Hip disarticulation: Factors affecting outcome Eric D. Endean, M D , Thomas H . Schwarcz, M D , D o n a l d E. Barker, M D , Nabil A. Munfakh, M D , R o b i n Wiison-Neely, R N , and G o r d o n L. H y d e , M D ,
Lexington, Ky. Hip disarticttlation, especially in patients with peripheral vascular disease, has been associated with high morbidity and mortality rates. This report describes patient characteristics that influence the clinical outcome of hip disarticulation. The medical records of all patients undergoing hip disarticulation from 1966 to 1989 were reviewed for surgical indication, perioperative wound complications, and postoperative deaths. Fifty-three patients underwent hip disarticulation for limb ischemia (10), infection (12), infection and ischemia (14), or tumor (17). The overall incidence of wound complications was 60%, and no significant differences were found among the groups. Prior above-knee amputation and urgent/emergent operations were significantly associated with increased wound complications (p < 0.05). The overall mortality rate was 21%, ranging from 0% (tumor) to 50% (ischemia) and differed significantly among the groups (p < 0.02). Mortality was significantly associated with urgent/emergent operations (p < 0.01). Age, diabetes mellitus, and previous inflow procedures did not influence mortality rates. The presence of limb ischemia influenced mortality rates to a greater extent than did infection, and a history of cardiac disease was statistically predictive of death. Wound complications frequently accompanied hip disarticulation, regardless of operative indication, and were significantly increased by urgent/emergent operations and prior above-knee amputation. Hip disarticulation can be performed with low mortality rates in selected patients. Both limb ischemia and infection substantially increase operative mortality rates. (J VAse Stm~ 1991;14:398-404.)
Successful lower extremity amputation can be achieved in most patients with below-knee or aboveknee amputations. However, in patients with proximal, life-threatening infections, ischemia, malignancy, or trauma, an above-knee amputation may be inadequate, necessitating a hip disarticulation (HD). Unruh et al. ~ recently reported on the clinical outcome of patients who underwent HD. They found an overall mortality rate of 44% and noted that preoperative infection had an adverse effect on mortality rates. In addition, their series had a postoperative wound infection rate of 63%. In an invited commentary, J. B. Aust stated that " . . . H D should be associated with a mortality approaching
From the Department of Surgery, Universityof KentuckyMedical Center, Lexington. Presented at the Fifteenth Annual Meeting of The Southern Association for Vascular Surgery, Palm Springs, Calif., Jan. 23-26, 1991. Reprint requests: Eric D. Endean, MD, Department of Surgery, University of KentuckyChandler MedicalCenter, 800 Rose St., Lexington, KY 40536-0084. 24/6/30623 398
0% and a morbidity related to an occasional wound infection...,2 Thus a discrepancy exists between the expected and the reported outcomes of patients who undergo HD. This study was undertaken to review the morbidity and mortality rates of H D at this medical center and to determine which patient characteristics influence the clinical outcome of this operation. PATIENTS AND METHODS The medical records of all patients undergoing H D between January 1966 and December 1989 at the University of Kentucky Medical Center were reviewed. Patients were grouped by operative indications, and their outcomes regarding perioperative wound complications (wound infection, superficial necrosis, deep tissue necrosis, persistent drainage, or wound dehiscence) and in-hospital postoperative deaths were recorded. Risk factors (Table I) including sex, presence of diabetes mellitus, hypertension, cardiac disease, and history of smoking were recorded. Prior operative intervention ipsilateral to the affected extremity was documented, including above-
Volume 14 Number 3 September 1991
H i p disarticulation
399
Table I. Patient characteristics by operative indication
Groups Patient characteristic
Ischemia N = 10
Infection N = 14
Ischemia and infection N = 12
Z/~/q/t0r N = 17
Mean age (yr) Diabetes mellitus Heart disease~ Smoking Sex (male)u Hypertension Prior AKAc Prior inflowc Urgent/emergent 1-IDa
60.0 4 (40%) 7 (70%) 7 (70%) 10 (100%) 3 (30%) 4 (40%) 6 (60%) 9 (90%)
47.3 4 (29%) 1 (7%) 9 (64%) 11 (79%) 2 (14%) 2 (14%) 0 9 (64%)
61.9 3 (25%) 9 (75%) 6 (50%) 8 (66%) 3 (25%) 12 (100%) 7 (58%) 11 (92%)
55.6 2 (12%) 2 (12%) 5 (29%) 6 (35%) 3 (18%) 1 (6%) 0 1 (6%)
A1CA,Above-knee amputation. ~O< 0.0002; Up < 0.005; ~p < 0.0001; ap < 0.001 chi-square analysisdetermining statistical significanceamong groups.
knee amputation and previous vascular inflow procedures. Hip disarticulations were classified as urgent/emergent if done for an acute life-threatening condition or elective if undertaken as a scheduled operation. The technique used in wound closure was noted. Finally, the frequency of HD at this medical center was documented by year to determine if there were any differences in the number of, or indication for, HDs done during the study period. Patient groups
Sixty patients required HD. Seven were done acutely for trauma, and these patients were excluded from the present study. The remaining 53 patients were categorized according to operative indication: ischemia, infection, ischemia and infection, and tulilor.
