Between January 1980 and December 1993, 1846 primary total knee arthroplasties were performed at our institution. Forty-three (2 per cent) of these procedures were in patients who had Workers' Compensation claims. In order to be included in the index group, a patient had to have been followed for a minimum of four years. One patient died from causes unrelated to the operation at seven months postoperatively. Although the knee had been functioning well clinically at the time of the six-month follow-up examination, this patient was excluded from the study, leaving forty-two patients. All patients were examined annually both clinically and radiographically, and the hospital records, including preoperative studies, operative reports, and clinical notes, were retrieved.
There were thirty-two men (76 per cent) and ten women (24 per cent). The mean age of the patients at the time of the total knee arthroplasty was forty-eight years (range, twenty-nine to sixty-eight years). The implant was inserted with cement in eleven patients and without cement in thirty-one. All patients received a porous-coated anatomic knee prosthesis (Howmedica, Rutherford, New Jersey)9. A tricompartmental arthroplasty was performed in all except eight patients, who had had a previous patellectomy.
Matched Group
The forty-two patients who were receiving Workers' Compensation were directly matched, with use of a computer-generated random-selection process, with a group of forty-two patients who were not receiving such compensation. The control group was drawn from the total group of 1846 total knee arthroplasties that were performed during the years of this study. The two groups were matched with regard to nine parameters: age at the time of the operation (within five years), gender, duration of follow-up (within two years), preoperative deformity in the coronal plane (within 5 degrees), obesity index (within 20 points and based on gender, height, and weight) as determined with use of the equation (weight in kilograms x 100)/(52.4 + [(height in centimeters - 150) x 0.67]) for women and with use of the equation (weight in kilograms x 100)/(59.7 + [(height in centimeters - 160) x 0.71]) for men and an index of greater than 120 categorized as obesity11, number of previous procedures before the index arthroplasty, method of fixation (with or without cement), preoperative Knee Society score10, and preoperative radiographic severity according to the criteria of Ahlbäck (Table I).
Diagnosis and Previous Procedures
All patients in both groups had a diagnosis of primary or post-traumatic osteoarthrosis.
Thirty-six (86 per cent) of the forty-two patients who were receiving Workers' Compensation had had a total of eighty-six procedures (mean, 2.4 procedures; range, one to seven procedures per patient) before the total knee arthroplasty. Forty-five of these procedures were arthroscopies that were performed in conjunction with a partial medial meniscectomy or a lateral meniscectomy, or both. Of the remaining procedures, eleven were proximal tibial osteotomies; ten, open arthrotomies; eight, patellectomies; four, Pridie procedures (drilling of the patellar cartilage); three, proximal soft-tissue patellofemoral realignments; two, Maquet osteotomies; two, reconstructions of the anterior cruciate ligament; and one was an open reduction with internal fixation of the patella. Twenty-three patients had not been managed with a previous patellectomy or proximal tibial osteotomy. Only six (17 per cent) of the thirty-six patients who had had a previous procedure had ever had sustained relief of pain—that is, no or only slight pain for at least a year afterward.
Thirty-six (86 per cent) of the patients who were not receiving compensation had had a total of eighty-two procedures (mean, 2.3 procedures; range, one to six procedures per patient) before the total knee arthroplasty. Forty-four of these procedures were arthroscopies. Of the remaining procedures, eleven were proximal tibial osteotomies; nine, open arthrotomies; eight, patellectomies; four, proximal soft-tissue patellofemoral realignments; two, reconstructions of the anterior cruciate ligament; two, open reductions with internal fixation of the patella and the tibial plateau; and two, Maquet osteotomies. As in the Workers' Compensation group, twenty-three patients had not had a previous proximal tibial osteotomy or patellectomy. Twenty-four (67 per cent) of these thirty-six patients had had sustained relief of pain for at least a year after a previous procedure.
Clinical Assessment
The most recent follow-up evaluation was conducted either at our clinic (forty patients in each group) or by telephone contact with both the patient and the patient's current orthopaedic surgeon (two patients in each group). All patients were assessed clinically with use of the Knee Society rating scale10. A score of 90 points or more was considered excellent; 80 to 89 points, good; 70 to 79 points, fair; and less than 70 points or a revision, poor. Patients who had an excellent or good result were considered to have a successful outcome, while those who had a fair or poor result were considered to have an unsuccessful outcome. All patients in the Workers' Compensation group had had a score of less than 70 points preoperatively (mean, 42 points; range, 17 to 69 points). In the group that was not receiving compensation, the mean preoperative Knee Society score had been 41 points (range, 14 to 68 points).
