Abstract
Study Design:
A retrospective case-control study.
Objectives:
The usefulness of a drain in spinal surgery has always been controversial. The purposes of this study were to determine the incidence of hematoma-related complications after posterior lumbar interbody fusion (PLIF) without a drain and to evaluate its usefulness.
Methods:
We included 347 consecutive patients with degenerative lumbar disease who underwent single- or double-level PLIF. The participants were divided into 2 groups by the use of a drain or not; drain group and no-drain group.
Results:
In 165 cases of PLIF without drain, there was neither a newly developed neurological deficit due to hematoma nor reoperation for hematoma evacuation. In the no-drain group, there were 5 (3.0%) patients who suffered from surgical site infection (SSI), all superficial, and 17 (10.3%) patients who complained of postoperative transient recurred leg pain, all treated conservatively. Days from surgery to ambulation and length of hospital stay (LOS) of the no-drain group were faster than those of the drain group (P < 0.001). In a multiple regression analysis, a drain insertion was found to have a significant effect on the delayed ambulation and increased LOS. No significant differences existed between the 2 groups in additional surgery for hematoma evacuation, or SSI.
Conclusions:
No hematoma-related neurological deficits or reoperations caused by epidural hematoma and SSI were observed in the no-drain group. The no-drain group did not show significantly more frequent postoperative complications than the drain use group, hence the routine insertion of a drain following PLIF should be reconsidered carefully.
Keywords
Introduction
Since 1971, when the neurosurgeon Jackson-Pratt first published a paper reporting the preventive effect of the drain for a subdural hematoma in a damaged brain, several types of drain, including closed-suction and simple conduit, have been believed to inhibit the accumulation of blood and fluid after spinal surgery also. 1 Based on this traditional belief in the need for a drain and the probability of serious complications caused by epidural hematomas, such as neural compression, neurological deficit, or reoperation for hematoma evacuation, a drain has been routinely used in spinal surgery, even in the minimally invasive surgery (MIS). 2 Especially, since the conventional open posterior lumbar interbody fusion (PLIF) with instrumentation includes more extensive procedures than does simple decompression, a drain insertion has been considered to be an essential and routine step at the end of the surgery. 3 This spine center had also performed routine insertion of drain for several decades. Then, our authors had participated in a discussion on the necessity of drains after spinal surgery at the international spine conference and we found that there is no clear scientific evidence for the routine insertion of the drain in aseptic spine surgery through a literature review for preparing conference discussion. These events led to the start of no-drain spine surgery in October 2016.
Contrary to the long-standing belief in the need for the drain, the risk of surgical site infection (SSI) or hematoma was not influenced by the use of a drain, and a systematic review concluded that spine surgeons should not routinely rely on the closed suction drain without a higher level of evidence.4,5 These reports addressed the efficacy of a drain after decompression alone, and only a few studies dealt with extensive lumbar surgery.4,6-11 To our best knowledge, there have been no studies dealing with the usefulness of a drain that included conventional PLIF surgery alone, even though it is the most commonly used fusion technique. Among patient demographics, comorbidities and obesity are known to be significant risk factors for increasing perioperative blood loss.12-18 Therefore, in this study, various factors such as obesity, use of medical drugs, and comorbidities were also evaluated.
The purposes of this study were (1) to determine the incidence of hematoma-related complications after PLIF without a drain; and (2) to evaluate the clinical necessity and usefulness of a routine insertion of a subfascial drain following PLIF based on the comparative analysis.
