Abstract
A prospective, randomized study was performed to assess the effectiveness of postoperative closed suction drainage. One hundred and twelve consecutive procedures involving autologous iliac-crest bone graft were performed, from December 29, 1992, to July 1, 1993, following a traumatic injury of the spine in 108 patients. Sixty of the sites from which the bone graft had been obtained were drained with a single large Hemovac device. The drains were maintained for two to five days postoperatively. The remaining fifty-two incisions were closed without a drainage device. All patients were evaluated clinically for problems with wound-healing. The incisions were considered to be healed when they had been asymptomatic for one year. Of eleven patients who had problems with wound-healing, six had been managed with a drain and five had not. The findings of this study do not support the routine use of drainage at the donor sites of iliac-crest bone grafts.
Surgeons have used postoperative drains for generations. Hippocrates drained an operative wound with a hollow wooden tube around 400 B.C.1. In 1961, Waugh and Stinchfield introduced perforated silicone-rubber tubing that was strong enough to withstand suction of greater than 100 millimeters of mercury (13.33 kilopascals). Despite the long history of drain usage, however, little in the scientific literature supports the use of drains. Some drainage systems have been associated with a high rate of infection8-10,13. A number of retrospective studies have suggested that drains may be unnecessary in operations on the hip and knee4-7,11, and the results of a large prospective, randomized investigation of total hip and total knee replacements by Ritter et al. concurred with the findings of those studies.
It is difficult to dispute the seemingly logical concept of wound drainage. Concerns that clotted blood is an excellent bacterial culture medium are valid. However, it is recognized that drains do not eliminate wound hematomas completely. Also, it is common practice to clamp or even remove drains when the drainage is excessive. The empirical use of drains to avoid postoperative hematomas, then, is called into question.
The purpose of our study was to evaluate the effect of postoperative suction drainage on wound-healing following the removal of bone from the iliac crest for use as a graft in patients who had an injury of the spine.
*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.
†Indianapolis Neurosurgical Group, 8402 Harcourt Road, Suite 500, Indianapolis, Indiana 46260.
‡Northwestern Memorial Hospital, Superior Street and Fairbanks Court, Chicago, Illinois 60611.
In this randomized, prospective study, 112 consecutive spinal procedures involving autologous iliac-crest bone graft were performed, from December 29, 1992, to July 1, 1993, by the senior one of us (P. R. M., Jr.), with the assistance of spine surgical fellows, on 108 patients who had a traumatic injury of the spine. There were seventy-three male and thirty-five female patients who had an average age of 38.4 years (range, thirteen to seventy-nine years). Bone graft was obtained through a separate incision from either the anterior or the posterior iliac crest, depending on the spinal operation being performed. Of the seventy-five donor sites of the grafts from the posterior iliac crest, forty-four were drained and thirty-one were not. Of the thirty-seven donor sites of the grafts from the anterior iliac crest, sixteen were drained and twenty-one were not. Management of the patient with or without a drain was assigned randomly by opening an envelope containing a card stating drain or no drain when the patient was wheeled into the operating room. These cards had been prospectively prepared and randomized at the outset of the study.
This study was evaluated by our Institutional Review Board, which determined that the use of a drainage tube caused no greater risk to the patient than procedures considered to be the standard of care at our institution. Thus, the Board decided that the use of an informed-consent form specific for the use of drainage tubes was not warranted.
The grafts from the posterior iliac crest were obtained at the beginning of the procedure through a longitudinal incision made two centimeters lateral to the posterior superior iliac spine, extending from the top of the iliac crest to two centimeters distal to the posterior superior iliac spine. With use of Cobb elevators and electrocautery, the gluteus maximus fascia was dissected subperiosteally from the outer table of the posterior iliac crest. The bone graft was obtained with use of osteotomes and gouges and then wrapped in gauze that had been soaked in blood from the patient. Before closure, a specimen was taken from the wound for culture. Thrombin-soaked Gelfoam (gelatin sponge; Upjohn, Kalamazoo, Michigan) was placed against the cancellous surface of the iliac crest, and the wound was packed with gauze while the bone graft was cut to the appropriate dimensions. The wound was then copiously irrigated with normal saline solution, and all previously placed Gelfoam was removed. Neither bone wax nor any other type of hemostatic material was used.
