Clear Search

The Untapped Potential of Observational Research in Clinical Care

By BCRF | May 22, 2017

BCRF researcher Dr. Visvanathan shares her thoughts on incorporating observational studies into evidence-based clinical care

The care a patient receives during and after cancer is based on clinical evidence obtained through precisely conducted clinical trials known as randomized controlled trials (RCT). The strength of RCT design is in the randomization of patients to interventions, the use of control groups and the levels of control imposed by the trial which reduce the effects of confounding factors and other biases. Despite a reliance on RCTs to guide clinical care, there are inherent limitations that affect the validity of clinical trial outcomes to real world interpretation.

A key limitation to RCTs is that only a subset of patients, a group usually enriched for patients with specific disease or treatment profiles, participate in clinical trials. That means that results are not necessarily translatable to a larger, more diverse patient population. The controlled nature of the treatment interventions, which are important to accurately interpret the trial outcomes also do not necessarily reflect the real-world experience, nor do surrogate endpoints that may not be validated with long term survival, quality of life, or patient reported outcomes data.

A recent report issued by the American Society of Clinical Oncology (ASCO) highlights the untapped potential of observational research to augment clinical trial results to better inform clinical decision making. The report explains how observational research, studies that observe outcomes as opposed to those with interventions designed to influence a specific outcome, can answer questions that are not possible in the structure of a clinical trial, generate new hypotheses that can be tested in clinical trials and conduct post market surveillance of new therapies. The report also acknowledged current challenges in utilizing observational research and offered several recommendations and action items to facilitate the incorporation of observational studies into recommendations of clinical decision making.

BCRF investigator, Dr. Kala Visvanathan, Professor of Epidemiology and Oncology at Johns Hopkins Bloomberg School of Public Health and Sidney Kimmel Cancer Center, Johns Hopkins School of Medicine is co-chair of the report and spoke to BCRF about the significance of the report, its recommendations and next steps.

BCRF: What do we mean by observational research and what spurred the report and recommendations from ASCO?

Dr. Visvanathan: Very simply, observational research is the study of populations over time. It’s been part of medical research, particularly cancer research for a long time, primarily in understanding the underlying relationships between exposure and outcomes, for instance risk factors for breast cancer, such as diet, exercise and other environmental exposures or population traits (age, race, etc.).

Well conducted randomized controlled trials (RCTs) are the gold standard for clinical care guidelines (i.e. decisions about treatment or interventions). By the use of randomization–a hallmark of this design–they are most often able to balance out unmeasured differences between the groups being compared. Observational studies on the other hand have to control for differences between individuals, so there has been some concern that they cannot provide the same level of evidence to guide clinical decision making.

Over time, we’ve come to appreciate that observation studies can generate new hypothesis, study novel endpoints, and help to study greater portions of the populations- racial differences or age differences in response to therapies or other interventions, for instance. They are more difficult to conduct because they do not take place under controlled conditions, but well done observational studies can be very informative- not just in understanding why things happen but in understanding outcomes and practices. They offer other benefits as well in that they can utilize existing data, allowing them to be more timely, efficient and less expensive, and they allow for repeated observations over time, evaluation of multiple outcomes and long term follow up. It can take 10-20 years before we can report on some outcomes.

BCRF: Now that ASCO has made its recommendations for incorporating observational studies into evidence-based clinical care, what’s next?

Dr. Visvanathan: We now need to think about how to implement the recommendations. The process is really in its early phases but needs to accelerate rapidly with the widespread introduction of electronic medical records (EMR) that can provide exciting new opportunities on a large scale. EMR can be a good input for observational studies, but there are still significant barriers including lack of standardization and data sharing that make the information housed in EMR and other databases difficult to extract.

We have the technology to standardize data collection but we need to do it quickly so that we have high quality data that can be used to address important clinical questions sooner rather than later.

We also need to educate clinicians on how to conduct good quality observational studies.  A lot of clinicians are still not thinking about well conducted observational studies as an adjunct or compliment to clinical trials.  What the ASCO report is saying is that there is value to this and we all have to work together. Ultimately the goal is to answer the question the best way possible and not all questions can be answered in the context of randomized trials.  This can be done either in the form of correlative studies imbedded in clinical trial design or in collaboration with epidemiologists conducting observational studies.

It is important to stress however, that a poorly designed observational study is not helpful. It takes the right expertise, as well as validated and standardized source information. The source of the data as well as the study design are vital to obtain strong evidence that can guide clinical care.

Another important piece to consider is the follow up data that must be collected at regular intervals. Those time points, often every 2-3 years, are important to mapping changes in multiple exposures over time so we can understand the effects of increases and decreasing in exposures as well as opportune times to intervene.

We need to encourage clinical journals to include experts in observation research on their editorial and reviewer boards so that these studies can be rigorously reviewed for quality, and ensure the privacy and confidentiality of the individual, while reducing the administrative burdens on the researchers.  

BCRF: As an epidemiologist and medical oncologist with a focus in cancer prevention on both cancer free women and survivors, how has observational research shaped the work that you do?

Dr. Visvanathan: Observational research has enabled our group to examine novel biomarkers in blood, urine and tissue, potential risk factors and preventive agents such as aspirin and statins over time, quickly and efficiently. It has helped to identify at-risk groups including groups that we might include in randomized studies.

The design of these studies has enabled us to compare women who develop cancer or have cancer to those who do not so we can identify changes to due cancer versus other causes. It has led to the formation of interdisciplinary teams within Johns Hopkins, as well as national and international teams to work around some of the questions in clinical care. We’ve used a variety of data sources to conduct our studies including clinic based cohorts in high risk women, population cohorts both national and international, EMR, and public databases such as NHANES,  and cancer and death registries.

One of the important questions we’ve been able to study is what happens over time- starting with before cancer begins. We’ve been using benign tissue from women without disease to identify early changes that differ between women who go on to get breast cancer compared to those who do not over a short interval. This will allow us to better assess risk in individuals and potentially intervene early. Another is the impact of inflammatory biomarkers and commonly used medications such as aspirin and statins on the development and progression of breast cancer.

You can read the full ASCO report here and more about Dr. Visvanathan’s BCRF research here.