Skip navigation
Link to Back issues listing | Back Issue Listing with content Index | Subject Index

Risk and obstetric epidurals

Systematic review

I doubt if any of us would ever feel the same confidence about safety as JE Salk, who, when questioned about the safety of his polio vaccine, is reputed to have replied
It is safe, and you can't get safer than safe.
The trouble with this concept of safe is that we know that some bad thing is likely to happen at some time, even if it is an extremely remote possibility. And if you multiply remote risk by many people, then some finite numbers are bound to pop up. And if the numbers have very dire consequences attached to them, the notion of safe tends to go out of the window.

Epidurals are routinely given during childbirth, to as many as 2.4 million US women every year. A new systematic review [1] of observational studies affords an insight into the safety of this procedure.

Systematic review

The basis of the review was an extensive search for observational studies regarding adverse events associated with epidural catheter use, not limited to obstetrics. The search included electronic searches, hand searching of major anaesthetic journals, reference lists, and reviews. Studies with fewer than 200 patients were not sought, as they were unlikely to report on rare events, and case reports would be unlikely to have a denominator to calculate rate.

Outcomes of interest were epidural haematoma, infection, and neurological injury in the shorter (less than one year) and longer term. There were various planned sensitivity analyses, including larger and smaller studies, and more and less recent studies.


In all, 27 observational studies reported on 1.37 million women. Only five studies were identified initially through electronic searches, with most (20) identified through reference lists. Larger studies with more than 10,000 women per study tended to have been published since 1990, and smaller studies tended to have been published before 1990.

Not all studies reported on all outcomes, so the number of patients varied somewhat. The main results were:

• Epidural haematoma occurred in six women, at a rate of about 1 in 170,000 in larger, more recent studies.

• Deep epidural infection occurred in 11 women in total, at a rate of about 1 in 145,000 women in larger, more recent studies.

• Persistent (more than one year) neurological injury occurred in three women, at a rate of about 1 in 240,000 women in larger, more recent studies.

Transient (less than one year) neurological injury occurred in 288 women in total, at a rate of about 1 in 6,000 in larger, more recent studies. In individual studies the risk of transient neurological injury varied between about 1 in 100 to more than 1 in a million (Figure 1), though larger studies tended to be more consistent.

Figure 1: Individual studies, showing the event rate and the individual risk for a woman for transient neurological injury following obstetric epidural


Many issues are highlighted by a study like this. Foremost is that of causation. While we are used to thinking that childbirth is safe, there are reports of spontaneous occurrence of epidural haematoma, deep epidural infection, and neurological problems with childbirth when epidurals have not been used. So while these outcomes may be related to the use of the epidural, we don't know this for sure.

Another is the need for very large numbers to estimate risk accurately. Even with the largest data set, for transient neurological injury, the degree of variability seen in Figure 1 was profound. Even with large numbers in the denominator, our estimate of the actual rate will be less than accurate, but the best we can do with what we have. A related point is the need to perform sensitivity analysis using larger, better, studies.

We can try and put the risks into the Paling Perspective Scale (Figure 2). There are two difficulties with this. We know only about the duration of these neurological injuries, not their severity. They are certainly not deaths, which is why this particular version of the Paling Scale may be misleading, other than tell us about the rate of event. We need relevant comparisons for these sorts of injuries.

Figure 2: Risk of neurological injury shown in a Paling Perspective Scale

Finally, though, what can we say about safety of epidurals used during childbirth? It is often said that adverse events are under-reported. Yet here the review has accumulated large numbers of women in larger, more recent, studies, where the importance of complete reporting was known. These results are probably a fair reflection of the real world.


  1. 1 W Ruppen et al. Incidence of epidural hematoma, infection, and neurological injury in obstetric patients with epidural analgesia/anesthesia. Anesthesiology 2006 105: 394-399.

previous or next story