Kaipel M, Kfapf D, Wyss.  Metatarsal length does not correlate with maximal peak pressure and maximal force.  CORR.  2011 Apr; 469(4).

WHY did the authors undertake this study?
We probably don’t know as much as we think we know about the metatarsal parabola.  It is generally accepted that a relatively “long” metatarsal compared to adjacent “shorter” metatarsals will result in increased plantar pressures and the possibility of pathology including pain, callus formation and other manifestations of lesser metatarsalgia. Previous studies and evidence-based medicine don’t necessarily support this supposition however. The objective of this study was to determine whether metatarsal length correlates with plantar-loading patterns of the metatarsal heads. 

HOW did they attempt to answer this question?
There were two primary outcomes measures of this study. The first was a static, transverse plane evaluation of second metatarsal length relative to the distal extensions of the first and third metatarsals with plain film radiography. The second was a measure of dynamic pressure under the metatarsal heads as assessed by pedobarography.

Two groups of patients were studied:  a control group without lesser metatarsalgia and an experimental group with lesser metatarsalgia. Exclusion criteria of both groups included those with fixed-foot deformities, forefoot malalignment (pes cavus, pes planus, hallux valgus, hammertoes, claw toes), previous foot surgery, major trauma, diabetes, rheumatic disease, nerve entrapment, and those who had a period of conservative treatment.

WHAT were the specific results?
Very few statistically significant results were found. The authors demonstrated no correlation between relative metatarsal length and pressure, and only found that maximal force under the first metatarsal head was decreased in the metatarsalgia group compared to the no metatarsalgia group (18.38% body weight vs. 21.36% body weight; p=0.0076).

HOW did the authors interpret these results?
From these results, the authors concluded  (rather definitively) that relative metatarsal length had no influence on plantar-loading patterns, and that surgical reconstruction of the forefoot based on the metatarsal parabola may not be supported.

Eckmann MS, Ramamurthy S, Griffin JG.  Intravenous regional ketorolac and lidocaine in the treatment of chronic regional pain syndrome of the lower extremity: a randomized, double-blinded, crossover study.  Clin J Pain.  2011 Mar/Apr; 27(3): 203-206.

WHY did the authors undertake this study?
Chronic regional pain syndrome (CRPS) is a significant and debilitating diagnosis. Although podiatric physicians are rarely involved in the definitive treatment of CRPS, we may be very much involved in the initial diagnosis and referral of patients suspected to have this condition, of which we may be the cause. The objective of this study was to examine the effect of intravenous regional ketorolac on the treatment of CRPS of the lower extremity. 

HOW did they attempt to answer this question?
The primary outcome measure of this study was overall pain with a 10-point numeric rating scale assessed one week following injection administration. Secondary measures of short-term pain (assessed one day following injection), pain with motion, limb volume difference, joint pain score, range of ankle motion and skin temperature were also assessed.

A total of 10 patients completed the study of a random treatment order of 4 intravenous regional blocks performed 1 week apart with 50mL of lidocaine 0.5% plus 0, 30, 60 and 120mg ketorolac. A randomized, double-blinded, crossover study design was utilized. 

WHAT were the specific results?
Statistically significant differences were found with only one outcome:  pooled data of all ketorolac groups (30, 60 or 120mg) showed 1 day reduction in pain scores. No differences were found at the 1 week mark with any of the outcome measures between groups. 

HOW did the authors interpret these results?
Based on these results, the authors concluded that intravenous regional ketorolac and lidocaine had limited effect on the treatment of lower extremity CRPS.

Let’s take a closer look at the topic of crossover study designs, particularly as it applies to the Eckmann, Ramamurthy and Griffin CRPS study. This paper has one of my favorite general types of titles because they included “a randomized, double-blinded, crossover study” as a tagline to their title. It’s like the authors are really trying to encourage readers to take a look at their paper by pointing out how excellently they designed it right from the start!

Regardless of whether or not this information is supplied in the title or methods section, it does provide us critical readers with some important information about the design of the study:

The crossover design of this study is somewhat unique to foot and ankle surgery, so let’s take a minute with a closer look.  In most of the interventional studies that we deal with, there are two general groups: (1) an experimental group that has some type of intervention and (2) a control group that doesn’t have this intervention. We examine the “effect” of the intervention by comparing outcome measures between the experimental and control groups. A crossover study design is basically the same thing, except that the same patient is serving as both the experimental and control groups. In the CRPS study, the patients served as the control group when they received an IVRB with 0mg of ketorolac, and they actually served as 3 different experimental groups when separately dosed with 30, 60 and 120mg of ketorolac. Had this been a non-crossover study, there would have been 4 distinct groups of patients: 1 group of ~10 patients receiving 0mg of ketorolac, 1 group of ~10 patients receiving 30mg of ketorolac, 1 group of ~10 patients receiving 60mg of ketorolac and 1 group of ~10 patients receiving 120mg of ketorolac. 

You can already plainly see one of the advantages of a crossover study design – it requires far less patients reach approximately the same degree of statistical power. In our example, a non-crossover study design would require about 40 patients to get similar results to the 10 patients required for the crossover design. 

Another advantage is that it decreases the chance of having some type of confounding variable affect the results because the patients in each of the groups are exactly the same. We know that each of the patients in the 4 groups have the same co-morbidities, same age, same diagnosis, etc because it’s essentially the same patient!

So why don’t we just always utilize crossover study designs? Well the answer is that often it is just not practical or realistic.  Let’s say we wanted to design a study comparing two different types of bunion procedures. It simply wouldn’t be possible to perform one type of osteotomy on a patient, let it heal and measure outcomes measures, and then go back in and perform a different type of osteotomy and repeat the process.  Most interventions have some degree of “carry over”, or long lasting effects where the patient will not return to exactly how they were prior to the first intervention. That isn’t exactly the case for our CRPS study.  The authors felt confident that 1 week was an appropriate “washout period” so that the effect of the first IVRB didn’t carry over to the second, third or fourth IVRB. 
 
The carry over effect isn’t the only potential limitation of crossover studies, but it does show that while crossover study designs do have potential benefits, they cannot be used for all situations.