SPECIFIC CLINICAL STUDY DESIGNS
A. Case report
1. Description. A case report is a brief, objective report of a clinical characteristic or outcome from a single clinical subject or event.
2. Study questions. A case report can address almost any clinical question )r issue, including screening test results or treatment outcomes or natural history findings. It is commonly used to report unusual or unexpected events, such as adverse drug reactions.
a. A single noteworthy event must first be identified.
b. Data collection is generally retrospective with a review and a descriptive summary of subjects or events.
c. No statistical analysis or comparison group is included in the design.
d. Although few conclusions can be drawn based on evidence from a single event or observation, they:
(1) Are valuable for first report of unexpected findings.
(2) Can generate hypotheses for testing.
(3) Can define issues for further study.
4. Strengths and limitations. A case report is often the first evidence of an unexpected or unusual event, but rarely are the results generalizable.
a. A report of advanced proliferative diabetic retinopathy in a patient with no other clinical evidence of diabetes is an appropriate subject for a case report.
b. The initial report of phocomelia in a newborn of a woman on thalidomide is an important example of how a single case report can trigger further important investigations.
B. Case series report
I .Description. A case series report is an objective report of a clinical characteristic or outcome
from a group of clinical subjects.
2. Study questions. A case series report can address almost any clinical problem, including screening test results or,treatment outcomes and natural history findings. However, it is most commonly used to describe clinical characteristics, such as signs and symptoms of disease or disease outcomes-for example, the natural history of a series of patients with specific disease.
a. Subjects must be identified with regard to the clinical events or characteristics in question.
b. Data collection may be retrospective or prospective, but a comparison or control group is usually not included.
c. Descriptive statistics are calculated to define the proportion of subjects with the characteristics under study. Results are often incorrectly compared with the results from other populations or studies of similar subjects.
d. Conclusions generally are limited because there is no comparison group within the study.
4. Strengths and limitations
a. Because the selection of study subjects is often unrepresentative or otherwise biased, the generalizability of the results is often limited. The denominator-that is, the population from which the subjects were drawn-rarely is defined, nor is the selection criteria for subjects, which further limits generalizability.
b. The lack of a control or comparison group also limits the generalizability of results.
c. Conclusions are often incorrectly considered to be generalizable and valid based on the large number of study subjects.
d. Study results are strengthened when a consecutive series of subjects, that is, all eligible subjects, are included over a specified period of time.
a. Intraocular pressure control in 100 consecutive patients with primary open-angle glaucoma who were treated with laser trabeculoplasty and followed for 1 year is an appropriate sub ect for a case series report.
-b. identification of several children born to mothers who had taken thalidomide, which gave evidence of possible association of this drug with birth defects, exemplifies a case series report.
C. Incidence and prevalence studies
1. Description. Incidence and prevalence studies are actually a type of case series report but differ in that the entire study population is well-defined and then uniformly surveyed regarding the parameters in question.
2. Study questions. Usually incidence and prevalence studies concern the occurrence of disease, but they also address the rate of other events, such as the adverse side effects of drugs or even death.
a. Incidence is the occurrence of an event or characteristic over a period of time-for example, the rate of staphylococcal food poisoning following lunch in a particular restaurant. Incidence is used to:
(1) Describe the rate of disease occurrence over time.
(2) Assess survival-that is, the incidence of death over time or at a specific time after follow-up-for example, the 5-year survival rate for breast cancer.
(3) Compare risk of disease between two or more populations.
b. Prevalence is the presence of an event or characteristic at a single point in time-for example, the rate of previously undiagnosed glaucoma in a population of elderly individuals screened for ophthalmic disorders . Prevalence is used to:
(1) Describe the burden of disease, especially undiagnosed disease.
(2) Define the rate of clinical characteristics in subjects with a specified disease-for example, weight loss in patients presenting with terminal disorders.
a. A target population to be followed over a period of time for an incidence study or to be surveyed at a single point in time for a prevalence study must be identified.
b. Events or characteristics must be measured and recorded by periodic reassessment over time for the incidence study and at a particular time for the prevalence study.
c. A rate, the occurrence of the measured target event over the entire population, must be calculated.
d. No statistical analytic component is included although rates within populations or subpopulations can be compared.
4. Strengths and limitations. True rates are determined, which can serve as comparison measures between populations. However, they may provide poor estimates of infrequent or rare events. They may be adversely affected by sampling, which may be unrepresentative. They usually require description of population characteristics, such as age, race, and sex, to be meaningful.
a. Incidence. The rate of colon carcinoma developing in an elderly population over I year is an appropriate subject for an incidence study.
b. Prevalence. The rate of unrecognized heart murmur in children presenting for their first preschool physical is an appropriate subject for a prevalence study.
D. Case-control study
1. Description. A case-control study is an observational or descriptive analytic study in which diseased and nondiseased or affected or nonaffected subjects are identified after the fact and then compared regarding specific characteristics to determine possible association or risk for the disease in question.
