Many factors influence the anesthesiologist's decision to perform neuraxial (spinal or epidural) anesthesia. Strong or absolute factors are the site of surgery, patient acceptance, local infection, and uncorrected hypovolemia. Other issues are preexisting neurologic disease, coagulopathy, bacteriemia, pulmonary disease, mental status, postoperative pain management, the anesthesiologist's experience, and the anticipated difficulty of the procedure itself. We were interested in identifying factors associated with technically difficult neuraxial blocks. Avoidance of a prolonged, difficult, and painful procedure is not the only reason for this interest. Needle trauma may cause neurologic complications, and multiple, traumatic attempts at needle placement have been associated with a higher incidence of epidural hematoma [1-4]. In fact, it has been suggested that an alternative technique should be considered for the patient with a "difficult back" and that a more experienced provider should take over a difficult procedure at an early stage . A literature search and a review of major textbooks failed to reveal any data relating patient characteristics to potential technical difficulties in performing neuraxial blockade. In an informal survey of 34 anesthesiologists, the overwhelming majority cited obesity as the most important predictive factor, with spinal deformity a distant second.
We designed this study to determine whether any patient characteristics allow the prediction of a technically difficult neuraxial block.
We studied 595 patients from two teaching hospitals over a period of 12 mo. Each institution's investigation review board approved this prospective observational study. Most neuraxial blocks (63%) were thoracic epidurals requested for postoperative pain management; they were not used as the primary anesthetic. All spinals and lumbar epidurals were used as the primary anesthetic. Before attempting the epidural or spinal, the attending anesthesiologist recorded the following patient data: sex, age, height, weight, body habitus (assessed subjectively as normal, thin, muscular, or obese), anatomical landmarks (good = easily palpable dorsal spinous processes, poor = difficult to palpate spinous processes, none = unable to positively identify spinous processes), and apparent spinal anatomy (assessed subjectively by inspection and examination as normal or deformed). Previous back surgery did not necessarily constitute a deformity. Body mass index (BMI) was calculated as: BMI = weight/height (2) (kg/m2). We recorded the training level of the provider as first-, second-, or third-year anesthesia resident; nurse anesthetist (CRNA); or fellow/staff. We did not collect data for residents with <6 mo of training. All trainees and CRNAs were supervised by the attending anesthesiologist. We also recorded needle type, needle gauge, spinal level used, and approach (midline or paramedian). All spinal and epidural blocks were performed with the patient in the sitting position. The responsible anesthesiologist determined the technique (epidural or spinal), the approach, the level, and the needle type and gauge. There was no time limit for completing the neuraxial block.
Two measures were used to assess the difficulty encountered in performing the block. First, was the needle placement successful at the initial spinal level? We termed this either first-level success or first-level failure. Second, we recorded the number of attempts required for successful needle placement. Each new skin puncture was considered another attempt, whether at the initial spinal level or at a second level. Simply redirecting the needle without a new skin puncture was not considered an additional attempt. Finally, the success or failure of the blockade was recorded. The neuraxial block was considered a failure if local anesthetic supplementation, a second neuraxial block, or general anesthesia was required. Epidurals placed for postoperative pain management were considered a failure if they could not be used as the sole source of analgesia within 24 h. When a second block was performed, the unsuccessful block was considered a first-level failure, and the number of attempts for the first-level failure and the second block were summed. Data were also collected for the number of complications, i.e., unanticipated dural puncture, traumatic puncture, postdural puncture headache severe enough to require treatment, or neurologic complications.
Logistic regression was used to test the association between age, spinal anatomy, type of block, or type of needle and first-level success. Odds ratio and 95% confidence intervals for first-level success of comparison groups relative to a reference group were calculated. Association with the total number of attempts was assessed using Poisson regression models, which allow for over-dispersion. Each factor was also studied for interaction with initial success to determine whether its effect on the total number of attempts was dependent on success or failure at the initial level. For all hypotheses, a significance level of 0.05 was used. Variables are reported as percentages and mean +/- SD.
We studied 595 patients, 381 male and 214 female. Ages ranged from 16 to 84 yr. There were 171 spinals and 424 epidurals (51 lumbar, 373 thoracic). All neuraxial blocks in this study were successful (590 of 595; 99.2%) or aborted (5 of 595; 0.8%) at either the first or second spinal level. No neurologic complications were observed.