Ischemia group. The ischemia group (n = 10) had tissue ischemia and/or gangrene as a result of peripheral vascular occlusive disease without clinically apparent infection in the affected limb. The patients' mean age was 60 years, with a range of 33 to 74 years. Six patients had previous operations to establish inflow including aortofemoral bypass grafting (3) and femoral thromboembolectomy (3). Aortofemoral bypass was done at least 2 years before HD. In two, further bypass from the graft to the deep femoral artery was attempted. Both of these failed acutely, and the patients underwent above-knee amputations, one requiring early revision to HD whereas the other revision was performed electively 3 months after above-knee amputation. Acute thrombosis of an aorto-femoral graft developed in one patient, and no attempt at revascularization was made before HD. Inflow was established in three patients with femoral thromboembolectomy. Because of extensive tissue necrosis, HD was required within 2
days in these patients. Four patients had acute limb ischemia judged to be unreconstructible and underwent urgent/emergent HDs. These included limb ischemia after abdominal aortic aneurysm repair, contralateral hypogastric and ipsilateral common iliac artery ligation for hemorrhage, and two above-knee amputations. Infection group. The infection group (n = 14) underwent HD for infection and had no ischemia. The average age of these patients was 47.3 years, with a range of 19 to 85 years. The infections included massive soft tissue infections (8), decubitus ulcers (3), or femoral osteomyelitis (3). Three of the soft tissue infections were clostridial gas gangrene. All three patients with decubitus ulcers were paraplegic, and one additional patient with a soft tissue infection was paraplegic. Ischemia and infection group. The ischemia and infection group (n = 12) had severe infection complicating an ischemic extremity. The mean age of patients in the ischemia and infection group was 61.9 years, with a range of 45 to 79 years. All patients in this group had a prior above-knee amputation. Six patients had undergone previous bypass operations to establish inflow including aortofemoral bypass (3), iliofemoral bypass (2), and aortoiliac bypass (1). Before liD, revision of the failing and/or infected grafts was attempted. One patient had a thromboembolectomy to reestablish lower extremity blood flow. Four prosthetic vascular grafts became infected. In two of these patients, the iliofemoral grafts were removed and replaced with a venous conduit. One patient had an attempted iliopopliteal obturator graft and in one patient the prosthetic graft was not removed. In four patients the infection was dostridial gas gangrene. In one patient the clostridial infection was
Journal of VASCULAR SURGERY
400 Endean et al.
a complication of an above-knee amputation done 5 days previously. In the three other patients an above-knee amputation was performed to treat lower extremity dostridial gas gangrene, and H D was required to control ascending infection. Tumor group. The fourth group was comprised of patients who had removal 0ftheir lower extremity, including the hip joint for tumor (n = 17). The mean age of these patients was 55.6 years, with a range of 13 to 87 years. These patients served as a reference group because they had neither preoperative infection nor ischemia. Twelve patients in this group underwent a hemipelvectomy, which is a more extensive procedure than HD and carries a higher wound complication rate? In our series, however, no differences in wound complications or mortality rates were noted between patients having a hemipelvectomy and those having an HD. To provide a larger reference group and to simplify comparisons, all patients with tumors who were undergoing removal of their entire lower extremity including the hip joint (HD and hemipelvectomy) were categorized as having undergone an HD. The tumor types were predominantly sarcomas and included chondrosarcoma (2), fibrosarcoma (1), fibrous histiocytoma (2), leiomyosarcoma (1), liposarcoma (3), osteosarcoma (1), rhabdomyosarcoma (1), synovial sarcoma (1), and unspecified sarcoma (2). In addition one patient each had melanoma, squamous cell carcinoma, and neurofibromatosis. Statistical analysis A one-way analysis of variance (ANOVA) was used to compare the mean age among the four surgical indication groups. Other patient characteristics were compared among the groups by use of a chi-square statistic for appropriate contingency tables. Univariate analyses to determine which risk factors were associated with the end points of wound complication and postoperative deaths were based on chi-square analyses. Multivariate analyses were based on a stepwise logistic regression model to identify which risk factors and which patient characteristics predicted the log odds of incurring a wound complication or postoperative death. Statistical significance was determined at thep < 0.05 level. RESULTS Overall, the number of HDs done during the study, as tallied in 5-year periods, remained stable (Fig.~l). In the first half of the study period (1966 to 1976), more HDs were done for patients with tumor and fewer for patients with an ischemic component.
In the second half (1977 to i989), the reverse was true. From 1966 through 1976, 13 HDs were done for tumor, whereas only four were done from i 9 7 7 to 1989. Before 1 9 7 7 eight HDs were done in patients with ischemia, compared to 14 after 1977. The demographic characteristics of the patients are shown in Table I. A 2:1 male to female ratio was observed with more than half (11/18) of the women being in the tumor group. This difference in distribution of sex among the groups was found to be statistically significant (2 < 0.005, chi square). The incidence of cardiac disease among groups was also found to be significantly different (p < 0.0002, chi square). The presence of smoking, diabetes mellitus, and hypertension (chi square) and the patients' age (one-way ANOVA) did not differ statistically among the groups. Differences in the distribution of prior above-knee amputations was found to be significant among the groups (2 < 0.0001, chi square). Because no patients in the rumor or infection groups had a vascular operation before HD, the distribution of prior inflow procedures was also found to be significantly different (2 < 0.0001) among the groups. Slightly more than half of the HDs (30/53, 57%) were done on an urgent or emergent basis, as opposed to elective operations (23/53, 43%). This distribution of urgent/emergent operations (Table I) among the groups was statistically different (2 < 0.0001, chi square), with only one patient in the tumor group needing an urgent HD. The indication for this urgent operation was rumor hemorrhage occurring after initiation of chemotherapy. Ten postoperative deaths occurred in patients taken to the operating room for an urgent/emergent HD, whereas only one patient undergoing elective HD died. Urgent/emergent operations were found to be significantly associated with both wound complications (2 < 0.04) and deaths (2 < 0.01, chi square).
