Updated: Dec 4, 2021
BJBJS - The Journal of Bone and Joint Surgery VOL 81-A, NO. 7. JULY 1999 begins on page 926
ANALYSIS OF ROUTINE HISTOLOGICAL EVALUATION OF TISSUES REMOVED DURING PRIMARY HIP AND KNEE ARTHROPLASTY
Investigation performed at the Department of Orthopaedic Surgery.
University of Virginia Health Sciences Center. Charlottesville, and Roanoke Memorial Hospital, Roanoke
Background: It has often hern hospital policy to send all resected specimens obtained during a total hip or knee arthroplasty for histological evaluation. This practice is expensive and may be unnecessary. We sought to determine the ability of surgeons to diagnose primary joint conditions correctly, and we attempted to identify any possible risks to the patient resulting from the omission of routine histological evaluation of specimens at the surgeon's discretion. Our objective was to ascertain whether routine histological evaluation could be safely omitted from the protocol for primary hip and knee arthroplasty without compromising the care of the patient. Methods: A total of 1388 consecutive arthroplasties in 1136 patients were identified from a database of primary total hip and knee arthroplasties that were prospectively maintained by the senior one of us. Followup data obtained at a mean of 53 years (range, two to ten years) were available after 92% (1273) of the 1388 arthroplasties. The preoperative diagnosis was determined from the history, findings on clinical examination. and radiographs. The intraoperative diagnosis was determined by gross inspection of joint fluid, articular cartilage, synovial tissue, and the cut surfaces of resected specimens. The combination of the preoperative and intraoperative diagnoses was considered to be the surgeon's clinical diagnosis. All resected specimens were sent for routine histological evaluation, and a pathological diagnosis was made. Attention was given to whether a discrepancy between the surgeon's clinical diagnosis and the pathological diagnosis altered the management of the patient. The original diagnoses were updated with the use of annual radiographs and clinical assessments. The cost of histological examination of specimens obtained at arthroplasty was determined by consultation with hospital administration, accounting. and pathology department personnel.
No benefits in any form have been received or will be received from a commercial party related directly or indirectly to the subject of this article. No funds were received in support of this study. Department of Orthopaedic Surgery. University of Virginia Health Sciences Center, Box 159. Charlottesville, Virginia 22908. Roanoke Orthopaedic Center. Box 21369, Roanoke. Virginia 24018-0546.
Results: A pathological fracture or an impending fracture was diagnosed preoperatively and confirmed intraoperatively during twelve of the 1388 arthroplasties. Histological analysis demonstrated malignancy in specimens obtained during eleven of these arthroplasties and evidence of a benign rheumatoid geode in the specimen obtained during the twelfth arthroplasty. The preoperative and intraoperative diagnoses made before and during the remaining 1376 arthroplasties were benign conditions, which were confirmed histologically in all patients. No diagnosis changed during the follow-up period. As demonstrated by a comparison with the histological diagnosis, the surgeon's clinical diagnosis of malignancy had a sensitivity of 100percent (95 % confidence interval. 74.0 to 100percent), a specificity of 99.9 % (95 % confidence interval. 99.6 to 100 %), a positive predictive value of 91.7 % (95 % confidence interval. 64.6 to 983 %), and a negative predictive value of 100 % (95 % confidence interval.99.7 to 100 %). There was a discrepancy between the preoperative and intraoperative diagnoses associated with eleven ararthroplasties. All eleven intraoperative diagnoses were correct, as confirmed histologically. Excluding the patients who had a pathological or impending fracture, the accuracy of the surgeon's preoperative diagnosis was 992 % (95 % confidence interval. 98.6to 993 %). When the intraoperative and preoperative diagnoses were combined, the accuracy was 100percent (95 % confidence interval. 99.7 to 100percent). Histological evaluation at our hospital resulted in total charges, including hospital costs and professional fees, of $19627 and a mean total reimbursement of $10239 per evaluation. In our series of 1136 patients with 1388 arthroplasties, these costs could have been eliminated for all but the twelve patients who had a suspected malignant lesion and the one patient in whom pigmented villonodular synovitis was found.
Conclusions: Benign conditions can be diagnosed accurately by an experienced surgeon. The preoperative diagnosis should be determined on the basis of a carefully obtained history; a detailed physical examination. and a thorough evaluation of radiographs. To complete the clinical diagnosis, the preoperative in formation should he combined with findings obtained through careful intraoperative observation of resected specimens and. in the case of hip arthroplasty, an inspection of the divided femoral head. In so doing, the surgeon can usually exclude clinically the possibility of malignancy or another condition that may alter the management of the patient and thus can reserve histological analysis for situations where the diagnosis is suspect or unexpected findings are noted intraoperatively. Illis selective, surgeon-directed use of histological analysis can result in notable financial savings.