Methods
Study Design, Setting, and Data Collection
This study was a retrospective case-control study. We retrospectively reviewed a prospective database on 347 consecutive patients from January 2015 to June 2018. There were 205 female and 142 male participants, with a mean age of 63.8 ± 11.6 years (range: 21 to 86). The mean follow-up period was 1.4 ± 1.3 years. Our sole criterion for inclusion was a degenerative lumbar disease that had undergone single- or double-level PLIF, completed the minimum 6-month follow-up. Only patients undergoing elective PLIF were included in the study. All surgeries had been performed by a single senior spine surgeon. The exclusion criteria were patients with infections, trauma, the presence of coagulopathies and bleeding disorders, or tumors. Patients with abnormal international normalized ratios or activated partial thromboplastin time or platelet count or bleeding time were excluded. The participants were divided into 2 groups by the use of a drain or not. Until October 2016, we routinely placed a drain (drain group, n = 182), and thereafter we did not use any drains in any degenerative spinal surgeries, including the PLIF cases of the subjects in this study (no-drain group, n = 165). Since the beginning of the no-drain spine surgery, we did not insert the drain in all of the spine surgeries. In the case of significant intraoperative bleeding, we used a topical gelatin-thrombin hemostatic matrix (a total of 5 mL of Floseal® (Baxter Healthcare, Deerfield, IL)) during surgery to control the bleeding.
The surgical techniques and perioperative patient care were as follows. We used the conventional open technique for PLIF, including periosteal muscle stripping, subtotal laminectomy using a high-speed burr and osteotome, bilateral interbody fusion using a titanium-cage filled with local bone graft, and bilateral pedicle screw instrumentation. Epidural venous bleeding was ceased by bipolar electrocoagulation and gelatin compressed sponge coverage. Oozing bony surface was treated with bone wax. In the drain group, we inserted a closed-suction drain (Jackson-Pratt drain, silicone type) below the deep fascia, over the exposed dural sac. The drainage line was linked to the 400-mL capacity rubber closed-suction bag with mild suction power (one-third negative pressure). We checked the amount of drainage every 24 hours after the end of surgery and removed the drain when the amount drained per day was less than 50 mL (postoperative days (POD) 2 or 3 in most cases). All patients were closed in exactly the same fashion, and meticulous bleeding control was done. We did not use additional supplements for perioperative bleeding control such as tranexamic acid and advanced electrocautery devices. Any anti-inflammatory or anticoagulant drugs were stopped in all patients before hospitalization for surgery (at least 7 days before surgery). Anticoagulants used for therapeutic purposes such as brain infarction treatment were restarted as soon as possible after the acute postoperative phase passed and the vital sign was stabilized (usually 1 or 2 days postoperatively). All patients received intravenously administered antibiotic prophylaxis just before and for 72 hours after surgery and prophylaxis for venous thrombosis with elastic stockings or pneumatic compressive devices. Blood transfusion was administered if the hemoglobin level was < 8 g/dL or for symptomatic patients with hemoglobin between 8 and 10 g/dL. Careful inspection of the surgical wounds was done during the acute postoperative hospitalization stay. We recommended discharge for all patients who showed good ambulation with or without the support and had tolerable pain (visual analog scale (VAS) less than 4) in this study. After the discharge, all patients were seen in an out-patient clinic for regularly scheduled follow-up at 2, 6, and 12 weeks after surgery, and the inspection of surgical wounds and blood tests were performed to evaluate hematomas and SSI. If a suspected infection was present, the wound was cultured and oral antibiotic treatment was begun.
Perioperative Factors
We recorded the number of fused segments, operating time, estimated blood loss (EBL), the rate of gelatin-thrombin hemostatic matrix (Floseal®) use, and postoperative blood transfusion requirements. Intraoperative EBL was measured by “Blood loss estimation protocol” (Table 1) made in our spine center based on previous studies about surgical blood-loss measurements. 19 Postoperative drainage was mainly measured by the total amount of closed-suction drainage. Intraoperative EBL and postoperative drainage amounts were divided by the number of fused segments.
Intraoperative Blood Loss Was Measured Using a Blood-Loss Estimation Protocol Made in Our Spine Center.
hr indicates hour.
Clinical and Laboratory Parameters
We observed the patients’ medical status related to bleeding, comorbidities, days from surgery to ambulation, length of hospital stay (LOS), any newly developed neurological deficit, reoperation rate, and SSI. We also observed postoperative recurred leg pain after the lucid interval. We observed the hemoglobin level preoperatively, immediately after the operation, and on POD 1, 3, 5, and 7. Final follow-up VAS for lower back pain and leg pain were evaluated.
