Measuring Performance of an ICU

Todd Fraser on 18-03-2012

An interesting article in a medical news digest recently described the impact of public reporting of hospital patient-centred outcomes. One of the measures used as a risk-adjusted hospital mortality rate. Despite expectations, the release of this performance information has not impacted on outcomes. It raised the question for me of how we should monitor the performance of an ICU. It seems in recent years the Standardised Mortality Ratio has been increasingly used to compare units, something that was considered taboo when I was in training. My understanding was that the SMR was heavily influenced by factors other than unit performance, and as such had little meaningful value in comparing units. Lead time bias, case mix, discharge planning (particularly approaches to palliation), performance of other services within the hospital, referral bias and a host of other issues supposedly prevent comparing SMRs between units being useful. But this appears to be occurring anyway. Is this a reasonable thing? Are there better ways to monitor a unit's performance?

3 Comments

Got something to say?

Log in to your account

Forgotten your password?

Matthew MacPartlin wrote 03-24-2012 06:02:27 pm
My understanding of the use of SMR is similar; ie it is virtually meaningless, but less so if you are comparing units with a similar role deliniation serving similar populations. And therein lies the difficulty, as has been the case when trying to apply trial findings to individual units.

I don't know what the best option is (and perhaps it is in our interest not to find one, lest it be twisted into a MyICU national league table) but maybe the SMR could be tailored into a relative-SMR-for-population-served, the trend of which might be more insightful than a one-off data point (think physiological static measurement versus trend).

Carl Horsley wrote 04-23-2012 07:24:46 am
Is mortality the only outcome of value for patients? While everyone would want the best chance of surviving if they or their family members were in an ICU, there are other aspects we would all value including not having harm done, LOS, communication, rehabilitation times etc. SMR is such a blunt tool for measuring outcomes and is certainly not the only patient-valued outcome.

Todd Fraser wrote 04-23-2012 10:31:55 am
No doubt you're right Carl.

This is becoming an area of interest for ICU research now, isn't it. The podcast I did with Neil some time ago reflects a lot of these issues - something I hadn't spent a whole lot of time thinking about. But the long term outcomes for survivors of ICU are not great. We need to put some effort into that.

As for measuring performance based on QAL outcomes, I'd imagine this would be even more difficult.