Escalating charges and declining reimbursements for total knee and hip arthroplasties have led to increasing demands for cost containment. Suggested measures for reducing costs have included decreasing the duration of the hospital slay*2’, standardization of prostheses and suppliers***, increasing the volume of procedures and establishing critical pathways*. However, these practices have been implemented in many hospitals and a plateau for savings may have been reached4. Charges for hospital services such as radiography and laboratory investigations account for about 4% of total costs1'. Many believe that the use of these services at the discretion of the clinician can result in notable savings without compromising patient outcomes (25,22,23,26). During the course of primary hip and knee arthroplasties, the femoral head or resected specimens from the knee are often sent routinely for histological evaluation. This practice may be in concordance with hospital bylaws or a result of the surgeon’s preference. DiCarlo el al. reported that, in a series of 1794 total hip arthroplasties, the rate of discrepancy between the clinical diagnosis and the histological diagnosis based on analysis of the resected femoral head was 5.4% (ninety-seven arthroplasties)’. Because morbid conditions were identified in seven patients, mandatory pathological evaluation of all resected specimens was advocated. However, recent evidence suggests that the surgeon can use his or her discretion in determining which specimens require a histological evaluation and that observed discrepancies between the preoperative and histological diagnoses have little bearing on patient outcome7**. It appears that selective use of histological examination of resected specimens could result in substantial savings. However, to our knowledge, guidelines for adequate clinical evaluation and the possible long-term risks to patients associated with the omission of routine histological evaluation remain to be established. The objective of the present study was to determine whether routine histological evaluation could be safely omitted from the protocol for primary hip or knee arthroplasty without compromising the care of the patient. The surgeon's preoperative and intraoperative diagnoses were compared with the histological diagnosis, and statistical relationships were established. Any change in diagnosis during the follow-up period was noted. Potential economic savings resulting from discretionary use of histological evaluation were also calculated.
Materials and Method
Between December 1987 and December 1995, the senior one of us (J. T. M.) performed 1388 consecutive primary total hip and knee arthroplasties in 1136 patients, and this group constituted the study population. All demographic and diagnostic data were recorded prospectively in a database. The group included 562 hips and 826 knees Three hundred and eighty-five of the patients were men. and 751 were women. The mean age at the time of the operation was sixty-nine years (range, twenty-two to 100 years). Preoperatively, a complete history was recorded for all patients and all patients were examined clinically and radiographically by the senior one of us who looked for evidence of trauma, tumour, infection, osteoarthritis (primary or secondary), inflammatory arthropathy (rheumatoid arthritis juvenile rheumatoid arthritis systemic lupus erythematosus spondyloarthropathies or crystalline arthropathies), or hemorrhagic arthropathy. Before all of the hip replacements, anteroposterior radiographs of the pelvis and the proximal part of the femoral shaft, as well as lateral radiographs of the involved hip joint, were made with the patient supine. Before the knee replacements, anteroposterior radiographs were made with the patient standing and lateral and Merchant patellar radiographs were made with the patient supine. Some radiographic features were characteristic of a particular diagnosis. For example, meniscal calcifications were seen in association with chondrocalcinosis; increased bone density and collapse, with avascular necrosis; sacroiliac joint fusions with ankylosing spondylitis; and erosive changes with inflammatory arthropathies12*. These findings were combined with information from other treating physicians to establish a preoperative diagnosis for the underlying joint abnormality, and this diagnosis was prospectively recorded in the database. Al the lime of the operation, the senior one of us inspected the joint fluid, the articular cartilage, and the synovial tissue for gross changes characteristic of the preoperative diagnosis In the total hip arthroplasties each femoral head was divided in half and viewed Intraoperatively for areas of collapse, cystic changes or abnormal architecture. Combining these observations al operation with the preoperative diagnosis allowed the determination of an intraoperative diagnosis which also was recorded prospectively. All specimens were then sent for histological evaluation, as required by the hospital bylaws and a pathological diagnosis was made. Preoperative, intraoperative, and histological diagnoses were made for all 1388 hips and knees treated with arthroplasty. The diagnoses were then reappraised al yearly follow-up intervals on the basis of an updated medical history, physical examination, and radiographs. Clinical evaluation included an assessment of joint pain and function with the use of the Knee Society1* or 11 arris14 hip-scoring system. Any change in medical history was noted. Anteroposterior radiographs of the pelvis and the proximal part of the femur and a true lateral radiograph of the hip were made with the patient supine al the follow-up evaluations after the hip replacements. Anteroposterior radiographs were made with the patient standing and lateral and Merchant patellar radiographs were made with the patient supine al the evaluations after the knee replacements. The prosthetic components were evaluated according to previously described radiographic crileriaMJ04UW\ Zonal analysis was used to determine the location, size, and any progression of radiolucent lines and osteolysis. Any changes in diagnosis during the follow-up period were recorded in the database.