Wound complications Wound complications occurred in 32/53 (60%) of patients (Table II). Most complications involved superficial and deep wound infections, skin necrosis, and/or deep tissue necrosis. An infectious wound complication developed in 26 patients. Twenty-two (85%) of these were associated with an additional wound complication including tissue necrosis, dehiscence, and hematoma. One graft infection occurred in the ischemia and infection group. Five wound hematomas developed, one in the infection group and four in the tumor group. Although the wound
Volume 14 Number 3 September 1991
Hip disarticulation 401
2O
t" O
(14)
[]
Ischemia and/or Infection
•
Tumor
-I O
(lo) FI
10
1
(10)
I
(10)
I
1
(9)
E z
0 1965-69
1970-74
1975-79
1980-84
1985-89
Years
Fig. 1. The number of liDs tallied in 5-year periods. The overallnumber remained stable over the 24 year study period. However, more operations were performed for patients with a component of limb ischemia in the second half of the study period than in the earlier half. Table II. Morbidity and mortality rates by operative indication Groups
Morbidity~mortality
Ischemia N = 10
Infection N = 14
Ischemia and infection N = 12
Tumor N = 17
Wound complications Mortality a
7 (70%) 5 (50%)
8 (57%) 2 (14%)
10 (83%) 4 (33%)
8 (47%) 0
~0 < 0.02 chi-squarc analysis determining statistical significance among groups.
complication rate in the ischemia and infection group was highest, comparison of the complication rates among the groups did not show statistically significant differences (chi square). In patients with preoperative infection, 14/26 developed postoperative wound infections (infection, 6/14; ischemia and infection, 8/12). The organism cultured before operation was recovered from the postoperative wound in seven (50%). In patients with clostridial gas gangrene (7), five had wound complications, four of which included infection. One patient with gas gangrene had no documented wound complications, and one died on the first postoperative day. The effect of wound closure technique on wound complications was evaluated. Primary closure was
used in 28 patients. A subsequent wound complication developed in 15 (54%) of these patients. The H D wounds were left open in 24 patients, and wound related complications developed in 18 (75%). One patient required placement of a free tissue flap, and a wound complication also developed in this patient. No statistically significant relationship was observed between wound complications and the method of wound closure. Prior ipsilateral above-knee amputation was found to be significantly associated with the development of wound complications. A postoperative wound complication developed in 15 (79%) of the 19 patients who had an ipsilateral above-knee amputation preceding HD. In contrast, postoperative
402
Endean et al.
wound complications developed in 17 of 34 (50%) patients who did not have a previous above-knee amputation. This difference was found to be statistically significant (p < 0.05, chi square). However, a multivariate analysis based on a stepwise logistic regression model showed that the only variable significantly predicting the log odds of a wound complication was urgent/emergent operation (p < 0.04). This suggests that it is the severity of the disease process, resulting in the need for urgent operation, that is most responsible for wound morbidity. Deaths Eleven patients died in the postoperative period for an overall mortality rate of 20.8%. Death was attributed to cardiac causes in three patients. Two patients had myocardial infarctions; one occurred after an upper gastrointestinal hemorrhage. One patient had exacerbation of congestive heart failure after HD. Seven patients died of infectious causes including sepsis resulting in multisystem organ failure (4), pneumonia (1), disseminated tuberculosis (1), and gas gangrene (1). One patient died after a stroke. Postoperative deaths for each group are listed in Table II. The incidence of deaths compared among groups was significantly different (p < 0.02, chi square). When mortality rates were compared between groups (chi square), significant differences were demonstrated between the ischemia group and the tumor group (p < 0.001), and the ischemia and infection group and the tumor group (iv < 0.01). No significant differences in mortality rates were found between other groups. Comparison of mortality rates between the ischemia and infection groups approached statistical significance (p = 0.058). The presence of a prior above-knee amputation was not predictive of postoperative death (chi square). Univariate analysis indicated that cardiac disease was significantly associated with increased mortality rates (p < 0.005). Postoperative death occurred in 8/19 (42%) patients with a history of cardiac disease, whereas only 3/34 (9%) deaths occurred in patients without a prior cardiac history. The incidence of a cardiac hiStOry is highest in patients in the ischemia (7/10) and ischemia and infection (9/12) groups, compared to only 1/14 patients in the infection group and 2/17 in the tumor group. Consequently, a multivariate analysis based on a stepwise logistic regression model showed that the only variable significantly predicting the log odds of a postoperative death was the presence of cardiac disease (p < o.ol).
