Keating JF, Will, EM. Operative versus non-operative treatment of acute rupture of tendo Achillis: a prospective randomised evaluation of functional outcome,  J Bone Joint Surg Br. 2011 Aug; 93(8): 1071-8.  (PubMed ID # 21768631)

WHY did the authors undertake this study?
The debate between operative vs. non –operative treatment of Achilles tendon ruptures has been contested over the years. Currently, the literature suggests that surgical treatment is optimal, based primarily on decreased rates of re-rupture. The authors of this study set out to evaluate the effect of surgical repair vs. cast immobilization on the functional outcome following an acute rupture of the tendo Achilles. In particular, the association of each treatment with quicker/better recovery of muscle function was explored.

HOW did they attempt to answer this question?
The primary outcome measure assessed was muscle function dynamometry specifically evaluating dorsiflexion and plantarflexion of the ankle, while the main clinical outcome measure was rate of re-rupture.  Additional outcome measures included clinical complications, range of movement of the ankle, and the Short Musculoskeletal Function Assessment (SMFA). 80 patients were randomized into 2 groups:  (1) operative repair using an open technique, or (2) non-operative treatment in a cast. The patients were then followed for approximately 1 year. All patients that presented to the author's department with acute tendo Achilles rupture between 2000 and 2004 were considered for the study. Exclusion criteria included age > 60, presentation > 10 days after injury, systemic disease including rheumatoid arthritis (RA) and chronic renal failure (CRF), and steroid treatment or other medication influencing soft-tissue healing, such as chemotherapy.

WHAT were the specific results?
The primary and secondary outcome measures failed to show a statistically significant difference between surgical repair and cast treatment. A statistically significant difference was shown with the mean SMFA scores at 3 months, in favor of the operative group (15 vs 20, p < 0.03). No significant differences were observed after this, and at one year the scores were similar in both groups. The rate of re-rupture was less in the operative group compared to the non operative group, but this difference was not statistically significant (p = 0.68).

HOW did the authors interpret these results?
From these results, the authors concluded that there was not a convincing advantage to surgical treatment of acute tendo Achilles rupture as compared to non surgical treatment in a cast.

Rajasekhar A, Lottenberg L, Lottenberg R, Feezor RJ, Armen SB, Liu H, Efron PA, Crowther M, Ang D. A Pilot Study on the Randomization of Inferior Vena Cava Filter Placement for Venous Thromboembolism Prophylaxis in High-Risk Trauma Patients.  J Trauma.  2011 Aug; 71(2): 323-29. (PubMed ID#: 21825934 )

WHY did the authors undertake this study?
Pulmonary embolism prevention via venous thromboembolism  prophylaxis  in high risk trauma patients (HRTP) who do not qualify for pharmacologic anticoagulation is often achieved with the placement of retrievable inferior vena cava filters (IVCFs). Despite conflicting data on their efficacy, prophylactic IVCFs have become part of the treatment algorithm for high risk trauma patients. There have yet to be any randomized control trials testing the efficacy of prophylactic IVCF placement, causing a discrepancy in standards of care. The authors undertook this study to determine the feasibility of performing a prospective randomized control trial of prophylactic IVCFs use in high risk trauma patients.

HOW did they attempt to answer this question?
The primary outcome of this pilot study was to assess feasibility objectives which included timely enrollment of high risk trauma patients, acceptable time to randomization, timely receipt of allocated treatment, and adherence to weekly lower extremity surveillance compression ultrasounds. Secondary feasibility objectives were to obtain preliminary data on VTE event rates at hospital discharge at 1 month and 6 months post discharge. Secondary outcomes included evaluation of incidence of PE, DVT, and death. In addition, the authors examined barriers to recruitment and randomization, use of pharmacologic prophylaxis and sequential compression devices (SCDs), compliance with follow-up, and removal rates of filters.

The authors were able to construct a single institution, prospective randomized controlled pilot feasibility study in a Level I trauma center. HRTPs were identified for pIVCF placement by the Eastern Association for the Surgery of Trauma guidelines. From November 2008 to November 2010, HRTPs were enrolled and randomized to either pIVCF or no pIVCF. All patients received pharmacologic prophylaxis when safe.

