Post-thoracotomy pain is considered one of the most intense types of postoperative pain because a thoracotomy incision involves multiple muscle layers and possible rib injury. Effective treatment of acute post-thoracotomy pain after lung resection is particularly important to keep the patient comfortable and to minimize pulmonary complications [1, 2]. Thoracic epidural analgesia (EDA) is the gold standard for pain control after thoracotomy. However, because of its severe side effects, it is contraindicated in patients taking anticoagulant or antiplatelet drugs, which have recently become more frequently used. In addition, some patients’ anatomy can make epidural catheter insertion challenging. Therefore, it is important to develop an alternative procedure.

Thoracic paravertebral block (PVB) blocks the nerves of multiple contiguous thoracic dermatomes above and below the infusion site [3] and has been shown to provide pain relief comparable to that of EDA but with fewer side effects [4, 5]. In addition, a paravertebral catheter can easily be inserted intraoperatively by the surgeon under direct vision, eliminating the need for a separate procedure.

The efficacy of thoracic PVB depends on accurate catheter placement and on the potency, concentration, and volume of local anesthetic. Our preliminary trials of thoracic PVB led us to suspect a strong correlation between a number of reported insertion methods [3, 4, 6–10] and substantial leak of local anesthetic into the pleural space, which results in less effective postoperative pain control. However, some authors have described the effectiveness of intraoperative percutaneous catheter placement during video-assisted thoracoscopic surgery (VATS) [11, 12]. This insertion method offers better pain control, as the extra-pleural space is closed above the catheter tip, allowing no leakage into the pleural space and facilitating easy coverage of more than one intercostal space. We applied this method because of its potential usefulness, not only in VATS, but also when used just before a thoracotomy incision.

At our institution, we have performed thoracic PVB using this insertion technique since March 2013, applying it to thoracotomy patients determined by the anesthesiologist to be contraindicated for EDA. In this study, using sample matching, we retrospectively analyzed the relationship between analgesic technique and pain-control outcome in thoracotomy patients.

Between March 2013 and February 2014, 244 patients underwent thoracotomy at our institution. Of these, 56 were contraindicated for EDA and therefore underwent the thoracic PVB procedure (PVB group). Contraindications were anti-platelet treatment (31 cases, 55.4%), therapeutic anticoagulation (nine cases, 16.0%), hemostatic disorders and/or coagulopathies (four cases, 7.1%), and potentially technically difficult epidural catheter insertion (12 cases, 21.4%). The types of surgery performed in this group were lobectomy (38 cases, 67.9%), segmentectomy (10 cases, 17.8%), wedge resection (seven cases, 12.5%), and thymectomy (one case, 1.8%).

The matched-control group (EDA group) was selected from 452 patients who received thoracotomy with EDA at our hospital from April 2011 to February 2013. After matching, this group consisted of 112 patients (Fig. 1).

In terms of complications, one patient in the PVB group, an 81-year-old female, suffered from respiratory depression immediately following surgery. This could have been caused by excessive use of ropivacaine; indeed, it resolved quickly after decreasing the rate of infusion of ropivacaine. No complications were observed in any other PVB-group patients. The EDA group displayed no severe complications such as epidural hematoma, post-dural puncture headaches, or spinal cord injury. However, at the end of the 2nd postoperative day, when the urinary catheter placed in the operating room was usually removed, 10 EDA patients (8.9%) had failed a voiding trial, which required reinsertion of the urinary catheter until voiding was successful. No patient in the PVB group had this problem (p = 0.03).

The aim of this study was to examine whether thoracic PVB could provide an alternative form of regional anesthesia when used in conjunction with general anesthesia. Our results demonstrate that thoracic PVB was not inferior to EDA in controlling post-thoracotomy pain.