Statistical Analysis
We compared the aforementioned variables between the 2 groups. We used SPSS 21.0 (SPSS, Inc., Richmond, CA, USA) for the statistical analyses, χ2 tests, and Student’s t-test including P-value to compare variables between the 2 groups. The normality of data distribution was assessed using a Shapiro-Wilk normality test. A 2-tailed P-value < 0.05 was considered significant for all statistical tests. Using a simultaneous multiple linear regression analysis, we assessed various potential confounding factors after adjustment for demographic data and perioperative variables. We identified candidate variables for each regression model via univariate screening with P < 0.20. Ahead of the logistic regression analysis, auto-correlations and multiple collinearity were evaluated by the Durbin-Watson test and the variance influence factor (VIF) analysis, respectively. For the goodness-of-fit test of the regression model, we used the Kolmogorov-Smirnov’s test and Breusch-Pagan’s test, respectively. Our hospital’s institutional review board approved this study protocol (no. OOO OOOO-OO-OOO).
Results
Surgical Results of No-Drain PLIF
A total of 165 no-drain PLIF surgeries were performed. The patients’ medical status was hypertension in 79 (47.9%), anticoagulant drug use in 34 (all patients stopped drugs at least 7 days before surgery), prior heart disease in 18, and prior brain infarct in 8. The single-level PLIFs were 89 cases (53.9%) and double-level PLIFs were 76 cases (46.1%), respectively. An intraoperative EBL was 573.9 ± 389.7 mL (351 to 1222 mL) (385.7 mL per fused segment). Days from surgery to ambulation was 2.2 ± 1.2 (1 to 6) days, and LOS was 9.2 ± 3.9 (5 to 18) days. There were 5 (3.0%) patients who suffered from SSI, all superficial and all treated conservatively without additional surgery. There was one case of a newly developed neurological deficit after surgery; a 64-year-old female patient who was treated with no-drain PLIF for L4-5-S1 levels. She complained of decreased motor power immediately after surgery; a CT scan showed the violation of the L5 pedicle screw. After the removal of the violated screw, the neurological deficit was dramatically recovered. This was the sole reoperation case. There was no newly developed postoperative neurological deficit caused by hematoma, nor reoperation for hematoma evacuation. Postoperative transient recurred leg pain after the lucid interval was observed in 17 (10.3%) patients, and the mean occurrence time of the recurred leg pain was 3.7 ± 1.2 (2 to 7) days after surgery, with a mean duration time of 2.4 ± 1.1 (1 to 5) days. All these symptoms were transient and relieved by conservative treatment, including medications and intravenous injections.
Comparative Analysis: No-Drain PLIF Versus PLIF Using Drain
The no-drain PLIF group included 165 patients, and the drain group included 182 patients. There was no significant difference in baseline demographics, diagnosis, medical status related to the bleeding, mean number of fused segments, mean operating time, the rate of gelatin-thrombin hemostatic matrix use, or intraoperative EBL (per level) between the 2 groups (Tables 2 and 3). The mean amounts of postoperative drainage (per level) of the drain group was 272.6 mL. The rate of postoperative blood transfusion requirements was significantly lower for the no-drain group (2.4%) than for the drain group (4.4%) (P < 0.001). Days from surgery to ambulation (2.2 days) and LOS (9.2 days) of the no-drain group was less than for the drain group (3.2 days and 11.7 days, respectively; P < 0.001; Table 4). In the no-drain group, the rate of the newly developed neurological deficit because of presumed hematoma, reoperation, or postoperative recurred leg pain after the lucid interval was less than for the drain group; however, there was no statistical significance. There was no case of additional surgery for hematoma evacuation in the no-drain group, but 1.6% (3 of 182) of the patients underwent removal of hematoma in the drain group (P = 0.191). For the SSI, there was no significant difference between the 2 groups (P = 0.392; Table 4). No significant differences existed between the drain and no-drain groups in the variables of perioperative hemoglobin levels at all measurement periods (Figure. 1) and final follow-up VAS for lower back pain and leg pain.
