Surgical site infection (SSI) is defined according to the criteria of the Centers for Disease Control and Prevention (CDC) [12]. A superficial infection of the incision site is diagnosed in the presence of clinical signs of infection that involves the skin, subcutaneous tissue, or muscle located above the fascial layer, and accompanied by at least one of the following criteria: purulent drainage located above the fascial layer and/or isolation of a microorganism from a wound culture showing signs of inflammation. A deep incisional infection is diagnosed when the infection is found to involve tissue below the fascia on opening and exploration by the surgeon. An organ space infection is diagnosed when clinical and imaging (CT scan or magnetic resonance imaging) signs are suggestive of an infection of a disc, a vertebra, or a paravertebral structure, and is associated or not associated with a positive culture of blood or of the material obtained by a needle aspiration of the vertebral space, needle biopsy of the bone, or open bone biopsy. Deep SSI are required to meet the following criteria: 1) the infection occurred within 30 days after the operative procedure if no implant was left in place and within 1 year if implant was in place; 2) it seemed to be related to the operative procedure; 3) it involved the deep soft tissue of the incision (fascial and muscle layer) or an organ space that was opened or manipulated during the operative procedure; and 4) the patient had either (a) purulent drainage from the deep incision; (b) deep incision spontaneously dehisced or was opened by a surgeon; (c) evidence of infection involving the deep tissue on direct examination; or (d) diagnosis of deep infection by a surgeon [12].

Postoperative infections other than surgical site infections were diagnosed according to the CDC criteria [12,13], which included a diagnosis of infection by the surgeon. All included patients are monitored for signs of drug toxicity and serious adverse events, such as Clostridium difficile associated colitis, allergic reaction or anaphylactic shock. Demographic and surgical data collected included the age at the time of surgery, gender, underlying medical conditions, history of previous spinal surgery and other surgeries, dosing and timing of preoperative antibiotic prophylaxis, diagnosis, type of surgery, use of instrumentation, use of allograft, number of vertebrae fused, estimated blood loss, and duration of surgery. For patients with an occurrence of surgical site infection, further data were collected on subsequent operations, including hardware removal. Postoperative data included symptoms and signs of infection with a follow-up of at least 6 months. Infections were determined by positive wound culture and sensitivity or by an attending spine surgeon's clinical impression that wound infection had occurred.

The mean age in our group was 47 years (range, 19-71 years). The male to female ratio was 49:41 and the average timing of administration of antibiotics was 77 minutes (range, 30-120 minutes). The average blood loss was 626 mL (range, 150-3,000 mL) with a mean duration of surgery for 3.2 hours (range, 1.5-6 hours). One case of superficial infection and one case of deep SSI was noted. Complex statistical analysis was avoided because for a wound infection rate of 10%, 5% and 2%, it would be necessary to enroll 474, 988, and 2,518 subjects, respectively, in each treatment group [14], 5,036 patients would need to be enrolled to show a decrease in the infection rate to 1%. Such large sample size establishes that a clinical trial is unlikely to occur in which the infection rate is as low as 2% [6,13].

Current practice for the choice of antibiotic is either a first or second generation cephalosporin, based on the fact that they have better coverage against gram positive organisms which are the most common cause for spine infection. Rubinstein et al. [19] conducted a double-blind, placebo controlled trial and recommended an administration of a single dose of cephazolin preoperatively for patients undergoing lumbar spinal surgery, provided that sensitivity monitoring is performed in all infected cases. An earlier animal study [20] concluded 1 g of intravenous cephazolin to be highly effective for prophylaxis against bacterial contamination [19-22]. In contradiction to this, a study by Walters et al. [3] concluded that the incidence of iatrogenic discitis cannot be eliminated with a broad spectrum antibiotic, such as cephazolin. Moreover, cephazolin appeared to be ineffective at preventing endplate destruction once discitis has become established. Current evidence exists to suggest that the charge on antibiotics, due to their ionisable groups, is important in determining their ability to diffuse into the disc [23]. Variation in the rate of diffusion for different antibiotics may be due to multiple factors, including molecular size and charge, serum protein binding, and antibiotic solubility in different tissues. Rhoten et al. [22] and Lang et al. [24] suggested that higher doses of cephalosporins are required to achieve therapeutic levels in the disc. Hoelscher et al. [25] have shown that high doses of antibiotics, including both cephalosporins and aminoglycosides, could have deleterious effects on the survival of cultured disc cells, cell proliferation and metabolic rates. In Gibson et al. [26], series, negatively-charged antibiotics were not detected in either the nucleus pulposus or the annulus fibrosus of a normal human intervertebral disc, despite high blood levels of antibiotics. It is therefore likely that the penetration of antibiotics is similar in normal and pathological discs. Therefore, the literature has conflicting evidence regarding the concept of higher dosing as well as the charge on the antibiotic as the determinant factor. Besides, it is notable that both cefazolin and cefuroxime are positively charged ions, but with a longer half-life of cefazolin. In a double-blind, placebo controlled clinical study by Petignat et al. [5], a single, preoperative dose of 1.5 g cefuroxime reduces the risk of organ space infection, most notably spondylodiscitis, after surgery for herniated disc in non-instrumented spine surgery. The overall rate of infection is 5.2% in the cefuroxime group and 6.3% in the placebo. However, it did not prove that cefuroxime was better than other antibiotics frequently used for prophylaxis, such as first-generation cephalosporin.

