A systematic review of comparative studies indicates that paravertebral block is neither superior nor safer than epidural analgesia for pain after thoracotomy

3.7.1 Matthews and Govenden (1989)

Matthews and Govenden compared the analgesic efficacy of continuous postoperative infusion of bupivacaine 2.5 mg/ml (0.25%) paravertebrally and epidurally. They aimed for similar loading doses and infusion rates in the two groups. Data from 19 thoracotomy patients for lung surgery are reported. At the T4/T5 level, 10 patients (PVB group) had a paravertebral catheter placed percutaneously during thoracotomy. Nine patients (TEA group) received an epidural catheter after wound closure. Both groups received an infusion of 3–10 ml/h of bupivacaine 2.5 mg/ml, preceded by a 10 ml bolus. Mean infusion rates were similar in the groups, as were pain scores, which they describe as a linear analogue pain scores, probably a visual analogue scale (VAS). Unfortunately they do not give results as pain at rest or pain on deep breathing and coughing, but they report that co-operation with physiotherapy and expectoration was adequate. Neither do they report any rescue analgesic medication, nor any method of randomization, nor how the patients were allocated to the PVB or TEA group. The authors concluded that both techniques provide good analgesia, although 30% of the PVB patients appeared to have lost their catheter already after 12 h of infusion. Hypotension and urinary retention were significantly more frequent in the TEA group.

Comments: This is a small study and any differences in efficacy would have to be large to be detected. Patients were followed only up to 24 h. The epidural infusion of bupivacaine 2.5 mg/ml up to 10 ml/h (average=7 ml/h/70 kg, or 17.5 mg/h/70 kg). This is overdosing, explaining the unnecessary haemodynamic and urinary side effects during TEA. We use about 5 mg/h/70 kg when fentanyl and adrenaline, 2 μg/ml each, are added to bupivacaine 1 mg/ml, and we almost never see hypotension or urinary retention caused by the TEA (Breivik, 1995; Niemi and Breivik, 2003; Niemi, 2004). Placing an epidural catheter after wound closure (with the patient still under general anaesthesia and curarized) is not optimal and it is not safe: The analgesic effect from TEA infusion during surgery, reduces the need for anaesthetic and opioid analgesics and gives the patient a “flying start” of postoperative analgesia after waking up after general anaesthesia. Siting an epidural catheter in a patient curarized under general anaesthesia has an unnecessary risk of spinal cord damage because the warning clinical symptoms of erroneous placement of needle and catheter are lost (Moen et al., 2004; Breivik et al., 2009).

3.7.2 Pertunnen et al. (1995)

Pertunnen et al. studied the analgesic effect of epidural (TEA), intercostal (IC), and paravertebral (PVB) block on post-thoracotomy pain, pulmonary functions, plasma concentration of bupivacaine, and adverse effects after lung surgery. Data from 15 patients in each of the three groups were compared. Before wound closure, the IC group was injected a total of 16 ml bupivacaine 5 mg/ml (0.5%) for unilateral T3–T7 intercostal nerve blocks, but not repeated thereafter. In the PVB group, a T2–T3 level catheter was placed intraoperatively under direct vision. The TEA group had their T5–T7 level catheter inserted preoperatively and tested with 4ml lidocaine 20 mg/ml (2%). Before wound closure, both catheter groups received a bolus of 8–12 ml bupivacaine 2.5 mg/ml (0.25%) followed by a 4–8 ml/h bupivacaine 2.5 mg/ml (0.25%) infusion, bolus volume and infusion rate defined by patient height. Rescue analgesic was provided as i.v. patient controlled analgesia (PCA) with boluses of morphine 30μg/kg. Pain at rest and pain during coughing were scored using VAS and verbal rating scale (VRS). Pain scores, morphine consumption and arterial plasma bupivacaine concentration were similar between the PVB and TEA groups. For the IC group, pain score during coughing immediately postoperatively was significantly lower compared with the TEA group. Also, for the IC group, significantly higher arterial plasma bupivacaine concentration compared with the other two groups was found immediately postoperatively. The concentration was significantly lower after 6 h. For plasma bupivacaine concentrations, great interindividual variations were observed, especially in the PVB group. Method of randomization was not reported. The authors conclude that neither method produced good post-thoracotomy analgesia.