If the envelope that had been chosen before the start of the procedure contained a drain card, a large Hemovac closed-suction drain was placed deep to the gluteus maximus fascia and carried out through a separate stab portal, from inside out, with use of the drain trocar. The drain was not sutured to the skin. Only one large drain was used in each wound; no drain was placed in the subcutaneous tissue. If the card read no drain, the wound was made and closed in an identical fashion except that no drain was inserted. All wounds were closed in a watertight manner with use of figure-of-eight 0 Vicryl sutures (polyglactin; Ethicon, Somerville, New Jersey) in the deep fascia. The subcutaneous fascia was closed in a simple, interrupted fashion with use of 2-0 Vicryl sutures. The incision was then closed with 3-0 Prolene sutures (polypropylene; Ethicon) in a running subcuticular fashion followed by reinforcement with Steri-Strips (3M, St. Paul, Minnesota) after application of benzoin. In the patients who had insertion of a drain, the closed self-suction apparatus was attached after the deep fascia had been closed.
The grafts from the anterior iliac crest were obtained as a tricortical strut. An incision was placed one centimeter posterior to the anterior superior iliac spine and extended posteriorly for approximately ten centimeters. The interval between the external oblique fascia and the gluteus medius fascia was developed. Subperiosteal dissection was carried out with use of Cobb elevators and electrocautery to expose both the inner and the outer table. An oscillating saw was used to make the anterior vertical cut two centimeters posterior to the anterior superior iliac spine; the cut was carried to a depth of two centimeters. The posterior vertical cut, which was placed at a distance posteriorly along the iliac crest according to the length of tricortical iliac-crest strut graft that was required, was also two centimeters in depth. A one-half-inch (1.27-centimeter) curved osteotome was used to score the outer table horizontally from the anterior to the posterior longitudinal cut. The osteotome was driven across to the inner table. The tricortical graft, wrapped in gauze that had been soaked in blood from the patient, was carefully placed on the back table. Thrombin-soaked Gelfoam was laid on the cancellous surface of the iliac crest. Gauze was packed into the wound while the tricortical graft was fashioned to the dimensions that were appropriate for the anterior spinal strut. The wound was irrigated copiously with normal saline solution, and all visible thrombin-soaked Gelfoam was removed. If a drain was used, it was placed deep to the gluteus medius fascia and carried out laterally through a separate stab incision from inside to outside.
As in the patients who were managed with the posterior graft, one large Hemovac drain was used if the blindly selected envelope contained a drain card but it was not placed if the card read no drain. The external oblique fascia and the gluteus medius fascia were approximated in a watertight fashion with figure-of-eight 0 Vicryl sutures. The subcutaneous fascia was approximated with 2-0 interrupted Vicryl sutures, and the skin was approximated with 3-0 Prolene sutures in a running subcuticular fashion. Benzoin was then applied, followed by Steri-Strips. If a drain was used, it was connected to its self-suction apparatus after the deep fascia had been closed.
Antibiotics were given prophylactically to all patients. In accordance with the orthopaedic protocol at Northwestern Memorial Hospital, one gram of vancomycin and 100 milligrams of gentamicin was given intravenously thirty minutes before the incision. One gram of vancomycin was then given intravenously every twelve hours for forty-eight hours, and 100 milligrams of gentamicin was given intravenously every eight hours for forty-eight hours. The sutures were removed as close to fourteen days postoperatively as possible.
The drain was maintained for at least two days. If the drain was inadvertently pulled out or fell out before the second postoperative day, the patient was disqualified from the study. One patient was disqualified for this reason. If, on the second postoperative day, the drainage had been less than thirty milliliters in the previous twenty-four hours, the drain was removed. If the drainage had been thirty milliliters or more, the drain was left in until the next day, when the drainage was reevaluated. Again, if the drainage had been less than thirty milliliters in the previous twenty-four hours, the drain was removed. All remaining drains were removed on the fifth postoperative day. Therefore, all drains were removed between two and five days postoperatively.
