Cover by Caroline Howard

The Statistics Behind Medicine

By Varsha Kumaravel

When was the last time you got a vaccine? How about the last time you swallowed an aspirin? Were you diagnosed with any kind of illness the last time you visited your doctor? At first thought, it is obvious that there is scientific study behind the medicine entering your body. What is usually not as obvious upon first thought, though, is the fact that statistics also play just as important of a role.

Chances are, you have probably been to a doctor. When you are a patient at a hospital or doctor’s office, a nurse checks your vital signs - your body temperature, blood pressure, respiration rate, and pulse. These statistics are entered into your patient chart, and then the nurse determines your overall wellness. But what is defined as “healthy” and “unhealthy” from person to person? What might be healthy for a five year old obviously might not be healthy for a sixty year old, so how do the people in charge of determining your well-being know what standards apply to your health specifically? Dr. Kiran Nihlani, a Statistics lecturer at the University of Pittsburgh, says, “We are surrounded by these imaginary or numerical cut-offs based on how we put people into one of the two categories. That cut-off value has to be decided based on some predictors and these cut-offs don't remain the same.”

Take blood pressure guidelines, for example. Every time you have been to the doctor’s office for an annual checkup, a nurse has taken your blood pressure and carefully recorded it in a chart. What you might not know is that the very guidelines that determine if your blood pressure is high or low were revised in 2017 for the first time in fourteen years. According to Harvard Health Publishing, the numbers for the diagnosis of hypertension - the medical term for high blood pressure - were lowered for all adults, with the purpose of helping doctors and patients address hypertension much earlier. The current evidence behind health guidelines, the available resources for healthcare, and the population’s general medical well-being all play a significant role in determining what these predictors are. Due to the guideline change, 70 to 79% of men age 55 and older were classified as hypertensive when they had previously been considered healthy.

In the past year especially, many people have led sedentary lifestyles, which will further affect the cut-offs for predictors.

Apart from receiving diagnoses from the doctor, you have likely also received a prescription or received a vaccine to protect you against an infectious disease or virus like influenza or COVID-19. How do you know that the vaccines have actual efficacy? How do you know that the medicine you are taking will not cause you to have a heart attack? From every vaccine you receive to every aspirin you swallow, the medicine you put in your body has been through countless trials, data collections, statistical analyses, clinical data assessments, and risk management strategies. “Clinical trials. That’s something I would say that’s the most visible use of statistics in medicine, and even more in [today’s] environment when you have the COVID vaccine,” says Dr. Nihlani. “A lot of planning [for a vaccine or drug] is based on sound statistical practices. The number of people to interview, what data to collect, what data can be used, and [seeing] if things are actually significant.”

Nearly all types of medicine, such as vaccines, antibiotics, and medical devices, must go through Food and Drug Administration (FDA) approval. Any approved product that is being sold or provided to the general population has gone through numerous stages of testing. During these stages, statistical tools are used to analyze how many people experienced side effects, what subpopulations these side effects impacted, and what the range of the effects are.

For vaccines seeking FDA approval, the clinical development stage is split into three phases. During Phase I, small groups of 20 to 80 adult subjects are given the trial vaccine. If the vaccine is intended for children, it will first be tested on adults, and then gradually tested on younger people until the target age is reached. The goal of Phase I is to determine the amount and type of immune response that the vaccine evokes. If the vaccine passes Phase I, it moves to Phase II, where the vaccine is given to several hundred participants who have similar characteristics (e.g., age) to those for whom the vaccine is being intended. These trials are carefully randomized and controlled using a placebo control group - a group that receives a treatment (in this case, the vaccine) that appears to be a real treatment, but in reality does not contain any actual medicine. The placebo is given to participants to monitor for potential signs of the placebo effect, which occurs when participants who are given a placebo treatment exhibit changes even though they are not receiving the true treatment. Overall, the purpose of Phase II is to research the vaccine’s safety, proposed dosage, immunization schedule, delivery method, and ability to provoke an immune response (immunogenicity).

