We present 3 cases of persistent cerebrospinal fluid (CSF) leak after combined spinal-epidural (CSE) anesthesia and postoperative epidural analgesia for elective cesarean delivery, which occurred over a 3-mo period in our obstetric unit. The CSE block was performed at the L3-4 interspace, with the parturient in the sitting position, and in each case appeared uncomplicated. The CSE kits used were those routinely available in our unit: a 16/26 gauge (G), 18/27 G (both Sims Portex, Kent, UK), and an 18/27 G (Abbott, Sligo, Republic of Ireland) kit. The epidural catheters were secured with transparent, occlusive dressings (Tegaderm®; 3M, St. Paul, MN).
Postoperatively, all three parturients received pethidine patient-controlled epidural analgesia (PCEA), our routine method of postcesarean delivery analgesia. This is achieved using a 60-mL disposable PCEA pump that delivers 4 mL of 5 mg/mL pethidine per bolus with a nominal 15-min lockout interval (Go Medical, Perth, Australia). In all 3 patients, use of PCEA was typical, with 60–120 mL of solution self-administered over a 24-h period and the epidural catheter removed 24–36 h postoperatively. Postoperative analgesia, epidural catheter position, and the epidural insertion site were checked daily by the Acute Pain Service.
A 36-yr-old nulliparous woman presented for an elective cesarean delivery for breech presentation at 38 wk gestation. She was a cigarette smoker but had no other relevant medical or obstetric history. Lumbosacral edema was noted during insertion but there was no evidence of preeclampsia. Surgery was uneventful. The first indication of a fluid leak was when she reported a “wet back” 2 days after surgery. Fifty milliliters of clear fluid had collected under the epidural dressing and her epidural catheter was removed. A hole of the size consistent with that of a 16-G epidural needle was noted at the insertion site and was weeping clear fluid. Laboratory analysis of electrolyte, glucose, and protein content of the fluid showed the composition was consistent with CSF.
The patient did not develop a postdural puncture headache (PDPH), fever, meningism, or neurological signs. She had no difficulty ambulating or nursing her newborn baby. However, based on neurosurgical advice, her skin puncture site was closed with 3-0 nylon sutures using local anesthesia, under sterile conditions in the operating room. The fluid leak was attenuated but persisted, so on the fifth postoperative day she was transferred to a neurosurgical unit for further suturing of the fistulous tract by the neurosurgeons. No antibiotics were given because she remained afebrile and had no signs of infection. She was discharged from the hospital on day 6 with no further leak or other sequela.
A 38-yr-old multiparous woman at 37 wk gestation presented for elective cesarean delivery for breech presentation. She had a history of asthma and preeclampsia. There was no reported technical difficulty with CSE insertion and surgery was uneventful. On the third postoperative day, a 3- to 5-mL fluid collection was noted under the dressing over the epidural site. The patient was afebrile, had no PDPH or neurological signs and symptoms. A 2-mL sample of this fluid was tested in the laboratory for β2-transferrin. β2-Transferrin was detected in the sample at a concentration strongly suggestive that the fluid was CSF. The epidural puncture site was closed with a 3-0 nylon suture and no further fluid leak occurred.
A 31-yr-old multiparous woman at 32 wk gestation, with twins showing discordant growth, presented for elective cesarean delivery. There was no reported technical difficulty with CSE insertion and surgery was uneventful. On day 3, a 2- to 3-mL collection of fluid was noted under the epidural site dressing. This fluid tested positive for β2-transferrin. The epidural puncture site was closed with 3-0 nylon and there was no subsequent fluid leak.
At discharge, each of the parturients was advised to return if symptoms developed and they were subsequently reviewed by telephone. The skin sutures were removed by a community midwife 5 days after insertion and no complications occurred.
We present three parturients with evidence of persistent fluid leak after CSE anesthesia for elective cesarean delivery, followed by postoperative epidural analgesia. This fluid contained, or was, CSF, the diagnosis being confirmed in two cases by β2-transferrin immunofixation assay. None of these women developed complications of CSF leak such as PDPH, fever, meningism, or neurological pathology. All were able to care for their newborn babies.
Our tertiary institution has performed approximately 1300 obstetric CSE techniques per year since 1993 and these cases are the first in which an apparent persistent leak of CSF has been diagnosed. The large volume of fluid consistent with CSF in the first case was surprising and this may have heightened our threshold for testing of smaller fluid leaks, because the next two cases occurred within three months of each other. We have been unable to find similar reports of CSF leak in this setting. Persistent CSF leak is not a normal event after neuraxial block and may be caused by fistula formation (1–5). Multiple attempts at dural puncture or epidural placement are considered a risk factor on the basis of fibrin deposition maintaining the patency of the fistulous tract (2,3). However, none of our cases had evidence of dural puncture by the epidural needle or of intrathecal catheterization. Persistent CSF leak suggests continuity is maintained between the subarachnoid space and the skin insertion site. In our first case, we postulate that lumbosacral edema may have prevented the normal skin elastic memory from closing the puncture site, but we have no explanation as to why fluid leak persisted in cases two and three.
We consider it unlikely that the type of CSE kits used contributed to the risk of persistent CSF leak, because all are in routine use in Australia. There is also no evidence in the literature that use of a 16-G epidural needle is associated with an increased risk of PDPH or persistent CSF leak after accidental dural puncture compared with an 18-G needle.
Equally, it is surprising that a small-gauge pencil-point spinal needle might cause a persistent CSF leak, and it is possible that there had been an undiagnosed dural puncture with the Tuohy needle. In approximately 30% of parturients with PDPH after epidural insertion, insertion is documented as uneventful (6). Against this possibility is the fact that none of our parturients experienced PDPH.
