You are on page 1of 2

404 LETTERS CORRESPONDENCE

unable to ‘‘comprehend overdose as does not conclude that the only thing ical, social, and psychiatric interven-
more than a pharmacologic event.’’ a heroin addict needs is an injection tions.—JIM CHRISTENSON, MD
We are offended by the statement. of naloxone. Our study merely iden- (jimchris@interchange.ubc.ca), JER-
Our inner-city hospital sees a large tifies objective and reliable criteria EMY ETHERINGTON, MD, GRANT
number of patients with addiction that tell us when it is reasonable to INNES, MD, ERIC GRAFSTEIN, MD,
problems and has a reputation for refer the patient to our social worker KAREN WANGER, MD, CHRIS FER-
compassionate care of these pa- and turn the ED stretcher over to NANDES, MD, JOHN J. SPINELLI,
tients. This does not mean that we the next patient in the waiting room. PHD, MIN GAO, MD, PHD, and
can admit every person presenting We fully endorse the active interven- SARAH PENNINGTON, RN, the St.
with addiction-related problems, nor tions in these areas once the medical Paul’s Hospital Department of Emer-
that we know the secrets to long- event is stabilized. gency Medicine, the Centre for
term maintenance of harm reduc- We hope that readers view our Health Evaluation and Outcome Sci-
tion and abstinence. We have study as one small component of the ences, University of British Colum-
opened beds in our ED specifically care of opioid overdose patients. We bia, Vancouver, BC, Canada
related to addiction, psychiatric ill- believe that it is usually unneces-
ness, and dual-diagnoses. We have a sary to monitor patients in an acute Key words. opioids; heroin; over-
24-hour social worker who assists dose; clinical prediction rule; dis-
care setting for recurrent apnea and
any patients wanting help beyond charge; prediction.
onset of adult respiratory distress
the ED to connect with other social
syndrome when they have normal
agencies. We are working on a study
to identify strategies to overcome
temperature, respiratory rate, oxy- Reference
the issues of undertreatment of pa- gen saturation, Glasgow Coma Scale
score, heart rate, and ability to mo- 1. Christenson J, Etherington J, Graf-
tients with HIV, particularly in the stein E, et al. Early discharge of patients
intravenous drug-using population. bilize one hour after administration
with presumed opioid overdose: devel-
We have community meetings with of naloxone. Such information does opment of a clinical prediction rule. Acad
social agencies, clinics, and primary not preclude other appropriate med- Emerg Med. 2000; 7:1110–8.
care physicians to identify how to
improve our care of inner-city pa-
tients, particularly those without
homes or those addicted to drugs
and alcohol. This study is only a Reliability and Validity of Diagnostic Tests
small piece of the overall under-
standing of these vexing problems,
Fridriksson et al. recently conducted (i.e., the proportion of subject with
but nevertheless, it is important to
a pilot study assessing serum neu- the disease who have a positive test)
understand the incidence of serious
complications of opioid overdose and ron-specific enolase (NSE) as a pre- and specificity (i.e., the proportion of
the predictors that determine dictor of intracranial leasions (ICL) subjects without the disease who
whether a patient is safe from the in children with head trauma.1 The have a negative test).
pharmacologic consequences of this authors make an important error It is desirable for a diagnostic
overdose event. We believe that pre- describing the objectives of their test to be both highly sensitive and
dicting future overdoses would be an study. They note, ‘‘The objective of highly specific. This is not usually
admirable goal but one that we did our pilot study was to evaluate the possible as increasing sensitivity de-
not attempt to tackle. reliability of serum NSE in predict- creases specificity, and vice versa.
Su and Hoffman justly state that ing ICL identified on head CT in Different testing strategies have dif-
these patients often require other children with acute blunt head ferent requirements for sensitivity
support. We agree. After the over- trauma.’’ However, the authors were and specificity.3 Since highly sensi-
dose, heroin addicts need psycholog- not assessing the ‘‘reliability’’ of this
tive tests are rarely negative in the
ical support, safe housing, counsel- diagnostic test. In fact, they were as-
presence of disease, these types of
ing, detoxification, rehabilitation, sessing the test’s ‘‘validity.’’
tests are particularly useful during
training, and reintegration into the Both reliability and validity are
early stages of a diagnostic workup
community. Someday, emergency measures of test performance, but
or when there is reason to suspect a
physicians may have the expertise they have separate and distinct
meanings.2,3 Validity is the degree to dangerous illness. Highly specific
and resources to provide these im-
which the data measure what they tests are rarely positive in the ab-
portant components of care. At pres-
were intended to measure (i.e., ac- sence of disease. Thus, these types of
ent, we do not, and these functions
curacy) when compared with a true tests are helpful for confirming a di-
do not occur in the ED. While they
are important, they are peripheral measure, otherwise know as a cri- agnosis that is suggested by clinical
to the study question. Our research terion standard or ‘‘gold standard.’’ data or other tests.
question was: Is it possible to iden- The term validity is also used when In contrast, reliability is the ex-
tify patients who are safe (based on referring to the ability of a test to tent to which repeated measure-
medical, not sociological, outcome classify subjects into dichotomous ments get similar results (i.e., pre-
measures) for early discharge after categories of diseased versus non- cision or reproducibility).2,3 It refers
opioid overdose? Contrary to what diseased. In this situation, validity to the capacity of a test to give the
Su and Hoffman suggest, our study is described in terms of sensitivity same result on repeated application.
ACADEMIC EMERGENCY MEDICINE • April 2001, Volume 8, Number 4 405

Reliability depends on 1) the varia- cific to be valid.2 Key words. reliability; validity; di-
bility in the specific manifestation Confusion between validity and agnostic tests; sensitivity; specificity.
on which the test is based, 2) the reliability, as with many clinical ep-
variability in the method of mea- idemiology terms, is very common. References
surement, and 3) the skill with Nevertheless, inaccurate use should
which the measurement is made. be avoided whenever possible as it 1. Fridriksson T, Kini N, Walsh-Kelly C,
Hennes H. Sterum neuron-specific eno-
One measure of reliability, for ex- clouds important distinctions of con- lase as a predictor of intracranial lesions
ample, would be the degree of dis- cepts concerning clinical research in children with head trauma: a pilot
persal of repeated measurements and medical decision making that study [brief report]. Acad Emerg Med.
2000; 7:816–20.
characterized as variance or stan- are crucial for veracious discourse
2. Morrison AS. Screening. In: Rothman
dard deviation. While reliability among clinicians. — DEMETRIOS N. KJ, Greenland S (eds). Modern Epide-
does not imply nor guarantee valid- KYRIACOU, MD, PHD, Division of miology, 2nd ed. Philadelphia: Lippin-
ity (i.e., high sensitivity and specific- Emergency Medicine, Northwestern cott-Raven, 1998.
3. Fletcher RH, Fletcher SW, Wagner
ity), unreliability will probably make University Medical School, Chicago, EH. Clinical Epidemiology, 3rd ed. Bal-
a test insufficiently sensitive or spe- IL timore: Williams & Wilkins, 1996.

You might also like