The donor sites of the iliac-crest grafts were meticulously observed by the spine surgical fellows for signs of infection or hematoma at two, six, and twelve weeks; six months; and one year. The monitoring was not blinded. The wounds were considered to be healed when they had been asymptomatic for a full year. If a postoperative infection developed, the patient was taken to the operating room for irrigation and débridement; antibiotic therapy was not started until cultures were performed. No antibiotics were given empirically if a wound appeared erythematous; a culture had to be performed.
Statistical analysis of the results included determination of the significance of mean differences with a two-tailed t test for uneven groups as well as a power analysis. A p value of less than 0.05 was required to assign significance.
The 108 patients were randomly separated into two groups: one group received a drain (sixty procedures), and the other did not (fifty-two procedures). Four patients had two spinal procedures, each of which required separate removal of a bone graft. In two of these patients, both wounds at the site of the iliac crest were drained. In the other two patients, one wound was drained and the second was not. The postoperative course for these eight iliac-crest donor sites was unremarkable.
As mentioned, one patient was removed from the study because the drain fell out postoperatively (in the operating room).
Only one of the empirically performed intraoperative cultures was positive, and it was false-positive. The wound was followed clinically, and it healed without evidence of infection.
Of the 107 patients (fifty-seven managed with a drain) available for study, eleven (10 per cent) had problems with wound-healing: six had been managed with a drain and five had not. A power analysis revealed no significant difference, with respect to the prevalence of problems with wound-healing, between the two treatment groups (value of beta, 0.95). Ten of the patients who had problems with wound-healing had had removal of the bone for the graft from the posterior iliac crest and one, from the anterior iliac crest. Of the ten patients who had the infection at the posterior donor site, four (two who had been managed with a drain and two who had not) had a complete neurological deficit; the other six patients had no neurological damage. The patient who had the infection at the anterior donor site (without a drain) had no neurological damage.
Three patients (two who had been managed with a drain and one who had not) had increasing pain as well as cellulitis and seropurulent drainage at the donor site within two months after the operation. They were taken back to the operating room for evacuation of the hematoma and débridement. All three of these patients had the infection at the posterior donor site. Cultures of specimens from these three wounds were positive for Staphylococcus species (coagulase-negative species in the undrained wound and Staphylococcus aureus in the two drained wounds). All three infections were treated postoperatively with intravenous administration of antibiotics followed by oral administration.
The remaining eight patients who had problems with wound-healing had mild-to-moderate dehiscence of the wound. Cultures of specimens from the four wounds that had not been drained were positive for Staphylococcus species (coagulase-negative species in two and Staphylococcus aureus in two). Of the four wounds that had been drained, one was found to be infected with Staphylococcus and another was found to be infected with Klebsiella on culture. Cultures of specimens from the other two wounds that had been drained were negative. Seven of these eight patients were managed with local wound care and oral administration of antibiotics. The remaining patient, for whom the culture was negative, was managed with local wound care only.
The time to presentation of the problems with wound-healing averaged nineteen days (range, six to thirty-four days) postoperatively for the patients who were managed with a drain and seventeen days (range, three to fifty-eight days) for those who were not. The time to wound-healing averaged forty-four days (range, twenty-two to eighty-two days) from the date of the operation for the former group and thirty-seven days (range, seventeen to sixty-two days) for the latter group.
The Injury Severity Score2 in this series averaged 21 points (range, 9 to 54 points). With the numbers available, we could not detect a significant difference (p > 0.05) in this score between the patients who had problems with wound-healing and those who did not, in either treatment group. We also found no significant difference when we compared the two treatment groups. In the group that was managed with a drain, the score averaged 28 ± 20.1 points (range, 9 to 50 points) for the patients who had problems with wound-healing and 22 ± 10.7 points (range, 9 to 54 points) for those who did not. In the group that was managed without a drain, the score averaged 15 ± 8.5 points (range, 9 to 29 points) for the patients who had problems with wound-healing and 18 ± 8.1 points (range, 9 to 38 points) for those who did not.