The patient's occupations were categorized, according to the method of Misamore et al., as involving strenuous, moderate, or non-manual labor. Eight patients in the Workers' Compensation group had had a preoperative occupation that involved strenuous labor; eighteen, moderately strenuous labor; and two, non-manual labor. The remaining fourteen patients had been unemployed. Twenty patients who were not receiving compensation had had a preoperative occupation that had been categorized as strenuous; eight, as moderately strenuous; and twelve, as involving non-manual labor. The remaining two patients were unemployed.
The level of activity was evaluated by grading distance and impact separately19. For distance, 1 point was assigned if the patient was housebound; 2 points, if the patient was able to engage in so-called normal activity, such as shopping or walking for approximately three to five miles (4.8 to 8.0 kilometers) per week; and 3 points, if the patient was able to engage in high-level activity, such as walking for at least one mile (1.6 kilometers) per day. For impact, 1 point was assigned if the patient was housebound or was able to take short walks only; 2 points, if he or she was able to perform low-level activity, such as playing golf, talking long walks, and occasional bicycling; and 3 points, if he or she could perform high-level activity, such as running, playing tennis, and aerobics.
Radiographic Evaluation
The radiographic evaluation included anteroposterior and lateral radiographs of the knee, made with the patient standing. The preoperative degree of osteoarthrosis was assessed with use of a modification of the grading scale of Ahlbäck1,17. In each group of patients, seven knees had mild osteoarthrosis, twenty-four had moderate osteoarthrosis, and eleven had severe osteoarthrosis.
Axial alignment was measured with use of anteroposterior radiographs, made with the patient standing and with use of a long cassette, as described previously4,19,23. Preoperative alignment was measured on the most recent radiograph that had been made before the operation, and postoperative alignment usually was measured on the radiograph that had been made at six weeks. A zonal analysis of all components was performed to assess the presence of progressive radiolucent lines and (in the case of prostheses that had been inserted without cement) shedding of the beads4. Fluoroscopic spot radiographs, made with the x-ray beam parallel to the tibial base-plate, were analyzed for all patients at each follow-up visit to obtain additional information concerning radiolucent lines.
Statistical Evaluation
The two matched groups were compared with regard to the use of narcotics for pain relief and the Knee Society scores as well as the specific scores for pain, stability, and range of motion. A 95 per cent confidence interval was calculated for absolute differences between the two groups.
After a mean duration of follow-up of eighty months (range, forty-eight to 178 months), the patients who were receiving compensation had a mean Knee Society score of 64 points (range, 25 to 100 points). The clinical result was excellent for ten patients (24 per cent); good for two (5 per cent); and fair, poor, or a revision for thirty (71 per cent). After a mean duration of follow-up of eighty-two months (range, forty-eight to 178 months), the patients who were not receiving compensation had a mean Knee Society score of 93 points (range, 57 to 100 points). The clinical result was excellent for thirty-two patients (76 per cent); good for five (12 per cent); and fair, poor, or a revision for five (12 per cent). The difference between the groups with regard to fair or poor results and revisions was significant (p < 0.01).
Four patients who were receiving Workers' Compensation had an early complication of the operation. Three of these patients, who had a minimum range of motion within twelve weeks after the arthroplasty, needed an open lysis of adhesions, and the fourth patient needed irrigation and débridement because of a superficial wound infection that did not extend into the knee joint. There were no postoperative complications in the patients who were not receiving compensation.
At the time of the latest follow-up visit, the Workers' Compensation group had a mean range of motion of 101 degrees (range, 65 to 125 degrees) and the comparison group had similar values (mean, 106 degrees; range, 80 to 125 degrees) (p = 0.9). Stability on anterior-posterior and medial-lateral clinical testing also was similar between the groups (p = 0.8). The differences in the overall Knee Society objective scores thus were based on differences in the pain scores.
Before the index operation, thirty-eight patients (90 per cent) in the Workers' Compensation group had used narcotic analgesics and four (10 per cent) had used anti-inflammatory or other non-narcotic medications. At the time of the latest follow-up visit, thirty-two patients (76 per cent) still needed narcotic analgesics, four (10 per cent) used non-narcotic medications, and six (14 per cent) did not use pain medication. In the group that was not receiving compensation, thirty-six patients (86 per cent) had used narcotic analgesics and six (14 per cent) had used non-narcotic analgesics before the procedure. At the time of the most recent follow-up, only four patients (10 per cent) used narcotic medications, four (10 per cent) used non-narcotic analgesics, and thirty-four (81 per cent) did not use pain medication.
Only two patients in the Workers' Compensation group had an occupation that was categorized as involving strenuous labor at the time of the latest follow-up visit (compared with eight patients preoperatively), six had an occupation that involved moderately strenuous labor (compared with eighteen patients preoperatively), eighteen had an occupation that did not involve manual labor (compared with two patients preoperatively), and sixteen were unemployed (compared with fourteen patients preoperatively). In the group that was not receiving compensation, twenty patients had an occupation involving strenuous labor (the same number of patients as preoperatively), ten had an occupation involving moderately strenuous labor (compared with eight patients preoperatively), nine had an occupation that did not involve manual labor (compared with twelve patients preoperatively), and two were unemployed (the same number as preoperatively).
