Abstract

Objective:

This single-center retrospective analysis was designed to evaluate the outcome of closed-suction wound drainage following posterior spinal fusion with internal instrumentation for mild to moderate adolescent idiopathic scoliosis (AIS).

Methods:

Eighty-six AIS patients undergoing posterior spinal fusion were divided into two cohorts: submuscular closed-suction wound drainage (n = 35) and simple compressed dressing clothes without wound drainage (n = 51). These two cohorts were thoroughly compared in terms of demographic distribution and perioperative blood loss, including hemoglobin and hematocrit levels and blood transfusion volumes. Additionally, the incidence of wound-related problems (pyrexia and wound complications), duration of hospital stay, and lumbar function evaluation (lumbar mobility and SRS-22 questionnaire scores) were annually assessed during at least 5-year follow-up.

Results:

The drainage group had significantly lower hemoglobin (93.73 g/L vs. 99.95 g/L, P = 0.01) and hematocrit levels (27.75% vs. 29.94%, P < 0.01) on the third postoperative day, as well as a significantly higher postoperative blood transfusion volume (40.0 mL vs. 23.5 mL, P = 0.011) compared to the non-drainage group. Furthermore, the duration of hospital stay was significantly longer in the drainage group than in the non-drainage group (10.9 d vs. 8.0 d, P < 0.01). In contrast, the two groups were statistically similar regarding duration of fever (0.9 d vs. 1.2 d, P = 0.268), incidence of wound problems, latest lumbar mobility (42.79° vs. 44.97°, P = 0.586), and scores of function/activity domain (16.74 vs. 16.08, P = 0.285) and pain domain (22.18 vs. 21.48, P = 0.374) in the SRS-22 questionnaire.

Conclusions:

Routine closed-suction drainage significantly increased blood loss and hospital stay without obviously improving wound healing or functional outcomes. Utilizing simple compressed dressings without drainage was a clinically superior and resource-efficient alternative for posterior AIS fusion, particularly in uncomplicated primary surgeries for mild to moderate deformities.

1 Introduction

Closed-suction wound drainage has been extensively applied in spine surgeries. Theoretically, wound drainage could effectively facilitate wound healing by providing a ready egress away from the surgical site, minimize the potential compression of exposed nerve structures, and reduce the risk of surgical site infection (1, 2). However, the aforementioned advantages have also been questioned by a number of researchers (3, 4). Some researchers argued that closed-suction drainage had limited benefits in minimizing hematoma as well as in preventing further compression of the spinal cord (4). In addition, the potential disadvantages of drainage, such as increased perioperative blood loss, canal-related contamination, prolonged hospital stay and increased nursing workload, have been frequently reported (5, 6). Hitherto, there has been no firmly established consensus on indications for drainage placement, drainage type and drainage depth and duration. Many spine surgeons even decided drainage placement and patterns based on personal preferences arbitrarily (5, 7).

Adolescent idiopathic scoliosis (AIS) is a complex 3-dimensional spine deformity involving coronal, sagittal and axial malalignment (8). Compared with regular spine surgery, the posterior spinal fusion with instrumentation in AIS is characterized by larger surgical sites, more blood loss and more complex postoperative complications. Moreover, the extensive dissection of paravertebral muscle and decortication of vertebral laminae make hemostasis more challenging (9). Particularly, effective postoperative drainage is capable of minimizing the hematoma formation and organization that could lead to reduced lumbar mobility (10). Nevertheless, the risk of drainage-related infection and potential blood loss cannot be neglected (11). To date, there are few high-quality studies discussing the influence of wound drainage on back pain and lumbar mobility in AIS patients. In this retrospective clinical study, we carried out a thorough comparison of perioperative drainage-related indices and long-term lumbar mobility.

2 Materials and methods

This retrospective clinical study was approved by the Institutional Review Board of the hospital where a number of sophisticated orthopedic surgeons were employed. Two groups of AIS patients undergoing posterior fusion and internal fixation were retrospectively and consecutively enrolled. Group A (drainage group) included 35 patients who received postoperative submuscular drainage accompanied with thick compressed dressings, and Group B (non-drainage group) consisted of 51 patients who received compressed dressings purely. All surgical procedures were performed by a single team of skilled orthopedic surgeons from June 2015 to September 2020.

2.1 Patients

The inclusion criteria for eligible patients were as follows: (1) aged 10–18 years and diagnosed with AIS according to the Lenke classification system (12), (2) presenting a major thoracic or lumbar/thoracolumbar curve less than 80°, (3) treated with pedicle screw and rod instrumentation via posterior approach, (4) followed up for at least 5 years at intervals of 6 months. Within the drainage group, there were 4 females and 31 males with an average age of 14.8 years and a mean follow-up period of 62.0 months (from 58 to 67 months). As for the non-drainage group, there were 43 females and 8 males, with an average age of 15.9 years and a mean follow-up period of 65.4 months (from 60 to 71 months).

