Risk factors associated with surgical site infections after thoracic or lumbar surgery: a 6-year single centre prospective cohort study

Background Surgical site infection (SSI) is a risk in every operation. Infections negatively impact patient morbidity and mortality and increase financial demands. The aim of this study was to analyse SSI and its risk factors in patients after thoracic or lumbar spine surgery. Methods A six-year single-centre prospective observational cohort study monitored the incidence of SSI in 274 patients who received planned thoracic or lumbar spinal surgery for degenerative disease, trauma, or tumour. They were monitored for up to 30 days postoperatively and again after 1 year. All patients received short antibiotic prophylaxis and stayed in the eight-bed neurointensive care unit (NICU) during the immediate postoperative period. Risk factors for SSI were sought using multivariate logistic regression analysis. Results We recorded 22 incidences of SSI (8.03%; superficial 5.84%, deep 1.82%, and organ 0.36%). Comparing patients with and without SSI, there were no differences in age (p=0.374), gender (p=0.545), body mass index (p=0.878), spine diagnosis (p=0.745), number of vertebrae (p=0.786), spine localization (p=0.808), implant use (p=0.428), American Society of Anesthesiologists (ASA) Score (p=0.752), urine catheterization (p=0.423), drainage (p=0.498), corticosteroid use (p=0.409), transfusion (p=0.262), ulcer prophylaxis (p=0.409) and diabetes mellitus (p=0.811). The SSI group had longer NICU stays (p=0.043) and more non-infectious hospital wound complications (p<0.001). SSI risk factors according to our multivariate logistic regression analysis were hospital wound complications (OR 20.40, 95% CI 7.32–56.85, p<0.001) and warm season (OR 2.92, 95% CI 1.03–8.27, p=0.044). Conclusions Contrary to the prevailing literature, our study did not identify corticosteroids, diabetes mellitus, or transfusions as risk factors for the development of SSI. Only wound complications and warm seasons were significantly associated with SSI development according to our multivariate regression analysis.


Background
Every surgery carries a risk of SSI, a complication which negatively impacts patient morbidity and mortality, increases financial demands by prolonging hospital stay, and may require further antibiotics and surgical procedures. SSIs are a significant group of healthcareassociated infections with high preventability [1,2]. Prudent preventive strategies have an important role in increasing postoperative patient safety and can limit the incidence of multidrug resistant strains. The elimination of this complication is a priority in all surgical management and is particularly important in spine operations, where these risks are heightened due to the frequent use of metallic implants, the nearby localization of the spinal cord, and the load-bearing function of the spine. The incidence of SSI in spine surgery varies from 2 to 13% according to literature of varying quality and methodology [3][4][5][6][7][8][9][10]. A protective protocol includes many strategies for reducing the risk of developing an SSI. This involves maintaining correct antibiotic prophylaxis, and proper hygiene throughout all stages of surgery and general care, not only in the operating theatre but crucially during the postoperative period until the wound has healed [1,2].
The aim of this study was to identify and analyse SSI in accordance with international definitions, and to search for risk factors associated with its onset in patients who had undergone thoracic or lumbar spine surgery.

Method
A 6-year single-centre prospective observational cohort study was conducted in the Neurocenter at the 900-bed Liberec Regional Hospital from 1 January 2005 to 31 December 2010. The incidence of SSI was monitored in 274 who fulfilled our inclusion criteria. These criteria were (1) planned operation; (2) thoracic or lumbar localization; (3) degenerative disease, trauma, or tumour; (4) short antibiotic prophylaxis defined as antibiotic administration before surgery and during long operations; and (5) patients who were recommended by neurosurgeons or anaesthetists for a postoperative stay in the eight-bed Neuro-intensive Care Unit (NICU) which has a multimodal preventive infection control and normoglycemia protocol. Exclusion criteria were as follows: (1) primary infection of the spine and (2) prolonged antibiotic prophylaxis defined as continued antibiotic administration before and/or after the operation.
SSIs were classified according to Horan et al. [11] as (1) superficial: skin and subcutaneous tissue; (2) deep: fascia and muscle; and (3) organ: organs and spaces. For superficial SSI without microbiology we included borderline cases. These were defined as either a large seroma, or spontaneous dehiscence with secretion and high CRP. Borderline cases were treated with local or systemic antibiotic therapy. The incidence of SSI was monitored for up to 30 days postoperatively and again after 1 year. The following prevention protocols were adhered to in order to conduct SSI analysis. Hygiene rules: (1) hand hygiene before and after patient contact and each procedure; (2) masks, surgical caps, sterile surgical gowns, and sterile insertion of systems in invasive medical procedures; (3) disinfection soap before entering the operating theatre; (4) principles for drainage and tubes: single-use products, closed systems, minimal necessary duration, minimal and only necessary disconnection using the port system, and regular and irregular exchange according to protocol; and (5) surgical wound fully covered and dry.
Antibiotic prophylaxis: We mainly used two types of antibiotics without rotation. The first choice antibiotic was cefazolin; in case of allergy, clindamycin was used.
(1) Short prophylaxis: Administered before and during the operation, without prolonging use after the operation. (2) Doses: Intravenous administration of the appropriate dose of antibiotics. In patients up to 120 kg either 2 g of cefazolin or 600 mg of clindamycin, over 120 kg either 3 g of cefazolin or 900-1200 mg of clindamycin, with repeated administration during high blood loss (over 1.5 l blood). (3) Timing: The correct timing before incision (30-60 min) and perioperative administration at the correct interval (cefazolin at 4 h, clindamycin without need for further administration) [1]. We studied the following risk factors of SSI: (1) parameters associated with operations (localization, number of vertebrae, reoperation, time of operation, use of graft and implant, ASA Score); (2) use of medical devices: drainage, airways, mechanical ventilation, and catheters (artery, central venous, urine); (3) administration of corticosteroids (methylprednisolone, hydrocortisone); (4) transfusions, blood loss, and haemoglobin; (5) ulcer prophylaxis; (6) diabetes mellitus; (7) Acute Physiology and Chronic Health Evaluation (APACHE) II score on admission; (8) C-reactive protein (CRP); (9) length of stay in the NICU and in our hospital; (10) non-infectious hospital wound complications; and (11) warm season (June, July, August).

