Postacute phase (incudes commentary and author reponse)(In Vitro Acetabular Contact Pressures During Rehabilitation,part 2).

To examine hip contact pressures during the acute and postacute phases of rehabilitation, we conducted a two-part study of a single subject with a pressure-instrumented femoral head prosthesis. This report compares the magnitudes of peak acetabular contact pressures measured during selected ambulation activities and therapeutic exercises in the postacute phase of rehabilitation following hemiarthroplasty. The postacute period for this study began on the day of discharge from an acute care hospital and continued to the final data-collection period, a total of 62 months, following which the patient died from unrelated causes.

Rehabilitation Programs During the Postacute Phase

Little information has been published on hemiarthroplasty rehabilitation treatment programs, and essentially no guidance exists to assist the clinician recommending activities following discharge from the acute care hospital. Although total hip replacement (THR) protocols serve as models for the acute postoperative rehabilitation of patients who have undergone hip hemiarthroplasty, even THR data are lacking for the postacute phase of rehabilitation. (1) Typically, these patients ambulate either with partial weight bearing (PWB) up to 3 months postoperatively (1-3) or with weight bearing to tolerance beginning shortly after surgery. (4,5) Most patients with a THR are full weight bearing (FWB) without a walking aid by 3 months postoperatively. (3,5) Touchdown weight bearing (TDWB) and non-weight bearing (NWB) are often not prescribed for patients following hip arthroplasty but are prescribed for the presumed load reduction benefit on hips with unstable fractures and painful osteoarthritis. (4,6) Specific exercise regimens for the postacute phase of rehabilitation following hip arthroplasty are lacking in the literature. Acute care exercises include isometric hip abduction (ABD) exercises (5,7) or active abduction exercises in the supine position. (1,3) Straight-leg-raising (SLR) exercises are components of some rehabilitation programs. (3) Others suggest that SLR exercises are too stressful for the hip and prohibit SLR exercises. (2) Decisions solely based on observation and experience create inconsistencies in hip arthroplasty rehabilitation programs. Comparative data on hip stresses during common rehabilitation activities could contribute to guideline formulation for postacute phase physical therapy.

Data reported in this article supplement those of prior reports for this patient. (7,8) No data have been published on hip pressures elicited during active range of motion in the acute postoperative period, nor were these prior data consistently reported from the same transducer or anatomical area. This study extended the level walking data to 62 months, followed a single femoral head area (transducer) over time, examined other activities for peak pressure changes over time, and compared peak pressures from activities measured during the same or similar test sessions.

The review of current clinical practices and data from instrumented hip studies (see the accompanying report on the acute phase of rehabilitation in this issue) suggested a hypothesized rank order, from lowest to highest peak pressure, of the activities studied: NWB ambulation, TDWB ambulation, SIR exercise, PWB ambulation, ABD exercise, and FWB ambulation. We further hypothesized that this order would not change over the 5-year data-collection period, but peak pressures were expected to increase most during the first year, followed by stabilization or decline over the 12-to-62-month period.

Method

Subject

The patient, an active elderly woman, was described in the part I report. The patient's posthospitalization rehabilitation and recovery were uneventful. She walked with crutches at 2.5 postoperative weeks and demonstrated a compensatory limp when asked to walk without crutches. The patient exhibited a clinically normal-appearing gait by 6 months, jogged on a treadmill at 2 m/s by 1 year, and could rise from a chair and negotiate stairs easily by 18 months. (8) Details of the patient's range of motion and strength have been reviewed and discussed in the part I report and in a prior report by Krebs et al. (7) Her height of 1.68 m and weight of 68 kg remained constant throughout the 62-month study.

Instrumentation

The pressure-instrumented endoprosthesis described in the part I report also provided data for this portion of the study. The effective sampling rate from each pressure transducer during this study was 253 Hz. A digital computer * was used for real-time data acquisition and storage. Pressure data were low-pass filtered at 10 Hz.

Immediately following implantation, 10 of the 14 transducers satisfied the calibration requirements of the project engineers. (8,9) The signal from transducer 1 (see part I report, Fig. 1) of the 10 original functioning transducers became unreliable in year 2 or 3 of the study. Only data from the 9 transducers (transducers 2-10) available throughout the test period are reported.

Procedure

The data collection was conducted in the MGH Biomotion Laboratory (Boston, Mass) between July 1984 and August 1989 at approximately 4- to 6-month intervals (Figs. 1 and 2). In addition to the six selected activities, other activities were studied during the test sessions. Often data for two or three of the six activities were collected during the same session; however, at no time were all six activities studied during the same test session (Figs. 1 and 2). Instructions and demonstrations provided to the patient for the activities were similar to those routinely used clinically.

[FIGURES 1-2 OMITTED]

FWB ambulation. The patient ambulated at free speed, without the use of a walking aid, placing full weight on the lower extremity with the implanted endoprosthesis. The walkway was approximately 10 m long, with the force platform located midway within the course.

PWB ambulation. The patient ambulated along the walkway, advancing the lower extremity with the implanted endoprosthesis forward with both crutches; the opposite lower extremity was then brought to or through the crutches. She was told to bear partial weight (approximately 13.6 kg [30 lb]) on the lower extremity with the implanted endoprosthesis. Prior to the PWB ambulation trials, the patient practiced placing 13.6 kg of her weight on the limb using a mechanical bathroom scale.

