Obstetric epidural catheter-related infections at a major teaching hospital: a retrospective case series

Introduction 

Epidural abscess and deep tissue infection after epidural catheter insertion are rare but serious complications, and may require surgical intervention and long-term antibiotic therapy and lead to permanent dysfunction or persisting pain. Spontaneous epidural infections without epidural puncture or epidural catheterization are reported at a rate of 2 per 10000 hospital admissions per year.1 The reported incidence of epidural abscess in the obstetric population varies widely, ranging from 0.2 to 48 per 100000 obstetric epidural procedures.2, 3, 4, 5, 6, 7

A retrospective analysis of prospectively collected data about complications of obstetric epidural analgesia and anaesthesia, conducted at our institution between July 1989 and August 1994 and involving 10995 cases, found no cases of superficial or serious deep tissue infection or abscess formation.8 However, a number of cases were known to have occurred in our hospital during the past decade, and there has been an increasing number of case reports of epidural-related infection in recent years.9

This retrospective study was designed to estimate the incidence and nature of epidural catheter-related infections in a large maternity hospital in Australia. We found four cases of serious infection in obstetric patients, all of whom had an epidural catheter inserted under full aseptic conditions, including the operator wearing a cap, mask, gown and sterile gloves, after repeated skin application of 0.5% chlorhexidine in 70% alcohol and after application of a transparent dressing.

Methods 

We conducted a retrospective chart review of epidural complications that occurred between January 2002 and December 2005 in our institution, King Edward Memorial Hospital for Women. The total number of births, elective caesarean sections and obstetric epidural techniques performed during this period was obtained from the hospital perinatal database (‘Stork’). The hospital has a tertiary referral maternity unit servicing a local metropolitan population of approximately 1.5 million and a rural population within the entire state of Western Australia (size approximately 2.5 million square kilometres) of approximately 500000.

Patients with complications were identified using the International Statistical Classification of Diseases (ICD) code 089.5. According to the International Statistical Classification of Diseases and Related Health Problems, 10th Revision, Australian Modification (ICD-10-AM), produced by the National Centre for Classification in Health (NCCH), code 089.5 is used to identify patients who have had complications of spinal and epidural anaesthesia during the puerperium. This code also identifies those women re-admitted to hospital with an epidural complication.

In each case data collected from the medical record included patient demographics and characteristics, the reason for insertion of the epidural catheter, the duration of epidural catheterization, potential risk factors for infection, and the signs and/or symptoms indicative of infection. Epidural catheter related infection was diagnosed initially on clinical grounds and usually included the presence of two or more of the following signs or symptoms: back pain, epidural insertion site tenderness, redness or purulent discharge, a suggestive neurological symptom or radicular pain, or fever. All patients with infective signs or symptoms were reviewed by the department of Anaesthesia and Pain Medicine and a management plan was established. The criteria for magnetic resonance imaging (MRI) were the presence of a neurological sign or symptom and increasing back pain or radicular pain despite antibiotic treatment.

Throughout the study period it was departmental policy for the operator to wear a cap and mask, wash hands, put on a sterile gown and gloves, to prepare the skin with at least two applications of 0.5% chlorhexidine in 70% alcohol, to cover the site with a sterile drape and to cover the insertion site and catheter with a sterile transparent dressing.

Continuous data were summarized using medians, interquartile ranges (IQR) and ranges. Categorical data were summarized using frequency distributions. Univariate analysis included Mann-Whitney tests for continuous data and χ2 or Fisher’s exact tests for categorical comparisons. SPSS 15.0 (SPSS Inc, Chicago IL) and Stata 8.2 statistical software were used for data analysis. All hypothesis tests were two-tailed and a P value <0.05 was considered significant.

Results 

During the study period, 17947 women gave birth (average 4487 deliveries per year). Of these women, 9482 (52.8%) had an epidural catheter inserted: 2727 for elective caesarean section and 6755 for labour analgesia.