Statistical analysis
Parametric t-tests or non-parametric Mann-Whitney U tests were used for comparison of continuous parameters. Comparison of categorical parameters was carried out using chi-square or Fisher's tests as appropriate. Univariate logistic regression was used for identifying prognostic factors of wound complications. Factors from our univariate analysis that met the significance threshold of p <0.1 were used for multivariate regression analysis; factors with p value <0.1 were left in the model. P values of less than 0.05 were considered significant. STATISTICA 13.2 (TIBCO Software Inc., Palo Alto, CA, USA) software was used for statistical analyses.
The study was conducted after the approval of the Regional Hospital Ethics Committee for Multicentric Clinical Trials.

Results
Of 286 consecutive patients treated at our centre, 274 met our inclusion criteria and were included in our study. We excluded twelve patients due to their prolonged antibiotic prophylaxis following the operation. The results of short antibiotic prophylaxis are shown in Table 1.
Over 6 years, we recorded 22 incidences of SSI (8.03%), the majority were superficial (5.84%), and a few were deep (1.82%) or organ (0.36%). When patients with SSI were compared with the control group, there were no significant differences in demographic data, diabetes mellitus, or ulcer prophylaxis ( Table 2). There was also no difference concerning corticosteroid use. The mean duration of corticosteroid use was 2.14±1.17 days. Results associated with operations are shown in Table 3. No differences were found in localization, number of vertebrae, duration of operation, or any other parameter associated with operations. Similarly, non-significant results were found in parameters associated with the immediate postoperative period in the NICU.
However, in the SSI group, we found more wound complications of other etiologies (such as dehiscence, secretion, seroma, or haematoma) ( Table 4). These complications together with incidence during the warm season (June, July, and August) were found to be the only significant predictors of SSI according to our multivariate logistic regression analysis (Table 5).

