Ravenell RA, Camasta CA, Powell DR.  The unreliability of the intermetatarsal angle in choosing a hallux abducto valgus surgical procedure. J Foot Ankle Surg. 2011 May-Jun; 50(3): 287-92. (PubMed ID: 21435913)

WHY did the authors undertake this study?
Generally speaking, our profession utilizes the first intermetatarsal angle (IMA) to help determine procedure selection when evaluating the hallux abducto valgus (HAV) deformity. Smaller IMAs can be corrected with a distal osteotomy, whereas larger IMAs typically require a more proximal procedure. The authors of this study aim to challenge this supposition by evaluating whether or not certain distal procedures can correct for larger IMAs.   One might say that they were attempting to determine the effect of multiple procedures on radiographic parameters in the severe radiographic HAV deformity. 

HOW did they attempt to answer this question?
The primary outcome measures of the study were pre- and post-operative transverse plane radiographic parameters (1st IMA and hallux abductus angle). A retrospective analysis of non-consecutive patients from an unknown quantity of surgeons over a 10-year data collection period was performed to select a total of 61 patients who had surgical correction of a HAV deformity with either a distal first metatarsal osteotomy, 1st MPJ arthrodesis or Lapidus arthrodesis, and who had a severe radiographic HAV deformity (defined as as1st intermetatarsal angle > 15 degrees and/or hallux abductus angle > 25 degrees). 

WHAT were the specific results?
All post-operative radiographic parameters were statistically improved compared to pre-operative parameters, and no statistical differences were found with respect to the amount of improvement between the groups. 

HOW did the authors interpret these results?
From these results, the authors concluded  that severely increased radiographic HAV deformities do not necessarily necessitate a proximal procedure.

Waterman BR, Belmont PJ, Cameron KL, Svoboda SJ, Alitz CJ, Owens BD. Risk factors for syndesmotic and medial ankle sprain:  role of sex, sport and level of competition.  Am J Sports Med.  2011 May; 39(5): 992-8.  (PubMed ID#: 21289274)

WHY did the authors undertake this study?
We know that the vast majority of ankle sprains (in the ballpark of 85%) occur within the lateral ankle complex.  The remainder (about 15%) occurs medially in the deltoids and superiorly within the syndesmosis. Although these happen less frequently, they tend to result in significantly greater disability and time to full recovery. The authors of this study aimed to identify risk factors associated with these injuries and describe epidemiologic data with respect to their incidence. 

HOW did they attempt to answer this question?
The United States Military Academy (USMA) utilizes a Cadet Illness and Injury Tracking System (CIITS) to document and record all patient encounters. Authors retrospectively examined this data over a 5-year data collection period to identify and investigate all medial and syndesmotic ankle sprains. 

WHAT were the specific results?
Syndesmotic injuries were slightly more common than medial ankle sprains (6.7% versus 5.1%, for a total incidence of 11.8% of all ankle sprains). Two risks factors were identified with respect to these injuries:  male sex and previous participation in high level sporting competition (such as intercollegiate athletics). They also found that people suffering from medial and syndesmotic injuries were more likely to have a higher BMI.

HOW did the authors interpret these results?
As is appropriate for a descriptive epidemiologic study, the authors did not draw definitive conclusions from this data, and rather reported their findings and risk factors only. 

Let’s take a closer look at the topic of selection bias, specifically as it applies to the Ravenell et al HAV study. I actually have several significant critiques of this article, but I’m going to try and be gentle because I think the authors started out with at least a good clinical question. A common theme in this section is that, generally speaking, prospective studies are preferable to retrospective studies when considering the level of clinical evidence. There are many reasons why this is the case, but one certainly is that with a prospective study design the investigators have much greater control about the specific patient characteristics that will constitute the different patient cohorts to be studied.  In most cases, we want to make direct comparisons between at least two groups of patients, so it helps if they are as similar as possible prior to the intervention, or whatever other variable is being studied.

In this study, we are presented with three groups of patients who had similar pre-operative transverse plane radiographic parameters (all significant hallux abductovalgus deformity defined as1st intermetatarsal angle > 15 degrees and/or hallux abductus angle > 25 degrees). The intervention or variable in the study was which surgical procedure was performed:  either distal first metatarsal osteotomy, 1st MPJ arthrodesis, or Lapidus arthrodesis. And finally we are presented with results in the form of post-operative transverse plane radiographic parameters which show that each of the three procedures had a similar effect.

The potential problem is that we have no idea why patients were placed into their respective groups.  Patients were not prospectively randomized into either distal first metatarsal osteotomy, 1st MPJ arthrodesis or Lapidus arthrodesis procedures. Instead, for whatever reason, the surgeon(s) at the time decided to perform one of the three procedures (or another procedure that was not studied here) because they felt it was indicated for that patient.

The “for whatever reason” in the preceding paragraph is a good example of selection bias.  There was likely a reason that the surgeon performed either a distal first metatarsal osteotomy, 1st MPJ arthrodesis or Lapidus arthrodesis at the time of the surgery, but because the study is retrospective in nature and we weren’t there to ask him/her, we have no way of knowing what that reason was. We don’t know why the patient was selected for that specific procedure. This “loss of control” of the study groups decreases our ability as critical readers to make direct comparisons between the groups after the intervention. 

Although this article provides a good example of selection bias, I don’t necessarily think it terribly affects the outcome or the general theme that the authors seemed to be presenting. Interestingly (at least I think it’s interesting, most people will probably think it’s a little too theoretical), the authors concluded that we don’t really need the intermetatarsal angle to determine procedure selection for the hallux abducto valgus deformity, but they used the intermetatarsal angle to demonstrate their argument!  It’s like they said “You don’t need the intermetatarsal angle, and I’ll use the intermetatarsal angle to prove that I’m right!”  That’s a little overly simplified of course, but funny when taken on its surface.