The surgeon’s diagnosis of underlying malignancy based on preoperative and intraoperative findings was compared with the pathologist's histological diagnosis for every specimen. The results were expressed in terms of sensitivity (the number of malignant lesions that were correctly diagnosed as malignant by the surgeon divided by the total number of malignant lesions identified histologically), specificity (the number of benign conditions that were correctly diagnosed as benign by the surgeon divided by the total number of benign conditions identified histologically), the positive predictive value (the number of malignant lesions that were correctly diagnosed as malignant by the surgeon divided by the total number of malignant lesions that were diagnosed by the surgeon), and the negative predictive value (the number of benign conditions that were correctly diagnosed as benign by the surgeon divided by the total number of benign conditions that were diagnosed by the surgeon). The accuracy of the surgeon's preoperative diagnoses of benign conditions was also calculated in relation to the histological diagnoses (the number of correct diagnoses that were made by the surgeon divided by the total number of the same diagnoses that were made on the basis of the histological evaluation). The overall accuracy of the surgeon’s diagnoses based on both preoperative and intraoperative findings was also calculated. 95 % confidence intervals were determined for all binomial proportions with the use of Wilson’s method.
Three patients died perioperatively; one each died from cardiac failure (no autopsy), pulmonary embolism (confirmed at autopsy), and myocardial infarction (confirmed at autopsy). An additional sixty-four patients (seventy-five arthroplasties) died during the follow-up period from causes unrelated to the index operation. Twenty-nine patients (thirty-seven arthroplasties) were lost to follow-up, although none were lost fewer than six months postoperatively. Thus, data obtained al a minimum of two years (range, two to ten years; mean.5.5 years) were available for 1273 (92 %) of the 1388 arthroplasties. The pathology department provided details of professional charges and reimbursements, received according to various insurance options, that were incurred for the gross and histological evaluation of the specimens. Hospital costs associated with histological analysis were calculated after consultation with hospital administration and accounting personnel. The hospital costs included those for slides, reagents and stain preparation, decalcification solutions, equipment (including depreciation), labour, and indirect expenses.
The surgeon’s clinical diagnosis (based on the preoperative history, clinical examination, and radiographic findings combined with the intraoperative observations) was most commonly primary osteoarthritis (949 arthroplasties), avascular necrosis (112). rheumatoid arthritis (ninety-two), traumatic arthritis (eighty-two), or acute fracture (eighty-three). The other diagnoses included congenital dislocation of the hip (sixteen arthroplasties), pathological fracture or impending fracture (twelve), ankylosing spondylitis (nine), slipped capital epiphysis (nine), Legg-Calvé-Perthes disease (six), gout (five), chondrocalcinosis (three), Pagel disease (three), a reversal of an arthrodesis (two), juvenile rheumatoid arthritis (two), pigmented villonodular synovitis (one), Gaucher disease (one), and ochronosis (one). A pathological fracture or an impending fracture secondary to an underlying malignant lesion was diagnosed before twelve (1 %) of the 1388 arthroplasties. Intraoperative findings confirmed the preoperative diagnosis but did not yield additional information regarding the pathological condition. The histological diagnoses associated with the twelve pathological or impending fractures included metastatic carcinoma from the breast (four), lung (two), prostate (one), and kidney (one) as well as multiple myeloma (three). One patient in whom a malignant lesion had been suspected on the basis of the preoperative and intraoperative findings had a benign rheumatoid geode. The histological reports were essential to establish the underlying diagnosis in these twelve patients. No occult malignancy was detectedtopologically in the remaining 1376 specimens. The ability of the surgeon to identify malignancy correctly was determined by comparing the preoperative and intraoperative clinical diagnoses of malignancy with the histological diagnoses. This analysis yielded a sensitivity of 100 % (eleven of eleven), with a 95 % confidence interval of 74.0 to 100 %; specificity of 99.9 % (1376 of 1377), with a 95 percent confidence interval of 99.6 to 100 %; a positive predictive value of 91.7 % (eleven of twelve), with a 95 % confidence interval of 64.6 to 98.5 %; and a negative predictive value of 100 % (1376 of 1376), with a 95 % confidence interval of 99.7 to 100%. Since no additional periprosthetic tumours developed during the study period, these values remained unchanged al the latest follow-up evaluation (al a mean of 55 years (range, two to ten years)).