From our point of view, the timing of administration is the most controversial to establish the exact safer limit for the timing of antibiotics administration. According to Classen et al. [27], the rate of wound infections are lowest if when the antibiotics are administered within 2 hours of surgery. The current opinion [13] is that patients should receive antibiotics 30 minutes prior to the skin incision and preferably before induction. This measure allows time for the antibiotic to reach tissues with optimal bactericidal serum concentration. Operative procedures proceeding more than 6 hours of procedures with instrumentation (implants prosthetic materials), diabetes, obesity, smoke, severe blood loss more than 1,500 mL, remote source of infection, and 3 or more co existing medical comorbids are additional risk factors for postoperative wound infection, thereby warranting redosing of antibiotic prophylaxis [17,18]. Fraser et al. [20] and Guiboux et al. [28], using sheep and rabbit models of discitis, respectively, demonstrated that the administration of intravenous cefazolin prevented the development of infection. These studies [20] demonstrated that the optimal time of administration was 30 to 60 minutes before surgery. Also, they found that the concentration of cefazolin in the disc was dependent on the maintenance of high serum antibiotic concentrations; thus, making it imperative to ensure the proper timing and dose of antibiotics. Once infection develops in the disc, intravenous antibiotics have little impact on the course of the disease. Some authors have advocated antibiotics 24 hours postoperatively [13]. Other studies have affirmed that additional administration have no value [27,29], even in the presence of wound drain and epidural catheter [30]. In our set up, we gave postoperative antibiotics for 48 hours due to the presence of drains and epidural catheter, as per a recent study by Kim et al. [9]. We do not consider fever alone as a parameter for continuing antibiotic as it is our clinical experience that it is surgical stress induced rather than a marker of infection. It has been our clinical experience that maintaining the therapeutic level is more important. In our series, out of 30 patients for whom antibiotics were given between 61 to 90 minutes, only 8 patients did not have intraoperative dosing as mentioned in Table 3. None of those patients developed an infection. These 8 patients had posterior decompression with instrumentation and fusion in all three different groups of antibiotics. Based on our findings that antibiotics are not detectable between 61 to 90 minutes can be challenged. Interestingly, all patients who received antibiotics beyond 90 minutes had an intraoperative dose (Table 3), given which may even challenge the role of prophylaxis itself and suggest that intraoperative dosing may be a better option. However, we have a very small subset of patients to come to this conclusion. Looking at the average duration among all three groups of antibiotics by data analysis, we came to conclusion that 75 minutes is safer and better in all groups where it is highly unlikely that patients will develop infection. Barker [6] conducted a metanalysis and concluded that antibiotic prophylaxis was beneficial for spinal surgery. The rate of infection was 5.9% in this metanalysis for patients without antibiotics, which was 2% lower in the control group (2.9%). The controversy has been that routine antibiotic prophylaxis leads to possible drug reactions, emergence of antibiotic resistance against bacteria and additional cost [15,16]. Others argue that prophylaxis for most spinal surgery should be used for internal fixation or fusion of spine as they are associated with higher risk of infection compared to surgeries without instrumentation and extensive surgical exposure. Our clinical observation was that if there is no break in asepsis during surgery, most of the superficial infection are acquired after surgery and reflects the standard of asepsis enforced during postoperative care. However, the data from our present study established that if high standards of asepsis are enforced, rate deep infections can also avoided. Our data reveals that classic microorganisms were recovered from one case of superficial infection. The culture and sensitivity was sent and patients were treated as per sensitivity of the organism. However, the limitation of our study was that we had no case of methicillin resistant Staphylococcus aureus (MRSA) or resistant organisms to cefazolin. We do not have data to support any specific guidelines regarding this particular infection and we still follow the standard practice as for treating any MRSA infected patients in any surgery worldwide. There was no adverse event reported in relation with the use of prophylaxis and is coherent with other studies.

The present study suffers from a few limitations. The present study is a randomized study, which was initially blinded to the nurse, patients and the resident collecting the data. However, the blinding was lost because a single spine surgeon involved in surgery was aware of the postoperative dose and the choice of antibiotic given. Due to the lack of a control group in this study of patients without instrumented surgery, the role of prophylaxis antibiotic may be challenged. However the economic, social and psychological burden far exceeds the consequences, if patients develop any surgical site infection. Another limitation comes from the fact that the diagnosis of SSI relied largely on the physician in charge. The impact of this possible bias is limited though not free because it was not a blinded study. One more limitation of our study regarding deep SSI is that as per CDC, the criteria minimum of a one year follow-up is necessary. Our study has conducted a follow-up only for 6 months. Although the initial results have been very encouraging, a double-blinded multicenter study across the subcontinent would be needed in order to answer other key issues and limitations.