During this week’s resident journal club for the Chino Valley Medical Center program, we will discuss a common problem for any of us that do foot and ankle surgery: pre-operative prophylactic antibiotics.

For the journal clubs that I run, I have a couple of requirements. First, the person assigned to provide the articles must create a PICO question based on a clinical experience he or she recently experienced. As a quick reminder, PICO (Patient Intervention Comparison Outcome) is a method of creating a clinical question, which is then used to search on PubMed or other medical search engine.

The PICO QUESTION

For example, in regard to preoperative antibiotics, we might ask the PICO question as follows: In patients undergoing foot and ankle surgery (P) do preoperative intravenous antibiotics (I) compared with no antibiotics (C) decrease postoperative infections (O)?

The next step is for the student or resident to document how they did their search. It’s at this point that we actually get to the article analysis part of the club. The two articles our clerk picked to answer this question were interesting to read, but one of them, a large retrospective cohort study, highlighted the major problem with so many research studies: the inappropriate mistaking of correlation for causation.

---

“The major problem with so many research studies: is the inappropriate mistaking of correlation for causation”

---

In a study by Hawn, et al, they performed a retrospective cohort study, and their objective was, “to determine whether prophylactic antibiotic timing is associated with SSI [skin and soft tissue infection] occurrence.” They reviewed 32,459 surgeries and found “the SSI risk varies by patient and procedure factors as well as antibiotic properties but is not significantly associated with prophylactic antibiotic timing”¹.

I’ll say at the outset that after reading the study, I found it convincing at the first pass. However, the authors of the study make a strong conclusion that they should not make because observational studies cannot make a conclusion. Their study actually can go only so far as to say there is a correlation between starting antibiotics within one hour of surgery and infection risk. They cannot actually say IV antibiotics do or do not prevent infection. The reason for this is that observational studies are not able to eliminate all of the lurking variables. For example, in this study the cohorts include patients undergoing orthopedic, gynecologic, colorectal, and vascular surgery. It’s pretty clear that these disparate and heterogeneous groups create bias.

---

“Observational studies cannot make a conclusion, as they do not eliminate the many variables. The disparate and heterogeneous groups create bias.”

---

A better study design would be to prospectively randomize patients undergoing one type of surgery (example - clean non-emergent elective abdominal surgery) into two groups in a double blind manner: one receiving antibiotics at a specific time before incision and the other group not receiving any antibiotics. Patients would then be observed for a designated time to determine if there’s an actual relationship between perioperative antibiotics and infection. The creators of this imaginary study would have control for a variety of confounding variables (such as patient-specific factors like diabetes) by creating specific inclusion and exclusion criteria before the study started and randomization of patients.

The obvious problem with this that you might be asking yourself is, “Given the number of retrospective studies in podiatric medicine and orthopedics, how can I make use of any of these studies?” It’s not an easy question to answer, but I have two comments to make on this front.

First, understand that retrospective studies cannot by their nature answer our research questions with finality. The best we can do with these studies is to say there is a correlation.

Observational Studies are Strengthened by the STROBE Standard

Second, it’s useful to know that this is a systemic problem in medicine and many people have worked to consider ways to make observational studies stronger. For example, experts have created the STROBE statement². This research group, after a consultative process created a 22-point checklist for authors of observational studies with the intent to maximize the utility of these studies.

If our journals mandated the use of the STROBE statement, we would have a larger number of high quality observational studies with which to make clinical decisions. Until then remember correlation is not causation!