Bower VM, Hobbs M.  Validation of the basic foot screening checklist: a population screening tool for identifying foot ulcer risk in people with diabetes mellitus. Journal of the American Podiatric Medical Association. 2009 Jul-Aug; 99(4): 339-47. (PubMed ID: 19605928)

WHY did the authors undertake this study?
Diabetic foot disease is a global public health concern with social, economic and moral implications.  One of the cornerstones of intervention is prevention, specifically through identification of at-risk patients and limbs. The primary objectives of this study were to determine the reliability and validity of the Basic Foot Screening Checklist (BFSC) utilized by the National Foot Care Project of Australia.

HOW did they attempt to answer this question?
The two primary outcome measures of this study were the reliability and validity of the BFSC.  The reliability was quantified through sensitivity, specificity and positive predictive value analyses.  The validity was quantified through measurement of the Kappa coefficient.

Twenty foot screeners performed testing on 500 participants as part of a larger “Living with Diabetes” course.  These screeners were non-physicians, but were employed in healthcare (nurses, dieticians and support officers) and had undergone a 1-day Foot Care Training Course.  112 patients were then re-examined by different screeners for inter-observer reliability analysis.  All participants were then screened by specialized podiatric physicians for “gold standard assessment” and inter-observer validity.  The BFSC identifies a high-risk foot as one with peripheral vascular disease, peripheral neuropathy, foot deformity, or previous ulceration/amputation

WHAT were the specific results? 
As a screening tool, the BFSC was 54% sensitive, 77% specific, and had a positive predictive value of 0.82.   The overall validity of the tool had a kappa coefficient of 0.35.

HOW did the authors interpret these results?
From these results, the authors concluded  that the sensitivity and specificity of the individual test components were good, but the overall validity and reliability of the BFSC was poor.  They attributed this to the screeners being able to correctly perform the BFSC, but not correctly interpret the results.

Malliaropoulos N, Ntessalen M, Papacostas E, Longo UG, Maffulli N.  Reinjury after acute lateral ankle sprains in elite track and field athletes.  The American Journal of Sports Medicine. 2009 Sep; 37(9): 1755-61. (PubMed ID#: 19617530)

WHY did the authors undertake this study?
The objective of this study was to determine the relationship between the grade of acute lateral ankle sprain and the risk of reinjury in elite athletes.  Lateral ankle sprains represent one of the most common athletic injuries, and may lead to long-term complaints, but there is a relative lack of clinical objective data to predict which injuries will lead to chronic pathology or recurrence. 

HOW did they attempt to answer this question?
Several outcome measures were undertaken to evaluate this clinical question.  The first was a classification of the acute ankle sprain on a 4-grade scale (I, II, IIIA, IIIB) taking into account ankle joint range of motion, ankle edema and anterior talar displacement with stress radiography.  The second was a quantitative assessment of return to activity.  The final was whether or not the athlete had a recurrence of the lateral ankle sprain within 24 months, and if so, how long post-initial injury.

Inclusion criteria of the population cohort were all athletes within the Greek Track and Field Federation presenting with an acute lateral ankle sprain.  Patients were excluded if they had a previous history of foot/ankle injury, any acute syndesmotic injury or any acute lower extremity fracture.

All patients were examined and classified by the same physician, and were prescribed the same rehabilitation protocol.  They were followed for a period of 24 months.

WHAT were the specific results?
Athletes with a grade II injury had a statistically significant higher rate of reinjury (29%; P < 0.05) when compared to grade I (14%), grade IIIA (5.6%) and grade IIIB (0%) injuries.  There was also a statistically significant positive association between the grade of the injury and the time to activity return. 

HOW did the authors interpret these results?
From these results the authors concluded that athletes with lower grade ankle sprains are less likely to experience a reinjury compared to athletes with high grade ankle sprains.  They also concluded that athletes with high-grade lateral ankle sprains are at a low risk of reinjury.

