Odds RATIOS and relative risks are commonly used to express results in clinical studies.

The
results of cohort studies and case-control studies are best expressed as relative risks and odds
ratios, respectively. The use and interpretation of these ratios are the subject of this brief
review. In a cohort study, a group of individuals exposed to an agent are compared with an
appropriately selected control group of individuals who are not exposed. Both groups are
observed until an event of interest occurs or for a prespecified time period. The association of
exposure and outcome is expressed as the relative risk (Table). Assuming that a study is
unbiased, the relative risk is interpreted as follows: A relative risk of 1 implies no association.
If the relative risk is greater than 1, and its 95% confidence interval does not include 1, then
the result implies a positive association between exposure and outcome at the 5%
significance level (ie, the outcome is more likely in the exposed cohort) (Table, study 3). If
the relative risk is less than 1, and its 95% confidence interval does not include 1, then the
result implies a negative association between exposure and outcome at the 5% significance
level (ie, the outcome is less likely in the exposed cohort).1,2 If the 95% confidence interval
includes the relative risk of 1, then an association of exposure and outcome is not proved by
the study at the 5% significance level (Table, studies 1 and 2).
Case-control studies are used when an outcome is recognized and the exploration of a
suspected causative agent is at an early stage, because case-control studies are cheaper and
easier to conduct than cohort studies. They are also useful if there is a very long time lag
between exposure and outcome or if the outcome of interest is very rare. In a case-control
study, patients with an outcome of interest are compared with appropriately selected controls
without the outcome. The odds of exposure to suspected etiologic agents are ascertained in
cases and controls (Table).
See also page 775
The odds of an event are the ratio of the number of events to the number of nonevents. The
odds are equal to the probability divided by 1 minus the probability (odds = probability/[1 −
probability]). The formula probability=odds/(odds+1) is used to convert odds back to
probability. An event is defined by the presence of an exposure to a suspected causal agent or
risk factor in a case-control study. For example, the odds that the cases in study 1 (Table)
were exposed to the factor is 4 (4/1). The probability that the cases in study 1 (Table) were
exposed to the factor is 0.8 (4/[4+1] or 4/5). The odds of exposure of cases are divided by the
odds of exposure among controls to derive the odds ratio. The association between exposure
and outcome is expressed as the odds ratio in case-control studies. Assuming that a study is
conducted in an unbiased manner, the odds ratio is interpreted as follows: An odds ratio of 1
implies no association. If the odds ratio is greater than 1, and its 95% confidence interval
does not include 1, then the result implies a positive association between exposure and
outcome at the 5% significance level (ie, the odds of exposure is greater in cases than in
controls) (Table, study 3). If the odds ratio is less than 1, and its 95% confidence interval
does not include 1, then the result implies a negative association between exposure and
outcome at the 5% significance level (ie, the odds of exposure is smaller in cases than in
controls).1,2 If the 95% confidence includes the odds ratio of 1, then an association of
exposure and outcome is not proved by the study at the 5% significance level (Table, studies
1 and 2). For example, suppose a case-control study was performed to study the relationship
between limb deformity and exposure to thalidomide. The results of the study indicated that
patients with limb deformities were more likely than controls to have been exposed to
thalidomide in utero. The odds ratio for thalidomide exposure was 3.5, and the confidence
interval was 1.8 to 6.6. Thus the odds that patients with limb deformities were exposed to
thalidomide in utero were 3.5 times the odds of thalidomide exposure in controls. Since the
odds ratio is greater than 1, and its 95% confidence interval does not include 1, the result
implies a positive association between thalidomide use and limb deformities. These results
(an odds ratio of 3.5 and confidence interval of 1.8 to 6.6) were actually the results of a study
by Wolf et al, who studied the relationship between sunscreen use and melanoma in a case-
control study in Austria. Their results indicated that patients with melanoma were more likely
than controls to have used sunscreen often.
Researchers often use the results of a case-control study to infer a causal relationship between
the exposure and the outcome. For example, the results of the sunscreen melanoma study
mentioned above, if deemed to be unbiased, can be interpreted to imply that the odds of
developing melanoma is 3.5 times higher in people who use sunscreen often compared with
those who do not use sunscreen. The practice of using a case-control study to infer the odds
ratio of outcomes in exposed and unexposed cohorts is valid because the odds ratio obtained
retrospectively in a case-control study is mathematically equivalent to the odds ratio that
would have been obtained in a prospective (cohort) study (Table, sample [ad/bc]). It is
important to remember that the ratio of the odds of exposure to a suspected etiologic agent in
cases and controls is actually determined in case-control studies. Ascribing a causal
relationship between exposure and outcome is an interpretation of the collected data. This
practice is valid only if the study is free of biases. Other features of case-control studies that
strengthen the validity of this interpretation include a strong association (ie, a high odds
ratio), a dose-response gradient, consistency among studies, and biological plausibility.
Technically, relative risk should not be used to express results in case-control studies because
the disease prevalence is not known and the apparent relative risk is dependent on the number
of controls chosen. However, the odds ratio is a reasonable approximation of the relative risk
when the outcome is relatively rare (eg, when less than 1% of people exposed to an agent
develop disease) (Table).4 Using the odds ratio as an approximation of the relative risk
produces progressively larger errors as the outcome rate rises above 1% (Table). To add
confusion to an already difficult area, clinical researchers will often report results of meta-
analyses, cohort studies, and randomized controlled trials using odds ratios. Odds ratios are
used because they have more desirable statistical properties than other measures.4 For
example, odds ratios can take any value between 0 and infinity, are symmetric, and can be
used to make adjustments for confounding factors using multiple regression.4 Unfortunately,
they are the measure least intuitively understood. If a meta-analysis, controlled trial, or cohort
study is reported using odds ratios, it is often possible to calculate the relative risk, difference
in response rates, or number needed to treat if the primary data are provided. Alternatively,
these more readily understood measures can be derived if the number of subjects in each
group, odds ratio, and overall outcome rate are provided.
remake nests in thesamelocation. Treatment of an affected gerbil is difficult, partly because
of the emerging resistance of avian mites to many of the insecticides currently available.
However,thepetownershouldbeencouragedtocontact a veterinarian familiar with exotic pets
and discuss the diagnostic and therapeutic options currently available. In addition, house
cleaning, fumigation, and removal of old cages and bedding material should be helpful. In
summary,webelieve that these are the first reports of clinical dermatitis from avian mites
acquired from pet gerbils.Wepostulate that the gerbils involved with patient 1might have
acquiredOsylviarum within the environment of the pet store in which they were kept. This
suspicion is fortifiedbythe observations ofoneofus (J.D.A.) whoknows of several other
subsequently affected gerbils in the same neighborhood in Cincinnati. In fact,2other
dermatologists in the Cincinnati area discoveredandtreated4families with similar “bites” from
infested gerbils (Debra Anderson,MD, and Paul Lucky,MD,oral communication, 1999-2000).
In the case of our patient 2, it is proposed that the classroom chickens shared their avian mites
with the classroom gerbils. We propose that because the association of avian mites and
gerbils has never been reported, such cases have gone unrecognized or misdiagnosed as bites
from other arthropods such as fleas or scabies mites.