After passing Phase II, the vaccine enters Phase III, where it is given to thousands of participants and is further tested for safety and efficacy. These trials are randomized and double blind - neither the researcher nor the participant know whether the treatment is a placebo or not. The objective of Phase III is to assess the safety of the vaccine in large groups of participants and to monitor for any rare side effects that were not detected in the previous phases. If the vaccine passes Phase III, the developer of the vaccine submits a license application to the FDA, and if approved, it is mass produced and distributed to the population at large.

With medical research outside of vaccine and drug development, however, it is often not as simple as just conducting an experiment with an imposed independent variable and a resulting dependent variable. Much of the research done in the medical science field cannot be experimental, as that would be unethical and immoral.

“So, we have a lot of observational studies,” Dr. Nihlani says. “But we give it an experimental flavor, just in the sense that you are still observing, no experimentation, but now you have two groups. One is the control group and the other is the case group. The case group has the presence of the exposure. The control group [does not] have the exposure.” For example, say a group of researchers want to investigate how different lifestyles influence cardiovascular health. It would be unethical to manipulate a participant’s lifestyle to be unhealthy, so researchers instead simply observe different lifestyles. They could have the control group consist of participants leading healthy, optimal lifestyles, while the case group would consist of participants leading lifestyles that are less than optimal. Researchers could measure the average heart rate of the participants in the healthy control group and the average heart rate of the participants in the less than optimal group, and then determine whether the average heart rates are significantly different or not.

Outside of the realm of medical research that continues to push future boundaries, medical trends govern the modern world. “I think a lot of statistics also goes into surveillance. There is a field of epidemiology called surveillance epidemiology, or sentinel epidemiology,” says Dr. Nihlani. According to the National Center for Biotechnology Information (NCBI), sentinel epidemiology monitors disease rates in particular geographical areas or population groups of interest. “Look at the COVID data coming in,” Dr. Nihlani continues. “There are hospitals, people go and get tested there, they come out positive, [hospital employees] report that data to the CDC, the CDC publishes that result… This can only happen because there is a surveillance going on. And there's a database that is being created."

But medical trends are not just about graphs. When it comes to the production of vaccines and drugs, how do officials know how much to disperse to different parts of the population? “How do public policymakers make these decisions? They have statistics,” explains Dr. Nihlani. “If they see the cases arising, then there are more [COVID-19] tests allocated to those places… if you have more medical emergencies in one place, the resources would go to that place.”

Statistics are as prevalent in your individual medical life as they are in the general population’s well-being. However, it might come as a surprise that statistics do not just play a role in the numerical side of medicine. Statistical analyses even extend into the visual side of medicine. For example, take cancer imaging. Machine learning has developed to the point where an image of potentially cancerous tissue can be taken at an X-ray level, and the machine can determine using statistics and an algorithm if a tumor is benign or malignant. “You keep feeding your algorithm multiple X-rays, and you'll let it make a decision about whether a tumor is benign, or malignant. Sometimes those variables would be numeric, say the size, the volume, if it is opaque or not, but also sometimes features will be just pictorial,” Dr. Nihlani says. “Based on pixels, you can unpack an image and we have algorithms that can use those pixels, which are also called features, and make a diagnosis.” 

Diagnostic imaging tests in medicine use unique statistical methods constructed for over a century to create prognostic models based on imaging results. Methods such as exploratory data analysis, which is the visualization and summarization of data, and data mining, which is the automatic discovery of patterns and relationships in data, has brought machines to the point where they detect tumors more accurately than trained radiologists. This is especially beneficial because the need to detect cancerous tumors as early as possible - before the cancer has progressed too far - is vital.

It is eye-opening just how much of medicine is as much about statistics as it is science. From an everyday prescription to a common diagnosis to a worldwide pandemic, statistics exist inherently right next to science. An important part of staying informed and educated about your medical life is understanding how statistics play a role in the vaccines you receive, the medicine you take, and the diagnoses you are given. The statistics behind medicine do their best to keep us safe and healthy, so take a moment, the next time you get a diagnosis, receive a vaccine, or swallow an aspirin, to remember the statistics. The future discoveries of medicine, and science in general, may remain a mystery to us for now, but the field of statistics helps to bring the future into a clearer focus.