2. Study questions. Most often a case-control study is used to address issues of the risk of association for disease-that is, the differences between diseased and nondiseased populations in the characteristic under investigation-for example, a comparison of the rate of cigarette smoking between those individuals who have lung cancer and those who are cancer free. They are also used in clinical decision analysis to assess the differences between diseased and nondiseased populations in test positivity-for example, the likelihood that patients with uri-
nary tract infections have greater than 10 white cells in unspun urine samples before treatment compared to those without infection.
a. Diseased and nondiseased populations must be identified, usually retrospectively.
b. The prevalence of characteristics under investigation and other characteristics that possibly may be related to the presence of disease and characteristic under study must be assessed.
c. Statistical comparison can be made between study groups of the characteristics under investigation to assess the likelihood that differences in these characteristics between diseased and nondiseased groups are real and not due to chance (alpha error).
d. Conclusions are useful for the generation of hypotheses and for initial evidence of putative risk associations. Results cannot be used for define causality.
4. Strengths and limitations
a. A case-control study is relatively easy and inexpensive to conduct since prospective or long-term follow-up is not required.
b. There is a potential for bias in the selection of subjects since a case-control study is not population based.
c. Bias in data collection may also occur since the presence or absence of disease is usually known to the subject and may be known to the study observer (unmasked). Bias may also influence the recall of previous exposure by the subject if possible associations are known to him or her, such as the association between cigarette smoking and lung cancer.
d. The incidence rate of disease in a population cannot be determined nor compared between populations to assess possible risk (relative risk). However, the odds ratio, which provides an estimate of the relative risk of characteristics between diseased and nondiseased populations can be calculated.
a. A comparison of prior estrogen use in uterine cancer cases compared to age-matched controls without cancer to assess possible risk for exposure to estrogens is an appropriate subject for a case-control study.
b. The case-control study comparing thalidomide ingestion in mothers of children with phocomelia with ingestion in mothers with normal children clearly demonstrates that thalidomide ingestions were more common in mothers with affected children than in mothers with children who did not have phocomelia.
E. Cohort study
1. Description. A cohort study is an observational or descriptive analytic study in which exposed and nonexposed populations are identified and followed prospectively over time to determine the rate of a specific clinical disease or event.
2. Study questions. Most often a cohort study is used to address issues of risk for the development of disease between populations exposed and not exposed to a factor under study-for example, a population of smokers and nonsmokers are followed over time to provide comparison rates between smokers and nonsmokers for lung cancer or heart disease. A cohort study also can be used in clinical decision analysis to assess the predictive value of test positivity or negativity-for example, to determine the proportion of patients with positive sedimentation rates at screening who have a diagnosis of cancer on a subsequent definitive workup.
a. A population of exposed and nonexposed (or test positive and negative) individuals must be identified and followed prospectively.
b. Exposure to the risk factors under study and other potentially associated variables must be quantitated initially and over time.
c. The incidence of the target event, such as cancer, over time must be measured.
d. The incidence rate of an event for both exposed and nonexposed populations must be calculated. This calculation can then be compared to determine the relative risk and incidence in exposed and nonexposed individuals. Statistical comparison of the rates of disease occurrence or outcome parameter between the exposed and nonexposed groups allows one to assess the likelihood that the observed differences are real and not due to chance (alpha error). In addition, the attributable risk can be determined which is simply the difference in the incidence of the event or disease between exposed and nonexposed populations. While relative risk is commonly used to assess possible etiology, attributable risk is used to measure the burden of disease in a population.
4. Strengths and limitations. Although costly and time-consuming due to a large number of subjects often needed and the prolonged duration of time required for follow-up of the disease occurrence, a cohort study allows for determination of a population-based rate of the event under question and the relative risk. Potential bias in recall and observations is lessened since exposure can be determined prior to the onset of disease or event. As with a case-control study, causality cannot be determined by results from this type of study since other important criteria must be met.
a. Follow-up in a population of adults exposed and not exposed as children to radiation of the neck to assess risk from thyroid cancer is an appropriate subject for a cohort study.
b. The cohort study of physicians comparing the incidence of lung cancer between individuals who smoked and those who did not was a landmark study in defining the association of cigarette smoking and lung cancer.
1. Description. A clinical trial is an experimental design used to assess differences between two or more groups receiving different interventions or treatments.
2. Study questions. A clinical trial is usually employed to compare outcomes between different treatments, such as antibiotic treatments for a specific disease or chemotherapy for a specific cancer. It can also be used in clinical decision analysis to compare outcomes, such as when comparing the differences in mortality from cancer among populations receiving different screening interventions.
a. A population with a clinical characteristic requiring intervention must be identified.
b. Subjects must be allocated, preferably randomly, to each of the treatment interventions. Treatments are administered in an identical or controlled manner to ensure uniformity of nontreatment covariants, which may affect outcomes.
c. Treatment outcomes and other results, such as side effects, costs, or benefits, must be measured. Preferably observations should be made while observers and patients are masked to the type of interventions (double masked or blind).
d. Rates of measured outcomes between the different treatment groups can be statistically compared.
e. The strongest evidence of differences in clinical outcome due to treatment effect are provided by the conclusion.
4. Strengths and limitations. A clinical trial allows for control of other clinical variables, which can also affect outcomes under investigation. Randomization minimizes the potential adverse effect from systematic error (bias). However, unmasking either the observer or subject often leads to biased observations and may invalidate the results. An insufficient number of subjects may lead to failure to detect true differences that may exist (beta error).
a. Comparison of chemotherapy versus chemotherapy plus radiation for laryngeal carcinoma is an appropriate topic for a clinical trial.
b. The recent randomized, controlled clinical trial for breast cancer showed that radical mastectomy was in some cases no more effective than lumpectomy.
From Cassens, Brett J., Preventive Medicine and Public Health, pg37