A total of 258 patients classified under code 089.5 were identified and the medical records reviewed. Forty-nine cases of infection considered to be related to epidural placement were identified (Table 1). This represents an incidence of epidural-related infection of 0.52% (95% CI 0.37-0.66%) or 1 in 194. Most were superficial and minor skin infections, but four women (0.04%, CI 0.01-0.11%; 1 in 2371) had deep tissue infection, two diagnosed as epidural abscess (0.02%, CI 0-0.08%; 1 in 4741) and two as paraspinal abscess. There were no cases of meningitis.

Table 1. Numbers of women who had an epidural catheter-related infection compared with the number giving birth and the number of epidural techniques performed

	Year	Women giving birth (n)	Women having an epidural technique (n)	Superficial infection (n)	Deep infection (n)
					Epidural abscess	Paravertebral abscess
	2002	4299	2180	8	1	0
	2003	4363	2248	14	0	0
	2004	4444	2367	6	1	2
	2005	4841	2687	17	0	0
	Total	17947	9482	45	2	2

The median age of the 49 women with infection was 30 (IQR 27-35) years and their median body mass index (BMI) 31kg/m2 (IQR 28-35). Twenty-eight (57%) had a BMI>30kg/m2 and 10 (20%) a BMI >40kg/m2. The insertion was rated as difficult in 13 cases and 11 women had risk factors for infection, mainly immunosuppression.

Thirty-three of the 49 women (67%) had a caesarean delivery (either elective or non-elective) and had the epidural catheter retained for postoperative pain relief (median duration 43h [IQR 30-48]). In 16 women (33%) the epidural catheter was used only during labour, for a median duration 11h (IQR 5-22). The median time between insertion and the onset of infective symptoms and/or signs was 55h (IQR 32-73). There were differences in the initial presentation of superficial infection and deep infections (Table 2). Among the four cases of deep tissue infection, all described back pain, two had deep tenderness, two had a focal neurological sign but none were febrile. The medical records of one patient were lost and the remaining three cases are described in more detail.

Table 2. Symptoms and signs at presentation of epidural catheter-related infection

	Symptoms/Signs	Superficial infection (n=45)	Deep infection (n=4)
	Back pain	19 (42%)	4 (100%)
	Epidural site redness	32 (73%)	1 (25%)
	Epidural site tenderness	31 (69%)	2 (50%)
	Epidural site localized swelling	8 (18%)	1 (25%)
	Epidural site purulent discharge	6 (13%)	1 (25%)
	Focal neurological sign	2 (4%)	2 (50%)
	Fever	6 (13%)	0

Case 1 

A healthy 29-year-old primigravid woman weighing 82kg experienced an uncomplicated pregnancy except for recurrent urinary tract infections. She was admitted with spontaneous onset of premature labour and breech presentation at 35weeks of gestation. Initially, labour analgesia was provided with a needle-through-needle combined spinal-epidural technique, performed without difficulty at the L3-4 interspace using a 16-gauge Tuohy needle and 27-gauge Whitacre spinal needle. Shortly thereafter she underwent an uneventful non-elective caesarean section under epidural anaesthesia, during which she received a single i.v. dose of cefazolin 1g for prophylaxis against wound infection. The epidural catheter remained in situ postoperatively for pethidine patient-controlled epidural analgesia. Approximately 12h postoperatively she was reviewed by the acute pain service (APS), who noted a small collection of fluid under the dressing (Tegaderm™, 3M Company, Maplewood, USA) at the epidural insertion site. This fluid was released and a new sterile dressing applied, but approximately 24h after epidural catheter insertion the nursing staff noted that the new dressing had been breached and the skin site exposed. The epidural catheter was removed and the following day the APS inspected the site, which appeared normal. She was discharged from hospital on the fifth postoperative day.

