Five men and five women who were nineteen to thirty-five years old volunteered for this study. They had no history of gait abnormalities, disease involving the motor system, or deformities of the lower extremities. This study was approved by the institutional review board, and informed consent was obtained.
Each person had F-Scan sensors (Tekscan, Boston, Massachusetts) placed on the plantar surfaces of both feet. F-Scan sensors are thin (0.178-millimeter), flexible Mylar (polyester) sheets incorporating 960 sensors that are evenly distributed at five-millimeter intervals in a grid-like pattern22. Individual sensors consist of intersecting rows and columns of conductors separated by a material that varies its electrical resistance with applied load. Sequential scanning of each row and column at 100 times per second allows the computer system to record and graph the amount of force at each point of intersection during the entire gait cycle. Individual sensing cells have a pressure-recording range of eight to 124 pounds per square inch (55.2 to 855.0 kilopascals).
The sheets may be trimmed to accommodate shoe sizes ranging from a man's size fourteen to a child's size two, and they are thin enough to be inserted as an insole into footwear without altering gait patterns. Only one set of sensors was used for each person so that the inherent variability between sensors, as demonstrated by Rose et al.14, would be eliminated. The sensors were trimmed to cover the entire plantar surface of the foot, and they were held in place with adhesive tape and an overlying stockinette.
The sensors were attached to a converter unit that was secured to the leg of each volunteer with a Velcro cuff (Fig. 1). A coaxial cable connected the converter to a 486 IBM-compatible computer, which provided graphs of force versus time (Fig. 2) for the loads at the cast-foot interface at heel-strike (T1) and toe-off (T3) with use of Tekscan systems software. Data were sampled at 300 hertz during the entire gait cycle. This provided measurements of the vertical ground-reaction force vector applied to the foot inside the cast.
Simultaneous recordings of the ground-reaction force vector at the cast-floor interface were made with two AMTI OR6-5 force-plates (Advanced Medical Technology, Watertown, Massachusetts). The two force-plates, which have a maximum capacity of approximately 900 kilograms for load, were arranged end-to-end to allow ample space for walking at a normal gait while measurements were obtained. These plates are specifically designed to measure ground-reaction forces along orthogonal axes x, y, and z through the use of strain-gauges mounted on four separate components within the platforms. However, only the values for the vertical ground-reaction force vector (axis z) from the first force-plate were used in our study because they correspond to simultaneously measured heel-strike values. Data were gathered from the gauges incorporated into the plates at a frequency of 1000 hertz.
The force-plates were recessed into an 18.3-meter walking platform so that no alteration in gait pattern was necessary to cross the plates. An amplifier (AMTI MCA series; Advanced Medical Technology) connected the force-plates to the 486 IBM-compatible computer. Force-plate data were analyzed with Ariel Systems software (Ariel Life Systems, San Diego, California). This generated graphs of force versus time for the loads at the cast-floor interface at heel-strike (F1) and toe-off (F3) during the stance phase of the gait cycle (Fig. 2).
Six sets of trials were performed three times each. The values that were obtained were normalized to the body weight of each volunteer, and the values for the three trials in each set were averaged for analysis. Each volunteer was asked to walk at his or her preferred rate over the force-plates for each trial.
The first set of trials, which consisted of walking while wearing regular shoes, was performed after calibration of the sensors to the body weight of the volunteer. For the second set of trials, each person wore a fiberglass patellar ligament-bearing cast that was well molded in the medial tibial plateau and infrapatellar regions, as described by Sarmiento17. All of the casts were applied and trimmed by the same experienced cast technician. A 4.0-by-0.5-centimeter window on the lateral aspect of the cast allowed a sensor to be connected to the converter box. Once sensors had been applied and calibrated, they were not disconnected until all of the recordings were obtained.
Patellar ligament-bearing casts work best when the knee is full extension, which maximizes contact between the cast and the patellar ligament and maximizes load-bearing by the medial tibial plateau and the patellar ligament. Therefore, for the third set of trials, the leg in the patellar ligament-bearing cast was kept locked in full extension with an externally applied Bledsoe brace (Medical Technology, Grand Prairie, Texas).
For the fourth set of trials, the cast was modified into a standard below-the-knee cast with its proximal extension just distal to the tibial tubercle. This was followed by the fifth set of trials, which measured the load on the foot of the volunteers while they wore a below-the-knee cast with an overlying Bledsoe brace locked in full extension. Finally, the cast was removed and the sixth set of trials was performed with the volunteers wearing only a Bledsoe brace locked in full extension.
Statistical Analysis
All of the values for the peak load at the cast-floor interface during loading response (F1) and the peak load at the cast-foot interface during loading response (T1) are expressed as a percentage of body weight. The mean differences between the F1 and T1 values in all six sets were compared with each other with a paired t test. A p value of less than 0.01 was considered significant. The reproducibility of the variables that were studied was evaluated with Spearman correlation coefficients. An r2 value of greater than 0.95 was considered significant. In this study, all six sets demonstrated reproducibility of greater than 0.95.
We measured the percent load reduction (F1 minus T1) at peak heel-strike because this phase corresponds to the weight-acceptance phase of gait and it is representative of maximum loading of the extremity in a cast. Toe-off was not used for analysis.
As mentioned, the ten volunteers performed each of the six trials three times. Previous authors have studied changes in total pressure applied by the involved extremity at the cast-floor interface9,15. However, because each person does not walk precisely the same way every time (that is, at the same speed or with the same movements of the upper and lower extremities), we noted that there was a considerable difference in the total load applied to the involved extremity by each person in the different sets. This was shown by differences in the peak loads at the cast-floor interface during the loading response. Therefore, only the differences between the F1 and T1 values were used for each analysis. In this way, we were able to isolate load measurements on the foot in the cast as a function of the cast design.