By the time of the latest follow-up, the mean scores for distance and impact in the group that was receiving compensation had not improved compared with the preoperative scores (mean, 1.84 and 1.71 points postoperatively compared with 1.76 and 1.67 points preoperatively). In the group that was not receiving compensation, the mean score for distance improved from 1.81 points preoperatively to 2.3 points postoperatively and that for impact improved from 1.69 points to 2.42 points.
Various parameters were analyzed in terms of their possible influence on the clinical outcome. Age did not appear to influence the results in the group that was receiving compensation. Six (27 per cent) of the twenty-two patients who were less than fifty years old had a good clinical outcome compared with six (30 per cent) of the twenty who were more than fifty years old (p = 0.9). The preoperative degree of osteoarthrosis also did not appear to influence the outcome in either group. In the Workers' Compensation group, two of the seven patients who had mild osteoarthrosis, seven of the twenty-four who had moderate osteoarthrosis, and three of the eleven who had severe osteoarthrosis had a successful result. In the group that was not receiving compensation, six of the seven patients who had mild osteoarthrosis, twenty-two of the twenty-four who had moderate osteoarthrosis, and nine of the eleven who had severe osteoarthrosis had a successful result.
In the Workers' Compensation group, of the six patients who had not had a previous operation, one had a good clinical outcome; two, a fair outcome; and three, a poor outcome. Of the twenty patients who had had one or two previous procedures, seven (35 per cent) had a good outcome. Of the sixteen patients who had had more than two previous operations, four had a good outcome. Of the eleven patients who had had a previous proximal tibial osteotomy, one had a good result. Of the eight patients who had had a previous patellectomy, two had a good result. When the nineteen patients who had had either a proximal tibial osteotomy or a patellectomy are excluded, nine (39 per cent) of twenty-three patients had a good or excellent result. Of the six patients who had had some relief of pain after a previous operation, one had a good result, one had a fair result, and four had a poor result. Of the nineteen patients who were not receiving compensation and had had a previous patellectomy or proximal tibial osteotomy, sixteen had a good or excellent clinical outcome.
Of the thirty patients in the Workers' Compensation group who had a fair or poor result, twenty had intractable pain; thirteen of these patients had a revision. Five tibial components were revised: one, because of loosening, and four, because of intractable pain. Six polyethylene tibial inserts were revised because of a subjective sensation of instability. In addition, a patellar component was revised because it had subluxated and an exploration was performed because of scarring of the extensor mechanism. After a mean duration of four years (range, two to eight years), only the patient who had had the revision because of loosening of the tibial component had a good clinical outcome (a Knee Society score of 86 points).
Three of the eleven patients in the Workers' Compensation group who had had fixation with cement and nine of the thirty-one who had had fixation without cement had a successful clinical outcome. In the group that was not receiving compensation, nine of the eleven patients who had had fixation with cement and twenty-eight of the thirty-one who had had fixation without cement had a successful clinical outcome. Thus, the rates associated with the two types of fixation were similar in each of the groups.
A separate analysis was performed for the twelve patients who had had an excellent or good clinical outcome in the Workers' Compensation group. On the basis of the numbers available, the only trend toward a difference between these twelve patients and the remaining thirty was in terms of the preoperative Knee Society score (mean, 55 points [range, 29 to 69 points] compared with 38 points [range, 17 to 64 points]; p = 0.07). This trend was not found in the group that was not receiving compensation. In that group, ten of the twelve patients who had had a preoperative Knee Society score of less than 38 points (mean, 30 points; range, 14 to 37 points) had a successful clinical outcome compared with twenty-seven of the thirty who had had a score of at least 38 points (mean, 45 points; range, 38 to 68 points); these two rates were similar.
Radiographic Results
One patient in the Workers' Compensation group was noted to have progressive radiolucent lines around the tibial component at the seven-year follow-up visit. This finding was indicative of loosening, and the knee was subsequently revised. No patient who had a good or excellent clinical outcome had impending radiographic failure. There were no differences between the two groups with regard to the shedding of beads or the presence of progressive radiolucent lines. Two patients in each group had progressive separation of beads from the patellar component; all four were asymptomatic. No patient in either group had progressive shedding of beads from, or progressive radiolucent lines around, the tibial or femoral component. Postoperatively, the mean femoral anteroposterior alignment was 98.8 degrees in each group (goal, 99 degrees; range, 95 to 102 degrees); only one knee in each group deviated more than 3 degrees from this goal. The mean anteroposterior tibial angle was 87 degrees in each group (goal, 87 degrees); no knee in either group deviated more than 3 degrees from this goal.