Journal of VASCULAR SURGERY
DISCUSSION The operative technique for H D has been well described in the literature, 4 but few articles have reported patients' clinical outcome. Hip disarticulation in patients with peripheral vascular disease is done infrequently since above-knee or below-knee amputations usually heal satisfactorily, even in patients with severely ischemic extremities. In the current study we have classified patients who underwent H D by operative indication and evaluated the effects of preoperative ischemia and/or infection on outcome. A high incidence of wound complications was noted in this series, regardless of operative indication. These complications ranged from skin necrosis and superficial wound infections to abscesses, hematomas, and deep tissue necrosis. Twenty-two of the 26 wound infections were associated with other wound complications including tissue necrosis, wound dehiscence, or hematoma. The frequency of multiple wound problems demonstrates the need for both meticulous surgical technique and adequate blood supply to the amputation site. The wound complication rates between the ischemia, infection, and rumor groups were similar (60%, 50%, and 47%, respectively), whereas the wound complications in the ischemia and infection group were more frequent (83%), although not statistically greater. Patients who had a prior above-knee amputation were found to be at a significantly increased risk for developing wound complications. This finding may be explained by the presence of preoperative ischemia of the extremity in most patients (16/19, 84%) in whom H D followed an above-knee amputation. It is interesting that all patients (12) in the ischemia and infection group had a prior above-knee amputation. Apparently, some above-knee amputations were done through marginally viable or infected tissue, anticipating that a higher amputation could be avoided. The subsequent development of frankly nonviable tissue in the above-knee amputation site and the inevitable delay in H D probably contributed to the associated high incidence of wound complications. The only factor predictive of wound complication by multivariate analysis was the urgency with which the operation was undertaken. Only 7/36 (19%) of operations done in patients with ischemia and/or infection were clone on an elective basis, whereas 16/17 (94%) HDs in patients with rumors were done electively. This indicates that HDs done for ischemia and/or infection were often done in life-threatening situations. Under such circumstances, marginally
Volume 14 Number 3 September 1991
viable tissue or residual infection remaining at the HD site would not be unexpected, which would lead to a high incidence of postoperative wound complications. Preoperative ischemia appeared to influence mortality rates to a greater extent than did infection. In the two groups of patients having a component of ischemia, the mortality rate in the ischemia group was higher (50%) than that of the ischemia and infection group (33%), suggesting that the presence of preoperative infection in patients with ischemic extremities did not increase mortality rates. In addition, the mortality rates of patients having only infection was lower (14%) than that when both infection and ischemia were present (33 % ). However, the younger age of patients in the infection group may also have contributed to decreased deaths. The increased mortality rate associated with urgent/emergent operations is probably a reflection of the acute nature and severity of the disease process. Multivariate analysis identified the presence of cardiac disease as the only factor predictive of death. By far the greatest incidence of cardiac disease was found in patients with ischemia (ischemia and ischemia and infection groups), and it has been well established that patients with peripheral vascular occlusive disease have an high incidence of cardiac disease.S'6 This association of extremity ischemia with cardiac disease underscores our observation that ischemia had a greater influence on mortality rates than infection. Other reports have described an association between perioperative ischemia of an above-knee amputation stump with deaths. Bunt 7 reported that the development of acute ischemic necrosis immediately after above-knee amputation had a high mortality rate (60%) despite attempts at revascularization. None of the patients in his series were reported to have undergone an HD. Kwaan and Co,molly8 also described a high mortality rate in patients in whom ischemia developed in an already healed above-knee amputation snmlp. The conclusion of both reports was that early revascularization is needed to treat an ischemic stump. We did not find a similar association between prior above-knee amputation and deaths rather in our patients undergoing HD, rather we found that above-knee amputation was associated with an increased incidence of wound complications. The only other recent report of patients undergoing H D described an experience of 34 patients over 11 years) Overall, the perioperative mortality rate was 44%, with the highest mortality rate
Hip disarticulation
403
occurring in patients treated for trauma (60%), followed by those operated on for ischemia (47%). A marked increase in deaths was noted when perioperative infection was present, and the overall wound infection rate in that series was 65%. The present report has notable similarities to these findings, especially with regard to morbidity and mortality rates.
In summary, we have reviewed a 24-year experience of HDs. The highest mortality rate occurred in patients undergoing HD for ischemia. Wound complications occurred frequently in all groups, ranging from 47% (tumor group) to 83% (ischemia and infection group). In the tumor group, all but one of the wound complications included infection, most (5/7) of which were associated with tissue necrosis at the amputation site. Nonviable tissue flaps may have contributed to this high incidence of wound infection. Urgent/emergent operations were predictive of a higher wound complication rate. Wound complications also occurred more frequently in patients who had undergone a prior ipsilateral above-knee amputation. It appears that a delay in performing HD may have contributed to this problem. Urgent/emergent operations were associated with increased mortality rates, whereas ischemia and a history of cardiac disease appeared to have the greatest influence on postoperative mortality rates. Thus in selected patients, that is, elective operations for tumor resection, HD can be done with a low mortality rate. However, regardless of indication, wound complications occur frequently. Therefore we recommend early operation (i.e., revascularization or above-knee amputation) for limb ischemia and/or infection to avoid the need for HD and its associated high morbidity and mortality rates. If an above-knee amputation is needed, a concomitant revascularization should be performed if any question of tissue perfusion exists at the amputation site. If revascularization cannot be accomplished because of infection or other technical difficulties, an HD may be indicated, rather than an above-knee amputation of questionable viability. Finally, prompt HD, without delay, should be considered when it becomes apparent that an above-knee amputation is nonviable. We are hopeful that these recommendations will improve the morbidity and mortality rates of HD. The authors thank Richard J. Kryscio, Department of Biostatistics, for the statistical analysis in this manuscript. REFERENCES 1. Unruh T, Fisher DF, Unnah TA, et al. Hip disarticulation: an 11-year experience. Arch Surg 1990;125:791-3.
404
Journal of VASCULAR SURGERY
Endean et al.
2. Aust JB. Invited commentary. Arch Surg 1990;125:793. 3. Douglass HO, Razack M, Holyoke ED. Hemipelvectomy. Arch Surg 1975;110:82-5. 4. Sugarbaker PH, Chretien PB. A surgical technique for hip disarticulation. Surgery 1981;90:546-53. 5. Beven EG. Routine coronary angiography in patients undergoing surgery for abdominal aortic aneurysm and lower extremity occlusive disease, l VASC SURG 1986;3:682-4. 6. Hertzer NR. Basic data concerning associated coronary disease in peripheral vascular patients. Ann Vasc Surg 1987;1:616-20.
7. BuntTJ. Gangrene ofthe immediate postoperative above-knee amputation stump: role of emergency revascularization in preventing death. J VAsc SURG 1985;2:874-7. 8. Kwaan JHM, Connolly JE. Fatal sequelae of the ischemic amputation stump: a surgical challenge. Am J Surg 1979;138: 49-52.