ROUTINE HISTOLOGICAL EVALUATION OE TISSUES REMOVED DURING HIP AND KNEE ARTHROPLASTY
Charges and Actual Reimbursement Per Payor Class, for Histological
Evaluation of Specimens Resected During Total Hip or Knee Arthroplasty Roanoke Memorial Hospital During 1998
The preoperative diagnosis was a benign joint condition before 1376 of the arthroplasties. Eleven preoperative diagnoses, all of the primary osteoarthritis, differed from the intraoperative diagnoses. The revised intraoperative diagnosis was avascular necrosis (grade 3 or 4, according to the system of Ficat el al. (12)) during nine arthroplasties. This avascular necrosis, which was identified after the femoral head was divided, had not been seen on the preoperative radiographs because of the advanced degenerative changes. Ochronosis was diagnosed during one arthroplasty, on the basis of black staining of the synovial tissue, and pigmented villonodular synovitis was diagnosed during the eleventh arthroplasty, on the basis of the characteristic brown pigmentation in hyperproliferative synovial tissue. All of these revised intraoperative diagnoses were the same as the subsequent histological diagnoses. The accuracy of the surgeon’s preoperative diagnosis compared with the histological diagnosis was 99.2 % (1365 of 1376), with a 95 % confidence interval of 98.6 to 99.5 %. The accuracy of the surgeon's combined preoperative and intraoperative diagnoses compared with the histological diagnosis was 100 % (1376 of 1376), with a 95 % confidence interval of 99.7 to 100 %. No diagnoses were revised during the study period (mean duration, 55 years); therefore, the rate of accuracy remained unchanged after more than two years of follow-up.
During the follow-up period, an infection developed in eleven patients (eleven arthroplasties), resulting in a two-stage revision in ten of them and a resection arthroplasty in one. Nineteen patients (nineteen arthroplasties) had a revision of one or more components because of aseptic loosening, osteolysis, polyethene wear, or periprosthetic fracture. Eight patients (eight arthroplasties) had an additional operation, without revision of components, because of periprosthetic fracture, instability, or rupture of the extensor mechanism. The findings al the subsequent operation did not alter the final diagnosis for any patient.
Radiographs made al the lime of follow-up after the unrevised arthroplasties were analyzed for zonal interface radiolucency^!<ulJ5. Nonprogressive radiolucent lines of no more than two millimetres in width were present in 12% (eighty-eight) of the 727 knees; the lines were seen in thirty-four femora, fifty-one tibiae, and three patellae. No radiolucent lines were complete, and no unrevised components were loose radiographycally1’. Radiographic analysis al the lime of follow-up after the hip arthroplasties revealed that six porous-coated femoral stems had a circumferential radiolucent line or evidence of subsidence10. In addition, three cemented femoral stems had radiolucent lines that were consistent with possible loosening and two had radiolucent lines that were consistent with probable loosening, according to the criteria of Harris and McGann \ A non-progressive radiolucent line was seen in zone 1 adjacent to 7% (thirty-six) of the 516 acetabular cups, in zone 2 adjacent to 24% (124) of the cups, and in zone 3 adjacent to 48% (248) of the cups*. An area of acetabular osteolysis developed in one patient, but it remained asymptomatic al ten years and the patient declined additional operative intervention.
At our private, university-affiliated hospital, the total charge per histological specimen analyzed after a total hip or knee arthroplasty in 1998 was $196.27. This includes $152.00 for the pathologist’s professional fees for gross inspection of the specimen and interpretation of the slides. The remaining $44.27 represents direct and indirect costs for decalcification and preparation of the specimen. These costs include those for labour, supplies, and administrative charges. During the same period, actual reimbursement, which was dependent on the type of health insurance (Table I), averaged $16.92 for preparation of the specimen and $85.67 for gross inspection and histological evaluation (a total of $10259)
.A malignant lesion was diagnosed before or during twelve arthroplasties, and pigmented villonodular synovitis was diagnosed during one. Histological analysis could have been omitted from the protocols for the remaining 1375 arthroplasties without compromising the outcome or prognosis in the long term. If a discretionary policy for histological evaluation had been implemented, total charges would have been reduced by $269,871.25 and total reimbursements would have been reduced by $141,061.25.
