National Breast Cancer Coalition

What are the different types of research studies?

Comparative Effectiveness Research (CER)

Comparative effectiveness research (CER) is a direct comparison of existing options available for treating a particular medical condition. It may compare similar treatments, such as competing drugs, or it may analyze very different interventions, such as surgery and drug therapy. It may also compare the effectiveness of how and when health care is delivered, such as different intervals of follow-up, or schedules of dosing. CER can use a range of research tools and methods, including systematic review of existing studies and evidence, modeling to simulate effects of interventions on different populations, or head-to-head clinical trials comparing one treatment to another.

There are many different types of research studies, some conducted in laboratories, and some in hospitals or clinics. Some studies are observational, while others are "experimental" and involve evaluating interventions. Each has a different design and methods, and each has its strengths and limitations. The type of research question being asked will help determine the best type of research study to conduct.

The descriptions below provide a basic overview of the different types of research studies that are used to collect evidence about breast cancer and its treatment.

Laboratory studies

Laboratory studies can be done using cells from animals or humans, or animal models. Research involving a controlled environment, such as cell cultures in a test tube or in a petri dish, are called in vitro studies. Studies done on living organisms are called in vivo studies.

Laboratory studies can also be referred to as

  • bench science because laboratory experiments are often conducted on a long "bench"
  • basic research because the studies are looking at the fundamentals of how things work and interact
  • preclinical research because the research is conducted prior to research in people in the clinic or in clinical trials

Human studies
There are several different types of research studies that involve humans. These include clinical trials and observational studies. Systematic reviews summarize the results from many different clinical trials and observational studies, and meta-analyses pool data collected from multiple smaller studies in addition to including a systematic review.

Clinical trials

Clinical trials are experimental studies that test new treatments in humans. Clinical trials are used to find out if new treatments work better, the same, or not as well as the standard treatment for the same disease. Clinical trials of experimental treatments (sometimes called "investigational treatments") are conducted in a series of steps called phases.

  • Phase I trials include only a small number of people and are designed to test a treatment's safety, and identify safe dosage levels and side effects.
  • Phase II trials include a larger number of people to test whether a treatment works (may be effective), and to further judge its safety.
  • Phase III trials are randomized trials that enroll hundreds or even thousands of people to test whether a drug is more effective than a placebo or than other drugs currently being used, and to monitor side effects and other information for using the new treatment safely.
  • Phase IV trials are also known as "post-marketing surveillance studies" and are used to monitor the long-term safety and effectiveness of a new treatment among a larger general population.

Randomization and masking are important aspects to clinical trials.

In most Phase III trials and some Phase II trials, patients are divided into at least two groups or "arms." One group of patients gets the new treatment, and is called the "investigational group." The treatment they get is called the "investigational treatment." Another group of patients gets the standard treatment, and is called the "control group." The standard treatment is the treatment you would get if you did not take part in the clinical trial. The two groups are compared to see which treatment works better. In well-designed clinical trials, patients are assigned to the different "arms" of the trial using a system similar to flipping a coin. In other words, patients do not choose whether they will get the new treatment as part of the investigational group or the standard treatment as part of the control group. Grouping patients by chance is called "randomization."

Randomized clinical trials
are considered the "gold standard" for obtaining evidence. The goal of a randomized trial is to enroll two groups of people who are as similar as possible, so that the only difference is whether they get the new treatment or not. Randomized trials are also designed to be ethical. They are only done if we really do not know whether a new treatment is better than the one currently being used.

In some trials, patients know which arm they are in. In other trials, patients do not know which arm they are in. This is called a "masked" or "blind" study.  In some trials, neither doctors nor patients know which arm the patients are in. These trials are called "double-masked" or "double-blind" studies.

Masking helps to minimize information bias. Information bias is when there is a systematic difference in the data collected between the arms of a study. For example, a doctor may provide different care to a patient known to be in the investigational group than to one known to be in the control group.  Or the investigator's expectations may bias the interpretation of results if they know who is receiving the treatment and who isn't.

Observational studies

Observational studies (also called epidemiology studies) are used to examine which risk factors (also called exposures) are associated with an increased or decreased risk that a person will develop a disease, like breast cancer. Unlike clinical trials, researchers just observe the participants and do not "intervene" by giving a treatment. There are several different types of observational studies.

Case-control study

These studies begin by identifying a group of people who already have a disease (cases). Next, the researchers find a comparison group without the disease that is as similar as possible to the people who got sick (controls). After these two groups are identified, researchers ask everyone in both groups the same question. For example, they might ask "Were you exposed to second-hand smoke as a teenager?" The researchers then compare how many of the cases were exposed to how many of the controls were exposed. If many more of the cases were exposed than the controls, then it might mean that exposure to second-hand smoke is linked to the disease.

The problem with case-control studies is that having the disease itself may affect how people answer questions about what happened to them in the past. For example, having been diagnosed with breast cancer may make women more likely to remember their exposure to second-hand smoke than women without breast cancer. This is recall bias. Another limitation of case-control studies is that while they can identify exposures associated with disease, exposure may not necessarily occur before disease. A true experimental clinical trial would be necessary to show that one factor caused another. This is often impractical, as in the case of second hand smoke - noone can be randomly assigned to be exposed to second hand smoke to see if breast cancer develops. However, the next level of epidemiology study - the cohort study - can help provide stronger evidence.

Retrospective cohort study

This type of observational study begins by identifying two groups of individuals who are alike in many ways but differ by a certain characteristic (exposure). For example, female nurses who smoke and those who do not smoke. Researchers then go back in time (retrospectively) and compare the medical records of these two groups of people to look for a certain outcome (disease), such as lung cancer. If many more female nurses who smoke are found to have developed lung cancer than those who did not smoke, then it might mean that smoking is linked to lung cancer.

Retrospective cohort studies have the same limitation that case-control studies do in that it is not always clear whether the exposure occurred before the disease started. In addition, if the people being studied are no longer alive, researchers may have to rely on their family members for information, which may be less accurate.

Prospective cohort study

This type of observational study involves identifying a large number of people, collecting information (exposures) about them at the beginning of the study, either from medical records or through surveys, and then following them over time to see what happens to their health. Prospective cohort studies do allow researchers to know that an exposure occurs before a disease is diagnosed, but that still does not mean that the exposure caused the disease. There could be a lot of other factors involved, including some that can't be measured.

It is always important to remember when looking at results from epidemiology studies - association does not mean causation.

Systematic reviews
These studies are summaries of results from lots of smaller studies. The smaller studies are very carefully chosen from the literature according to strict rules, and the quality of the studies is measured. Systematic reviews are given a lot of weight, and are at the top of the hierarchy of evidence because they look at a whole range of studies and make conclusions about them. However, a review cannot eliminate bias present in the original studies.


A meta-analysis combines the numbers from smaller studies and re-analyzes them. A meta-analysis is often part of a systematic review, but does not have to be. By combining smaller studies, a meta-analysis gives researchers a way to see an effect that smaller studies might not be able to see.

However, it's important that the researchers include only the numbers that are appropriate to combine. For example, a meta-analysis that pools data across all modes of cancer therapy for all types of cancer would be inappropriate because some cancer treatments are effective for certain cancers whereas others are harmful. Combining the results of these studies would yield a meaningless result not applicable to any of the treatments. In addition, sometimes a meta-analysis is not advisable. For instance, when there are too many differences between the studies, there aren't enough studies, or the studies are missing a lot of data.