Submitted Feb. 13, 1991; accepted May 2, 1991.
Lexington, Ky. Hip disarticttlation, especially in patients with peripheral vascular disease, has been associated with high morbidity and mortality rates. This report describes patient characteristics that influence the clinical outcome of hip disarticulation. The medical records of all patients undergoing hip disarticulation from 1966 to 1989 were reviewed for surgical indication, perioperative wound complications, and postoperative deaths. Fifty-three patients underwent hip disarticulation for limb ischemia (10), infection (12), infection and ischemia (14), or tumor (17). The overall incidence of wound complications was 60%, and no significant differences were found among the groups. Prior above-knee amputation and urgent/emergent operations were significantly associated with increased wound complications (p < 0.05). The overall mortality rate was 21%, ranging from 0% (tumor) to 50% (ischemia) and differed significantly among the groups (p < 0.02). Mortality was significantly associated with urgent/emergent operations (p < 0.01). Age, diabetes mellitus, and previous inflow procedures did not influence mortality rates. The presence of limb ischemia influenced mortality rates to a greater extent than did infection, and a history of cardiac disease was statistically predictive of death. Wound complications frequently accompanied hip disarticulation, regardless of operative indication, and were significantly increased by urgent/emergent operations and prior above-knee amputation. Hip disarticulation can be performed with low mortality rates in selected patients. Both limb ischemia and infection substantially increase operative mortality rates. (J VAse Stm~ 1991;14:398-404.)
Successful lower extremity amputation can be achieved in most patients with below-knee or aboveknee amputations. However, in patients with proximal, life-threatening infections, ischemia, malignancy, or trauma, an above-knee amputation may be inadequate, necessitating a hip disarticulation (HD). Unruh et al. ~ recently reported on the clinical outcome of patients who underwent HD. They found an overall mortality rate of 44% and noted that preoperative infection had an adverse effect on mortality rates. In addition, their series had a postoperative wound infection rate of 63%. In an invited commentary, J. B. Aust stated that " . . . H D should be associated with a mortality approaching
From the Department of Surgery, Universityof KentuckyMedical Center, Lexington. Presented at the Fifteenth Annual Meeting of The Southern Association for Vascular Surgery, Palm Springs, Calif., Jan. 23-26, 1991. Reprint requests: Eric D. Endean, MD, Department of Surgery, University of KentuckyChandler MedicalCenter, 800 Rose St., Lexington, KY 40536-0084. 24/6/30623 398
0% and a morbidity related to an occasional wound infection...,2 Thus a discrepancy exists between the expected and the reported outcomes of patients who undergo HD. This study was undertaken to review the morbidity and mortality rates of H D at this medical center and to determine which patient characteristics influence the clinical outcome of this operation. PATIENTS AND METHODS The medical records of all patients undergoing H D between January 1966 and December 1989 at the University of Kentucky Medical Center were reviewed. Patients were grouped by operative indications, and their outcomes regarding perioperative wound complications (wound infection, superficial necrosis, deep tissue necrosis, persistent drainage, or wound dehiscence) and in-hospital postoperative deaths were recorded. Risk factors (Table I) including sex, presence of diabetes mellitus, hypertension, cardiac disease, and history of smoking were recorded. Prior operative intervention ipsilateral to the affected extremity was documented, including above-
Volume 14 Number 3 September 1991
H i p disarticulation
399
Table I. Patient characteristics by operative indication
Groups Patient characteristic
Ischemia N = 10
Infection N = 14
Ischemia and infection N = 12
Z/~/q/t0r N = 17
Mean age (yr) Diabetes mellitus Heart disease~ Smoking Sex (male)u Hypertension Prior AKAc Prior inflowc Urgent/emergent 1-IDa
60.0 4 (40%) 7 (70%) 7 (70%) 10 (100%) 3 (30%) 4 (40%) 6 (60%) 9 (90%)
47.3 4 (29%) 1 (7%) 9 (64%) 11 (79%) 2 (14%) 2 (14%) 0 9 (64%)
61.9 3 (25%) 9 (75%) 6 (50%) 8 (66%) 3 (25%) 12 (100%) 7 (58%) 11 (92%)
55.6 2 (12%) 2 (12%) 5 (29%) 6 (35%) 3 (18%) 1 (6%) 0 1 (6%)
A1CA,Above-knee amputation. ~O< 0.0002; Up < 0.005; ~p < 0.0001; ap < 0.001 chi-square analysisdetermining statistical significanceamong groups.
knee amputation and previous vascular inflow procedures. Hip disarticulations were classified as urgent/emergent if done for an acute life-threatening condition or elective if undertaken as a scheduled operation. The technique used in wound closure was noted. Finally, the frequency of HD at this medical center was documented by year to determine if there were any differences in the number of, or indication for, HDs done during the study period. Patient groups
Sixty patients required HD. Seven were done acutely for trauma, and these patients were excluded from the present study. The remaining 53 patients were categorized according to operative indication: ischemia, infection, ischemia and infection, and tulilor.