2.2 Strategy of fusion extent

All patients were treated with pedicle screws and rod instrumentations through posterior approach. The upper end vertebra (UEV) was selected as upper instrumented vertebra (UIV). The lower instrumented vertebra (LIV) was the last substantially touching vertebra (LSTV) as was described by Zhu's group (13).

2.3 Surgical procedures

Exposure: We placed the patient in a prone position and made a midline incision on the back. The epidermis, subcutaneous tissue and paravertebral muscles were cut to expose the vertebrae at the planned fusion levels.

Pedicle screws placement: Pedicle screws were placed at each level on the convex side and at every 1–3 levels on the concave side using a freehand technique. Then the inferior articular process was removed for stress release and better fusion.

Rod placement and locking: We installed a prebent correcting rod on the convex side of the major curve, secured the nuts onto the screws, and rotated the rod to rebuild the normal sagittal alignment. Appropriate distraction and compression force were applied on the major curve and then screws were locked after each distraction or compression maneuver. Subsequently, we embedded a prebent maintenance rod on the concave side after achieving a satisfactory outcome. The rod on the concave side was tightened after appropriate maneuvers in a similar way. Necessary adjustments were performed via in situ benders according to intraoperative shoulder balance assessment and the C7-central sacral vertical line (C7-CSVL).

After finishing the correction, one or two transverse connectors were installed. The laminae of instrumented levels were decorticated and autograft and allograft bones were added to promote bone fusion.

Intraoperative motor evoked potential (MEP) and somatosensory evoked potential (SSEP) were used in all patients to monitor spinal cord function.

2.4 Postoperative management

All patients were equipped with thick compressed dressing clothes on surgical sites (Figure 1) and were encouraged to perform bedside rehabilitation exercise with the protection of hard bracing. The drainage tube was removed when daily drainage volume was less than 50 mL/d. Red blood cell (RBC) transfusion therapy was applied when Hb level was less than 70 g/L or typical anemia-related dizziness, dyspnea or tachycardia were observed (14). All patients were instructed to take off the bracings after 3 months and started physical exercise training under the guidance of a single group of physical therapists.

(A) The surgical incision was covered with the typical compressed dressing clothes, the upper border of which was approximately paralleled with the horizontal line. (B) The thick dressing clothes were stuck to the skin by elastic adhesive tapes in a symmetric pattern.

All patients were followed up for at least 5 years and were annually guided to complete the SRS-22 questionnaires independently according to their subjective feelings. We also measured lumbar mobility at each follow-up repetitively with the help of a digital inclinometer following the method reported by Sanchez's (15). A typical case was shown in Figure 2.

(A) A 14-year-old female patient was diagnosed as Lenke 5C AIS, with a major lumbar curve of 51°. (B) The major cobb angle was corrected to 5.7° immediately after surgery and no submuscular drainage was placed. (C) The major cobb angle was spontaneously improved to 2.8°at the latest follow-up. No myositis ossificans was observed on x-ray films.

2.5 Statistical analysis

The perioperative parameters included: demographic data (including age, gender, height, weight, body mass index), preoperative data (pre-op Cobb angle, Lenke classification, pre-op Hb and Hemocrit level), intraoperative data (operation time, fusion segments, estimated blood loss, intra-op blood transfusion volume), postoperative data (post-op blood transfusion volume, post-op Hb and Hemocrit level, duration of pyrexia, peak body temperature, surgery-related complication, length of hospital stay). The long-term parameters included: follow-up times, function/activity domain and pain domain scores in SRS-22 questionnaires.

Data were analyzed using Statistical Package for the Social Sciences software (IBM SPSS Statistics, RRID: SCR_016479). Since this study was a single-center retrospective cohort study with minimally 5-year follow-up, it was difficult to estimate the sample size in advance. Therefore, we performed a post-hoc power analysis for the key outcomes (e.g., intraoperative data and postoperative clinical and hematologic data, etc.) to calculate the study's power based on the final sample size. The normal distribution of continuous variables was assessed using the Kolmogorov–Smirnov test, while homogeneity of variance was evaluated using Levene's test. Continuous variables with normal distribution and homogenous variance were analyzed using bilateral independent t-tests if the corresponding variables were measured simultaneously. Otherwise, Wilcoxon rank-sum tests were applied. For SRS-22 questionnaire scores, we utilized MANOVA analysis given that each domain was influenced by multiple factors (e.g., ages, subjective feelings, missing values, etc.). Moreover, Fisher exact test and Pearson Chi-square test were used for categorical variables.

