Our data indicate that advanced age was associated with a higher risk of sanguineous and dural puncture, although advanced age was also associated with a lower incidence of insufficient analgesia. The level chosen for epidural puncture influenced the rate of sanguineous puncture, unsuccessful catheter placement and insufficient analgesia. Height was the only anthropometric index correlated with immediate complications, the risk of unsuccessful catheter placement being less in taller individuals.
The incidence of sanguineous puncture (SP) has previously been reported at between 0.7 and 15.7% [15–19]. A recent meta-analysis comparing air and liquid as a medium for loss of resistance found an overall incidence of 5.9% in parturients . Most of the published data relates to obstetric EA and considering that the probability of complications is estimated to be higher in pregnancy (3 to 15.7% in obstetric EA [16–19] vs. 0.7 in EA for lower abdominal procedures ), our finding of 4.4% while broadly consistent with previous reports is somewhat higher than in non-obstetric EA.
Our findings of more SP in the lumbar spine may be due to the anatomy of the posterior venous spinal plexus. This plexus occupies a considerably larger volume in the lumbar area than further cranial . Moreover, this venous system has no valves, which causes an additional distension of the caudal vascular bed when the patient is in an upright position . An analysis investigating potential factors reducing the risk of intravascular catheter placement demonstrated that the lateral position was beneficial (3.7% vs. 15.7%) . In addition, a thoracic approach seems to reduce the risk of SP  and this finding is consistent with our data.
Neither the risk of SP in elderly patients nor age-based changes of the intraspinal venous plexus have been previously evaluated systematically. Therefore, we can only speculate as to whether varicose changes and/or a higher vulnerability in older people may contribute to the higher incidence of SP in our study population.
Accidental dural perforation (ADP)
A rate of 1.6% ADP is in accordance with data published previously (0% and to 2.7% [14, 15, 17, 19, 20]. Two studies found a higher incidence of ADP in lower spinal levels – one study evaluated only the thoracic region , while the other study investigated the entire spine . Our results show similar trends, with a higher incidence of ADP in the lumbar spine. An explanation for this observation could be related to the reduction in thickness of dura mater around lumbar level L2/3 . To our knowledge, a more frequent occurrence of ADP in older patients has not been described before. A potential hypothesis to explain this might be that typical age-related changes like degenerative disc disease, facet joint arthritis and progressive kyphosis may make the identification of the epidural space more difficult than in younger patients. Furthermore, stenosis of the spinal canal, a typical disease of the elderly, might result in adhesive processes which could reduce the epidural space . In summary, these factors may explain the higher incidence of ADP.
Unsuccessful catheter placement (UCP)
The incidence of unsuccessful catheter placement (0.9%) is in agreement with previously published data: among obstetric patients the incidence of abandoned trials to establish EA was estimated between 0.5% to 4.6% [19, 21], among non-obstetric patients 0.13%  and 1.1% for thoracic EDA . Previous studies did not detect any relation between level and UCP incidence . In a regression analysis of our data, a correlation was found between punctured level and UCP: the lower lumbar and middle thoracic spine were associated with a discernably higher risk than punctures at the reference level (TH9-11).
Inadequate analgesia (IA)
The observed incidence (8.8%) (at least partially) insufficient EA is commensurate with published data (0% to 12% [15–17, 19, 21]. The wide range of IA reported in the literature might be due to the absence of a commonly accepted definition of insufficient analgesia. Some authors claim a failure of the method only when an unscheduled additional general anesthesia was required [17, 19], while others use the visual analog scale to measure pain intensity . Harney et al. classified an asymmetric effect as failure to insert . Our concept, to classify EA as insufficient on the basis of the postoperative opioid requirement may have led to our relatively elevated incidence of IA.
The regional distribution of IA is remarkable in that we observed a higher incidence in the high thoracic and the low lumbar regions. Our presumption is that in many of these cases the catheter was placed correctly but the local anesthetic did not reach all spinal segments needed for analgesia of the entire site of operation. Therefore choosing a very high or low spinal region for epidural puncture may lead to analgesic insufficiency.
The risk of paresthesia is lower in both male patients and those of advanced age. The overall incidence of 2.8% is lower compared to data acquired from spinal anesthesia (6.3% ). However, reports of transient paresthesia are a rare feature potentially influenced by subjective bias of the anesthesiologist in reporting such events. In the current study, reports of persistent paresthesias were not encountered.
The appropriate identification of the level punctured during EA and the experience and subjective bias of anesthesiologists may have considerable impact on the quality of the data collected and therefore these features need to be controlled for.
The spinal level is usually identified on the basis of imprecise landmarks and there are certainly a number of cases where the punctured segment is not identified correctly. To reduce systematic error associated with misjudged level the technique of grouping segments by anatomical area was used .
It is more difficult to account for differences in the expertise of the anesthesiologists performing EA. One might assume that most trainees gain first EA experience with lumbar punctures due to the absence of the spinal cord in this area. This could be a reason for some complications occurring more often in caudal areas and explain similar findings from other publications. However, in the case of inability to place the catheter, every trainee of our department is required to consult a senior consultant before aborting the procedure of EA. Therefore, only experts are responsible for the higher incidence of UCP in the lumbar area. The higher incidence of sanguineous puncture can be accounted for anatomic reasons and is rather unlikely to be related to the anesthesiologist’s experience. Whether training of the anesthesiologist influences the rate of accidental dural puncture cannot be retrospectively examined using our data because the data are anonymized and the anesthesiologist who performed the epidural puncture cannot be directly identified.