Ischemia group. The ischemia group (n = 10) had tissue ischemia and/or gangrene as a result of peripheral vascular occlusive disease without clinically apparent infection in the affected limb. The patients' mean age was 60 years, with a range of 33 to 74 years. Six patients had previous operations to establish inflow including aortofemoral bypass grafting (3) and femoral thromboembolectomy (3). Aortofemoral bypass was done at least 2 years before HD. In two, further bypass from the graft to the deep femoral artery was attempted. Both of these failed acutely, and the patients underwent above-knee amputations, one requiring early revision to HD whereas the other revision was performed electively 3 months after above-knee amputation. Acute thrombosis of an aorto-femoral graft developed in one patient, and no attempt at revascularization was made before HD. Inflow was established in three patients with femoral thromboembolectomy. Because of extensive tissue necrosis, HD was required within 2
days in these patients. Four patients had acute limb ischemia judged to be unreconstructible and underwent urgent/emergent HDs. These included limb ischemia after abdominal aortic aneurysm repair, contralateral hypogastric and ipsilateral common iliac artery ligation for hemorrhage, and two above-knee amputations. Infection group. The infection group (n = 14) underwent HD for infection and had no ischemia. The average age of these patients was 47.3 years, with a range of 19 to 85 years. The infections included massive soft tissue infections (8), decubitus ulcers (3), or femoral osteomyelitis (3). Three of the soft tissue infections were clostridial gas gangrene. All three patients with decubitus ulcers were paraplegic, and one additional patient with a soft tissue infection was paraplegic. Ischemia and infection group. The ischemia and infection group (n = 12) had severe infection complicating an ischemic extremity. The mean age of patients in the ischemia and infection group was 61.9 years, with a range of 45 to 79 years. All patients in this group had a prior above-knee amputation. Six patients had undergone previous bypass operations to establish inflow including aortofemoral bypass (3), iliofemoral bypass (2), and aortoiliac bypass (1). Before liD, revision of the failing and/or infected grafts was attempted. One patient had a thromboembolectomy to reestablish lower extremity blood flow. Four prosthetic vascular grafts became infected. In two of these patients, the iliofemoral grafts were removed and replaced with a venous conduit. One patient had an attempted iliopopliteal obturator graft and in one patient the prosthetic graft was not removed. In four patients the infection was dostridial gas gangrene. In one patient the clostridial infection was
Journal of VASCULAR SURGERY
400 Endean et al.
a complication of an above-knee amputation done 5 days previously. In the three other patients an above-knee amputation was performed to treat lower extremity dostridial gas gangrene, and H D was required to control ascending infection. Tumor group. The fourth group was comprised of patients who had removal 0ftheir lower extremity, including the hip joint for tumor (n = 17). The mean age of these patients was 55.6 years, with a range of 13 to 87 years. These patients served as a reference group because they had neither preoperative infection nor ischemia. Twelve patients in this group underwent a hemipelvectomy, which is a more extensive procedure than HD and carries a higher wound complication rate? In our series, however, no differences in wound complications or mortality rates were noted between patients having a hemipelvectomy and those having an HD. To provide a larger reference group and to simplify comparisons, all patients with tumors who were undergoing removal of their entire lower extremity including the hip joint (HD and hemipelvectomy) were categorized as having undergone an HD. The tumor types were predominantly sarcomas and included chondrosarcoma (2), fibrosarcoma (1), fibrous histiocytoma (2), leiomyosarcoma (1), liposarcoma (3), osteosarcoma (1), rhabdomyosarcoma (1), synovial sarcoma (1), and unspecified sarcoma (2). In addition one patient each had melanoma, squamous cell carcinoma, and neurofibromatosis. Statistical analysis A one-way analysis of variance (ANOVA) was used to compare the mean age among the four surgical indication groups. Other patient characteristics were compared among the groups by use of a chi-square statistic for appropriate contingency tables. Univariate analyses to determine which risk factors were associated with the end points of wound complication and postoperative deaths were based on chi-square analyses. Multivariate analyses were based on a stepwise logistic regression model to identify which risk factors and which patient characteristics predicted the log odds of incurring a wound complication or postoperative death. Statistical significance was determined at thep < 0.05 level. RESULTS Overall, the number of HDs done during the study, as tallied in 5-year periods, remained stable (Fig.~l). In the first half of the study period (1966 to 1976), more HDs were done for patients with tumor and fewer for patients with an ischemic component.
In the second half (1977 to i989), the reverse was true. From 1966 through 1976, 13 HDs were done for tumor, whereas only four were done from i 9 7 7 to 1989. Before 1 9 7 7 eight HDs were done in patients with ischemia, compared to 14 after 1977. The demographic characteristics of the patients are shown in Table I. A 2:1 male to female ratio was observed with more than half (11/18) of the women being in the tumor group. This difference in distribution of sex among the groups was found to be statistically significant (2 < 0.005, chi square). The incidence of cardiac disease among groups was also found to be significantly different (p < 0.0002, chi square). The presence of smoking, diabetes mellitus, and hypertension (chi square) and the patients' age (one-way ANOVA) did not differ statistically among the groups. Differences in the distribution of prior above-knee amputations was found to be significant among the groups (2 < 0.0001, chi square). Because no patients in the rumor or infection groups had a vascular operation before HD, the distribution of prior inflow procedures was also found to be significantly different (2 < 0.0001) among the groups. Slightly more than half of the HDs (30/53, 57%) were done on an urgent or emergent basis, as opposed to elective operations (23/53, 43%). This distribution of urgent/emergent operations (Table I) among the groups was statistically different (2 < 0.0001, chi square), with only one patient in the tumor group needing an urgent HD. The indication for this urgent operation was rumor hemorrhage occurring after initiation of chemotherapy. Ten postoperative deaths occurred in patients taken to the operating room for an urgent/emergent HD, whereas only one patient undergoing elective HD died. Urgent/emergent operations were found to be significantly associated with both wound complications (2 < 0.04) and deaths (2 < 0.01, chi square).