Comments: A well-designed study. Unfortunately, less than optimal TEA is used, omitting opioid and α₂ -agonist. Also, TEA was started after completing surgery, instead of exploiting the epidural analgesia during surgery. Pulmonary functions tended to be lower in the PVB group and plasma concentrations tended to be higher in the PVB group compared with TEA. Patients in all three groups had high morphine PCA-consumption causing many of the observed adverse effects, especially nausea, vomiting, pruritus, drowsiness, confusion, hallucinations, and difficulties in breathing.

3.7.3 Kaiser et al. (1998)

Kaiser et al. evaluated PVB, which they call extrapleural intercostal analgesia and post-thoracotomy TEA after lung surgery. They randomized 30 patients to two groups, but had 13% catheter failure in each group. The TEA group (13 patients) received a T5–T6 level catheter preoperatively and bupivacaine 5 mg/ml (0.5%) 4–6 ml/h intraoperatively, followed by 4–8 ml/h of bupivacaine 2.5–3.75 mg/ml (0.25–0.375%) plus fentanyl 2 μg/ml postoperatively. The PVB group (13 patients) had a catheter placed paravertebrally at incision level under direct vision before wound closure and received 0.1 ml/kg/h bupivacaine 5 mg/ml (0.5%) plus ornipressin 0.05 U/ml postoperatively, preceded by 20 ml bupivacaine 5 mg/ml (0.5%) over 20 min towards end of surgery. Both groups got meflumenaminic acid 500 mg every 6 h and were provided additional rescue analgesic as s.c. nicomorphine 0.1 mg/kg every 4–6 h. Unspecified (rest or coughing?) pain intensity was scored on a 5 categories verbal rating scale. TEA was only provided during ICU-stay and was discontinued after 2.7 days (mean). PVB catheter infusion on surgical wards following 2 days in the ICU was continued for an average of 6.6 days. Pain scores were similar in the two groups until the TEA was discontinued on the 2nd or 3rd post-operative day when the patient left the ICU. The TEA patients then reported more pain. Nicomorphine consumption was higher in the TEA group throughout the study. There was no statistical difference in complications between the groups, although one patient in the TEA group, after discontinuation of the TEA, developed atelectasis, pneumonia, sepsis, multiorgan failure and died.

Bolus dose for the TEA was not reported, neither was total volume of infused bupivacaine solution, nor whether pain scores represented pain at rest or pain during coughing. Serum bupivacaine concentrations were similar on the first postoperative day, on the 3rd postoperative day measured only in the PVB group, on average = 1.3 mg/l but up to 3.5 mg/l in one patient. Toxic concentration is 2.6–4.6 mg/l (Rosenberg et al., 2004). Method of randomization was not reported. The authors conclude that TEA and PVB are equipotent and safe methods for post-thoracotomy analgesia, and that PVB should be the preferred analgesic approach for patients who do not qualify for or do not accept TEA. They presume PVB to be safe enough for use on surgical wards for up to a week, whereas their TEA method required ICU-monitoring, they kept their TEA patients in the ICU and therefore TEA was a more costly method than PVB in their hospital. They therefore suggested that PVB should be considered as the method of choice in all patients.

Comments: This is a study highly biased toward PVB, using bupivacaine 5 mg/ml with a vasopressor for almost 7 days on surgical wards, and 2.5–3.75 mg/ml without a vasopressor but with fentanyl 2 μg/ml for less than 3 days of TEA in the ICU. They were convinced that PVB was more effective and safer than TEA. Their TEA infusions have a too high bupivacaine and too low fentanyl concentration, not adding adrenaline. Thus, their TEA infusion was neither effective nor safe. They especially recommended PVB when TEA is contraindicated when haemostasis is disturbed. However, PVB infusions also enter the epidural space in most cases (Conacher and Kokri, 1987; Purcell-Jones et al., 1989). The risk of bleeding and infections in the spinal canal must be equal in the two catheter methods.

