Case-control studies:   A case-control study uses subjects who already have a disease (or condition) and subjects free from the disease (or condition), and then investigates if the two groups have similar or different exposures to risk factors associated with the disease (or condition). With the introduction of controls, these studied have a decreased opportunity for bias.  However, they are not randomized so they are still susceptible to confounders and different types of biases such as selection bias.  These studies are important when studying rare diseases or when a randomized clinical control trial would be unethical.

Case-control study example:

A researcher identifies 100 patients who needed root canal treatments (RCTs) in the past year.  She then identifies another 100 patients who did not need RCTs in the past year.  While comparing past similarities and differences of the 2 groups she finds that the patients who had RCTs in the past year had an increased rate of visiting the dentist.  She then concludes that visiting the dentist is a risk factor for RCTs!

NO…. That’s silly! J

In fact, this is one of the flaws of this type of research.  It cannot adequately show cause and effect.  This is very important to understand:  association does NOT equal causation!

About Confounders Let’s imagine that a study finds that women who have had dental treatment while pregnant have fewer preterm delivery babies.  It might be because having dental care while pregnant gives better birth outcomes.  However, an alternative theory might be that women who seek dental care while pregnant are more health conscious and therefore take better care of themselves resulting in less preterm births as well. This is called a confounder.  A confounder is when the association of an exposure (dental care) and the outcome (preterm birth) is mixed up with the real effect of another exposure (more health conscious) and the same outcome (preterm birth).

Case series studies:  A study on a series of patients with an outcome of interest.  These studies do not have controls.  For example:

A researcher identifies 100 patients who have had tooth colored fillings placed.  She want to know how many of them needed crowns within 5 years.   So she goes back and reviews patient charts to find out.

There are many flaws with this research, but we will stick to just 2 for now including:

1.      No control: A control group (a group of patients who does not get the therapy, has an alternative therapy or does not have the same exposure) is important because without one we do not know if needing a crown after a tooth colored filling is just a normal happening after placing any filling or if it’s specific to tooth colored fillings.

2.      No randomization: Randomization is when you assign people to a test group (i.e. people who got tooth colored fillings) or to a control group (i.e. people who got silver fillings) randomly.  Since case series studies by definition do not have controls, they do not have randomization either.  Randomization is a common way researchers help diminish the impact of bias and confounders (more on randomization, bias and confounders later). 

Animal studies: Again these study designs result in surrogate outcomes that may not translate to human subjects.  Animal physiology can be substantially different than a human’s and therefore we cannot assume that results will be the same in humans.  One familiar example of this is dogs and chocolate.  The world’s most amazing chocolate comes from http://www.chatelainechocolate.com  I can and do eat massive quantities of La Chatelaine chocolates, but my dog Olive cannot.  Chocolate contains theobromine which in large quantities is poisonous to dogs.

Human and animal physiology is different. Research in animals does not always translate to humans (or visa versa).

Expert opinion: This type of “evidence” has a strong potential for bias and may or may not be based on the best available science.

Ideas based on plausibility or physiology: There are many examples of ideas that seem plausible, but in fact turn out not to be true when examined closely. For instance, giving prenatal fluoride was once thought to impart caries (cavity) resistance to children.  When I think about it, it seems plausible that fluoride would cross the placenta and then be incorporated into the developing teeth, there by imparting caries resistance. However, the current body of evidence suggests this technique has no therapeutic benefit.

Bench top research: These types of studies result in surrogate (alternative) outcomes which may not translate to human subjects. For example, a researcher may bond (glue) a tooth colored filling to an extracted tooth and then pound away on the filling with a machine designed to mimic chewing. However, there is no guarantee that this is even remotely similar to the human mouth. Furthermore, information gathered in these studies such as “bonding agent X withstands 1000 cycles” does not necessarily translate into clinical significance such as “Does this help the patient keep their tooth longer? Yes or No?”

Our knowledge of the world changes quickly and the science that was valid and important yesterday may be outdated tomorrow.  EBD is a way to implement advances in the science of oral health care while at the same time viewing these advances with curiosity and skepticism.  

The concept of EBD was born from a similar movement in medicine in the 1990’s led by Dr. David Sackett called “evidence-based medicine” (EBM) and is definded by the American Dental Association (ADA), which is fast becoming one of the leading forces in EBD, as: 

“…an approach to oral health care that requires the judicious integration of systematic assessments of clinically relevant scientific evidence, relating to the patient’s oral and medical condition and history, with the dentist’s clinical expertise and the patient’s treatment needs and preferences.”

Pictorially, it can be deconstructed as:

Simply stated, it’s a way to do all the things we do in dentistry (crowns, choosing cements, providing periodontal therapy) that uses your own professional expertise, your patient’s individual values and circumstances and that takes strongly into consideration the best and most current research available. I would guess that most providers are already engaged in items #1 and #2 to a very high level.  Though some practitioners may be incorporating principle of #3 (utilizing the best and most current research), it is likely that even more of us are not (more on this in a later post).

All 3 of these components are essential.  The science must be balanced with professional judgment and a patient’s desires and individual circumstances.  As a side note, (and this is not based on science and is only my personal opinion) I believe that the fact that EBD must incorporate ALL 3 components is the reason that no other party besides dental practitioners can truly “practice” EBD.  For example, a 3^rd party payer (insurance company) is missing 2 of the essential components and therefore, can never be a part of the health care decision making process.  That’s just my opinion though.

Related to EBD is evidence-based decision making (EBDM). EBDM is solving clinical questions and problems with principles of EBD.  EBD is the noun and EBDM is the verb.   Though EBDM includes finding the best objective evidence on a given topic it is important to understand that:

1) Evidence by itself is not adequate to make clinical decisions.

2) All evidence can be graded according to validity, importance and potential for bias.  This graded evidence can then be used to assisit in EBDM.