Three weeks later the patient presented at the emergency department, complaining of increasing back pain and leg weakness that interfered with ambulation as the day progressed. She was afebrile and had a normal leucocyte count and C-reactive protein (CRP) level. Examination revealed slight swelling and tenderness at the epidural site, no demonstrable motor weakness in either limb, but decreased sensation in an L4-5 dermatome distribution on the right leg. MRI performed the same day indicated enhanced pre-fascial uptake in the paraspinal muscles. On microbiological advice she was given a single 1-g i.v. dose of ceftriaxone and regular i.v. flucloxacillin 2g 6-hourly. Subsequently, blood cultures and screening for multiresistant Staphylococcus aureus proved negative.

After 5days she continued to experience the same mild symptoms but as her examination and all laboratory tests remained normal, she was discharged from hospital, changing to oral flucloxacillin for a further two weeks. On outpatient review after a week she remained systemically well and active, her leg symptoms had resolved and on deep palpation the epidural site was slightly tender. She failed to attend another appointment one month later.

Case 2 

A healthy 28-year-old G2 P1 woman weighing 86kg, with a past history of repeat ovarian cystectomy for benign cysts, presented in spontaneous labour at term with a breech presentation. She had been booked for operative delivery so had a non-elective caesarean section and at the same time another ovarian cystectomy. The operation was performed under combined spinal-epidural anaesthesia, established by insertion of a 16-gauge Tuohy needle and 27-gauge Whitacre spinal needle at the L3-4 interspace. The insertion was technically easy and the 2.5-h surgical procedure uncomplicated. She was given i.v. cefotetan 2g according to hospital infection control guidelines. Postoperatively she received pethidine patient-controlled epidural analgesia for 48h, with good effect. On review at this time by the APS it was noted that the epidural site appeared normal but was tender and the epidural catheter was removed. The next morning, on postoperatively day 3 the APS noted an epidural site infection. Purulent discharge was expressed from a 5-cm swollen subcutaneous area, microbiological swabs were taken and i.v. flucloxacillin 2g 4-hourly started. The patient was afebrile, systemically well and without neurological symptoms or signs, although on examination altered sensation to light touch was detected on the dorsal aspect of the right foot. MRI showed abnormal enhancement of the right paraspinal muscles and a 6-mm anterior-posterior diameter epidural abscess in the posterior space at L3-4, with no mass effect. Gram positive cocci were seen on microscopy (subsequently identified as Staphylococcus aureus) and after microbiological and neurosurgical consultation her antibiotic therapy was changed to i.v. merepenum 1g 8-hourly, i.v. vancomycin 1g 12-hourly and oral rifampacin 600mg daily. Within 12h, however, she deteriorated clinically, experiencing worsening back pain, fever to 38.5°C, meningism, photophobia and increased swelling of the epidural skin site. After further blood cultures she was transferred to another facility with neurosurgical services, for continued observation. Her CRP was 81mg/L (normal range <15mg/L) and leucocyte count 13.9×106/L. A peripherally inserted central catheter was placed and i.v. flucloxacillin recommenced. Over the next 11days she made a full recovery. Antibiotic therapy was changed after 8days to a 3-week course of oral rifampicin and i.v. teicoplanin 400mg daily. She was booked for further MRI 6weeks later but after discharge was lost to follow-up.

Case 3 

A 20-year-old G2 P0 woman who had her antenatal care at another hospital was admitted in early labour. She had a history of asthma, epilepsy, anxiety and depression, urinary tract infections and smoking (6 pack-years). The reason for admission at our tertiary obstetric centre was that she was morbidly obese, with a BMI of 51kg/m2. At 3cm dilatation she requested epidural analgesia and an epidural catheter was inserted without difficulty at the L3-4 interspace using a 16-gauge Tuohy needle. She had several hours of effective patient-controlled epidural analgesia during labour before a spontaneous vaginal delivery. The epidural catheter was removed intact approximately 10h after insertion and she was transferred back to her parent hospital.

Based on the limited information available to us, she experienced increasing back pain and difficulty ambulating over the next few days. Epidural abscess was diagnosed and she was transferred to another hospital with neurosurgical facilities, where on the seventh postpartum day she underwent decompressive laminectomy at L3. Frank pus was drained from an encapsulated epidural abscess and antibiotic therapy was continued for six weeks.

