, ramp or 40-at-40, it remains part of the ARDS package, has basic science backing, and doesn’t seem harmful.

Nasal high-flow oxygen prevents re-intubation

NHF oxygenMaggiore et al have looked at the use of nasal high-flow oxygen in 100-ish pneumonia/trauma patients and found, most significantly, a reduction in re-intubation rate. You also get more sats for your FiO2, and fewer mask-now-on-their-ear moments. We know it gives flow-related 2-5 cmH2O of CPAP and is preferred by claustrophobes, but this small study suggests NHF is now fully-fledged and finding it’s fortes. The re-intubation prevention aspect deserves a closer, bigger look.

Older posts «

Follow on Feedly