TDWB ambulation. The patient ambulated along the walkway using the same crutch-gait sequence described for PWB ambulation, with the exception of the weight-bearing instructions. She was instructed to touch the foot of the lower extremity with the endoprosthesis on the ground, allowing only 4.5 kg (10 lb) of her weight on it. The patient practiced using the bathroom scale prior to the TDWB ambulation trials.

NWB ambulation. The patient assumed a tripod stance with the crutches and the left lower extremity. She then ambulated the length of the walkway by swinging to or through the crutches with the left lower extremity. The patient was instructed to keep the foot of the right lower extremity with the implanted endoprosthesis off the floor.

ABD exercise. The patient performed a maximal voluntary isometric contraction (MVIC) with the lower extremity with the implanted endoprosthesis. A researcher opposed the abducting limb by placing his hand lateral to the supine patient's ankle. The patient was asked to push as hard as she could against the researcher's hand.

SLR exercise. The patient performed the SLR exercise with the limb with the implanted endoprosthesis while lying supine on a hard, flat surface. She was instructed to keep the knee straight while raising and lowering the limb.

Data Analysis

Data collected in the MGH Biomotion Laboratory from 2.5 weeks to 62 months postoperatively were analyzed retrospectively. Fifteen test comparisons were available for analysis and were numbered consecutively starting with comparison 1 at 2.5 weeks. Because many activities were tested in each session, only one trial of each activity was analyzed in this study. Because only one activity of interest to this investigation was studied in each session at 53, 57, and 60 months and because the patient was likely to be physiologically stable over 4 months during the fourth year, the data from month 53 were grouped with the data from month 57 (comparison 14) and the data from month 60 were grouped with the data from month 62 (comparison 15).

The transducer selected for each activity was the transducer that most frequently recorded the maximum peak pressure across test sessions for the given activity. The designated transducer peak pressures were then compared for each activity across trials. When more than one trial for an activity was conducted during a given comparison period, the trial with the maximum peak pressure for the selected transducer was recorded. Except for one activity (PWB ambulation) in which the maximum peak pressures occurred with equal frequency between two transducers, a single transducer was selected for each given activity for study across sessions. Hence, the data reported in this article may differ from the previously reported data for this patient. Based on frequency for recording the maximum peak pressure, all FWB and SIR trial data are from transducer 3, all ABD and NWB trial data are from transducer 5, all PWB trial data are from transducer 3 or 5, and all TDWB trial data are from transducer 2 (see part I report, Fig. 1). The activities were ranked and compared across trials to determine whether the hypotheses were supported or rejected. Causes for rejection of the hypotheses were one activity out of the predicted rank order and differences of less than 0.2 MPa between successive activities in the hypothesized rank order. Because the study involved a single subject, it was believed prudent to use this conservative basis for acceptance or rejection of the hypotheses.

Results

Results are summarized in the Table and Figures 1 and 2. The Table shows that the rank-order hypothesis was supported on nine occasions and rejected on six occasions. The greatest rise in peak pressures occurred between 2.5 weeks and 6 months for the FWB, PWB, and SIR activities (Figs. 1 and 2). By 1 year postdischarge, FWB, PWB, NWB, and SLR contact pressures reached maximum levels, with a tendency to stabilize or decline. Isometric hip abduction exercise, an effort-dependent activity, did not demonstrate this trend, particularly after 36 months. The maximum peak pressure for ABD exercise was observed at 53 months, and peak pressures for the ABD activity varied by as much as 1.9 MPa among trials.

At 15 months, NWB trials performed with the knee flexed and the hip neutral resulted in a lower peak pressure (2.4 MPa) than did NWB trials performed with the knee extended and the hip slightly flexed (3.5 MPa). The NWB activity was performed with the knee extended at 11, 12, and 15 months, which may explain the higher pressures recorded on these occasions.

The highest peak pressure recorded for the six activities was 6.5 MPa; this peak pressure, however, occurred during a putative TDWB ambulation trial at 11 months when the patient incorrectly performed this method of ambulation. The patient exerted 710 N of force, or 1.06 times body weight, on the involved limb during this TDWB ambulation trial, according to force-plate analysis. This trial was omitted from the data analysis because force-plate criteria for TDWB ambulation were not met. The lowest peak pressure recorded for the six activities was 1.1 MPa during NWB ambulation at 28 months.

Discussion

The magnitude of in vivo hip contact pressures followed the predicted rank-order relationship on 9 of the 15 test comparisons. The NWB ambulation trials most frequently violated the predicted rank order with higher-than-expected pressures. The SLR exercise trials least often violated the predicted rank order. These results indicate that prescribed weight bearing and exercise type were not good predictors of hip contact pressure magnitudes across activities. Factors such as walking speed and muscle forces may explain the unexpected results.

FWB Ambulation

Compensatory limp and velocity changes were probably responsible for differences in peak pressures observed between FWB and PWB ambulation at 2.5 weeks. The patient was observed to have a slow gait and compensatory limp when walking unsupported at 2.5 weeks, a time when ordinarily she relied on crutches for functional ambulation. Increased walking velocity results in increased hip moments and increased hip contact forces. (10) Likewise, decreased velocity decreases the magnitude and rate of muscle force development during walking. (11(p327)) In vitro studies using this prosthesis design have shown that peak contact pressures vary directly with applied force. (12) Hip peak pressures, therefore, probably vary directly with velocity and muscle activity during FWB ambulation.