Comparison of Baseline Demographics Between the Groups With No-Drain PLIF and PLIF Using a Drain.
Mean ± standard deviation with a range.
Student t-test and Pearson χ2 test were used.
The normality of data distribution was assessed using a Shapiro-Wilk normality test.
CI indicates confidence interval.
Comparison of Perioperative Factors Between No-Drain PLIF and PLIF Using Drain Groups.
Mean ± standard deviation with a range.
Student t-test and Pearson χ2 test were used.
The normality of data distribution was assessed using a Shapiro-Wilk normality test.
* Significance, P < 0.01.
CI indicates confidence interval; EBL, estimated blood loss; EBL / level, estimated blood loss per fused segment.
Comparison of Clinical Parameters Between the No-Drain PLIF and PLIF Using Drain groups.
Mean ± standard deviation with a range.
Student t-test and Pearson χ2 test were used.
The normality of data distribution was assessed using a Shapiro-Wilk normality test.
* Significance, P < 0.01.
CI indicates confidence interval; VAS, visual analog scale; POD, postoperative day.
Comparison of the perioperative changes of hemoglobin levels between the no-drain posterior lumbar interbody fusion (PLIF) and PLIF using drain groups.
Multiple Regression Analyses
Based on the results of the comparative analysis identified 2 clinical variables with significant differences between the 2 groups (days from surgery to ambulation and LOS), we performed multiple linear regression analyses for these 2 variables to assess potential confounding factors. First, multiple regression analysis for the variable of days from surgery to ambulation was performed after adjustment for age, sex, diagnosis, medical status, revision surgery, number of fused segments, operating time, and transfusion. A drain insertion was found to have a significant effect on the increased days from surgery to ambulation (P < 0.001). The Kolmogorov-Smirnov normality test (P = 0.200 > 0.10) and Breusch-Pagan heteroscedasticity test (P = 0.785 > 0.05) also satisfied the goodness-of-fit test, and our regression analysis fitted the appropriate model (Table 5). The second multiple regression analysis was performed for the variable of LOS after adjustment. A drain insertion was found to have the most significant effect on the increased LOS, (P < 0.001) (Table 6). The Kolmogorov-Smirnov normality test (P = 0.310 > 0.10) and Breusch-Pagan heteroscedasticity test (P = 0.876 > 0.05) also satisfied the goodness-of-fit test, and our regression analysis fitted the appropriate model.
Associations of Days From Surgery to Ambulation With Drain Insertion by Multiple Linear Regression Analysis.
Durbin-Watson’s d = 1.924 (du = 1.822), Breusch-Pagan’s χ2 = 0.894 (P = 0.785), Kolmogorov-Smirnov’s z=2.011 (P = 0.200).
SE indicates standard errors; VIF, variance inflation factor.
Associations of the Length of Hospital Stay With Drain Insertion by Multiple Linear Regression Analysis.
Durbin-Watson’s d = 1.892 (du = 1.822), Breusch-Pagan’s χ2 = 0.745 (P = 0.876), Kolmogorov-Smirnov’s z=2.119 (P = 0.310).
SE indicates standard errors; VIF, variance inflation factor.