Although poor results after operations on the back have been well documented in patients who were receiving compensation, the results after operations on the extremities have rarely been studied. Some authors have reported better outcomes after treatment of injuries of the extremities compared with those after treatment of injuries of the axial skeleton in patients receiving compensation. Beals and Hickman found that forty-five (64 per cent) of seventy patients who had had an injury of an extremity returned to work compared with only forty-two (38 per cent) of 110 patients who had had a low-back injury. The workers with a back injury had psychopathological problems that were markedly greater in magnitude and complexity than those of the workers with an injury of an extremity.
Misamore et al. recently reported a good or excellent clinical result for thirteen (54 per cent) of twenty-four shoulders in patients who were receiving Workers' Compensation and for seventy-six (92 per cent) of eighty-three shoulders in patients who were not receiving compensation. The shoulder scores in the former group were significantly inferior to those in the latter (p < 0.0004). Despite the difference in outcome, the objective findings, which included the active range of motion, were similar between the two groups, although the patients who were receiving compensation had significantly poorer subjective results (p < 0.002 for pain and p < 0.001 for function). The findings of Misamore et al. are similar to those of the present study, in which the two groups of patients had no detectable differences in terms of certain objective findings, such as stability of the knee, range of motion, and alignment. However, we noted a significant difference (p < 0.01) in the subjective findings (pain scores) between the two groups. In addition, no differences in radiographic findings (progressive radiolucent lines or shedding of beads) were detected. A discrepancy between subjective and objective findings in patients receiving compensation has been documented in other studies6,8,21.
The poorer results in the patients who were receiving compensation in the present study might partially be explained by the types of previous operations that these patients had had. A number of studies have shown poor results after total knee arthroplasty in patients who have had a previous patellectomy13,14,20. In a recent study, Martin et al. reported a satisfactory result after primary total knee arthroplasty in only thirteen (59 per cent) of twenty-two patients who had had a previous patellectomy. The results were better when a posterior stabilized prosthesis was used14. In the present study, only two of eight patients who had had a previous patellectomy and were receiving compensation had a satisfactory result. It should be noted that a non-constrained prosthesis was used in the present study; perhaps the use of a more constrained prosthesis would have improved the results. However, of the eight patients who had had a previous patellectomy and were not receiving compensation, seven had a satisfactory outcome, indicating that the effect of this factor was less important than the factor of compensation. In addition, the results of total knee arthroplasty after a previous proximal tibial osteotomy may not be optimum, as was shown in a study in which twenty-four (33 per cent) of seventy-three knees had a poor result after a mean duration of follow-up of seven years (range, two to eleven years)16. All ten patients who were receiving compensation in that study had a poor result. This finding is similar to that of the present study, in which only one of eleven patients who had had a previous proximal tibial osteotomy and were receiving Workers' Compensation had a satisfactory outcome. In the group that was not receiving compensation, nine of eleven patients who had had a previous proximal tibial osteotomy had a good or excellent clinical outcome. Even when the nineteen patients in the Workers' Compensation group who had had a previous proximal tibial osteotomy or a patellectomy are excluded, only nine (39 per cent) of twenty-three patients had a successful outcome compared with twenty-one (91 per cent) of twenty-three in the group that was not receiving compensation.
We attempted to define subsets of patients within the Workers' Compensation group who did poorly or well; however, no risk factors were found to be associated with a successful outcome, with the numbers available. The only marginally significant factor that could be related to success in this group was a higher preoperative Knee Society score (p = 0.07). The patients who had a successful outcome had had a mean preoperative score of 55 points, whereas those who had an unsuccessful outcome had had a mean score of 38 points. This might reflect a difference in the pain threshold or in other psychological factors. Similarly, the patients who were receiving compensation and had had at least one procedure before the total knee arthroplasty had poorer results with regard to relief of pain compared with the patients who were not receiving compensation and had had at least one previous procedure (six [17 per cent] of thirty-six compared with twenty-four [67 per cent] of thirty-six; p < 0.05). This further substantiates the hypothesis that patients who are receiving compensation may have a different pain threshold, secondary gain, or other psychological factors that play a role in determining the subjective outcome after an operative procedure involving the knee.
The futility of performing a revision operation when there is no identifiable mechanical cause of symptoms must be emphasized. Only one of the eleven patients in the Workers' Compensation group who had a revision but no identifiable mechanical problem had a good clinical outcome at the time of the latest follow-up. This finding is similar to that of a study in which only eleven (41 per cent) of twenty-seven good clinical outcomes were in patients who had had exploration of a radiographically normal total knee prosthesis because of unexplained pain18.
In summary, the surgeon should be aware that Workers' Compensation is one of several variables that may have an untoward influence on the outcome of a knee replacement operation.