|
|
|
Indian Pediatr 2009;46: 765-766 |
 |
Severity of Illness Scoring: One Step Closer
to "the Trenches" |
|
Gabriel J Escobar
Perinatal Research Unit, Kaiser Permanente Division of
Research, 2000 Broadway Ave, 021 R10, Oakland, California 94612, USA.
|
|
Although not yet employed in routine
clinical practice, severity of illness scoring has become an essential
tool for health services research involving ill newborns(1,2),
children(3), critically ill adults(4,5), and now even adults admitted to
general medical-surgical wards(6). One way to conceptualize these scores
is that they provide us with a method to learn, within a short time, from
the combined experience of tens of thousands of patient encounters,
something no physician could hope to achieve in a lifetime of practice. In
an era when electronic medical records are starting to become more
widespread, it is conceivable that these scores will eventually become
part of how all of us practice medicine. However, before these scores are
transitioned from health services research papers into "the trenches,"
considerable methodological work still needs to be performed.
The article by Sundaram, et al.(7) in this issue
of Indian Pediatrics is an example of some of this necessary work.
In their study, the authors introduced a novel adaptation of an existing
tool, the Score for Neonatal Acute Physiology, version II (SNAP-II)(1):
varying the T 0. Instead of employing
the original scoring time frame (from NICU admission until 12 hours into
said admission), they employed a variable T0, which they set as the onset
of clinical signs of septicemia. Importantly, they found that, when
employed in this fashion, SNAP-II not only retained its ability to predict
mortality (with a very respectable area under the receiver operator
characteristic curve of 0.82), but also predicted organ dysfunction.
The methodological implications of this study for
future research extend beyond neonatology. Current severity of illness
scores are based on fixed time frames, which are typically bounded on one
end by a highly discretionary event (typically, time of admission to some
hospital unit) and by some time interval (typically 1, 12, or 24 hours) on
the other. Although some work has been done on serial assignment of
severity scores(8,9), variable time frames (particularly when the
variability is based on disease onset) have not been employed. Implicit in
the work by Sundaram, et al.(7) is the notion that future studies (e.g.
randomized clinical trials or retrospective case-control studies) could
employ variable severity scoring time frames. This would be highly
desirable in settings where a significant – and variable – degree of
physiologic instability (with its attendant confounding) is already
present at the time of enrolment, and where enrolment is subject to the
vagaries of "real world" medicine (e.g., future studies on
nosocomial sepsis). Also implicit in their results is the notion that, in
the near future, ad hoc severity scoring time frames could be
employed (e.g., in a hospital with an electronic medical record,
the computer could generate a severity score based on some time frame of X
hours preceding the moment when the clinician hit the "assign severity
score" button).
Certain limitations must temper our excitement. First,
use of variable scoring time frames presumes that a degree of physiologic
uniformity is present in the population being studied. In the case of this
study, this assumption seems solid, and not just on account of the
results: the babies were all of relatively similar chronological and
gestational ages. Second, the score or scores being used must have some
plausibility for the population being studied. Thus, for example, while
using SNAP-II for 3 day old 32 weekers may be reasonable, employing it for
risk adjusting outcomes in 3 month old ex-32 weekers with bronchopulmonary
dysplasia is probably not warranted. However, in principle, the approach
described by Sundaram, et al.(7) could also be used in that
setting, albeit with a different severity score (yet to be invented!).
Funding: None.
Competing interests: None declared.
References
1. Richardson D, Corcoran J, Escobar GJ, Lee S. SNAP-II
and SNAPPE-II: simplified newborn illness severity and mortality risk
scores. J Pediatr 2001; 138: 92-100.
2. Parry G, Tucker J, Tarnow-Mordi W. CRIB II: an
update of the clinical risk index for babies score. Lancet 2003; 361:
1789-1791.
3. Pollack MM, Patel KM, Ruttimann UE. The Pediatric
Risk of Mortality III—Acute Physiology Score (PRISM III- APS): a method of
assessing physiologic instability for pediatric intensive care unit
patients. J Pediatr 1997; 131: 575-581.
4. Zimmerman JE, Kramer AA, McNair DS, Malila FM. acute
physiology and chronic health evaluation (APACHE) IV: hospital mortality
assessment for today’s critically ill patients. Crit Care Med 2006; 34:
1297-1310.
5. Zimmerman JE, Kramer AA. Outcome prediction in
critical care: the acute physiology and chronic health evaluation models.
Curr Opin Crit Care 2008; 14: 491-497.
6. Escobar GJ, Greene J, Scheirer P, Gardner M, Draper
D, Kipnis P. Risk adjusting hospital inpatient mortality using automated
inpatient, outpatient, and laboratory databases. Medical Care 2008; 46:
232-239.
7. Sundaram V, Dutta S, Ahluwalia J, Narang A. Score
for neonatal acute physiology II predicts mortality and persistent organ
dysfunction in neonates with severe septicemia. Indian Pediatr 2009; 46:
775-780.
8. Ferreira FL, Bota DP, Bross A, Melot C, Vincent JL.
Serial evaluation of the SOFA score to predict outcome in critically ill
patients. JAMA 2001; 286: 1754-1758.
9. Fowlie PW, Gould CR, Tarnow-Mordi WO, Strang D.
Measurement properties of the Clinical Risk Index for Babies—reliability,
validity beyond the first 12 hours, and responsiveness over 7 days. Crit
Care Med 1998; 26: 163-168.
|
|
|
 |
|