3.7.4 Richardson et al. (1999)

Richardson et al. compared the effects of preoperative and continuous paravertebral (PVB) and epidural (TEA) bupivacaine on pain, pulmonary functions, and stress responses after thoracotomy. Mostly lung surgery and a few oesophagastrectomy procedures were performed. Randomization was by sequential allocation of eligible patients via computer-generated random numbers. In the TEA group (49 patients), an epidural catheter was placed preoperatively at T7–T10 and 3 ml bupivacaine 5 mg/ml (0.5%) followed by a bolus of 10–15 ml bupivacaine 2.5 mg/ml (0.25%) before and up to 10 ml at chest closure. The PVB group (46 patients) preoperatively received a bolus of up to 20 ml bupivacaine 5 mg/ml (0.5%) at T6–T8 level, had a catheter placed under direct vision intraoperatively, through which a bolus of up to 20 ml bupivacaine 2.5 mg/ml (0.25%) was injected at chest closure. Postoperatively, bupivacaine 2.5 mg/ml (0.25%) for TEA and bupivacaine 5 mg/ml (0.5%) for PVB were infused at a rate of 0.1 ml/kg/h. All patients received diclofenac 50 mg every 8 h, orally or rectally, and i.v. PCA morphine 1 mg boluses with 5 min lockout time. Outcome variables were documented for 2 days: Pain was scored using VAS at rest and during coughing. Pain scores were low in both groups, but pain scores and cumulative morphine consumption were statistically lower in the PVB group. Peak expiratory flow rate was somewhat better preserved in the PVB group, and increases in plasma cortisol and glucose were less in the PVB group. Incidence of nausea, vomiting and postoperative respiratory morbidity was lower in the PVB group. Compared with the TEA group, the PVB group received about twice as much bupivacaine. The authors suggested that the observed difference in pulmonary function and nausea and vomiting might have been related to the higher consumption of rescue morphine in the TEA group. Three PVB patients developed temporary confusion, possibly due to bupivacaine accumulation. The authors concluded that TEA and PVB are effective for post-thoracotomy analgesia, and that PVB is superior for analgesia, pulmonary function, neuroendocrine stress responses, side effects, and postoperative respiratory morbidity.

Comments: This is a well-done study with a good number of patients. However, in this study as in all the other 9 studies, TEA is far from “optimal” thoracic epidural analgesia. Epidural catheters were sited below T7, and although they advanced the catheter 5 cm, one never knows in which direction the epidural catheter goes: It may curl up or turn down (caudad) as often as cephalad. Extent of segmental sensory losses to ice were checked, but not documented. The study may therefore have had a major, fatal flaw, in favour of PVB. The PVB catheters were always placed in segmentally optimal position, whereas the TEA catheters may have been far too caudad. This could explain the differences between the PVB and TEA in this study. Generalization to thoracic epidural analgesia as less effective and with more pulmonary complications than PVB cannot be made from this study.

