risk factors Nested A case control study performed case-control on subgroups within a cohort study study EXPERIMENTAL STUDIES Randomized A group of subjects is controlled randomly assigned to an trial experimental arm or a comparison arm The comparison arm may be placebo or an accepted, standard treatment Metaanalysis and systematic review Data from multiple clinical studies are pooled and assessed for similar findings Results from each study are critically appraised, compiled, and analyzed prior to making a summary conclusion Allows better ascertainment of risk factors leading to less bias in sampling Ability to study rare occurrences within a cohort Highest grade of evidence Can establish causality Prospective allowing standard definitions and measurement Randomization markedly reduces risk of bias or systematic differences between the groups May allow blinding to further reduce bias Any outcome difference between groups should be attributable to the experimental intervention Able to raise the certainty of a borderline finding in multiple smaller studies Decrease the effect of studies with limited validity and inconsistent results Similar to standard case control limitations Expensive and resourceintensive Require additional expertise in study design Often require large group to be powered to detect differences in outcomes, adding to cost and resource utilization Specialized, complex statistical analysis required Subject to the quality of studies that comprise them Observational Studies Observational studies are nonexperimental in nature (see Table 24.2) The phenomenon of interest is observed without imposing experimental or controlled conditions They are variable regarding the quality and quantity of evidence they provide, and they have the potential for bias and error, both known and unknown However, they are often more efficient, less expensive, and more achievable than a clinical trial Carefully performed observational studies may be valuable in determining long-term outcomes and associated factors, in describing rare adverse events or rare disorders, and in determining the characteristics of diagnostic tests Case Reports and Case Series These studies can be useful in identifying unique clinical observations (see Table 24.2) Case reports are often short communications, detailing either a single patient or a limited group of patients They can provide significant detail into individual or small group experience that may provide helpful information in very rare occurrences However, they have no control group and limited ability to apply the findings to a larger population An example of this would be a report detailing the full medical and surgical history, catheterization hemodynamics, and echocardiographic findings for two patients with plastic bronchitis and protein-losing enteropathy after a Fontan procedure These patients may provide clues into links between these two diseases, but it is unclear if any of this information is applicable to other Fontan patients or just represent very unusual circumstances in these individuals Cohort Studies These studies are often employed to describe the characteristics, management, and outcomes for a defined population (see Table 24.2); for example, all Fontan patients followed at a single center Cohort studies can be solely descriptive in nature, or they can look for associations between characteristics and outcomes; for example, examining late-term survival in all Fontan patients at a single center stratified by systemic right versus left ventricle Some of the characteristics may be factors in management; for example, late-term survival in all Fontan patients stratified by lateral tunnel Fontan versus extracardiac conduit Fontan Because these factors are not randomly assigned, important bias is often introduced into any comparison of strategies for treatment since many factors also influence decisions regarding the selection of the treatment and possibly outcomes For instance, it is unlikely, outside of an experiment, that surgeons at the center would randomly choose a lateral tunnel Fontan over an extracardiac conduit Fontan Instead, anatomic variants and surgical expertise with a particular procedure are likely involved in the decision making; this introduces bias and nonrandom error into the study, since these factors may also be associated with the outcome Several types of statistical adjustments are available to help minimize bias in these comparisons, but they often provide incomplete adjustment and adjust only for characteristics known and measured by the investigators The first step in designing a cohort study is to define and assemble the cohort The cohort must be relevant to the question at hand and representative of the target population for which the investigator wishes to extrapolate the results The study may be prospective or retrospective in nature In a prospective study, the investigator defines the cohort and then follows it over time, with tracking and assessments made to identify those subjects who develop outcomes This design has advantages in that multiple potential factors can be assessed at baseline with standardized measures, ensuring some control of quality Risk factors can be assessed dynamically, identifying those that are acquired over time, particularly those relating to management In addition, multiple types of outcomes can be assessed dynamically This type of design is useful for determining the incidence of outcomes and for determining potential causal associations between risk factors and outcomes The design is efficient if the prevalence of risk factors and outcomes is relatively common and the time over which outcomes may occur is short An example of this would be as indicated earlier, identifying all Fontan patients who had surgery in the previous 5 years at a center and then following them forward in time (concurrently) to determine what anatomic and surgical factors are associated with mortality, as well as tracking secondary outcomes such as heart transplantation, the development protein-losing enteropathy, or liver fibrosis Studies of cohorts can also be retrospective in nature In a retrospective cohort study, the outcome status of a group of patients is determined, and then an assessment of risk factors is pursued through the nonconcurrent assessment of available sources of data (the “chart review”) or recall of the subjects themselves These studies are useful in giving an estimate of the prevalence of outcomes and conditions They are often more efficient than prospective studies in that the cohort has usually been assembled and characterized in the past and the follow-up has occurred The potential biases, however, are much greater, and the quality of the available data may be suboptimal An example of a retrospective cohort study would be examining all Fontan patients in the last 25 years at a center, assessing their current living status, performing a chart review to characterize factors of interest in each patient, and then seeking associations between patient characteristics and survival Cross-Sectional Studies Cross-sectional studies are a type of study whereby a cohort is assembled, and