Henning JA, Jones CB, Sietsema DL, Bohay DR, Anderson JG. Open reduction internal fixation versus primary arthrodesis for lisfranc injuries: a prospective randomized study. Foot Ankle Int. 2009 Oct; 30(10): 913-22.   (PubMed ID: 19796583)

WHY did the authors undertake this study?
Definitive intervention for traumatic injuries of the Lisfranc tarsometatarsal complex remains a controversial surgical challenge with several treatment options available. The authors of this study attempted to evaluate outcomes of primary arthrodesis (PA) versus primary open reduction internal fixation (PORIF), interestingly with a prospective and randomized study design.

HOW did they attempt to answer this question?
The primary outcome measures of this study were the Short Form-36 (SF-36) and Short Musculoskeletal Function Assessment (SMFA) scores at regular intervals up to 24 months following surgery.  Additional measures were also recorded including secondary surgeries, reduction quality, malreduction, malunion, nonunion, delayed union, fixation failure, hardware failure, incision healing, infection, amount and duration of pain medications, usage of usual shoes, independent ambulation, and time until return to work.  It should be noted that all patients in the PORIF groups were advised to have a scheduled hardware removal at the 3-4 month mark.  And finally, an overall measure of patient satisfaction was recorded on a Likert scale by phone interview. 

Inclusion criteria of the population cohort were skeletally mature patients with Lisfranc injury of less than 3 months duration.  Exclusion criteria included patients with major intra-articular fracture pattern, prior foot pathology/intervention, and a list of associated medical comorbidities.  Patients were randomized with a random number generator, and standard operative and post-operative protocols were followed.  A total of 14 patients were in the PORIF group, and 18 patients were in the PA group.

WHAT were the specific results? 
No statistically significant differences were found  between the PA and PORIF groups at any time interval when examining the primary outcomes of the SF-36, SMFA, and patient satisfaction scores.  There was a statistically significant difference in secondary surgeries with the PORIF group more frequently requiring additional surgery (78.6% versus 16.7%; p<0.05), although this is unsurprising considering that hardware removal was planned in this group.

HOW did the authors interpret these results?
From these results, the authors concluded that there is no functional or patient satisfaction difference between primary arthrodesis and primary open reduction internal fixation of Lisfranc injuries, and that primary arthrodesis may be a better surgical option because it reduces the need for a secondary surgery in the form of hardware removal. 

Schouten O, Boersma E, Hoeks SE, Benner R, van Urk H, van Sambeek MR, Verhagen HJ, Khan NA, Dunkelgrun M,  Bax JJ, Poldermans D, Dutch Echocardiographic The Cardiac Risk Evaluation Applying Stress Echocardiography Study Group.  Fluvastatin and perioperative events in patients undergoing vascular surgery. N Engl J Med. 2009 Sep 3; 361(10): 980-9.  (PubMed ID:  19726772)

WHY did the authors undertake this study?
Podiatric surgeons routinely treat patients with extensive peripheral arterial disease and work closely with vascular surgeons in the in-patient and out-patient settings.  This study investigates outcomes following vascular surgery, specifically post-operative adverse cardiac events following initiation of statin therapy pre-operatively. 

HOW did they attempt to answer this question?
The primary outcome measures of this study were postoperative myocardial ischemia and death from myocardial causes following vascular surgical intervention.  In addition, levels of total cholesterol, low-density lipoprotein cholesterol, interleukin-6 and C-reactive protein were measured peri-operatively.

Four hundred and ninety-seven patients were randomized in this double-blind design to receive fluvastatin or a placebo at a median of 37 days before noncardiac vascular surgery.  Outcomes were measured up to 30 days post-operatively.

WHAT were the specific results?
Statistically significant differences were found with respect to postoperative myocardial ischemia and death from myocardial infarction between the two groups.  The statin therapy was not associated with the increase in the rate of adverse events.

HOW did the authors interpret these results?
The authors concluded that preoperative statin therapy improved postoperative cardiac outcomes in patients undergoing vascular surgery.                                                                                                                    

Let’s take a closer look at the topic of statistical power analysis.  Both articles from the Podiatric and Medical Literature Digests actually have similar study designs.  In each, a group of patients undergoing a surgical intervention are identified pre-operatively.  The authors then randomly assign the patients to one of two groups, and then plan on following the patients post-operatively with certain outcome measures.  The outcome measures and statistical analyses are planned in advance to determine differences (or a lack of differences) between the two groups in the post-operative period.   Take a minute to look back on the two articles and see how both fit this generalized design.

With this type of study design, statistics can actually be performed before the very first piece of data is collected to figure out how many patients should be studied to determine clinically and statistically significant differences between the two groups.  This technique is called using power analysis to determine sample size.  Without getting too technical, this allows a researcher to estimate how large a sample size is required to detect an effect large enough to be clinically significant.  The study can then be planned based on this sample size determination.    One can certainly appreciate how it would be “better” from a critical analysis standpoint to estimate this information at the start of the study, as opposed to collecting a bunch of data and then after-the-fact trying to see if you can find any statistically significant differences.

Both of our studies performed this pre-study analysis.  The Henning et al Lisfranc study estimated that 60 patients would be needed to determine a statistically significant difference between the two groups, while the Schouten et al statin study estimated that a sample of 500 patients would determine appropriate statistical power.  The Henning et al group should actually be commended for putting together a solid study design.  It is relatively rare in the lower extremity orthopedic literature to have a prospective and randomized study about a surgical intervention.  Although the majority of studies have to do with surgical interventions, most are retrospective in nature and only examine one group.  They even took it a step further by using power analysis to determine sample size.  In other words, before the study even began they figured out what a clinically significant difference would be between the two groups, and then determined how many patients they should study to detect statistically significant differences.

Unfortunately, they didn’t necessarily follow through on their design.  In fact, the authors stopped collecting data when the sample size reached 40 patients because preliminary analysis determined “no difference in function” between the two groups, but statistically significant differences in “hardware removal rates, and secondary surgeries” (although this was expected because of planned hardware removal in the PORIF group).  While it is reasonable to “discontinue the study to avoid further potentially unnecessary surgeries” in light of no functional differences between the two groups, they had already determined that they would need 60 patients to even detect a functional differences between the two groups!  So while there was probably no statistically significant functional difference between the two groups, the study did not reach the statistical power the authors originally decided was acceptable.

This obviously just briefly touches on the subject of power analysis within the medical literature.  I’m sure we’ll get into it in more detail as this journal club evolves, but for those interested in learning more now, I would recommend my favorite biostatistics book:  Glantz’s Primer of Biostatistics.  

Please join us for an online discussion of these topics:

 -Do you feel that primary arthrodesis is a better surgical option?

-Has statin therapy improved postoperative cardiac outcomes in
your patients? 

-What sample size do you think would have been adequate for the author's study?