(Table 1 and Table 2) show the relationship of patient characteristics and technical factors to first-level success. First-level success was not affected by age, sex, spinal anatomy, or body habitus for either spinal or epidural blocks. First-level success was, however, significantly more likely for patients with good landmarks than for patients with poor or no landmarks (Table 2). The difference in first-level success between poor and no landmarks (82% vs 71%) was not statistically significant. First-level success was not different for spinal versus epidural and was not related to approach for either technique, or, in the case of spinals, to needle type or gauge. However, first-level success was significantly more likely with thoracic than lumbar epidurals (P = 0.048).
We found that the provider's level of training did not affect first-level success (P = 0.53) (Table 3). Even when the more experienced providers (staff/fellows, third-year residents, and CRNAs) were combined and compared with the less experienced providers (residents with 7-25 mo of training), the difference in first-level success was not statistically significant. In 33 instances, a senior provider took over and performed the block at the initial level. The rate of first-level success was 87%.
(Figure 1) shows the distribution of the total number of attempts. Only one attempt was necessary for 380 (64%) patients; two attempts were necessary in 104 (17%) instances. Three patients required 10 attempts. Despite presumed proper needle placement, epidural catheters could not be advanced in 5 of 424 (1%) patients; however, the number of attempts for needle placement was included in the analysis.
(Table 1 and Table 2) show the relationship of patient characteristics and technical factors to the total number of attempts. There were no differences related to sex or age and no differences in the overall number of attempts required for epidurals and spinals. For spinal blocks, neither needle type nor gauge affected the number of attempts. Although spinal deformity did not affect first-level success, it did significantly increase the number of attempts (Table 2). Thoracic epidurals required fewer attempts than lumbar epidurals (Table 1). The median versus the paramedian approach did not affect number of attempts for either spinal or epidural blocks.
The quality of landmarks and body habitus were both significantly associated with the number of attempts required for both spinals and epidurals (Table 2). This was true only if there was first-level success. If it was necessary to use a second level, there was no relationship between the number of attempts and either body habitus or landmarks. Given successful needle placement at the first level, the relationship between the number of attempts and the quality of landmarks was significant (P < 0.001). Fewer attempts were required for patients with good landmarks than for patients with poor or no landmarks, and no difference could be found between poor and no landmarks. Likewise, there was a significant relationship between the number of attempts and body habitus, proceeding normal-thin-muscular-obese (P = 0.001), but there were no detectable differences between adjacent groups.
In patients with initial-level success, the number of attempts increased with BMI (P < 0.001). Although statistically significant, the expected number of attempts increased by one attempt for every 20-kg/m2 increase in BMI (equivalent to a 140-lb weight gain for an average man). In cases of failure at the initial spinal level, no association between the number of attempts and BMI was present.
We were unsuccessful in placing epidural catheters in five patients (1%) (three lumbar and two thoracic). Despite apparently correct placement, 16 (4%) epidural catheters did not provide adequate analgesia; 1 was replaced immediately and 5 were replaced postoperatively for pain management. For the remaining 10 patients, another attempt to place the catheter was deemed either unnecessary or contraindicated. Therefore, the total initial failure rate (considered unsuccessful catheter placement or nonfunctional catheter) for epidural anesthesia/analgesia was 5% (21 of 424). Twelve patients (3%) had subarachnoid punctures. Two were given prophylactic blood patches and did not develop headaches; two other patients later required blood patches for severe headache. Of the remaining eight patients, three developed mild headaches that resolved with conservative treatment within 48 h, and the remaining five did not develop headaches. Continuous aspiration of blood, indicating an intravascular catheter position, occurred in nine patients (2%).
Free-flowing cerebrospinal fluid was obtained before the injection of local anesthetic from all patients who received spinal anesthesia. In three patients (2%), the anesthetic level was inadequate for surgery; two required general anesthesia, and one patient's spinal anesthesia was supplemented with a local anesthetic.
Anesthesiologists have recognized the importance of identifying a priori patients who may be at greater risk in other aspects of management, such as the difficult airway , malignant hyperthermia , or perioperative adverse outcome of general anesthesia . However, we are unaware of any published evidence or study that has examined factors to identify potential technical problems in the performance of neuraxial blockade.
Others have emphasized the importance of avoiding traumatic neuraxial block procedures [5,9]. For example, the Norwegian Association of Anaesthesiologists' guidelines for central blockade in patients with potential bleeding problems specifically mention a competent and "atraumatic" anesthesiologist . Horlocker et al.  suggest that operator expertise may relate to the incidence of minor neuraxial hemorrhagic events and emphasize "atraumatic" technique for patients receiving antiplatelet medications. Owens et al.  reviewed six reports of spinal hematomas after spinal anesthesia. In the five cases for which comments were available, four of the five were termed "difficult tap."