She returned to our hospital two years later with a twin pregnancy. At the antenatal anaesthesia clinic she complained of chronic mild back pain with lifting and bending and was counselled about analgesia and anaesthesia options for labour and delivery. Subsequently, she underwent induction of labour at 36weeks of gestation and had a vaginal delivery of each twin, having used fentanyl patient-controlled intravenous analgesia for pain relief. At her next pregnancy two years later, she again had vaginal delivery of twin fetuses, on this occasion using remifentanil patient-controlled i.v. analgesia.

Discussion 

In this retrospective study of women delivering in a single maternity hospital over a four-year period, we found epidural catheter-related infection at a rate of 1 in 200 catheter insertions and four cases of deep tissue infection, at a rate of approximately 1 in 2500. Clinically overt infections of the skin at the epidural insertion site have been reported to occur in 0.8-6% of non-obstetric patients after a few days of catheterization,10, 11, 12, 13 and the rate of superficial infection of 0.5% in our study was consistent with this range.

Although deep tissue infections arise through a number of mechanisms,9, 14 local skin infections at the catheter entry site may progress to subcutaneous, epidural, paravertebral, paraspinal or meningeal infections. Nevertheless these deep tissue infections are rare and the reported incidence of epidural abscess in the obstetric population specifically is 0.2-48 in 100000.2, 3, 4, 5, 6, 7 Estimates at the lower end of this range are likely to be significant underestimates, being based on surveys and estimates from the literature or where under-ascertainment was likely due to restricted follow-up. More reliable estimates come from prospective studies. Wang et al. prospectively evaluated a one-year period in Denmark in which 17372 epidural catheter placements were reported and the incidence of epidural abscess was 1 in 1930,15 but there did not appear to be any obstetric cases. Some single-centre studies have reported similar rates in non-obstetric populations13, 16, 17 and obstetric populations.5 Our results support an incidence of neuraxial infection at the upper end of the previously reported range. This may in part be due to better surveillance, but may also reflect an apparent increase in the frequency of this complication in recent years. One author (MP) is not aware of any cases of epidural site infection in our hospital during the period 1988 to 1995 when a prospective audit was in place. Although our study is retrospective, our APS follows all women having an epidural technique until discharge from hospital. In addition, all patients are informed about the signs and symptoms of epidural-related infections, provided with written information and contact forms, and are advised to contact the hospital or their general practitioner if any such symptoms or signs arise. Follow-up home visits by the visiting midwifery service were also documented in the medical record. Further, as this is the only tertiary obstetric hospital servicing a population of approximately 2 million, most serious obstetric epidural-related complications are referred to us. On the other hand, it is also possible that we did not have complete ascertainment of all cases of deep tissue infection, because these infections often present days or weeks later,9 (for example case 1 above) by which time they may present or be admitted to another hospital.

Epidural infections arise spontaneously as often as 2 per 10000 hospital admissions per year,1 but haematogenous spread may not be the most frequent mechanism of epidural infection related to epidural techniques. Contamination of epidural equipment or solution is possible, and skin bacteria can gain access to the epidural space along the catheter insertion track.9 The latter mechanism is supported by a study that traced the source of an epidural abscess to Staphylococcus aureus strains from the patient’s skin flora,18 as well as the finding that 83% of epidural catheter-related infections are caused by various Staphylococcus species, typical of skin flora.9, 19, 20 As skin bacteria are the main source of epidural-related infections, close attention to all details of an aseptic procedure during insertion is of primary importance, but may be poorly performed.21 The importance of an aseptic technique that includes wearing of a mask is highlighted in a survey by Sprigge and Harper, which described a case of meningitis due to a pharyngeal organism following spinal anaesthesia for caesarean delivery, where the anaesthetist had not been wearing a mask during insertion.22 Reynolds believes that in the face of all the evidence, the wearing of masks is mandatory for spinal and epidural insertion.23

Daily monitoring of the catheter insertion site for signs of infection should be mandatory and, because presentation is often delayed, increasing patients’ and medical practitioners’ awareness of this possible diagnosis may also be valuable.