Discussion
Contrary to the unconscious belief for the drain, the risk of SSI or hematoma-related complications in single-level lumbar decompression surgery were not influenced by the use of a drain. 4 A systematic review concluded that spinal surgeons should not routinely rely on the closed suction drain without a higher level of evidence. 5 Previous reports addressed the efficacy of a drain after discectomy or laminectomy, which is decompression alone.4,7-9,11 Only a few studies dealt with fusion surgery.6,10 In this study, more than 160 consecutive cases of PLIF without drain showed no increase in hematoma-related complications. Our study supports prior studies that questioned the necessity for the routine use of a drain.4,6,9-11,20-22 Notable findings of this study were several clinical advantages of the no-drain PLIF, such as reduced days from surgery to ambulation and the LOS. The majority of prior studies addressing the efficacy of a drain after spinal surgery reported outcomes about SSI,4,6,9,10 incidence of postoperative symptomatic epidural hematoma,23,24 new neurological deficit,6,9 or recurrent pain. 8 Only a few studies reported clinical data related to ambulation outcome and LOS after MIS without a drain. 20 An early start of ambulation is clinically crucial for preventing post-surgical complications that can result from long-standing bed rest. Delayed mobilization after spinal surgery has been linked to an increased risk of postoperative pneumonia and respiratory decompensations.25-27 The reason why the days from surgery to ambulation of the drain group was delayed than that of the no-drain group is thought to be due to discomfort or inconvenient feeling from drains or lines connected to them. 20 A reduced hospital stay (almost 2.5 days faster) is the notable advantage of no-drain surgery in our case series. Prolonged hospital stays have numerous drawbacks: an increase in health-care costs and delayed return to daily living. 28 The mean LOS of this study (9.2 days in the no-drain group and 11.7 days in the drain group) was relatively longer than those of the previous studies. Multiple previous studies defined prolonged or extended LOS as the increased LOS by 7∼9 days.29-31 This is because of South Korea’s medical insurance characteristics (national health insurance system, extensive medical support, and health care services). However, the timing of the discharge was determined using a unified standard, and there was a significant difference between the 2 groups. Evaluation of the postoperative recurred leg pain was one of the differential findings of our study. This symptom could be caused by several conditions, such as neural compression by epidural hematoma, perioperative nerve-root swelling, or unexplained nerve-root irritation.32,33 When a patient complains about persistent recurred leg pain without definite technical errors, neural compression by hematoma should be presumed in general. This study showed no significant difference in recurred leg pain between the 2 groups.
In the setting of spinal surgery, one of the biggest concerns is the postoperative epidural hematoma.7,32-37 Hematoma could be the feasible medium for bacterial colonization, and the space-occupying lesion compressing the neural tissue might subsequently induce abrupt re-aggravation of leg pain, neurological deficit, or more seriously, cauda equina syndrome.7,33,35,36 In general, a surgical drain has been widely used with the theoretical expectation of preventing these complications.1,7,34 Actually, one prospective, randomized clinical study showed a decrease in the incidence and size of a hematoma in the drain group on POD 1, as detected by MRI. 8 However, the presence of an epidural hematoma after the spinal surgery itself directly showed no obvious effect on the postoperative clinical results.6,10,11,21 Several randomized studies and systematic reviews have suggested that the use of drains is not effective in preventing hematoma-related complications. Furthermore, spinal surgery using a drain might need additional time and cost, because of paraspinal muscle injury and patients’ discomfort due to the drain insertion. Even though no medical report has proved a significant effect of the drain, many spinal surgeons routinely use various types of drain purely based on the traditional belief, even in the MIS.20,22 We have occasionally encountered patients who developed postoperative epidural hematoma with cauda equina syndrome or neurologic deficits, even when the drain was used, and who needed additional surgery for hematoma evacuation in the end. These facts suggest that drain insertion could not guarantee complete prevention of hematoma-related complications. Even in a case using multiple or larger-diameter drains, complications from postoperative epidural hematoma were noticed also. 34 The authors hypothesized that the continuous negative pressure of a closed suction drain could be what induces constant oozing of tissue fluid and bleeding from the damaged surgical wound while suppressing hemostasis. This hypothesis could be indirectly sustained by the higher rate of postoperative blood transfusion in the drain group from our results. Of course, it is difficult to conclude that the frequency of transfusions was low simply because the drain was not used, or whether a drain did lead to an increased need for transfusions. However, the statistical difference of the transfusion requirement between the 2 groups was clear and obviously significant.