Wound complications Wound complications occurred in 32/53 (60%) of patients (Table II). Most complications involved superficial and deep wound infections, skin necrosis, and/or deep tissue necrosis. An infectious wound complication developed in 26 patients. Twenty-two (85%) of these were associated with an additional wound complication including tissue necrosis, dehiscence, and hematoma. One graft infection occurred in the ischemia and infection group. Five wound hematomas developed, one in the infection group and four in the tumor group. Although the wound
Volume 14 Number 3 September 1991
Hip disarticulation 401
2O
t" O
(14)
[]
Ischemia and/or Infection
•
Tumor
-I O
(lo) FI
10
1
(10)
I
(10)
I
1
(9)
E z
0 1965-69
1970-74
1975-79
1980-84
1985-89
Years
Fig. 1. The number of liDs tallied in 5-year periods. The overallnumber remained stable over the 24 year study period. However, more operations were performed for patients with a component of limb ischemia in the second half of the study period than in the earlier half. Table II. Morbidity and mortality rates by operative indication Groups
Morbidity~mortality
Ischemia N = 10
Infection N = 14
Ischemia and infection N = 12
Tumor N = 17
Wound complications Mortality a
7 (70%) 5 (50%)
8 (57%) 2 (14%)
10 (83%) 4 (33%)
8 (47%) 0
~0 < 0.02 chi-squarc analysis determining statistical significance among groups.
complication rate in the ischemia and infection group was highest, comparison of the complication rates among the groups did not show statistically significant differences (chi square). In patients with preoperative infection, 14/26 developed postoperative wound infections (infection, 6/14; ischemia and infection, 8/12). The organism cultured before operation was recovered from the postoperative wound in seven (50%). In patients with clostridial gas gangrene (7), five had wound complications, four of which included infection. One patient with gas gangrene had no documented wound complications, and one died on the first postoperative day. The effect of wound closure technique on wound complications was evaluated. Primary closure was
used in 28 patients. A subsequent wound complication developed in 15 (54%) of these patients. The H D wounds were left open in 24 patients, and wound related complications developed in 18 (75%). One patient required placement of a free tissue flap, and a wound complication also developed in this patient. No statistically significant relationship was observed between wound complications and the method of wound closure. Prior ipsilateral above-knee amputation was found to be significantly associated with the development of wound complications. A postoperative wound complication developed in 15 (79%) of the 19 patients who had an ipsilateral above-knee amputation preceding HD. In contrast, postoperative
402
Endean et al.
wound complications developed in 17 of 34 (50%) patients who did not have a previous above-knee amputation. This difference was found to be statistically significant (p < 0.05, chi square). However, a multivariate analysis based on a stepwise logistic regression model showed that the only variable significantly predicting the log odds of a wound complication was urgent/emergent operation (p < 0.04). This suggests that it is the severity of the disease process, resulting in the need for urgent operation, that is most responsible for wound morbidity. Deaths Eleven patients died in the postoperative period for an overall mortality rate of 20.8%. Death was attributed to cardiac causes in three patients. Two patients had myocardial infarctions; one occurred after an upper gastrointestinal hemorrhage. One patient had exacerbation of congestive heart failure after HD. Seven patients died of infectious causes including sepsis resulting in multisystem organ failure (4), pneumonia (1), disseminated tuberculosis (1), and gas gangrene (1). One patient died after a stroke. Postoperative deaths for each group are listed in Table II. The incidence of deaths compared among groups was significantly different (p < 0.02, chi square). When mortality rates were compared between groups (chi square), significant differences were demonstrated between the ischemia group and the tumor group (p < 0.001), and the ischemia and infection group and the tumor group (iv < 0.01). No significant differences in mortality rates were found between other groups. Comparison of mortality rates between the ischemia and infection groups approached statistical significance (p = 0.058). The presence of a prior above-knee amputation was not predictive of postoperative death (chi square). Univariate analysis indicated that cardiac disease was significantly associated with increased mortality rates (p < 0.005). Postoperative death occurred in 8/19 (42%) patients with a history of cardiac disease, whereas only 3/34 (9%) deaths occurred in patients without a prior cardiac history. The incidence of a cardiac hiStOry is highest in patients in the ischemia (7/10) and ischemia and infection (9/12) groups, compared to only 1/14 patients in the infection group and 2/17 in the tumor group. Consequently, a multivariate analysis based on a stepwise logistic regression model showed that the only variable significantly predicting the log odds of a postoperative death was the presence of cardiac disease (p < o.ol).
