Lenarz CJ, Watson JT, Moed BR, Israel H, Mullen JD, Macdonald JB.  Timing of wound closure in open fractures based on cultures obtained after debridement.  J Bone Joint Surg Am.  2010 Aug 18; 92(10): 1921-6.   (PubMed ID: 20660225)

WHY did the authors undertake this study?
Exact protocols for the appropriate treatment of open fractures remains an area that has continued to elude orthopedic science.  Although we are well aware of the potential complications and sequelae of treatment failure, we have yet to arrive at the best way to prevent their occurrence.  The objective of this study was to retrospectively evaluate a specific treatment protocol with a variable of wound closure timing.

HOW did they attempt to answer this question?
The primary outcome measures of the study were largely descriptive, and consisted of deep infection rate, number of irrigation/debridement procedures, and days to closure.  Secondary comparative statistical analysis was performed between upper and lower extremity injuries, as well as between different Gustillo-Anderson fracture classification types. 

Inclusion criteria of the population cohort were all patients presenting over an approximate 2.5 year data collection period with open extremity fracture.  All fractures were initially irrigated, dressed and stabilized in the emergency room prior to surgical debridement, and treated with tetanus and antibiotic prophylaxis as per established protocols.  In the operating room, debridement, irrigation and stabilization (either temporary or definitive) was performed at the discretion of the surgeon, but aerobic and anaerobic cultures were always taken following the irrigation.  Wounds were generally left open with a planned return to the operating room in 48 hours where debridement, irrigation and culturing were again performed.  This process was repeated until negative wound cultures were obtained, at which point definitive closure was performed.  Antibiotics were continued or discontinued at the discretion of the treating physician based on culture results.

WHAT were the specific results?
The overall deep infection rate was 4.3% which compares favorably to historical controls ranging from 7-22%.  Specific to the lower extremity, the deep infection rate was 3.6%, mean number of debridements was 2.58, and the mean days to closure was 6.15.  Risk of infection increased in the presence of diabetes and with higher BMI values.  

HOW did the authors interpret these results?
From these results, the authors concluded  that a relatively low rate of deep infection could be achieved with the described treatment protocol.

Svensson L, Bohm K, Castren M, Pettersson H, Engerstrom L, Herlitz J, Rosenqvist M. Compression-only CPR or standard CPR in out-of-hospital cardiac arrest.  N Engl J Med.  2010 Jul 29; 363(5): 434-42.  (Pubmed ID#: 20818864)

WHY did the authors undertake this study?
We have certainly been learning a lot of information with respect to CPR over the course of the last year.  I feel bad for the poor instructors who have to (first of all) keep up-to-date with their course material/lectures, and (second of all) attempt to teach old dogs new tricks!  This study out of Sweden adds to our body of knowledge and was very specific.  They utilized a prospective, randomized study design to compare compression-only CPR and standard CPR when performed (1) on the basis of instruction [essentially from someone calling 911 – or the Swedish equivalent!], (2) before the arrival of EMT, and (3) in witnessed cases of out-of-hospital cardiac arrest.

HOW did they attempt to answer this question?
The primary outcome measure of the study was 30-day survival, but secondary outcome measures included 1-day survival, first detected cardiac rhythm and survival to discharge from the hospital.

Following a rapid determination of specific inclusion and exclusion criteria, emergency dispatchers randomized a total of 1276 patients to receive instruction for either standard CPR (2 ventilations per 15 compressions) or compression-only CPR (no ventilations).

WHAT were the specific results?
There was no significant difference in the rate of 30-day survival between the two groups (8.7% versus 7.0%), nor the other secondary outcome measures.

HOW did the authors interpret these results?
Based on these results, the authors concluded that this provides evidence that compression-only CPR should be considered the preferred method for CPR performed by bystanders in patients with cardiac arrest.

Let’s take a closer look at the Lenarz et al open fracture article and discuss how retrospective studies can quickly lead to the loss of the total number of patients available for analysis.  In PJC#22, we introduced the concept of levels of evidence employed by numerous journals, and specifically JBJS where this article comes from.  The Lenarz et al article is categorized as a Level IV study, which is obviously a little bit down on the totem pole.  Although the authors attempted to initiate a “prospective protocol”, the study design was in fact retrospective in nature.  Let’s go through the article in detail and attempt to figure out exactly how many patients actually underwent the prospective protocol, and how many we lost because of the design of the study.

First, during the data collection period a total of 422 open fractures were treated by study authors.  Of those, only 346 were had complete data for final analysis.  So right off the bat, we lose 76 patients (or 18%) to follow-up.  This isn’t an outrageous number, and certainly a prospective study will also lose a certain amount of patients to follow-up.  Although this can be a little bit misleading, a general rule of thumb is that approximately 80% of patients need to complete follow-up for a study to retain its validity.  It’s close, but we’re within that range here.

The primary variable of this prospective study was timing of wound closure, and the primary point of emphasis was to not close any wounds until negative wound cultures (at the 48-hour mark) had been achieved.  We learn however, that 73 wounds were primarily closed following initial irrigation and debridement or allowed to heal by secondary intention without a planned return to the OR (32 Type Is, 32 Type IIs, and 9 Type IIIs).  That’s another 17% (73/422) of the total population cohort, and over one-fifth (21%; 73/346) of the total number analyzed that weren’t actually treated by the protocol at all!  So now we out about 35% (149/422) of the patients that met inclusion criteria.

We then discover information about several other unique sets of patients.  Forty-three patients (10%; 43/422) had wounds closed with “presumably” negative culture results, but cultures results that were never confirmed.  Twenty-four patients (6%; 24/422) had wounds closed “despite the presence of positive wound cultures (a protocol breach)”.  And approximately twenty-three patients (5%; 23/422) had “initially negative culture results that converted to positive later after closure”.   It’s a little unclear how much overlap we are dealing with here, but these numbers could potentially bring our total patient loss up to 57% (239/422).
 
We then lose a little bit of the standardization of the protocol following initial debridement.  We learn that Type I wounds were “usually” packed open with a wet-to-dry dressing, and that “most” Type II and II wounds were treated with negative-pressure wound therapy (an intervention known to decrease infection rates).  In addition the lack of consistency with post-operative dressing, we also don’t have standardization in terms of duration and course of antibiotic therapy.   We don’t have specific numbers, and only know that antibiotics were “routinely” discontinued 24 hours after wound closure, but with “some” being continued for six weeks, and that the “average” duration of treatment was three weeks.

I’m obviously being a little bit harsh on this article.  I do not at all mean to imply that we should totally discount the results and conclusions of this investigation, simply to raise the point that you really can’t initiate a protocol unless it is done prospectively.  When a protocol is evaluated in a retrospective manner, you typically end up with a situation like we are dealing with here, where you try and cram “square” patients into the “round” hole of the study design.  When a protocol is instituted prospectively, we generally only end up with “round” patients for our “round” holes!