Hodge et al (8) reported that contact pressures measured during level walking over the first 3 years peaked by the first year, stabilized, then declined. Full weight-bearing ambulation peak pressures measured over time coincide with time-related changes in gait characteristics previously reported in THR patients. (13) The greatest increase in velocity, cadence, and stride length occur during the first 6 postoperative months, with further improvement up to 2 years. Little change in gait is observed between postdischarge years 2 and 4.

PWB Ambulation

Decreased velocity and decreased muscle forces during PWB ambulation may explain the lower peak pressures relative to FWB ambulation in trials conducted from 6 to 28 months, when the patient's gait appeared clinically normal. Velocity of ambulation is significantly less when walking with crutches than when walking unassisted in healthy subjects. (14) With the exception of the 2.5-week trial, PWB peak pressures ranged from 65% to 80% of those for FWB ambulation trials recorded on the same occasion. Although PWB ambulation apparently does reduce the internal pressures on the hip relative to FWB ambulation, FWB ambulation in this patient did not reduce the pressure on the hip to levels sought clinically (ie, 25%-50% of weight bearing). Opila et al (15) found that patients with total hip replacements with identically prescribed aids and gait patterns adopt idiosyncratic styles when using a walking aid. The belief that patients are placing a specific prescribed load on the hip is questionable in light of the results of our study and the study by Opila et al. Further studies of a larger number of patients with instrumented hips could elucidate the variability of PWB ambulation and other prescribed gait patterns.

DWB Ambulation

The TDWB ambulation trial data should be interpreted with caution because the TDWB pressures were collected long after the acute phase and may not be representative of the acute or subacute postoperative hip. These trials, however, do illustrate differences that can occur within one subject performing the same prescribed gait on different occasions. The high peak pressure of 6.5 MPa and the low peak pressure of 2.0 MPa portray the variation possible with presumed TDWB ambulation. The two trials representative of TDWB ambulation at 12 and 15 months show that TDWB ambulation can decrease stress on the hip relative to FWB and PWB ambulation, providing the TDWB ambulation is performed correctly.

NWB Ambulation

Non-weight-bearing ambulation generated peak pressures that exceeded TDWB peak pressures and equaled PWB peak pressures. This relationship is probably a result of the muscle forces acting across the hip to maintain the position of the non-weight-bearing limb. When the knee of the NWB limb is extended (as in the 11-, 12-, and 15-week trials), the effective lever arm of the limb's weight is longer compared with when the knee is flexed, resulting in a larger moment about the hip joint. A larger required moment would result in higher muscle forces. Joint compression increases with increased muscle force, (11(p185)) and such increases may explain the higher peak pressures during ambulation with the knee extended.

Rydell (16) also found that the position of the non-weight-bearing limb affected force measurements made in vivo. Differences in instrumentation methods prevent making direct comparisons between our study and that of Rydell. The low peak pressures during NWB ambulation with the knee flexed warrant further study in future investigations with instrumented hips.

ABD Exercise

The high peak pressures measured during resisted isometric ABD exercise were likely a result of high muscle force production. In a previous study involving this patient, Krebs et al (7) compared contact pressure magnitude during resisted isometric ABD exercise to force transducer and electromyographic recordings in the third and fourth postoperative years. Increases in contact pressure magnitude coincided with increased abductor torque and electromyographic activity. Increasing the angular velocity during exercise increased the magnitude of hip contact pressures.

The rise and fall in ABD peak pressures observed in the later years of this study probably resulted from variations in effort rather than changes in abductor torque capability. Murray and colleagues (13) found that maximum voluntary dynamometer abductor torque production usually peaks by the second postoperative year in patients with a THR.

The results of this study support the recommendations of Krebs et al. (7) Magnitude and rate of muscle force development should be considered when selecting exercises for the postoperative hip. If control of acetabular contact pressures is desired in the postoperative hip, exercises such as slow, supine active abduction would be favored over resisted isometric abduction.

SLR Exercise

Peak pressures during SLR exercises showed little variation over repeated trials. A surgically induced decrease in the ability of the abductor muscles to generate torque may explain the finding that ABD peak pressures were no different from SLR peak pressures at 2.5 weeks. On two later comparisons, when ABD and SLR peak pressures were measured on the same day, ABD peak pressures exceeded SLR peak pressures by as much as 0.9 MPa. Peak pressure magnitudes during SIR consistently fell below PWB and FWB ambulation peak pressures. The evidence supports including SLR exercises in rehabilitation protocols when PWB or FWB ambulation is permitted.

Clinical Implications

Data from a single subject were used for this study, and the results may not be generalizable to others. This study, however, provides unique data of the pressures exerted on a femoral endoprosthesis over 5 years. Further studies with a larger number and variety of patients are needed to confirm or refute these results.

The results of this study support the evidence that aided gait can reduce the stress on acetabular cartilage in a hip with a femoral endoprosthesis. Partial weight-bearing, TDWB, and NWB ambulation reduced hip stress in this patient but did not always follow a hierarchical relationship based on degree of prescribed weight bearing. Walking aids may be indicated for up to 30 months following femoral arthroplasty if high articular contact pressures are to be avoided during ambulation (Fig. 2). Exercises such as SIR and gravity-reduced abduction that do not produce high muscle forces could be favored over resisted isometric exercise to reduce acetabular contact pressures, but they might produce lower muscle performance gains.