Journal of VASCULAR SURGERY
DISCUSSION The operative technique for H D has been well described in the literature, 4 but few articles have reported patients' clinical outcome. Hip disarticulation in patients with peripheral vascular disease is done infrequently since above-knee or below-knee amputations usually heal satisfactorily, even in patients with severely ischemic extremities. In the current study we have classified patients who underwent H D by operative indication and evaluated the effects of preoperative ischemia and/or infection on outcome. A high incidence of wound complications was noted in this series, regardless of operative indication. These complications ranged from skin necrosis and superficial wound infections to abscesses, hematomas, and deep tissue necrosis. Twenty-two of the 26 wound infections were associated with other wound complications including tissue necrosis, wound dehiscence, or hematoma. The frequency of multiple wound problems demonstrates the need for both meticulous surgical technique and adequate blood supply to the amputation site. The wound complication rates between the ischemia, infection, and rumor groups were similar (60%, 50%, and 47%, respectively), whereas the wound complications in the ischemia and infection group were more frequent (83%), although not statistically greater. Patients who had a prior above-knee amputation were found to be at a significantly increased risk for developing wound complications. This finding may be explained by the presence of preoperative ischemia of the extremity in most patients (16/19, 84%) in whom H D followed an above-knee amputation. It is interesting that all patients (12) in the ischemia and infection group had a prior above-knee amputation. Apparently, some above-knee amputations were done through marginally viable or infected tissue, anticipating that a higher amputation could be avoided. The subsequent development of frankly nonviable tissue in the above-knee amputation site and the inevitable delay in H D probably contributed to the associated high incidence of wound complications. The only factor predictive of wound complication by multivariate analysis was the urgency with which the operation was undertaken. Only 7/36 (19%) of operations done in patients with ischemia and/or infection were clone on an elective basis, whereas 16/17 (94%) HDs in patients with rumors were done electively. This indicates that HDs done for ischemia and/or infection were often done in life-threatening situations. Under such circumstances, marginally
Volume 14 Number 3 September 1991
viable tissue or residual infection remaining at the HD site would not be unexpected, which would lead to a high incidence of postoperative wound complications. Preoperative ischemia appeared to influence mortality rates to a greater extent than did infection. In the two groups of patients having a component of ischemia, the mortality rate in the ischemia group was higher (50%) than that of the ischemia and infection group (33%), suggesting that the presence of preoperative infection in patients with ischemic extremities did not increase mortality rates. In addition, the mortality rates of patients having only infection was lower (14%) than that when both infection and ischemia were present (33 % ). However, the younger age of patients in the infection group may also have contributed to decreased deaths. The increased mortality rate associated with urgent/emergent operations is probably a reflection of the acute nature and severity of the disease process. Multivariate analysis identified the presence of cardiac disease as the only factor predictive of death. By far the greatest incidence of cardiac disease was found in patients with ischemia (ischemia and ischemia and infection groups), and it has been well established that patients with peripheral vascular occlusive disease have an high incidence of cardiac disease.S'6 This association of extremity ischemia with cardiac disease underscores our observation that ischemia had a greater influence on mortality rates than infection. Other reports have described an association between perioperative ischemia of an above-knee amputation stump with deaths. Bunt 7 reported that the development of acute ischemic necrosis immediately after above-knee amputation had a high mortality rate (60%) despite attempts at revascularization. None of the patients in his series were reported to have undergone an HD. Kwaan and Co,molly8 also described a high mortality rate in patients in whom ischemia developed in an already healed above-knee amputation snmlp. The conclusion of both reports was that early revascularization is needed to treat an ischemic stump. We did not find a similar association between prior above-knee amputation and deaths rather in our patients undergoing HD, rather we found that above-knee amputation was associated with an increased incidence of wound complications. The only other recent report of patients undergoing H D described an experience of 34 patients over 11 years) Overall, the perioperative mortality rate was 44%, with the highest mortality rate
Hip disarticulation
403
occurring in patients treated for trauma (60%), followed by those operated on for ischemia (47%). A marked increase in deaths was noted when perioperative infection was present, and the overall wound infection rate in that series was 65%. The present report has notable similarities to these findings, especially with regard to morbidity and mortality rates.
In summary, we have reviewed a 24-year experience of HDs. The highest mortality rate occurred in patients undergoing HD for ischemia. Wound complications occurred frequently in all groups, ranging from 47% (tumor group) to 83% (ischemia and infection group). In the tumor group, all but one of the wound complications included infection, most (5/7) of which were associated with tissue necrosis at the amputation site. Nonviable tissue flaps may have contributed to this high incidence of wound infection. Urgent/emergent operations were predictive of a higher wound complication rate. Wound complications also occurred more frequently in patients who had undergone a prior ipsilateral above-knee amputation. It appears that a delay in performing HD may have contributed to this problem. Urgent/emergent operations were associated with increased mortality rates, whereas ischemia and a history of cardiac disease appeared to have the greatest influence on postoperative mortality rates. Thus in selected patients, that is, elective operations for tumor resection, HD can be done with a low mortality rate. However, regardless of indication, wound complications occur frequently. Therefore we recommend early operation (i.e., revascularization or above-knee amputation) for limb ischemia and/or infection to avoid the need for HD and its associated high morbidity and mortality rates. If an above-knee amputation is needed, a concomitant revascularization should be performed if any question of tissue perfusion exists at the amputation site. If revascularization cannot be accomplished because of infection or other technical difficulties, an HD may be indicated, rather than an above-knee amputation of questionable viability. Finally, prompt HD, without delay, should be considered when it becomes apparent that an above-knee amputation is nonviable. We are hopeful that these recommendations will improve the morbidity and mortality rates of HD. The authors thank Richard J. Kryscio, Department of Biostatistics, for the statistical analysis in this manuscript. REFERENCES 1. Unruh T, Fisher DF, Unnah TA, et al. Hip disarticulation: an 11-year experience. Arch Surg 1990;125:791-3.
404
Journal of VASCULAR SURGERY
Endean et al.
2. Aust JB. Invited commentary. Arch Surg 1990;125:793. 3. Douglass HO, Razack M, Holyoke ED. Hemipelvectomy. Arch Surg 1975;110:82-5. 4. Sugarbaker PH, Chretien PB. A surgical technique for hip disarticulation. Surgery 1981;90:546-53. 5. Beven EG. Routine coronary angiography in patients undergoing surgery for abdominal aortic aneurysm and lower extremity occlusive disease, l VASC SURG 1986;3:682-4. 6. Hertzer NR. Basic data concerning associated coronary disease in peripheral vascular patients. Ann Vasc Surg 1987;1:616-20.
7. BuntTJ. Gangrene ofthe immediate postoperative above-knee amputation stump: role of emergency revascularization in preventing death. J VAsc SURG 1985;2:874-7. 8. Kwaan JHM, Connolly JE. Fatal sequelae of the ischemic amputation stump: a surgical challenge. Am J Surg 1979;138: 49-52.
Submitted Feb. 13, 1991; accepted May 2, 1991.