These cases raise several questions, for example, does the epidural catheter sometimes impede the dural inflammatory response, preventing normal closure and healing of the puncture site, and does injection of a large volume of epidural solution influence fistula formation? The large volume of fluid noted in our first parturient was alarming, but we suspect that pethidine solution contributed substantially to the amount. It is possible that only a small amount of CSF was actually lost, or that the laboratory results suggesting the presence of CSF represented a false-positive test.
In 1994, Howes and Lenz (2) prophesied that, with the increasing popularity of CSE, “CSF fistulae may become more common…particularly when they are performed through the same interspace” because of “presence of an epidural catheter close to the site of the dural puncture.” However, Rawal et al. (7) suggest that many factors associated with CSE reduce the likelihood of PDPH, presumably by minimizing CSF leak. These include the use of fine spinal needles (26–32 G); meticulous dural penetration because the Tuohy needle acts as an introducer; deflection of the spinal needle by the Tuohy needle tip leading to reduced overlap of the dura mater and arachnoid mater; and the effect of epidural opioid. Some studies suggest PDPH rates are less with CSE than epidural techniques and that unintended dural puncture is less frequent (7,8).
Given the absence of reports, we believe that persistent CSF leak after CSE anesthesia must be very rare. However, it may also be that the leakage of small volumes of CSF is under-diagnosed or under-reported. We routinely use transparent, occlusive dressings to cover epidural insertion sites to aid the ease of inspection. This approach may allow fluid to collect under the dressing, where it is easily detected. In the absence of a transparent dressing and vigilant nursing staff, a small CSF leak along the insertion track after its removal could easily be missed or the extravasating fluid ascribed to analgesic solution or edema, other possibilities in the differential diagnosis (9,10).
In our latter two cases, β2-transferrin immunofixation assay confirmed the presence of CSF. This assay almost unequivocally identifies the presence of CSF and can be performed on very small samples of fluid (<0.1 mL). β2-Transferrin is an iron transport protein found only in CSF, aqueous humor, and perilymph (11). If the β2-band is present in the electrophoretic gel, in addition to the β1-band from tissue fluid, then CSF is present (Royal Perth Hospital protein laboratory communication). The sensitivity and specificity of this assay depends on dilution and contamination of the CSF sample by other fluids. Clinical use of the β2-transferrin assay has previously been restricted to detection of CSF rhinorrhea (12). A more rapid, sensitive, and specific assay for CSF, detecting β-trace protein, has also been described (13).
Given an apparent lack of experience with this event, the optimal management of persistent CSF leak from an epidural insertion site is uncertain. The rationale for skin closure was to reduce the risk of meningitis, although there is no evidence to support this intervention. There is also no evidence on which to base recommendations such as epidural blood or fibrin glue patching, or expectant treatment with observation.
The natural history of this complication is unknown and if the course is one of spontaneous closure of the fistulous track and a benign outcome, it is possible that patients with undiagnosed small CSF leaks are currently being discharged without consequence. We recommend testing for β2-transferrin or β-trace protein, to confirm or refute the presence of CSF, when persisting fluid leak occurs from the insertion site after neuraxial block. This may guide diagnosis, help to determine the prevalence of this complication, and assist with decision-making about subsequent management. The pathophysiology and optimal management of the condition, however, remain to be determined.
We thank Dr. R. Scarff of Royal Perth Hospital Department of Biochemistry for his advice regarding diagnostic tests for cerebrospinal fluid.
1. Hullander M, Leivers D. Spinal cutaneous fistula following continuous spinal anesthesia. Anesthesiology 1992; 76: 139–40.
2. Howes J, Lenz R. Cerebrospinal fluid cutaneous fistula: an unusual complication of epidural anaesthesia. Anaesthesia 1994; 49: 221–2.
3. Priestley GS, Brighouse D. Cerebrospinal cutaneous fistula. Anaesthesia 1994; 49: 744.
4. Holden S, Hedley RM, Wheilden M. Cerebrospinal fluid cutaneous fistula. Anaesthesia 1994; 49: 836–7.
5. Joseph D, Anwari JS. Cerebrospinal fluid cutaneous fistula after labour epidural analgesia. Middle East J Anesthesiol 2001; 16: 223–30.
6. Paech M, Banks S, Gurrin L. An audit of accidental dural puncture during epidural insertion of a Tuohy needle in obstetric patients. Int J Obstet Anesth 2001; 10: 162–7.
7. Rawal N, Holmstrom B, Crowhurst JA, Van Zundert A. The combined spinal-epidural technique. Anesthesiol Clin North America 2000; 18: 267–95.
8. Norris MC, Grieco WM, Borkowski M, et al. Complications of labor analgesia: epidural versus combined spinal epidural techniques. Anesth Analg 1994; 79: 529–37.
9. Kirby IJ, Ryan TDR. Generalised oedema and epidural anaesthesia. Anaesthesia 1985; 40: 709–10.
10. Douglas J, Choi D. Immediate complications of regional blockade in obstetrics. In: Reynolds F, ed. Regional analgesia in obstetrics: a millennium update. London: Springer-Verlag, 2000: 285–306.
11. Reisinger PWM, Hochstrasser K. The diagnosis of CSF fistulae on the basis of detection of beta2
-transferrin by polyacrylamide gel electrophoresis and immunoblotting. J Clin Chem Clin Biochem 1989; 27: 169–72.
12. Marshall AM, Jones NS, Robertson IJ. An algorithm for the management of CSF rhinorrhea illustrated by 36 cases. Rhinology 1999; 37: 182–5.
13. Petereit HF, Bachmann G, Nekic M, et al. A new nephelometric assay for beta-trace protein (prostaglandin D synthase) as an indicator of liquorrhoea. J Neurol Neurosurg Psychiatry 2001; 71: 347–51.