Conclusions

Rehabilitation programs for the patient with a subacute postoperative hip warrant scrutiny considering the present biomechanical findings. Data obtained from the single subject in this study suggest that (1) PWB ambulation is one method to lessen both muscle forces and velocity, and therefore peak acetabular contact pressures, during ambulation; (2) TDWB ambulation, when performed correctly, generates less acetabular stress than does PWB, FWB, or NWB gait with the knee extended; and (3) active submaximal exercises produce lower acetabular contact pressures than those produced by maximal resisted hip isometric exercises. Straight-leg-raising exercises can be included in rehabilitation programs when PWB or FWB ambulation is permitted.

Postoperative gait recovery and hip pressures in this patient coincide with typical changes in the gait characteristics of the patient with a THR. Resisted isometric abduction, which is patient effort dependent, showed the least consistency over time. Further study of a larger number of patients with instrumented hip prostheses is needed before these findings can be generalized to all patients in hip rehabilitation programs.

Table. Comparison of the Predicted Rank Order with the
Resultant Rank Order During Test Comparisons

   Test Comparison (b)

Activity (a)     1       2   3       4   5   6   7

NWB              1           1       2   2   1
TDWB                                 1   1
SLR              2 (c)   1
PWB              5       2   2 (c)       2   2
ABD              3 (c)                           1
FWB              4       3   3           3   3   2
Hypothesis (d)   R       S   R       R   R   S   S

   Test Comparison (b)

Activity (a)     8   9   10   11   12   13   14   15

NWB              1
TDWB
SLR                                     1         1
PWB              2
ABD              3   1   1    2    1    2    2    2
FWB              4   2   2    1    2    3    1    3
Hypothesis (d)   S   S   S    R    S    S    R    S

(a) Listed top to bottom according to hypothesized order:
NWB=non-weight-bearing ambulation; TDWB=touch-down
weight-bearing ambulation; SLR=straight-leg-raising exercise;
PWB=partial weight-bearing ambulation; ABD=abduction exercise;
FWB=full weight-bearing ambulation.

(b) Test comparison numbers refer to time after surgery as
indicated in the text. Peak pressure magnitudes are ranked
from lowest to highest; see Figures 1 and 2 for absolute
pressure magnitude data for each activity.

(c) Differences between successive activities are <0.2 MPa.

(d) S=supports hypothesis; R=rejects hypothesis.

* PDP 11/60, Digital Equipment Corp, 146 Main St, Maynard, MA 01754.

References

(1) Kisner C, Colby LA. Therapeutic Exercise: Foundations and Techniques. Philadelphia, Pa: FA Davis Co; 1985:345.

(2) Chandler JL. Postoperative rehabilitation of the total hip patient. In: Stillwell WT, ed. The Art of Total Hip Arthroplasty. Orlando, Fla: Grune & Stratton Inc; 1987:372-401.

(3) Eftekhar NS. Total hip replacement using principles of low-friction arthroplasty. In: Evarts CM, ed. Surgery of the Musculoskeletal System. New York, NY: Churchill Livingstone Inc; 1983:188-190.

(4) Rockwood CA, Green DP. Fractures. Philadelphia, Pa: JB Lippincott Co; 1975;2:1026-1027.

(5) Beber CA, Convery FR. Management of patients with total hip replacement. Phys Ther. 1972;52:823-828.

(6) Amstutz HC, Kim WC. Osteoarthritis of the hip. In: Moskowitz RW, Gelberman RH, Mankin HJ, Treadwell BV, eds. Osteoarthritis.. Diagnosis and Management. Philadelphia, Pa: WB Saunders Co; 1984:427.

(7) Krebs DE, Elbaum L, Riley PO, et al. Exercise and gait effects on in vivo hip contact pressures. Phys Ther 1991;71:301-309.

(8) Hodge WA, Carlson KL, Fijan RS, et al. Contact pressures from an instrumented hip endoprosthesis. J Bone Joint Surg [Am]. 1989;71: 1378-1386.

(9) Hodge WA, Fijan RS, Carlson KL, et al. Contact pressures in the human hip joint measured in vivo. Proc Natl Acad Sci USA. 1986;83: 2879-2883.

(10) Crowninshield RD, Brand RA, Johnston RC. The effects of walking velocity and age on hip kinematics and kinetics. Clin Orthop. 1978;132: 140-144.

(11) Soderberg GL. Kinesiology: Application to Pathological Motion. Baltimore, Md: Williams & Wilkins; 1986:327.

(12) Rushfeldt PD, Mann RW. Improved techniques for measuring in vitro the geometry and pressure distribution in the human acetabulum, II: instrumented endoprosthesis measurement of articular surface pressure distribution. J Biomech. 1981;14:315-323.

(13) Murray MP, Gore DR, Brewer BJ, et al. Joint function after total hip arthroplasty: a four-year follow-up of 72 cases with Charnley and Muller replacements. Clin Orthop. 1981; 157:119-124.

(14) McBeath AA, Bahrke M, Balke B. Efficiency of assisted ambulation determined by oxygen consumption measurement. J Bone Joint Surg [Am]. 1974;56:994-1000.

(15) Opila KA, Nicol AC, Paul JP. Forces and impulses during aided gait. Arch Phys Med Rehabil. 1987;68:715-722.

(16) Rydell NW. Forces acting on the femoral head prosthesis: a study on strain gauge supplied prostheses in living persons. Acta Orthop Stand. 1966;37(suppl 88):1-132.