There is increasing evidence that histological evaluation of specimens resected during primary hip and knee arthroplasties should not be mandatory. However, discrepancies between the surgeon’s preoperative diagnosis and the histological diagnosis have been reported after 0.4% (eight) of 2035 total knee arthroplasties24 and after 5.4% (ninety-seven) of 1794 total hip arthroplasties(9). In a series including both hip and knee arthroplasties, in which the protocol was similar to our own, the discrepancy rale was 0.84 percent (six of 715)7, which is comparable with our rate of 0.80% (eleven of 1376). A discrepancy rate of 10% (sixteen of 168) in a series of total hip and knee arthroplasties was reported by Raab cl al.25. However, in that series, pathology requisition forms contained clinical data on only 60% (101) of the 168 hips and knees and little attention was given to intraoperative findings. Unlike the authors of the other studies, we also documented the surgeon’s intraoperative diagnosis. Additional diagnostic information can be obtained by careful inspection of the joint surfaces; synovial tissue; and, when a hip replacement is performed, the cut surfaces of the divided femoral head. In our series, this practice reduced the rate of discrepancy between the surgeon’s combined preoperative and intraoperative diagnoses and the histological diagnosis to zero for benign etiologies.
We noted eleven discrepancies between the preoperative diagnosis and the intraoperative diagnosis for patients who had benign conditions. In nine instances, a preoperative diagnosis of primary osteoarthritis was revised to an intraoperative diagnosis of avascular necrosis on the basis of the gross appearance of the cut surfaces of the femoral head. Ochronosis was evident because of the black appearance of the synovial tissue, which is characteristic of this condition. Pigmented villonodular synovitis was similarly diagnosed on the basis of the characteristic appearance of proliferating pigmented synovial tissue. All of the surgeon’s intraoperative diagnoses were confirmed by the histological evaluations. No morbid conditions would therefore have been misdiagnosed had the surgeon decided which specimens were to be sent for histological evaluation.
The clinical relevance of discrepancies between the surgeon’s diagnosis and the histological diagnosis has not been fully established. The situation is further complicated by an 8% intraobserver variance (fourteen of 168) in the histological evaluation of resected specimens by pathologists3. This variance raises the question of whether the histological diagnosis is an appropriate so-called gold standard for the final diagnosis.
DiCarlo cl al. reported that two of 1794 resected femoral heads had evidence of a malignant condition that had not been suspected preoperatively’. The postoperative management of the two patients was affected by the histological diagnosis. The overall prevalence of this potential disaster is unknown but appears to be very low. In total hip arthroplasties, division and inspection of the femoral head intraoperatively may further reduce the risk of such a missed diagnosis. We are not aware of any other reported series in which a discrepancy between the clinical and histological diagnoses affected the management of the patient. In the present study, at a mean of 5.5 years after the arthroplasty, no diagnosis was revised, the management of the patients was not altered by discrepant histological diagnoses, and no occult periprosthetic malignant lesions developed.
Our indications for histological evaluation during primary total hip and knee arthroplasty include a preoperative or intraoperative suspicion of malignancy, pathological fracture, impending pathological fracture, or the need to identify a pathological entity. If those criteria had been used in the present study, a histological evaluation would have been necessary after only 1 percent (thirteen) of the 1388 arthroplasties. Important underlying pathological conditions would not have been missed in any of the remaining patients had histological evaluation been omitted, and the outcome of the operation would not have been compromised. Another potential indication for histological evaluation may be for instructional purposes, including clinical correlation with interpretations of radiographs, in a teaching hospital. In conclusion, our findings support the contention that analysis of histological specimens can be safely omitted from the protocols of most total hip and knee replacements without fear of compromising the outcome and with considerable savings in both hospital charges and costs. It is possible that other studies involving more patients and longer follow-up may lead to different conclusions. Histological evaluation should be omitted only when the surgeon has diagnosed a benign condition on the basis of a carefully obtained history, a carefully performed physical examination, and a thorough review of the radiographs and when the preoperative diagnosis has been confirmed by intraoperative inspection of the resected specimens and, in cases of total hip arthroplasty, the cut surfaces o