Although the Bower and Hobbs article provides several good statistical teaching points, I think we’ll save those for the online discussion and instead spend the time here to discuss the more general topic of author conclusions and confounding variables.  The semantics we use for the Podiatric and Medical Literature Digests were specifically chosen:  How did the authors interpret these results?  Critical analysis of the medical literature actually involves you interpreting the results, and avoiding the potential for author bias.  Unfortunately, all too often authors take advantage of the “Discussion” portion of an article to extrapolate and insert their own opinions that aren’t grounded in anything actually measured in the article.  Let’s use the Malliaropoulos et al article to expand on this concept.

Each conclusion reached in the “Discussion” portion of an article should relate directly back to one of the outcome measures.  The conclusions are usually pretty easy to pick out of the “Discussion” as stand-alone statements.  When reviewing articles, I usually underline these statements as I’m reading, and then look back to the specific outcome measures and results to ensure that it is an appropriate conclusion.  Here are four of the statements that I underlined from the Malliaropoulos et al Discussion, and what specific results were used to reach the conclusions:

Discussion, First Paragraph:  “Elite track and field athletes with high-grade lateral ankle sprains are at a low risk of reinjury after their original injury” and Discussion, Twelfth Paragraph “This investigation show that elite track and field athletes with lateral ankle sprains experience a significant difference in the rate of reinjury according to the different grades of injury.”  These conclusions are supported by the outcome measure whether or not the athlete had a recurrence of the lateral ankle sprain within 24 months.  The authors had previously defined “high grade lateral ankle sprains” as Grade IIIA or IIIB, which had 5.6% and 0.0% recurrence rates respectively.  I think that most people would agree that these numbers are absolutely low by themselves, and are relatively low compared to the recurrence rates of Grade I (29%) and II (14%) sprains in this study.  The statistical analyses also supported this relative difference.

Also notice how the authors specifically referred to “elite track and field athletes” in this conclusion, and not “athletes” or “patients”.  Think about how this conclusion would change if they had used one of these different wordings.  The inclusion criteria of the population cohort was specifically elite track and field athletes, and thus we can’t conclude that the results are applicable to different populations.  The authors do a good job of pointing this out in the subsequent paragraphs.

-Discussion, Ninth Paragraph:  “The method outlined in this study, combined with close follow-up, allows the physician to predict with some degree of certainty the risk of reinjury for track and field athletes who suffer an acute lateral ankle sprain.”  We need to be a little careful here as critical readers.  This statement implies that the grade of initial injury is either the only factor which will predict recurrence, or is the primary factor that will predict recurrence.  Just because the grade of ankle injury was the only variable studied, does not mean that it is the only variable that will have an effect on the endpoint.  The type of athletic activity, athlete perception of the injury, foot type, obesity, and proprioception baseline are all variables that may be predictive for reinjury, but not studied here.  These are examples of confounding variables, or things that may influence the outcome, but that are not directly measured in an investigation.

We also need to be a little careful with the phrase “combined with close follow-up”.  While the authors did an excellent job of describing their rehabilitation protocol, this was not necessarily a variable in the study as all patients underwent the same protocol.  They also didn’t describe what the “follow-up” was after completion of the rehabilitation.  While this conclusion may be correct, it was not directly investigated in this study with either the variables or the outcome measures. 

-Discussion, Eleventh Paragraph:  “The results of our study provide cost-effectiveness and general applicability predictive criteria for the evaluation of patient with acute lateral ankle sprain.”  There was no outcome measure examining “cost-effectiveness” in this study, and so this is another example of a conclusion that is not supported by the outcome measures.

For the most part, the authors of this study did a good job of not extrapolating or inserting their own opinion into the discussion.  The final two paragraphs provide good examples of this.  Take a look at some of the following phrasing that they use when presenting ideas that are not supported by any of the outcome measures in the study:  “We cannot offer explanations to account for differences…”  “We can speculate….” “We cannot exclude the possibility…”  “The design of our study does not allow determination of…”, etc.  These are the types of terminology one would like to see when authors are presenting their own ideas as opposed to drawing specific conclusions from the outcome measures of a study. 

Please join us for an online discussion of these articles:

 -What objective criteria do you use when assessing athletes
for return to activity?

-What about diabetic foot screening? 

What objective criteria do you use when a diabetic patient with no complaints presents on initial referral from a PCP for screening?