Meta-analyses [15, 16] have shown that PVB provides analgesia comparable to that of EDA and has a better side-effect profile, because it is associated with less postoperative urinary retention, nausea and vomiting, and hypotension. In the present study, no one in the PVB group experienced any of these complications. Reducing the occurrence of these complications, especially in elderly patients, is required for quick postoperative recovery. In addition, Horlocker et al. reported that one of the most devastating complications of epidural anesthesia is spinal cord injury, a rare but catastrophic complication that can result from instrumentation of the epidural space [9]. From an anatomic perspective, PVB has by definition a lower risk of these complications.

While meta-analysis demonstrated a tendency toward less effective pain control with PVB, the difference was not significant. In addition, although some randomized controlled trials showed that PVB was less effective than EDA in controlling postoperative pain [17–19], others showed equivalent pain control in the two groups [4, 20]. The cause or causes of theses discrepancies must be determined, and to do so, we must focus on the technique of catheter insertion.

The meta-analyses included various techniques of catheter insertion, but we believe our current catheter-insertion method of catheter insertion offers better pain control with three reasons. First, this method helps us to avoid massive leakage of local anesthetic from the paravertebral space. Kanazi et al. reported that the failure of pain control using PVB could be attributed to inadequate diffusion of local anesthetic into the paravertebral space [18]. This method enabled us to insert the catheter without the injury of pleura surrounding the paravertebral space; therefore, intraoperative leakage was highly unlikely. Second, this method allows us to start PVB before creating the thoracotomy skin incision. Kotze et al. mentioned in their systemic review that that the group who received PVB before skin incision tended to show better pain control [21], and we also found in our clinical experience that a bolus infusion of local anesthetic before skin incision offered the better pain control. Third, the additional safety of this method is supported by the fact that the paravertebral catheters are placed under direct vision via the thoracoscopy monitor. Although ultrasound-guided catheter insertion has been reported to be safe and reliable [22], continuous PVB may also be problematic in patients with spinal anomalies, trauma, or a history of spine surgery [23]. However, this method makes it easy to confirm visually that the paravertebral space is filled with anesthetic throughout the surgery. Thus, this method of catheter insertion will offer better pain control and safety, so that we suggest that the efficacy of PVB should be reevaluated using this method.

Here, the question may be raised, if PVB is really effective in thoracotomy cases, although more ropivacaine with fentanyl was used in the PVB group than in the EDA group. Messina et al. also observed that a significantly larger amount of local anesthetic and opioid medication was required to achieve the same level of pain control in PVB group [17]. However, this result does not necessarily mean that PVB is less effective than EDA. Because in the present study fentanyl was used intravenously in the PVB group but directly introduced through the epidural catheter in the EDA group, comparison of the two groups from this perspective is difficult. On the other hand, regarding the amount of local anesthesia, we decided to use a higher concentration of local anesthesia when we initiated the use of thoracic PVB at our hospital, because several studies have shown that a greater amount of local anesthetic was required in the PVB group than in the EDA group [4, 8, 24–26]. However, some reports have already demonstrated that higher concentration and rate were not required in PVB groups to achieve the same effect [27–29]. Therefore, the amount of ropivacaine and fentanyl in the PVB group can be reduced to the same level as in the EDA group.

The limitations of this study—sampling bias, selection bias, and recall bias—are usually present in a retrospective case-control study. Regarding sampling bias, there were significant differences in the frequency of three factors in the backgrounds of patients: BMI, the presence of acquired heart disease, and the presence of ischemic cerebrovascular disease. These factors were strongly related to the way in which both groups were chosen, because patients who used anticoagulant or antiplatelet drugs regularly always allocated to the PVB group. As for selection bias, it is difficult to avoid this bias, however, anesthesiologists and surgeons always decided beforehand who would receive thoracic PVB by considering the contraindications for insertion of an epidural catheter. With regard to recall bias, all relevant medical records were used for this retrospective study. However, we acknowledge that, because the completeness of the medical records varied, some patient data were not available.

In conclusion, thoracic PVB with this thoracoscopic method, which reduced the frequency of urinary retention, was at least as effective as EDA for the postoperative pain control after thoracotomy with lung resection. We believe this procedure can be a good anesthetic alternative, especially for patients with contraindications for EDA.