Local skin infection and lack of hygienic discipline in handling epidural catheters are likely risk factors for epidural infection, as are patient factors that impair immunological defence, such as malignancy, diabetes mellitus, renal impairment, steroid therapy and alcohol or intravenous drug abuse.1, 9, 13, 19, 20 In our series, three of 49 patients had gestational diabetes and one of these also had a history of intravenous drug use, but we cannot comment on risk factors in this study because we do not have comparative data from the non-infected population.

Many patients who develop an epidural abscess have received prophylactic doses of heparin before epidural catheterization13, 15 and it is possible a small epidural haematoma may predispose to abscess formation, but such findings do not establish a causal relationship.

Provided rigorous skin asepsis is used, an increase in the number of skin passes or technical difficulty with epidural catheter insertion does not appear to be a risk factor for an infectious complication.24 Loo et al.2 recommended the routine use of bacterial filters and that the duration of epidural catheterization in obstetric patients should be limited to less than 24h. Although epidural analgesia after caesarean section is well accepted practice, a factor that might explain the incidence noted in this study was the duration of catheterization, which exceeded 24h in the majority of this cohort. It is known that bacterial colonisation of the skin insertion site increases significantly after 48h and that the risk of superficial infection increases with the duration of catheterization.11, 25 One of our patients developed a superficial infection after a filter disconnection. The catheter tip grew coagulase-negative Staphylococcus and Enterococcus. Although it has been suggested that epidural catheters may be retained after filter disconnection,26, 27 one author (MP) is aware of a local case of epidural abscess after a catheter disconnection and we consider the catheter should always be removed after such an event.

Some signs and symptoms of epidural-related infection are common in the peripartum period, making the clinical diagnosis of deep tissue infection more difficult. Back pain is one of the commonest complaints in pregnancy and the puerperium. Local tenderness after epidural insertion is usually due to minor tissue trauma and temporary. In one review the classic triad of epidural abscess, namely back pain, fever and variable neurological deficit, was present in only 13% of patients at the time of diagnosis,9 and our findings support the lack of consistency of presentation with those features. Other diagnostic confounders are local skin reactions, for example to chlorhexidine or adhesive dressings, and maternal fever from unrelated causes such as endometritis, mastitis and wound infection. Consequently, the diagnosis of epidural-related deep tissue infection requires a high index of suspicion and imaging should be performed when there is clinical concern. Computed tomography scan is less accurate than MRI, especially with gadolinium.2, 3, 9, 28 In organizing imaging, it is also important to remember that, as noted in our series, infection may be located in tissues adjacent to the epidural space, including the paravertebral space, paraspinal and adjacent muscles,29, 30, 31 and nearby bone or disc.32, 33, 34

Treatment of deep tissue infection involves aggressive antibiotic therapy, with or without surgical decompression within 6-12h.2 Empirical antibiotic treatment should cover species such as Staphylococcus aureus, Streptococci and Gram-negative bacilli,3, 9 and antibiotic treatment should continue for up to 6weeks, extending to 8weeks if osteomyelitis is present.35 Conservative management without surgical drainage is an option in cases without neurological symptoms.36 Once neurological symptoms develop, severe disability and fatalities have been reported in up to 50% of patients.3 Of the four cases with deep tissue infection in our series, one (an epidural abscess) was treated surgically and the other three were managed conservatively with intravenous antibiotics. None of these women had permanent neurological sequelae.

In conclusion, this study was undertaken with the aim of estimating the incidence of obstetric epidural catheter-related infection in our hospital over a 4-year period. We found that the incidence of superficial infection was consistent with previous reports, but that the incidence of deep infection (0.04%, CI 0.01-0.11%; 1 in 2371) was higher than most previous series in obstetric populations. Nevertheless, the number of women studied was relatively small and the confidence intervals wide, so more complete analysis of data from national audits,37 and from multicentre prospective audits such as organized by the Society for Obstetric Anesthesia and Perinatology,38 are awaited.