Types of Epidemiologic Studies

Cohort. Cohort studies examine the difference in disease occurrence over time between exposed persons and unexposed persons. Cohort studies may be prospective (exposure information is collected at the beginning of the study and continued until the end of the follow-up period) or retrospective (exposure information is collected from historical records) (lARC 1999). Cohort studies of cancer tend to be retrospective because most cancers have long latency periods between exposure and disease onset. This study design is commonly used for cancer investigations 

TABLE 15.5. Five-Year Survival Rates for Selected Cancers—United States, 1996-2004 (SEER 2008) 

Mortality data are frequently used in cohort studies as opposed to incidence data because of the relative ease of obtaining information on deaths. In particular, the advent of the National Death Index has made mortality data more readily available. While all states maintain registries of incident cancer cases, many of the registries are relatively new and data quality can vary from state to state. Investigators conducting follow-up studies are required to comply with each state s requirements for use of the data. For cancers such as pancreatic cancer where survival is poor, mortality data is an excellent surrogate for the risk of the disease. For other cancers where the survival is much better, such as testicular cancer, mortality is a poor estimator of incidence. Table 15.5 describes the 5-year survival for several selected cancer sites for the period 1996-2004. As evident from the table, there is a considerable difference in survival rates for cancer of different sites. The 5-year survival rate for pancreatic cancer was only 5.1% compared to a survival rate of 98.9% for prostate cancer (SEER 2008). 

To compare the incidence or mortality for a disease of interest between two populations, the data are adjusted to allow for differences between the two populations with respect to age, gender, race, and other variables. The adjustment may be direct or indirect. The reader is referred to WHO (1999) for a description of the two methods. The indirect method, which is more commonly used, generates measures of association known as standardized mortality ratios (SMRs) and standardized incidence ratios (SIRs). These ratios compare the mortality or incidence for the disease of interest in the population of interest with that which would be expected based on an external population. 

In a proportionate mortality or proportionate incidence study, one compares the proportion of deaths or incident cases due to a condition of interest with that expected based on deaths or incident cases in an external, usually the general, population. When the proportions of causes of death are compared, the ratio is known as the proportionate mortality ratio (PMR). The comparison of proportions of incident cases is known as the proportionate incidence ratio (PIR). Proportional measures can be misleading since a decrement in the proportion of deaths or incident cases due to a particular cause will de facto lead to an increase in the proportion due to another cause. The commonly held view with regard to PMRs is that they are good approximations to SMRs from cohort studies when the cohort s all-canses combined SMR is equal to 1.0 (Checkoway et al. 1989). The odds ratio has also been used as a measure of association in PMR or PIR studies and may be a more appropriate measure of association for evalnating proportional measures (WHO 1999). 

---

Cohort Study—A type of epidemiological study of a specific group or groups of people who have had a common insult (e.g., exposure to an agent suspected of causing disease or a common disease) and are followed forward from exposure to outcome. At least one exposed group is compared to one unexposed group. 

Cross-sectional Study—A type of epidemiological study of a group or groups which examines the relationship between exposure and outcome to a chemical or to chemicals at one point in time. 

Case-Control Study—A type of epidemiological study which examines the relationship between a particular outcome (disease or condition) and a variety of potential causative agents (such as toxic chemicals). In a case-controlled study, a group of people with a specified and well-defined outcome is identified and compared to a similar group of people without outcome. 

Three types of epidemiological studies of cancer contribute to the assessment of carcinogenicity in humans—cohort studies, case-control studies and correlation (or ecological) studies. Rarely, results from randomized trials may be available. Case series and case reports of cancer in humans may also be reviewed. 

---