Commentary

The following commentary is on parts I and II of "In Vivo Acetabular Contact Pressures During Rehabilitation."

The authors of these two articles have presented an interesting paradigm for investigating some of the implicit assumptions on which we base much of our clinical practice as well as a provocative model for clinical studies. This study was based on limited data from a single subject. These data were derived from the sustained work of a team with expertise and resources rarely available to the clinical investigator or clinician in physical therapy. Consequently, the physical therapist in a more traditional clinical setting may be tempted to dismiss the study and the results as being so far from the mainstream of physical therapy as to be irrelevant, or, conversely, the therapist may be so impressed by the sophisticated technology that treatment protocols will be revised as suggested by Strickland et al. The goal of this commentary is to suggest that a third alternative--to ponder and discuss--may be more suitable at this time.

These two articles provide an enlightening perspective that we certainly should consider, but they do not provide the final answers regarding hip joint pressures during typical rehabilitation activities. Instead, the work generates many questions for all of us to ponder as we prescribe exercises and teach activities of daily living to our patients. One of the goals of any research project is to generate new ideas and discussion, and these authors are to be commended for doing just that.

As the report by Strickland et al and that by Givens-Heiss et al represent two parts of a larger body of work, I will discuss them as though they are a single report. These two articles describe the use of the same technology, report data from the same subject, and discuss some of the same limitations. Thus, I feel justified in directing my comments to both reports simultaneously. I must note, however, that I believe Givens-Heiss et al have been considerably more circumspect in their conclusions, and I believe that similar caution is required of the reader. This commentary will address issues raised by the design of this two-part study, the methodology used, and finally the conclusions drawn.

The most obvious concern regarding design in this body of work is the use of a single subject. The authors certainly acknowledge this limitation, but I believe further discussion is warranted. The use of a single subject prevents generalizing the results in any way to any other patient or patient population. In addition, the use of an endoprosthesis for treatment of a hip fracture clearly precludes the acquisition of pretreatment data. The absence of such baseline data hinders the generalizability still further because it is impossible to determine whether the pressure patterns exhibited by this patient are indicative of the patient's prefracture or postfracture movement patterns or result from this device or class of devices. Many of the tasks analyzed were measured only twice, so the normal variability of the measurements cannot be assessed. Without knowledge of the normal distribution of the measurements, it is impossible to ascertain the significance of the observed differences. Despite the inability to generalize from the data reported, the ability to record in vivo joint pressures is exciting and worthy of an audience by raising our consciousness regarding the possible impact of our interventions.

Another interesting issue raised by the design of this investigation is the use of retrospective data to draw correlations between function and standard clinical measures of pain, strength, and flexibility. The authors tried to retrieve clinical data on pain, range of motion (ROM), muscle strength, and ambulatory status from the patient's medical record. Such information should be standard elements of a physical therapy assessment, reassessment, and discharge summary and therefore should be readily accessible retrospectively. The authors, however, reported that pain medication levels were the only indication of pain; manual muscle test grades were only available for the quadriceps femoris and hip abductor muscles and only reported once; and active hip flexion ROM (apparently the only ROM studied) was only reported four times, the last of which occurred 1 week prior to discharge. Admittedly, teamwork and foresight among the investigators and the clinicians might have improved documentation, but we clinicians must also recognize that quantifiable documentation of standard clinical measures can and should serve as the best source of data for the evaluation of interventions.

Reliability of clinical measurements is an important issue to be addressed in any investigation. Any physical therapist who wants to report change or needs to see another therapist's patient must use measures that yield reliable data. The authors did not address the reliability of the clinical measurements they obtained, but precious few data were available. The reliability of data obtained with analog scales of pain--visual or not--is well established, and the use of such a measure need not impinge on treatment time. Strength and ROM measures are well known to us and should not be reserved for the patient with adhesive capsulitis or the patient with a spinal cord injury, or even for the research situation. Retrospective use of the clinical record is a very sensible approach to studying the effects of interventions. Its value, however, is directly dependent on the adequacy of the clinical record. The regular use of standardized, repeatable measures will allow each of us to critically evaluate our own interventions and will also allow us to participate as integral members of an investigatory team when new technology, as demonstrated in these reports, becomes available.

The methodology used in this work also deserves some discussion. The variable reported by the authors is pressure derived from 10 pressure transducers embedded in the head of a femoral endoprosthesis. As the authors noted, pressure is force per unit of area and should equal the hip joint reaction force if the pressures of infinitesimally small areas over the entire surface of the femoral head were summed. The authors report data based on 10 working transducers (out of 14 originally applied), which cover roughly 3.5% of the surface area of the femoral prosthesis (calculated from the dimensions of the prosthesis and transducers). The authors apparently assumed that the peaks recorded from those 10 transducers were indeed the peak pressures sustained by the femoral head. It is possible, however, that peak pressures actually occurred somewhere else on the femoral head, where no recording was made.

Data from only one transducer were used in part I of the study, which examined the acute phase of recovery. The authors ranked the functional tasks in order of the peak pressures. They state, however, that the analyzed data were collected from the transducer that "most frequently recorded the maximum peak pressure for all test sessions and for all activities [italics added]7' Their statement can actually be interpreted several ways. One interpretation is that as the data reported were from the transducer that recorded maximal peaks most frequently, other transducers may actually have recorded a peak in at least one activity; however, these data were not utilized in this study. The authors' statement also leaves open the possibility that the only analyzed data were from the transducer that reported the most frequent peaks. This interpretation would leave open the possibility that when the rest of the data are analyzed, new maxima may be identified. In either case, the reader is left wondering whether the peak pressures reported are truly the peaks that were recorded or whether other maxima are just not yet identified.