Discussion
The incidence of SSI is an important mark of quality management in every surgical procedure. Since these infections are preventable, it is important to take an interest in their monitoring [2]. SSIs can worsen the final results of operations, and additionally in spinal surgery, a patient's mobility can be affected due to the close proximity of the spinal cord and neural structures. This will raise the costs of care for the spine operation.
Since new therapeutic approaches are limited, the basis of SSI management is prevention. This means maintaining an aseptic environment, a standard which is followed closely in the operating theatre. This standard is also important throughout every stage of the postoperative period, and especially in the initial phase until the wound has healed.
The biggest challenge to the implementation of a preventative care protocol is compliance of the entire team of doctors, nurses, and technicians.
An important component of our SSI prevention strategy was the correct antibiotic prophylaxis [1]. Antibiotic prophylaxis is based on the principle of eliminating any bacterial contaminant by administering a suitable antibiotic so that it is present in the surgical site, even in blood clots, in an effective bactericidal concentration throughout the entire operation. One common mistake which has significant epidemiological consequences is the inappropriate prolongation of antibiotic prophylaxis. All our prophylaxes were short-term, with the exception of 12 (4.20%) of our 286 consecutive patients, who received prolonged antibiotic prophylaxis. These 12 patients were excluded from our study for this reason. Another common error that substantially impairs the effectiveness of antibiotic prophylaxis is the incorrect timing of preoperative administration. We resolved this issue by giving antibiotics immediately in the preoperative prep-room, thus achieving the appropriate level of protection at the time of incision. Last but not least, failure to administer appropriate doses when operations are prolonged can result in excessive or insufficient prophylaxis.
To interpret the quality and effectiveness of antibiotic prophylaxis properly, patient populations should be stratified according to risk and outcomes, and interpretation should take into account the influence of other risk factors. For each procedure, process indicators and audit methodology should be defined in the quality assessment of antibiotic prophylaxis. In order to draw a statistically meaningful conclusion, a minimum of 100 homogenous procedures should be evaluated. Our study population of 274 patients (defined by 6-year period) fulfils this criterion with a high safety margin. For each operation, the  [3][4][5][6][7][8][9][10]. This wide variation is because a large proportion of the studies are of a retrospective design, the criteria for the definition of SSI is inconsistent, and a lack of meticulous reporting, as cited in Spine (Boody and Vaccaro, 43) [9]. Our results are based on prospectively collected data, a consecutive population, and the internationally accepted definition of SSI according to Horan et al. [11]. During this 6-year monitoring period, we identified 22 patients with SSI, which is at the upper limit (8.03%) of the results reported in literature. However, we emphasize that we included every single incidence of SSI, including borderline cases; we decided to include these due to our experience of non-purulent secretion with positive microbiology.  ASA American Society of Anesthesiologists, NICU neurointensive care unit, TISS Therapeutic Intervention Scoring System, APACHE Acute Physiology and Chronic Health Evaluation, CRP C-reactive protein, OP operation various causes including the quality of wound care, are a significant risk factor. Established procedural methods in the preoperative and perioperative period confirm a high degree of preventability of SSI [2]. For this reason, it is important to carefully monitor all such non-infectious wound complications. Another significant factor (in fact a leading factor in cases of delayed SSI) appears to be patient compliance. However, this factor is extremely hard to describe using numerical methods and it was not evaluated in our study.
Contrary to the prevailing conclusions in literature [12][13][14][15], some of the anticipated risk factors were not confirmed by our study, namely diabetes mellitus, the use of corticosteroids, transfusions, or ulcer prophylaxis. Concerning diabetes mellitus, we attribute the result to our strict maintenance of normoglycemia in our patients. Two reasons for the insignificant influence of corticosteroids as a risk factor may be the short duration of immunosuppressive therapy during the preoperative and postoperative period (mean duration was 2.14±1.17 days) and the size of the dose administered. The prevailing corticosteroid was hydrocortisone in a substitute dose (150-300 mg), which was not indicated as a risk factor for SSI in the results. The insignificant influence of transfusions can probably be attributed to our transfusion strategy with a low trigger of haemoglobin levels (70 g/L), which resulted in fewer transfusion cases. This study found that besides non-infectious wound complications as a general category, the only significant risk factor was operation during the warm season, in our region from June to August. In these months, the local average temperature is 20.9°C in June, and 18.7 in both July and August. The reason for an increase in SSI during the warm season is primarily due to the lack of  air-conditioning in our wards at the time of the study. The higher ambient temperature during the summer leads to increased sweating of the skin and results in less favourable conditions for wound healing, requiring more frequent re-dressing of the wound. Skin irritation is more likely, leading to decreased compliance by some patients (keeping the dressings clean and avoiding mechanical stimuli to the wound). However, this explanation is purely observational, as such factors are nearly impossible to evaluate numerically. There is a slight increase in the rate of SSI during January and February. We did not identify any objective factors for such an increase. The only possible explanation we could think of is speculative in its nature. Considering our system of reimbursement, the surgeons always face restrictions of budget and resources at the end of the fiscal year. They admitted they tend to operate on patients with more favourable radiological findings and less risk in general terms during December. They also tend to postpone more difficult cases to the beginning of the new fiscal year. Such selection is based on personal, nonparametric experience and subjective evaluation. Our assumption is further supported by zero SSI during the final month of the year; however, we have no means for any kind of numerical evaluation.
Our study has several limitations. We do not evaluate some factors of potential significance, namely smoking and nutrition. Despite the majority of elective operations, these factors cannot be assessed properly due to lack of data on admission and the short period before surgery. All the patients had glucose values tested during the period before the operation; however, immediate preoperative values were measured only in patients with diabetes mellitus.

Conclusions
Contrary to the prevailing literature, our study on a population of planned thoracic or lumbar spine surgery patients with short antibiotic prophylaxis hospitalized postoperatively in the NICU did not identify corticosteroids, diabetes mellitus, or transfusions as risk factors for SSI. Our study concludes that any kind of noninfectious wound complication and operation during the warm season represent independent risk factors for developing such infections.