WHAT were the specific results?
Results of the feasibility objectives showed the following:  time from admission to enrollment was on average 47.4 hours ± 22.0 hours, time from enrollment to randomization was on average 4.8 hours ± 9.1 hours, time from randomization to IVCF placement  was on average16.9 hours ± 9.2 hours, adherence to weekly compression ultrasound within first month was achieved better in the non IVCF (62.5%) vs the IVCF group (44.4%), and 1-month clinical follow-up was well noted in both groups VCF group (83.3%) and non-IVCF group (100%). At 6-month follow-up, one PE in the non-filter group and one DVT in the filter group had occurred. One non-PE-related death occurred in the filter group.

HOW did the authors interpret these results?
The authors concluded that this pilot study exhibited that it was feasible to set up a randomized study that compared placement of IVCFs to no placement of IVCFs in high risk trauma patients. They noted that future studies needed to continue to address recruitment issues, protocol adherence, and follow-up issues. The use of pilot studies in preparing for larger studies was also validated.

Let’s take a closer look at the topic of appropriate tables, specifically as it applies to the Keating and Will Achilles article. As we discussed in the last PJC, it is very important for authors to clearly identify their outcome measures and results so that readers don’t have to go searching through the paper in order to find the most important information.  One of the easiest ways to do this is by setting up a good table. A good table should tell me:  (1) what outcome measure is being studied, (2) what the descriptive statistics of that outcome measure were, and (3) what comparative statistics of that outcome measure were. And it should do this in a clearly organized way so that the numbers really “jump” out at the reader.

Let’s take a look at how our two papers do this.  The Keating and Will Achilles article sets up Table III (page 1075) with 6 columns detailing the mean, STD and SEM of each of the two patient populations (operative and non-operative groups).  First off, the title of the table does tell us the specific outcome measure being studied, range of movement, which is a good start.  Second, it does provide us with some descriptive statistics of the patient population with the mean and STD. Unfortunately, it also includes the SEM, which is really not necessary to report in this case. The STD is a valuable descriptive statistic that is used to show how much variability exists around the mean of a single sample of the population. Whereas the SEM is a statistic that should be used as a measure of precision of the estimated mean. The SEM should not be used to describe/illustrate data variability, and generally a 95% confidence interval is a more appropriate statistic to evaluate this. And finally, the table itself fails to show us any information about comparative statistics between the two groups. To find this information, I have to go searching through the results section to find that there was no statistically significant difference in the range of movement (dorsiflexion and plantarflexion) between the operative vs. non-operative group.

In the Rajasekhar PE study, we see the tables set up slightly differently. Let’s use Table 4 on page 326 as an example. Here we appreciate from the title and labeling of the columns that (1) we are evaluating the outcome measure of pharmacologic prophylaxis, (2) there are two separate columns detailing the descriptive statistics of the two patient populations, and (3) we have a separate column detailing the results of comparative statistics that were performed. This is done so in a clear and concise manner which provides the reader with essentially all the information they could hope to glean from the results section.

A good table should contain the following information: First and foremost, the table should have a purpose, and should be used to convey information more efficiently than could otherwise be accomplished by text alone. Secondly, the type of information being conveyed should dictate the form of the table. For example, a table which is being used to illustrate a specific trend will be set up differently than a table which is simply being used to show recorded data. Third, the table should display the information of choice in a clear and organized manner so that the reader can quickly make comparisons and extract the necessary information. Fourth, as a general rule of thumb, values which are being compared should be placed adjacent to one another in the table. This helps the reader to assess the information quickly. Fifth, visual cues within the table like bold type and spacing will help to contribute to the functionality of the table. Finally, the data which is presented in the table should not be duplicated in other places within the article. It is fine to discuss the outcomes, but the tables should be where the data is displayed.

A great resource that I’ve mentioned in this space before is the book “How to report statistics in Medicine” by Lang and Secic. In addition to a lot of other information, there is a whole chapter on the thought process behind setting up a good table.