Further investigation is needed to determine whether the peaks recorded by any of the 10 transducers reflect the pressure patterns of the rest of the femoral head. Additional clarification is also necessary to explain the relationship of the pressures recorded by the 10 transducers before definitive conclusions can be drawn about the stressfulness of given activities. Although the data the authors have presented are an important reminder to all clinicians that some apparently innocuous exercises and activities of daily living may be more stressful than we have ever appreciated, it is probably premature to conclude whether or how treatment approaches should be revised. Clinicians must certainly consider the effect of our exercise protocols, modalities, and other interventions on underlying structures for each patient. Yet we must remain cautious when trying to apply results from new and limited studies directly to the clinical setting. Regardless of the final results of this investigation, these data will hopefully make us more temperate in our acceptance of intuitively appealing ideas. If isometric exercises yield large joint pressures and if toe-down weight bearing generates smaller stresses than non-weight bearing, then we should be stimulated to test other comfortable assumptions. For instance, we could question the importance of lower-extremity strength in walking or the validity of isokinetic testing for assessing functional capacity.

In looking at the meaning of the data reported in the present work, we also need to consider the physiological implications of "pressure." The word itself has taken on very negative connotations within society as well as in physical therapy. Consider "job pressures," "pressure sores," and "joint pressures." Although the authors point out that cartilage lubrication and cartilage synthesis are dependent on joint pressures, the focus of the work appears to be on minimizing joint pressures. The pressures reported for both the acute and chronic phases (<6.0 MPa) are well below the reported ultimate strength of articular cartilage (10.0 MPa). (1) These pressures also appear to be transient in nature, although some pressures such as those produced during stair climbing and walking occur cyclically. Hodge et al (2) note the potential dangers of repeated low-level stresses contributing to fatigue and degeneration of cartilage, subchondral bone, and prostheses and their fixations. Although such negative effects are possibilities, the logical solution--minimization of joint pressures through more prolonged bed rest--may have equal or more deleterious effects. Perhaps prolonged unloading of the joint will interfere with cartilage metabolism of the residual cartilage in hemiarthroplasties or of the surrounding joints. (3) Sustained unloading might inhibit the actual healing of the surrounding bone, slowing the fixation of the uncemented prosthetic device. (4) Prolonged unloading will assuredly have a negative effect on muscle strength, which has been demonstrated to be important in shock absorption. (5)

Thus, in an attempt to minimize joint pressures, we might actually be setting the stage for further joint destruction. The authors do acknowledge that joint pressure reduction is one alternative to be considered, but the proposed protocol revision implicitly accepts the need for pressure reduction. Just as lightweight prostheses might be desirable in some patient populations and not in others, joint pressure reduction or enhancement may be patient specific. In short, although it is appealing to immediately seek rehabilitation approaches to minimize joint pressures, we need look no farther than the National Aeronautics and Space Administration's (NASA) and the former Soviet Union's efforts to minimize space sickness to realize that pressure is an essential force in tissue metabolism. (6) In this day of weight reduction and cost containment, less is not always better.

Regardless of whether one accepts or refutes the questions raised about the methodology of this research, the authors draw conclusions from these data that must also be questioned. As stated earlier, the data reported in this work are derived from a very limited sample of pressure sites of unknown variability in the subject. The reader must also ask whether the differences in pressures, even if found to be statistically significant, are of physiological significance. In the acute phase of recovery, the maxima of all the activities varied from 3.44 to 4.79 MPa, a difference of 1.35 MPa. Similarly, the minima varied from 1.76 to 2.50, a difference of 0.74 MPa. Are these differences really physiologically important? I cannot personally answer that question, but I believe it must at least be considered before treatment protocols are established based on the belief that these values are physiologically meaningful.

In conclusion, this work demonstrates the usefulness of technology to address basic questions in physical therapy. The design and methodology have significant limitations, which reduce the immediate clinical applications of the results. The importance of this work, however, lies not in the specific data reported, but in its demonstration of the usefulness of sophisticated technology to ask questions intrinsic to physical therapy. Of equal importance are the questions that the reported results raise about some basic assumptions we make in the clinic. The authors are to be applauded for their efforts and for the very fact that their work generated further discussion.

Carol A Oatis, PhD, PT
Associate Professor
Department of Physical Therapy
Beaver College
Glenside, PA 19038-3295

References

(1) Cochran GVB. A Primer of Orthopaedic Biomechanics. New York, NY: Churchill Livingstone Inc; 1982.

(2) Hodge WA, Fijan RS, Carlson KL, et al. Contact pressures in the human hip joint measured in vivo. Proc Natl Acad Sci USA. 1986;83: 2879-2883.

(3) Volpi M, Katz EP. On the adaptive structures of the collagen fibrils of bone and cartilage. J Biomech. 1991;24(suppl 1):67-77.

(4) Simske SJ, Guerra KM, Greenherg AR, Luttges MW. The physical and mechanical effects of suspension-induced osteopenia on mouse long bones. J Biomech. 1992;25:489-499.

(5) Paul IL, Munro MB, Abernathy PJ, et al. Musculo-skeletal shock absorption: relative contribution of bone and soft tissues at various frequencies. J Biomech. 1978;11:237-239.