3 Results3.1 Preoperative data

The preoperative data are shown in Table 1. No significant difference was found between the two groups in age, gender, height, weight, body mass index (BMI), preoperative Cobb angle and Lenke curve type.

Demographic dataGroup AGroup BP value(n = 35)(n = 51)Age (y)14.8 ± 3.215.9 ± 2.60.081aMale/Female4/318/430.256aHeight (cm)155.3 ± 8.7156.8 ± 10.50.533aWeight (kg)43.2 ± 7.244.8 ± 7.30.319aBMI (kg/m2)17.8 ± 2.218.5 ± 5.00.459aPreoperative main curve Cobb angle (°)54.6 ± 14.350.9 ± 12.20.055aLenke curve type, n (%)0.455a Type I17 (48.6)26 (51.0) Type II1 (2.8)5 (9.8) Type III1 (2.8)2 (3.9) Type IV2 (5.7)2 (3.9) Type V8 (22.9)11 (21.6) Type VI6 (17.2)5 (9.8)

Preoperative characteristics of two groups (mean ± SD).

Group A, compressed dressing clothes with submuscular closed suction drainage; Group B, compressed dressing clothes with no drainage.

a

Compared with group A, P > 0.05.

3.2 Intraoperative data

We calculated the effect size of each representative indicator of intraoperative and postoperative data using Cohen's d , . Based on the effect size, we further calculated a statistical power of at least 0.802 to detect the observed difference at α = 0.05. The normal distribution and homogeneity of the key perioperative data were confirmed by the Kolmogorov–Smirnov test and Levene's test, respectively. Generally, there were no significant differences between the two groups regarding operation time, estimated blood loss (EBL), autologous blood transfusion volume, allogenic blood transfusion volume, number of fusion segments and pedicle screws (Table 2). No osteotomy was performed and no intragenic complications such as cerebrospinal fluid leakage or pleural rupture occurred during surgeries. No abnormal MEP or SSEP signal change was observed during pedicle screw insertion and correction maneuver.

Intraoperative dataLIV ≥ L1P value(LIV < L1)P valueDrainageNon-drainageDrainageNon-drainage(n = 8)(n = 9)(n = 27)(n = 42)Operation time (min)248.1 ± 54.7235.7 ± 45.10.614a266.8 ± 46.6253.2 ± 41.90.205aEBL (mL)612.5 ± 231.7640.2 ± 203.60.347a676.3 ± 356.9664.9 ± 293.60.885aAutologous blood transfusion (mL)475.0 ± 204.8448.9 ± 175.30.459a535.0 ± 244.8488.9 ± 205.30.300aAllogenic blood transfusion (mL)40.0 ± 106.327.4 ± 80.20.534a38.7 ± 108.626.3 ± 68.50.566aAllogenic blood transfusion cases, n (%)1 (12.5%)1 (11.1%)0.692a4 (14.8%)5 (11.9%)0.492aNO. of fused segments11.9 ± 0.811.8 ± 1.80.894a12.8 ± 2.412.4 ± 2.80.527aNO. of pedicle screws15.00 ± 2.5615.65 ± 2.300.644a15.55 ± 2.8515.68 ± 3.050.850aNO. of reserved lumbar segments5.00 ± 0.764.74 ± 0.530.296a2.29 ± 0.692.32 ± 0.660.877a

Intraoperative data of the patients with different fusion extent (based on lumbar fusion) (mean ± SD).

a

Compared with drainage group, P > 0.05.

To investigate the underlying mechanism of wound drainage in different surgical sites, we further divided the patients into two subgroups according to whether the lumbar segments were extensively fused (LIV < L1) or not (LIV ≥ L1). When evaluating the aforementioned intraoperative data in the patients whose LIVs were above L1 (n = 17), the difference between the drainage (n = 8) and non-drainage (n = 9) cohorts was insignificant (Table 2). A similar outcome was obtained in the patients whose LIVs were below L1 (n = 69) (Table 2).