3.7.5 Bimston et al. (1999)

Bimston et al. compared the efficacy and complications of post-thoracotomy PVB and TEA after lung surgery. 30 patients intraoperatively received a paravertebral catheter under direct vision with a loading dose of 18 ml bupivacaine 5 mg/ml (0.5%) with adrenaline 5 g/ml plus 2 ml fentanyl 50 μg/ml. 20 patients had a thoracic epidural catheter placed preoperatively. However, neither test dose nor loading dose nor level of insertion of the catheter is reported. One PVB patient, who had the catheter accidentally removed at the end of surgery, received an epidural catheter and was transferred to the TEA group. Both catheter groups received a solution of bupivacaine 1 mg/ml (0.1%) plus fentanyl 10 g/ml postoperatively, titrated to provide adequate analgesia, the infusion rates are not reported in detail but were 10–15 ml/h initially (although they write 10–15 ml/min, which must be an error, unless this was intended as a very rapid bolus!), gradually reduced to 5–10 ml/h after 48 h. They do not report figures, but it appears that the infusion rates were similar in the two groups. Additional analgesics were provided, the frequency of requests, which was equal between the groups, but not the actual drugs and their amounts are reported. Pain was measured with a VAS. Immediately postoperatively and at 40 h and later VAS-values were similar for the groups. Between 8 and 32 h after surgery the TEA group had a significantly lower VAS than the PVB group. Whether pain while coughing or pain at rest was evaluated is not reported. Less urinary retention, defined as Foley catheter replacement after initial discontinuation, was seen in the PVB group. However, they report that urinary catheters were removed on the 2nd postoperative day in the PVB group, whereas not until the epidural catheter was removed in the TEA group! In our studies, using optimal TEA, we routinely remove (successfully!) urinary catheters in the morning of the 1st postoperative day (Niemi and Breivik, 2003; Niemi, 2004). Other side effects, including hypotension, nausea, vomiting and pulmonary complications were equally distributed between the groups.

Randomization was “blindly, by random units table” method, but this resulted in 30 patients in the PVB group and only 20 in the TEA group. The authors conclude that both methods provide excellent pain control. However, they save US$ 500 on the patient’s bill by avoiding TEA, which is managed by the Acute Pain Service of the anaesthesia-department. The corresponding author, a surgeon, prefers to be independent of the pain service and manage post-thoracotomy pain on surgical wards by using PVB.

Comments: A correct method of randomization should always give an equal number of patients in treatment and control group, unless otherwise planned, e.g. a placebo-group can be smaller than an active treatment group, for ethical reasons. Their method of randomization is therefore suspect. They do not report segmental site of the TEA catheter, or any details of the infusion rates. They used a very large amount of fentanyl in both groups, and many of the observed side effects were undoubtedly due to systemic fentanyl-effects, not due to the block techniques. It is not a laudatory illustration of the US health care system, that the surgeons prefer the least effective analgesic technique because the anaesthesia department’s Acute Pain Service charges US$ 500 for providing safe epidural analgesia.

3.7.6 Dhole et al. (2001)

Dhole et al. compared TEA and PVB for postoperative analgesia after minimally invasive direct coronary artery bypass (MIDCAB) performed through a minithoracotomy. Data from 20 patients in each group are reported. An epidural catheter was placed at T4–T5 level in the TEA group, the same level was used for percutaneous placement of catheter in the PVB group. Both groups received a 3 ml test dose of lidocaine 20 mg/ml (2%), followed by a bolus of 8 ml bupivacaine 5 mg/ml (0.5%) and a 6 ml/h infusion of bupivacaine 2.5 mg/ml (0.25%). Segmental spread of needlestick-hypoaesthesia confirmed correct catheter placement. Rescue analgesia was i.m. ketorolac 30 mg. Respiratory rate was significantly lower in the PVB group. VAS at rest and during coughing was not different. No difference in consumption of rescue analgesia with ketorolac was observed. No complications were seen in the PVB group, in the TEA group, one patient suffered transient hypotension, another pain at the site of catheter insertion. Cardiac index was higher in the TEA group. Method of randomization is not reported. The authors concluded that PVB is as effective as TEA for post-thoracotomy analgesia in MIDCAB patients, and that PVB may be safer than TEA.

Comments: Although no bleeding in the spinal canal or paravertebral space was observed, the heparin regimen during CABG and NSAID for rescue analgesia may disturb primary and secondary haemostasis enough so that TEA is unsafe in these patients. However, the paravertebral space is part of the epidural space, making the infrequent, but serious bleeding and infectious complications in the epidural space a possibility during and after PVB, as with traditional TEA.

