LWBK1006-10 LWW-Govindan-Review November 24, 2011 11:21
Chapter 10•Health Services Research and Economics of Cancer Care 119ANSWERS
Answer 10.1. The answer is C.
In an effort to assess effectiveness in a real-world setting, health ser-
vices research studies are frequently observational in nature. Random-
ized trials typically establish efficacy of a specific intervention, and
meta-analysis of randomized trials enhances the precision of efficacy (or
toxicity) estimates.Answer 10.2. The answer is B.
Multivariable statistical models will not control for unmeasured variables.
This can introduce confounding, which can result in abberant conclusions
in an observational study. Patient consent is often not required in health
services research studies, as their completion would frequently be impos-
sible with a consent requirement. Randomization is seldom used in health
services research, instead researchers must try to control for the factors
that predicted the nonrandom assignment of treatment. Sample sizes from
administrative databases are typically large, allowing sufficient statistical
power to detect differences between groups.Answer 10.3. The answer is B.
SEER provides adequate baseline information for most observational
studies and is valuable in that it can be linked to Medicare claims data
for treatment details and correlated to populations to allow for estimates
of incidence rates. However, one limitation of SEER is that limited longi-
tudinal follow-up information is available in the registry.Answer 10.4. The answer is D.
Quality can be measured by “structural” metrics, which include charac-
teristics of organizations and providers, as well as by “process” metrics,
which are usually defined as adherence to established clinical guidelines.
Only rarely are clinical outcome measures used to evaluate quality of care
delivery.Answer 10.5. The answer is B.
One of the biggest threats to volume-outcome studies is the inability to
control for hospital case-mix. In some cases, high-volume hospitals are
likely selecting for better-risk patients, thereby improving survival out-
comes. There are often no clear thresholds to differentiate high- and low-
volume centers. Strongest associations between volume and outcome are
seen with high-risk procedures. Associations are seen with both immedi-
ate and long-term outcomes.Answer 10.6. The answer is D.
Because many cancer treatments often have significant toxicity with only
modest impact on overall survival, outcomes of interest to patients,