3.3 Postoperative data

Table 3 presents the postoperative data of the two groups in general. Characterized by an obviously longer hospital stay (10.9 ± 3.6 d vs. 8.0 ± 2.6 d, P < 0.01), the Group A also received a greater postoperative blood transfusion volume (40.0 ± 24.7 mL vs. 23.5 ± 15.1 mL, P = 0.011) and exhibited a higher blood transfusion rate (7/35 vs. 5/51, P = 0.028). We also recorded and listed the hemoglobin and hemocrit levels as the two important indices that indirectly reflected postoperative blood loss (Table 4). The hemoglobin and hemocrit values in the drainage group were significantly lower compared with non-drainage group on the third postoperative day (93.73 ± 10.08 g/L vs. 99.95 ± 9.45 g/L, 27.75 ± 2.55% vs. 29.94 ± 3.15%, P < 0.01). Similar outcomes were observed in the patients whose lumbar segments were extensively fused (LIV < L1) (Table 5). The hospital stay duration and postoperative blood transfusion were significantly higher in the drainage cohort than in the non-drainage cohort (10.9 ± 3.8d vs. 7.7 ± 2.1d, 45.16 ± 88.84 mL vs. 21.05 ± 62.20 mL, P < 0.01). Meanwhile, the differences in the 3-day hemoglobin and hemocrit values were statistically significant (92.51 ± 7.64 g/L vs. 99.12 ± 9.81 g/L, 27.52 ± 1.93% vs. 29.79±3.35%, P < 0.01) (Table 4). Conversely, in the patients whose lumbar segments remained unfused (LIV ≥ L1), no significant difference existed between the drainage and non-drainage group with respect to postoperative indices (Table 4). Generally, no significant difference was found in the duration of postoperative pyrexia (axillary temperature ≥ 37.3 ℃) (1.1 ± 0.7d vs. 1.2 ± 0.6d, P = 0.268) and the peak temperature value within 72 h (37.5 ± 0.4 °C vs. 37.6 ± 0.4 °C, P = 0.523), indicating that the two groups presented an approximately similar degree of aseptic inflammatory reaction as well as febrile reaction (Table 3). The incidences of hematoma-related complications were similarly low in both groups (3 swellings in Group A, 2 swellings and 2 ecchymosis in Group B). All these cases received no targeted intervention and recovered spontaneously before discharge (Table 3).

Postoperative dataGroup AGroup BP value(n = 35)(n = 51)Length of stay (days)10.9 ± 3.68.0 ± 2.6*<0.01bDuration of pyrexia (days)1.1 ± 0.71.2 ± 0.60.268aPeak temperature within 72 h (°C)37.5 ± 0.437.6 ± 0.40.523aAllogenic blood transfusion cases, n (%)7 (20%)5 (9.8%)*0.028bBlood transfusion volume (mL)40.0 ± 24.723.5 ± 15.1*0.011b3rd-day Hemoglobin (g/dL)93.73 ± 10.0899.95 ± 9.45*<0.01b3rd-day Hematocrit (%)27.75 ± 2.5529.94 ± 3.15*<0.01bIncision-related complications, n (%)3 (8.57%)4 (9.80%)0.347a

Postoperative clinical data of the patients of two groups (mean ± SD).

a

Compared with group A, P > 0.05.

b

Compared with group A, P < 0.05.

*

indicated significant difference.

Hematologic dataLIV ≥ L1P valueLIV < L1P valueDrainageNon-drainageDrainageNon-drainage(n = 8)(n = 9)(n = 27)(n = 42)Hemoglobin (g/dL) Preoperative131.5 ± 10.4128.7 ± 9.30.562a130.36 ± 7.27128.94 ± 8.400.504a 0 days postoperative106.06 ± 9.40108.11 ± 6.580.607a103.50 ± 11.84104.18 ± 12.320.836a 3 days postoperative102.50 ± 14.37100.89 ± 8.620.780a92.51 ± 7.6499.12 ± 9.81*0.008b 7 days postoperative109.62 ± 13.76107.33 ± 6.780.664a108.25 ± 6.28107.85 ± 9.360.847aHematocrit (%) Preoperative38.65 ± 2.3438.48 ± 2.970.897a38.50 ± 3.1738.82 ± 2.210.670a 0 days postoperative31.43 ± 2.5832.38 ± 2.130.417a30.48 ± 3.3831.04 ± 3.500.547a 3 days postoperative29.91 ± 3.7428.74 ± 2.850.918a27.52 ± 1.9329.79 ± 3.35<0.01b 7 days postoperative32.60 ± 4.3031.96 ± 1.390.676a32.45 ± 2.2632.73 ± 2.700.679a

Perioperative hematologic data of non-transfused patients with different fusion extent (based on lumbar fusion) (mean ± SD).

a

Compared with group A, P > 0.05.

b

Compared with group A, P < 0.05.

*

Indicated significant difference.

Postoperative dataLIV ≥ L1P valueLIV < L1P valueDrainageNon-drainageDrainageNon-drainage(n = 8)(n = 9)(n = 27)(n = 42)Length of stay (days)8.6 ± 2.09.8 ± 4.00.464a10.9 ± 3.87.7 ± 2.1*<0.01bDuration of pyrexia (days)0.9 ± 0.41.2 ± 0.70.243a0.9 ± 1.11.2 ± 0.10.335aPeak temperature within 72 h ( °C)37.1 ± 0.337.3 ± 0.30.413a37.6 ± 0.437.7 ± 0.30.226aAllogenic blood transfusion cases, n (%)1 (12.5%)1 (11.1%)