Your intensive care unit collects APACHE III and mortality data and derives the  Standardized Mortality Ratio (SMR) every 3 months as a quality control measure. The SMR for your unit normally ranges between 0.65-0.7. In the latest 3 month figure, the SMR for your unit was noted to be 1.2. Outline, what are the possible reasons for the change in the SMR?

[Click here to toggle visibility of the answers]

College Answer

SMR is the ratio of the observed hospital mortality and the actual hospital mortality. A ratio of > 1 implies a mortality higher than expected. Potential explanations:

a)  Ensure data entry is correct and accurate and consistent with prior practice (ie comparable)
b)  Issues like quantifying GCS accurately will have an impact on APACHE scores and consequently SMR. Quantification of GCS is a major source of inaccuracy. Also source of admission and diagnosis
c)  SMR reflects system wide performance rather than ICU performance alone, because based upon hospital mortality, not ICU mortality. Look at pre ICU and post ICU facilities in the hospital

d)  SMR affected by case-mix, so changes in case mix may account for increase in SMR and increased other hospital admissions

e)  One needs to examine if there has been a deviation from clinical protocols in the ICU
f)   Lead time bias (pre ICU care) has been shown to impact on SMR and this neds to be factored into.
g)  Are there new inexperienced staff in ICU who might need training?

Discussion

LITFL have a point-form summary. In this summary, there is an excellent final paragraph, which discusses the reasons as to why the SMR might be changing.

The SMR is the ratio of the observed hospital mortality vs. predicted hospital mortality for a specified time period. An SMR of 1 means the mortality is as expected, and an SMR of >1 is worse than expected. The massive jump in SMR as described in the college question is indeed a disturbing development, one which has prompted LITFL authors to blame influenza pandemics and terrorist attacks.

So. Why might the SMR be on the rise?

It would for two possible reasons.

Either the observed mortality rate is increasing, or the predicted mortality rate is decreasing.

Increase in the observed mortality rate

  • Change in protocols
  • Change in admission practices -i.e. more patients being admitted who have little chance for survival
  • Change in discharge practices - i.e. more patients remaining in hospital to be palliated instead of being discharged with community services (this is what happens when you cut funding to community palliative care nurses).
  • Change in staff (inundation by incompetent staff?)
  • Unstable transfer patients received from other hospitals
  • Hospital performance as a whole may be affected by system-wide policy or staff changes (i.e. did all the senior nursing staff suddenly go on annual leave?)

Decrease in the predicted mortality rate

  • Change in illness severity scale encoding - i.e. the encoding has been omitting factors which might otherwise have increased the APACHE and SOFA scores.
  • "Lead time bias" - treatment received prior to ICU admission may result in artifically normalised acute physiology scores
  • "Healthy worker effect" - a change towards selective ICU admission practices may be favouring patients who score low on illness severity scales, eg. young elective surgical patients.

References

References

Fletcher, John. "Standardised mortality ratios." BMJ 338 (2009).

Liddell, F. D. "Simple exact analysis of the standardised mortality ratio."Journal of Epidemiology and Community Health 38.1 (1984): 85-88.

Wolfe, Robert A. "The standardized mortality ratio revisited: improvements, innovations, and limitations.American Journal of Kidney Diseases 24.2 (1994): 290-297.

Gaffey, William R. "A critique of the standardized mortality ratio." Journal of Occupational and Environmental Medicine 18.3 (1976): 157-160.

McMichael, Anthony J. "Standardized mortality ratios and the'healthy worker effect': scratching beneath the surface." Journal of Occupational and Environmental Medicine 18.3 (1976): 165-168.

Tunnell, R. D., B. W. Millar, and G. B. Smith. "The effect of lead time bias on severity of illness scoring, mortality prediction and standardised mortality ratio in intensive care—a pilot study." Anaesthesia 53.11 (1998): 1045-1053.

Rosenberg, Andrew L., et al. "Accepting critically ill transfer patients: adverse effect on a referral center's outcome and benchmark measures." Annals of internal medicine 138.11 (2003): 882-890.