3.7.7 Luketich et al. (2005)

Luketich et al. studied the data from 91 lung surgery patients to compare post-thoracotomy TEA with paravertebral analgesia. Preoperatively, the TEA group (44 patients) had a catheter inserted at the T3–T6 level, which was tested with a 3 ml lidocaine 15 mg/ml (1.5%) plus adrenaline 5 μg/ml, followed by an infusion of bupivacaine 1.25 mg/ml (0.125%) plus morphine 50 g/ml, 4–8 ml/h. No TEA loading dose is reported. The paravertebral group (47 patients) received 10 ml bupivacaine 2.5 mg/ml (0.25%) percutaneously through a needle before thoracotomy, followed by an under direct vision placement of a paravertebral catheter at T8–T9 level by the surgeon. Another loading dose—10 ml of bupivacaine 5.0 mg/ml (0.5%) was given before chest closure. Postoperatively the PVB group received bupivacaine 2.5 mg/ml (0.25%) 0.1 ml/kg/h plus i.v. morphine patient controlled analgesia (PCA) for 4–6 h. Pain was reported by a composite score and no difference between the groups was identified. Whether dynamic pain or pain at rest was assessed was not specified. Additional analgesics were provided; the exact amounts and drugs are not reported. Complications, including pneumonia, were similar between the groups. Foley catheter days were significantly fewer in the PVB group, definition of need for Foley catheter is not reported. Method of randomization is not reported. The authors concluded that satisfactory pain control was achieved with both regimens, and they recommended PVB plus i.v. PCA for patients in whom epidural catheter placement is not feasible.

Comments: The PVB group received about twice as much bupivacaine as the TEA group, but no opioid in their postoperative infusion. Their “composite pain score” is unfortunately only an average of three measurements of pain intensity scores and therefore not a composite score—at all. A meaningful composite pain score must comprise more variables than merely pain intensity scores, e.g. a pain intensity score and the amount of rescue analgesic consumed (Silverman et al., 1993; Romundstad et al., 2006). The surgeons express preference for PVB because “daily consulting fees to the Acute Pain Team” are avoided and no time is lost in the operating room.

3.7.8 Casati et al. (2006)

Casati et al. compared the analgesic efficacy of continuous PVB and TEA for post-thoracotomy pain for 2 days after lung surgery with “blinded” observers. Each group included 21 patients. The TEA group received a T5–T7 level catheter and a loading dose of 5 ml ropivacaine 7.5 mg/ml (0.75%). As loading, the PVB group had three paravertebral injections, each 5 ml ropivacaine 7.5 mg/ml (0.75%), into T4, T5 and T6 level. Thereafter, a PVB catheter was percutaneously inserted at the T4 level. Postoperatively, both groups received paracetamol 1 g every 8 h plus ropivacaine 2 mg/ml (0.2%) continuously. Infusion rate was 5–10 ml/h, adjusted to maintain pain score on a visual analogue scale (VAS) ≤4/10. Also, i.v. morphine 5 mg was provided as rescue analgesic to maintain VAS ≤4/10. VAS was scored at rest and during coughing. For total postoperative ropivacaine volume, amount of rescue analgesicand VAS at rest and when coughing, no difference was found between the groups. Hypotension, defined as >30% reduction in systolic blood pressure, was more frequent in the TEA group. No severe complications occurred in either group. For randomization, computer-generated sequence and sealed envelopes were used. The authors concluded that PVB is as effective as TEA for analgesia, but has less haemodynamic side effects and is an alternative to TEA for post-thoracotomy pain control.

Comments: This study is one of 3 of the 10 included studies that placed the PVB-catheter percutaneously, 2.5 cm lateral to the spinous process, below the transverse process. A Tuohy needle was advanced 1 cm deeper than first contact with the transverse process and catheter inserted 2–3 cm beyond the tip of the needle. This allowed blinding of the outcome variableobservers, as type of analgesic block could not be identified. However, in our setting bilateral upper and lower segmental levels (hypoaesthetic areas) are documented regularly, thus any “blinding” of the postoperative observers of effects would be impossible. They have randomized properly, and observed dynamic pain during coughing. We regret the use of a too concentrated local anaesthetic only, without opioid and adrenaline, for the TEA. An “optimal” three component epidural mixture would have avoided hypotension and urinary retention (Niemi and Breivik, 2002, 2003).