(6) Carter DR, Wong M, Orr TE. Musculoskeletal ontogeny, phylogeny, and functional adaptation. J Biomech. 1991;24(suppl 1):3-16.

Author Response

We thank Dr Oatis for her commentary. We are pleased to engage in further discussion of our work and agree that more work must be completed before optimal rehabilitation programs can be promulgated.

The historical basis of many of today's hip rehabilitation programs can be traced to inferences from Rydell's 1966 study, (1) which reported data from two subjects during a single session at the sixth postoperative month. Our work challenges some clinical assumptions that resulted from Rydell's work. We address here four of Dr Oatis's comments in an effort to stimulate further "ponder[ing] and discuss[ion]," anal, we hope, action.

1. Do any of the 10 transducers reflect the pressure patterns of the rest of the femoral head? Extensive in vitro testing preceded the implantation. (2,3) In the MIT Hip Simulator, (4) rapid manipulation of a femoral sphere, identical to the implant used in our two-part study, provided 253 virtually simultaneous pressure data readings for each load condition and time. These extensive data permitted spatial description of the overall pressure field and direct comparison with the applied load measured with a separate transducer. Mathematical spatial integration of the pressures into force agreed with the directly measured force, always within 5%. For a metal ball well fitted to the acetabular effective diameter (such fit being critical to clinical longevity of the surgical procedures (5)), the overall pressure distribution, although exhibiting surprisingly high pressure gradients with distance and high maxima, is always smooth. Thus, in vivo, the highest-reading transducer is close to the highest actual pressure experienced by the acetabular cartilage. We reported only the highest-recording transducer as a measure of the relative force experienced by acetabular cartilage. Our extensive in vitro series supports the meaningfulness of this approach.

In vivo, with at most 10 active transducers reporting, we have shown that we cannot reconstruct the overall pressure distribution to the confidence level desired (ie, within 5%, as in the in vivo studies (6)). We have underway in vitro experiments in the MIT Hip Simulator on this patient's retrieved prosthesis and explant acetabulum in which we can simulate the in vivo conditions in order to generate more detailed pressure distribution data and indirectly establish the overall force that was experienced by the hip joint in life. In addition, a direct force and pressure measuring implant is under development, (7) using force transducers in the neck of an endoprosthesis that also carries the same pressure instrumentation. This new Moore-type prosthesis should shed more light on Dr Oatis's question.

2. Fixation of the uncemented prosthetic device. An important specific correction of the commentary is that the implant used in our two-part study is cemented into the femoral shaft.

3. How pressures affect cartilage and what pressures are tolerable or therapeutic. Cartilage in diarthrodial joints must experience cyclical variations in pressure, induced, for example, by gait. These variations serve to augment diffusion processes in transporting nutrients through the collagen-proteoglycan matrix to the chondrocyte cells, removing metabolic products from the cells, and distributing the cellular products to replenish the matrix. The important clinical question is, What are acceptable, or desirable, pressures?

Prior to our in vitro and in vivo data, only estimates of the average pressure in a joint such as the hip were available, achieved by dividing estimates of the load across the joint by estimates of the apparent area of contact in the joint. Pressures in the human hip joint were thought to be 2 to 3 MPa at most. By comparison, our measurements of maximum pressure in gait of typically 5 MPa and in rising from a chair of at most 18 MPa were surprising. But are even such pressures injurious to the cartilage? Dr Oatis's reference to the ultimate strength of articular cartilage (10 MPa) is for the failure of cartilage tissue in tension, which is not directly relevant to clinicians because the major loading of cartilage is compressive. (8)

Our research (8) based on mathematical models of the joint, using measurements of the specific joint geometry and based on knowledge of cartilage physical properties, indicates that high pressures will not damage a natural joint, where (we have shown) the opposing cartilage surfaces are congruent, or, in the case of hemiarthrosis, where the fit of the ball to the acetabulum is good. In such a well-fitted joint, we have shown that 90% to 95% of the load is not supported by the cartilage at all, but rather by the hydrostatic pressure of the fluid in the tissue (typically 70% to 80% by weight). (8) If the fluid in cartilage can escape from the tissue upon loading, however, the matrix cannot support much more than 1 MPa and will be squeezed flat and destroyed in the process. Thus, our reports (9,10) of pressures up to 18 MPa, and routinely 5 MPa, are strong evidence that fluid pressure and not the cartilage tissue carries the preponderance of the applied compressive load. Note, however, that this analysis applies only if the joint fits well, for if it does not, fluid leaks out easily at the interface between the components. This explains in part why a well-fitted endoprosthesis is vital to the device's longevity and suggests why osteoarthritis induces arthropathy. In osteoarthritis, tissue fibrillation and the crevassed surface of the cartilage layer destroy the effectiveness of the seal, leading to the expulsion of fluid normally internal to the matrix. Thus, the matrix is "squashed," innervated bone is abraded, and the patient reports pain--the clinical symptom of osteoarthritis.

We are therefore not arguing that therapists should strive to eliminate acetabular pressure. Rather we report pressures from the highest-reading transducer to suggest that the order and the severity of certain rehabilitation protocols should be considered in designing rehabilitation programs. (9) As stated previously, we do not believe that acetabular contact pressures should be the sole consideration in designing rehabilitation programs. Understanding hip stresses, however, should allow therapists to balance the presumed benefits of, for example, strengthening exercises against their physiological costs. Using a standard activity such as hip stresses during normal gait puts other activities into context.