SSI is the other critical issue for the use or non-use of a drain in spinal surgery. In general, for the treatment of SSI after orthopedic surgery, a drain insertion is a standard way to evacuate the infected material. 38 However, a lack of prophylactic efficacy of a drain in the aseptic clean surgery has been scientifically established at present in the orthopedic field.6,10,11,21,39,40 On the other hand, because of the possibility of communication between external and internal spaces through the drainage line or conduit tube, a drain itself could be the source of contamination and an inlet of infection. In fact, several kinds of literature have suggested a higher rate of SSI in patients undergoing drainage after surgery.41,42
The decision-making around whether or not to place a drain is necessary to decide carefully, which is a critical question that takes place during the surgery itself. In the present study, the no-drain group showed an early postoperative recovery (reduced time of postoperative ambulation and LOS) but did not significantly increase the hematoma related complication rate. Of course, in the face of a complex decision-making tree, more detailed and unmentioned various factors including unrecorded anticoagulant use, surgeon’s impression of intraoperative bleeding, hemostasis, and increased morbidity (blood transfusion requirements) with the use of a drain should be considered in terms of risk-benefit scale. However, it is clear that a reconsideration of routine insertion of drains in vague expectations is essentially necessary, and a no-drain PLIF surgery is thought to be tried thoroughly worthwhile based on our data.
Our study is the first retrospective and consecutive case-control series on the incidence of hematoma-related complications and surgical results following PLIF, a single type of spinal surgery by a single surgeon. This could minimize the heterogeneity from patient demographics and surgical factors.
Our study had several limitations. First, it was a retrospective comparative study without randomization. The second was the relatively small size of the study population. Since SSI and additional surgery for hematoma evacuation have extremely lower incidence, multi-center studies with larger populations would be needed to solve these critically determinant issues about the efficacy of a drain. Our data may contribute to a common consensus in the future regarding drain placement after lumbar fusion surgery. Third, there is a possibility that potential compounding factors related to perioperative bleeding such as the surgeon’s impression of intraoperative bleeding and hemostasis technique were overlooked. Fourth, we did not use any imaging modality to assess epidural hematoma in our cases, mostly because of the presence of metallic implants that may interfere with the accuracy of measurements. The fifth was the relatively short follow-up period. As hematoma can theoretically contribute to epidural fibrosis and scaring which could lead to recurrent back and leg pain in the long-term, a sufficient period of follow-up evaluation to look for neurological symptoms due to epidural fibrosis would be ideal.
It was difficult to draw definite conclusions with a retrospective and relatively small sample size. However, routine insertion of a drain following PLIF is not always required, and early postoperative results (in-hospital results) of no-drain PLIF surgery were comparable to those of the drain group.
Conclusions
In our consecutive case series without the use of a drain, there was neither any hematoma-related neurological deficit nor any reoperation. The no-drain group did not show a significantly higher incidence of postoperative complications, and a drain group has not been shown to provide a benefit or better surgical results. When adequate hemostasis is achieved during uncomplicated single- or double-level PLIF for degenerative lumbar disorders, the routine insertion of a drain should be reconsidered carefully.
Footnotes
Authors’ Note
Acknowledgments
The authors would like to thank Su-Jin Yoon, RN, and Sukyung Kang, TD, for assisting with the preparation and revision of the paper.
Declaration of Conflicting Interests
This manuscript does not provide information regarding medical devices or drugs. The first author (H.D.J) has received research support funding from the Soonchunhyang University Research Fund. There are no relevant financial activities outside the submitted work to declare. The authors report no conflict of interest concerning the materials or methods used in this study or the findings specified in this paper.
Funding
The author(s) received no financial support for the research, authorship, and/or publication of this article.
Research Ethics and Patient Consent
Soonchunhyang University Hospital Institutional Review Board approved this study protocol (Soonchunhyang University Hospital, SCHUH 2018-12-002). Informed consent was exempted from the institutional review board due to the retrospective design.