3.7.9 Mehta et al. (2008)

Mehta et al. compared continuous TEA and PVB for postoperative analgesia after robotic-assisted coronary artery bypass surgery accompanied by minithoracotomy. The TEA group (19 patients) had an epidural catheter inserted at C7–T1 level. The PVB group (17 patients) had a catheter inserted percutaneously at T4–T5 level into to the left PVS. Both groups received a catheter test dose of 3 ml lidocaine 20 mg/ml (2%), followed by a bolus 8 ml bupivacaine 5 mg/ml (0.5%) and an infusion of bupivacaine 2.5 mg/ml (0.25%) 0.1 ml/kg/h. VAS was scored at rest and during coughing, and diclofenac sodium 75 mg i.m. was given on patient demand or at VAS >5/10 at rest. No significant differences in pain scores, consumption of rescue analgesics, mean time to extubation, haemodynamic or respiratory variables were found. Two patients in the TEA group experienced upper limb numbness, which recovered after stopping the infusion of local anaesthetic. Method of randomization is not reported. The authors concluded that PVB appears to be safe, effective and comparable to TEA for postoperative analgesia after robotic-assisted coronary artery bypass surgery. They consider PVB safe despite recent anticoagulation.

Comments: Same solution and regimen used for PVB and TEA, both placed percutaneously, with blinded observers. TEA catheter placed much too high (at C7–T1), causing numbness of arms in 10% of patients. PVB-catheters were placed at appropriate level T4–T5. Combination of heparin for LIMA and diclofenac rescue analgesic potentially increases risk of spinal bleeding—in PVB as well as TEA. Again we register that TEA would have performed better with lower concentration of bupivacaine (1 mg/ml instead of 2.5 mg/ml) with fentanyl (2 g/ml), and adrenaline (2 g/ml) (Niemi and Breivik, 2002, 2003).

3.7.1 Gulbahar et al. (in press)

Gulbahar et al. compared TEA (19 patients) and PVB (25 patients) for post-thoracotomy analgesia after lung surgery. Data from 13 patients in the TEA group are reported. Six were excluded due to catheter misplacement or too early removal. TEA catheter at T7–T10 level was inserted preoperatively and 5 ml bupivacaine 2.5 mg/ml (0.25%) injected at chest closure. The PVB group included 25 patients who had their catheter placed at thoracotomy level under direct visual guidance prior to chest closure. Postoperatively, both groups received 0.10 ml/kg/h bupivacaine 2.5 mg/ml (0.25%) plus patient controlled bolus of 2 ml with a lock out time of 1 h. Morphine was provided for rescue analgesia. No significant differences between the groups were found for VAS, morphine consumption, attempted PCA boluses, or respiratory variables. Urinary retention, nausea, vomiting, and hypotension were significantly more frequent in the TEA group, the authors suggested this could be attributed to the epidural administration of too concentrated local anaesthetic. No side effects were noted in the PVB group. Method of randomization is not reported. The authors concluded that the methods are equally effective for post-thoracotomy analgesia. They concluded that PVB is preferable as catheter placement procedure was quicker and PVB had fewer adverse effects postoperatively.

Comments: The TEA catheters were sited much too low (below T7) for relieving pain after thoracotomy, explaining part of the problems with patients randomized to TEA. We are not informed of whether pain scores were assessed at rest or when coughing. The epidural infusion contained too high concentration of bupivacaine, no fentanyl, and no adrenaline. Therefore this was not a fair comparison of thoracic epidural analgesia with paravertebral block. Still they did not find any difference in effect on pain during 3 days following thoracotomy. We agree with the authors that the higher incidence of adverse effects in the TEA group was due to too high concentrations of local anaesthetic, as well as too low segmental placement of the epidural catheters. Unequal numbers of patients in the two groups indicate that the, unreported, method of randomization was not appropriate.