Nutritional parameters were examined to determine if the patients who had problems with wound-healing were more depleted nutritionally than those in whom the wound healed without problems. With the numbers available, we could not detect a significant difference (p > 0.05), with respect to nutritional parameters, between the patients who were managed with a drain and those who were not or between the patients who had problems with wound-healing and those who did not, in either treatment group. In the group that was managed with a drain, the patients who had problems with wound-healing had an average total lymphocyte count of 1387 ± 6.7 per cubic millimeter (1.4 ± 0.0067 x 109 per liter) (range, 384 to 2256 per cubic millimeter [0.4 to 2.3 x 109 per liter]), an average serum albumin level of 34 ± 9 grams per liter (range, twenty-eight to forty-two grams per liter), and an average total protein level of 66 ± 6 grams per liter (range, sixty to seventy grams per liter). The patients in that group who had no problems with wound-healing had an average total lymphocyte count of 1243 ± 100.8 per cubic millimeter (1.2 ± 0.1 x 109 per liter) (range, 309 to 3776 per cubic millimeter [0.3 to 3.8 x 109 per liter]), an average serum albumin level of 39 ± 6 grams per liter (range, twenty-three to forty-six grams per liter), and an average total protein level of 65 ± 8 grams per liter (range, forty to eighty-three grams per liter). In the group that was managed without a drain, the patients who had problems with wound-healing had an average total lymphocyte count of 996 ± 422 per cubic millimeter (1.0 ± 0.4 x 109 per liter) (range, 640 to 1552 per cubic millimeter [0.6 to 1.6 x 109 per liter]), an average serum albumin level of 40 ± 6 grams per liter (range, thirty-three to forty-eight grams per liter), and an average total protein level of 68 ± 8 grams per liter (range, fifty-nine to seventy-eight grams per liter). The patients who had no problems with wound-healing in that group had an average total lymphocyte count of 1314 ± 1117 per cubic millimeter (1.3 ± 1.1 x 109 per liter) (range, 120 to 4290 per cubic millimeter [0.1 to 4.3 x 109 per liter]), an average serum albumin level of 39 ± 6 grams per liter (range, twenty-four to forty-nine grams per liter), and an average total protein level of 67 ± 8 grams per liter (range, forty-four to eighty grams per liter).
This study is not the first to suggest that the use of drains may be unnecessary after uncomplicated orthopaedic procedures3,14. Retrospective studies4-7,11 supporting either one position or the other clouded the issue until randomized, prospective studies by Beer et al. and Ritter et al. suggested that drains offered little advantage with respect to the outcome of total joint replacement.
Beer et al. evaluated thirty-eight patients who had had a bilateral total knee arthroplasty and twelve patients who had had a bilateral total hip arthroplasty; in both groups, both sides had been operated on during the same operative session and a drain had been used on only the right side. The authors found no difference, between the groups, with regard to incisional drainage, swelling, or motion of the joint. They concluded that routine use of a drain was unnecessary after total joint arthroplasty.
To our knowledge, the largest prospective, randomized study of the postoperative use of drains was performed by Ritter et al. In their study of 415 total joint replacements, 215 wounds were drained with a closed suction system and 200 were not drained. The authors found no significant differences between the two groups with regard to excessive postoperative drainage requiring cessation of range-of-motion exercises, the amount of transfused blood, or the preoperative and postoperative hemoglobin levels. In addition, they noted an expenditure of $21,500 (215 drainage units at $100 per unit) to provide closed suction drainage to the patients. Their preliminary conclusion was that closed suction drainage offers little advantage to patients who have a total joint replacement.
The findings of the present prospective, randomized investigation of wound-healing after removal of bone from the iliac crest for use as a graft concurred with the results of the previous prospective studies3,12: we found that closed suction drainage played no role in wound-healing. There was no significant difference between the number of drained wounds (six) and the number of undrained wounds (five) that had problems with healing (value of beta, 0.95). Problems with wound-healing are frequent at the donor sites of bone grafts, as illustrated by the retrospective review by Younger and Chapman. Of 243 donor sites of autologous bone grafts, six (2 per cent) were complicated by infection; two (1 per cent), by prolonged wound drainage; and eight (3 per cent), by a large hematoma. Interestingly, although Younger and Chapman were not looking specifically at the role of drains, they reported that the rate of complications was slightly higher among patients who had a suction drain than among those who did not.