The mean decrease in load transmission (F1 minus T1) was 26 percent (range, 7 to 53 percent) when the patellar ligament-bearing cast was worn with an overlying Bledsoe brace locked in full extension, whereas the mean decrease was only 11 percent (range, 6 to 25 percent) when the patellar ligament-bearing cast was worn without the brace (Table I). Similarly, when the below-the-knee cast was worn alone, load transmission decreased a mean of 13 percent (range, 1 to 37 percent), whereas the below-the-knee cast with a brace provided a mean decrease of 7 percent (range, 1 to 17 percent). Footwear alone caused a mean reduction in load transmission of 5 percent (range, 0 to 10 percent). There was a mean reduction in load transmission of 4 percent (range, 1 to 8 percent) when a Bledsoe brace locked in full extension was worn alone. The difference between the reduction in load provided by the patellar ligament-bearing cast with an overlying Bledsoe brace and the reductions measured during the other trials was significant (p = 0.007). No significant difference was demonstrated between the reductions measured during any of the other trials (p = 0.03) (Fig. 3).
Early functional loading of stable fractures of the lower extremity has gained widespread acceptance since it was described by Dehne et al. in 19614. Patellar ligament-bearing casts have been said to reduce ultimate loads at the fracture site, although the importance of the amount of reduction has been disputed16-18.
In our study, we did not find, with the numbers available, that the patellar ligament-bearing cast alone significantly reduced loading at the cast-foot interface relative to that at the cast-floor interface. The peak load at the cast-foot interface at heel-strike is representative of the loading of the lower extremity that would be seen at the site of a tibial fracture.
Use of the locked extension brace maximized load-bearing by the medial tibial plateau and the inferior patellar pole. Normal walking, with its typical 20 degrees of knee flexion during the early stance phase20, theoretically negated this distribution of load to the proximal aspect of the tibia and the patellar ligament; this accounted for the difference between the results with the patellar ligament-bearing cast alone and those with the patellar ligament-bearing cast with an extension brace.
In light of our findings, clinical applications of patellar ligament-bearing casts for reduction of load must be questioned. A decrease in the transmission of load to the foot was found only when a patellar ligament-bearing cast (which allows for early functional weight-bearing) and a knee brace locked in extension were worn simultaneously. The use of patellar ligament-bearing cast with a knee brace locked in extension minimizes the potential benefits of maintaining a greater range of knee motion and the documented return of function after an injury7.
We did not directly address the use of patellar ligament-bearing casts for the treatment of disorders of the forefoot and midfoot in our study because we were concerned primarily with loads transmitted to the hindfoot and the tibial shaft, as represented by the differences between the F1 and T1 values. Saltzman et al., using F-Scan sensors in a study of Charcot changes in the midfoot, were not able to detect a significant difference in the mean peak loads under the midfoot when the patients wore a patellar ligament-bearing brace15. Thus, they did not recommend the use of a patellar ligament-bearing brace. The same study demonstrated a 37 percent decrease in peak load transmission to the hindfoot. This differs from the 11 percent decrease that was found in our investigation. However, the free-motion ankle hinge that was used in their study allowed for the dispersion of generated force through eccentric plantar flexion of the ankle; this is not possible with the patellar ligament-bearing cast.
A more important aspect of our study was the fact that the values for load transmission were not recorded at the cast-floor interface only. Through the simultaneous use of force-plates and F-Scan sensors, we were able to obtain a ratio between the total load applied to the affected lower extremity (the cast-floor interface) and that applied to the foot in the cast (the cast-foot interface). We believe this to be a more accurate representation of the function of the cast design, independent of changes in gait patterns. Changes in gait patterns could decrease the load applied to the extremity in the cast, thereby falsely decreasing load measurements within the cast.
The limited durability of the F-Scan insole sensors may be a confounding variable in our investigation. According to Rose et al., the sensors may be used for approximately thirty gait cycles; after this, the accuracy of the pressure recordings begins to decrease steadily secondary to wear on the individual pressure-sensing cells14. We therefore limited the recording during each trial to approximately three or four gait cycles. We also completed the trials with the patellar ligament-bearing cast and the patellar ligament-bearing cast with a brace early, when the F-Scan sensors were deemed most accurate.
Our decision not to exchange the F-Scan sensors was based on the work of Rose et al., which demonstrated considerable variability in pressure recordings between different sensors14. Our results demonstrated a reduction in the pressure recordings when the patellar ligament-bearing cast was worn with a Bledsoe brace but an increase in the pressures recorded in subsequent trials with the patellar ligament-bearing cast alone, the below-the-knee cast with a brace, and the below-the-knee cast without a brace. If limited durability (decreased sensitivity of the cells) were indeed a confounding factor, the later trials would have shown a reduction in load, not the increased values that we observed. If anything, the decreasing sensitivity of the F-Scan sensors falsely narrowed the difference between the patellar ligament-bearing cast with a Bledsoe brace and the other braces that we tested.
Maintenance of rotational control of the proximal aspect of the tibia is another proposed benefit of a patellar ligament-bearing casts. This issue was not addressed in the present study, and it remains a theoretical advantage that needs further investigation. However, many authors have reported excellent clinical results after the treatment of low-energy fractures of the tibial shaft with a patellar ligament-bearing cast1,12,23. The results of our study do not refute those reports, and we are not proposing a new method of conservative management. However, we did not find that a patellar ligament-bearing cast used without maintenance of full extension of the knee afforded a greater reduction of load to the lower extremity than did a traditional below-the-knee cast. We therefore agree with Svend-Hansen et al. that the choice of treatment must be based on other factors21.