4. The contribution of muscle force to isometric versus extension exercises. The results of our two-part study suggest that voluntary, though not necessarily conscious or perceived, co-contraction is the major contributor to the high pressures we observe. Previously reported electromyographic data from the same subject of our two-part study support this assertion (see Fig. 2 in the article by Krebs et al (9)).

In summary, we agree with many of Dr Oatis's comments and that more subjects with instrumented femoral endoprostheses need to be studied before such data can be generalized to all patients with hip implants. A second pressure-instrumented endoprosthesis has already been implanted in an 82-year-old male subject, and data collected will be compared with data reported in the preceding papers. Preliminary results (up to 9 months postimplantation) from this second implant are quite consistent with the conclusions presented in the present reports. Until more data are available, clinicians need to make treatment choices based on the best available evidence and other factors. In short, if the practical question is, "Are these data infallible?" then the answer is, "We think not." But to the question, "Are the data meaningful?" we answer, "We think so."

The philosophical question raised by Dr Oatis's commentary is whether to continue physical therapy practice based on current customs only loosely related to objective data or to change our practice as new information comes to light. We believe that to "ponder and discuss" such information is necessary but probably not sufficient.

David E Krebs, PhD, PT
Robert W Mann, ScD
Deborah L Givens-Heiss, PT, OCS
Ellen M Strickland, PT

References

(1) Rydell NW. Forces acting on the femoral head prosthesis: a study on strain gauge supplied prostheses in living persons. Acta Orthop Scand 1966;37(suppl 88):1-132.

(2) Rushfeldt PD, Mann RW, Harris WH. Influence of cartilage geometry and the pressure distribution in the human hip joint. Science. 1979;204:413-415.

(3) Rushfeldt PD, Mann RW. Improved techniques for measuring in vitro the geometry and pressure distribution in the human acetabulum, II: instrumented endoprosthesis measurement of articular surface pressure distribution. J Biomech. 1981;14:315-323.

(4) Mann RW, Rushfeldt PD, Palmer DW, et al. Experimental facility for synovial joint research in situ. In: Advances in Bioengineering. New York, NY: American Society of Mechanical Engineers; 1983:108-109.

(5) Harris WH, Rushfeldt PD, Carlson CE. Pressure distribution in the hip and selection of hemiarthroplasty. In: The Hip: Proceedings of the 3rd Annual Open Scientific Meeting of the Hip Society. Philadelphia, Pa: CV Mosby Co; 1975:93-98.

(6) Carvajal T. In Vivo Pressure Distribution Data from the Human Hip Joint. Cambridge, Mass: Massachusetts Institute of Technology; 1985. Thesis.

(7) Ito K. In Vivo Measurement of the Joint Reactive Force in the Human Hip Joint. Cambridge, Mass: Massachusetts Institute of Technology; 1989. Thesis.

(8) Macirowski T, Tepic S, Mann RW. Cartilage stresses in the human hip joint. Journal of Mechanical Engineering. In press.

(9) Krebs DE, Elbaum L, Riley PO, et al. Exercise and gait effects on in vivo hip contact pressures. Phys Ther 1991;71:301-309.

(10) Hodge WA, Carlson KL, Fijan RS, et al. Contact pressures from an instrumented hip endoprosdiesis. J Bone Joint Surg [Am]. 1989;71: 1378-1386.

DL Givens-Heiss, PT, OCS, is Senior Clinician, Harmarville Outpatient Rehabilitation Center, 351 W Beau St, Washington, PA 15301-4613, and Instructor, Department of Physical Therapy, Wheeling Jesuit College, Wheeling, WV 26003. She was a master's degree candidate at the time of this study at the MGH Institute of Health Professions, Boston, MA.

DE Krebs, PhD, PT, is Associate Professor, Program in Physical Therapy, MGH Institute of Health Professions, 15 River St, Boston, MA 02108-3402 (USA), and Lecturer, Massachusetts Institute of Technology, Cambridge, MA 02139. Address all correspondence to Dr Krebs at the first address.

PO Riley, PhD, is Technical Director, MGH Biomotion Laboratory, Massachusetts General Hospital, Fruit St, Boston, MA 02114.

EM Strickland, PT, is Staff Coordinator, Orthopedics, Department of Physical Therapy, Department of Veterans Affairs Medical Center, Boston, MA 02130. She was a master's degree candidate at the time of this study at the MGH Institute of Health Professions.

M Fares, PT, is a doctoral candidate at Boston University. She was a master's degree candidate at the time of this study at the MGH Institute of Health Professions.

WA Hodge, MD, is Assistant in Orthopaedics, Massachusetts General Hospital, 5 Longfellow Pl, Boston, MA 02114.

RW Mann, ScD, is Whitaker Professor of Biomedical Engineering, Department of Mechanical Engineering, Massachusetts Institute of Technology; Director, MIT Newman Laboratory for Biomechanics and Human Rehabilitation; and Director, Harvard-MIT Rehabilitation Engineering Center, Cambridge, MA 02139.

This study and the surgical procedure were both approved by the human studies committees of Massachusetts General Hospital and the Massachusetts Institute of Technology.

This research was supported in part by US Department of Education Grant #1 H133P90005, Veterans Administration Medical Center Grant #REUVDHT9898, and the Arthritis Foundation.

This article was submitted September 27, 1991, and was accepted May 18, 1992.