It is difficult to challenge an idea that seems logical and is steeped in tradition. One such idea is the empirical use of closed suction drainage in order to assist wound-healing by alleviating postoperative hematoma. The assumption that hematomas that come into contact with bacteria play a role in wound infection is probably valid14. However, it is the association of bacteria with the hematoma, not the hematoma itself, that causes problems with wound-healing. On the basis of the results of the present study and those of other recently reported prospective investigations3,12, it is necessary to question whether drains indeed alleviate wound hematomas.
In conclusion, we believe that closed suction drainage has no effect on wound-healing following the removal of bone from the iliac crest for use as a graft in patients who have a traumatic injury of the spine.
Adams, F.: The Works of Hippocrates. Baltimore, Williams and Wilkins, 1939.
Baker, S. P.; O'Neill, B.; Haddon, W., Jr.; and Long, W. B.: The Injury Severity Score: a method for describing patients with multiple injuries and evaluating emergency care. J. Trauma,14: 187-196, 1974.14187
1974
[PubMed]
Beer, K. J.; Lombardi, A. V., Jr.; Mallory, T. H.; and Vaughn, B. K.: The efficacy of suction drains after routine total joint arthroplasty. J. Bone and Joint Surg.,73-A: 584-587, April 1991.73-A584
1991
Browett, J. P.; Gibbs, A. N.; Copeland, S. A.; and Deliss, L. J.: The use of suction drainage in the operation of meniscectomy. J. Bone and Joint Surg.,60-B(4): 516-519, 1978.60-B(4)516
1978
Bryan, R. S.; Dickson, J. H.; and Taylor, W. F.: Recovery of the knee following meniscectomy. An evaluation of suction drainage and cast immobilization. J. Bone and Joint Surg.,51-A: 973-978, July 1969.51-A973
1969
Cobb, J. P.: Why use drains?. J. Bone and Joint Surg.,72-B(6): 993-995, 1990.72-B(6)993
1990
Hadden, W. A., and McFarlane, A. G.: A comparative study of closed-wound suction drainage vs. no drainage in total hip arthroplasty. J. Arthroplasty,5 (Supplement): 21-S24, 1990.5 (Supplement)21
1990
Healy, D. A.; Keyser, J., III; Holcomb, G. W., III; Dean, R. H.; and Smith, B. M.: Prophylactic closed suction drainage of femoral wounds in patients undergoing vascular reconstruction. J. Vasc. Surg.,10: 166-168, 1989.10166
1989
[PubMed]
Jepsen, O. B.; Larsen, S. O.; and Thompson, V. F.: Post-operative wound sepsis in general surgery. II. An assessment of factors influencing the frequency of wound sepsis. Acta Chir. Scandinavica, Supplementum 396, pp. 80-90, 1969.
Lidwell, O. M.: Sepsis in surgical wounds. Multiple regression analysis applied to records of post-operative hospital sepsis. J. Hyg.,59: 259-270, 1961.59259
1961
Reilly, T. J.; Gradisar, I. A., Jr.; Pakan, W.; and Reilly, M.: The use of postoperative suction drainage in total knee arthroplasty. Clin. Orthop.,208: 238-242, 1986.208238
1986
[PubMed]
Ritter, M. A.; Keating, E. M.; and Faris, P. M.: Closed wound drainage in total hip or total knee replacement. A prospective, randomized study. J. Bone and Joint Surg.,76-A: 35-38, Jan. 1994.76-A35
1994
Stevens, D. B.: Postoperative orthopaedic infections. A study of etiological mechanisms. J. Bone and Joint Surg.,46-A: 96-102, Jan. 1964.46-A96
1964
Waugh, T. R., and Stinchfield, F. E.: Suction drainage of orthopaedic wounds. J. Bone and Joint Surg.,43-A: 939-946, Oct. 1961.43-A939
1961
Younger, E. M., and Chapman, M. W.: Morbidity at bone graft donor sites. J. Orthop. Trauma,3: 192-195, 1989.3192
1989
[PubMed]