Of all the factors we examined, the quality of landmarks was the most significant independent predictor of difficulty as measured both by first-level success and number of attempts. Body habitus was an independent predictor only with regard to the number of attempts. The significance of this finding is that body habitus does not seem to be as important as many seem to think, and it certainly is not as important as the quality of landmarks. Unfortunately, the only objective, numerical patient characteristic, BMI, was very weakly related to neuraxial block difficulty. However, landmarks and body habitus may have some association. The obese patient group had the lowest proportion of patients with good landmarks (34%) compared with >90% in the normal, thin, and muscular groups. However, we found that, for patients with good landmarks, there was no significant difference in difficulty as measured by both criteria, regardless of body habitus. The findings were the same for the patients with poor landmarks body habitus groupings. In this regard, our results show that the only other significant predictor was the presence or absence of spinal deformity. We therefore conclude that the most effective way to determine beforehand the potential for a difficult neuraxial block is to examine the back for landmarks or obvious deformity.
Our results show that, when a neuraxial block is difficult enough that a second level is tried, there is no correlation between the number of attempts at the second level and landmarks, body habitus, or any of our other observations. The first level was abandoned in 76 (13%) cases, usually after two or three attempts. Success at the second level usually required two attempts (1.9 +/- 1.1). In this context, the number of attempts may not completely reflect the effort expended. Multiple redirections of the needle or adjustments in patient position may have occurred; however, we did not try to capture these data because we believe that it is too subjective and open to observer interpretation. Therefore, when there is a first-level failure, our data do not predict the extent of difficulty at a new level.
Although our two measures of difficulty, first-level success and number of attempts, are clearly arbitrary, at least one (number of attempts) has been used by other investigators [10-12]. One could evaluate other measures, such as number of needle redirections, elapsed time for the procedure, and patient discomfort; however, these measures are subjective and imprecise. Inasmuch as 97% of the blocks were ultimately successful, using the success or failure of the block as the measure of technical difficulty would require a prohibitively large study. First-level success and number of attempts as we defined them are precise, discrete, and not subject to interpretation. We believe that they represent a good estimate of the technical difficulty of a neuraxial block.
Results from our study both confirm and differ from conclusions reached by other authors. In a study of 300 spinal anesthetics using 25-, 27-, and 29-gauge needles, 15% were judged difficult and 10% required more than five attempts, but there were no differences as a function of needle gauge . However, in our study, 2% of spinal blocks required more than five attempts, and overall (spinal and epidurals), 6% required more than five attempts. In a study of 100 spinals, Harrison and Langham  reported a 75% success rate on the first attempt, whereas our first attempt success rate was 64%. However, our overall successful spinal anesthetic rate (98%) is comparable to that of others [10-12].
Considering safety and trauma during epidurals, Renck  emphasized that the operator's experience may result in a less traumatic procedure. However, we did not find a statistically significant association between the level of training and first-level success. Kopacz et al.  found that first-year residents successfully located the epidural space in 76% of cases. Our first-year residents' success rate was 73%. Further, for Kopacz et al. , the overall success rate was 99%, comparable to our overall 98% success rate.
Kopacz et al.  reported that the midline approach had a higher success rate and required fewer attempts than the paramedian approach. We found no difference in first-level success or number of attempts for either type of block, regardless of the approach. However, like others, we found that neither needle type nor gauge affected the difficulty of subarachnoid blocks [10,11].
Our finding that, by both difficulty measures, thoracic epidurals were less difficult than lumbar epidurals is surprising. The fact that most (93%) thoracic epidurals were performed by more senior providers is, however, not the explanation. In fact, the staff/fellow group alone required significantly more attempts for lumbar than thoracic epidurals (3.3 vs 1.9; P = 0.001).
From our study of nearly 600 neuraxial blocks, we conclude that the most reliable method of determining beforehand the potential for a technically difficult block is an examination of the patient's back for the quality of landmarks and for obvious deformity. Body habitus is a less reliable predictor. Other factors, such as technique (spinal or epidural), approach, equipment, and provider experience, seem to have little or no effect. Further, when the first level is abandoned, there are no predictors of difficulty at the second level. If the block was difficult at the first level, there is no way to predict that it will be easier at a second level. Finally, under the conditions of our study and for reasons we cannot explain, thoracic epidurals were less difficult than lumbar epidurals.
1. Vandermeulen E, van Aken H, Vermylen J. Anticoagulants and spinal-epidural anesthesia. Anesth Analg 1994;79:1165-77.
2. Auroy Y, Narchi P, Messiah A, et al. Serious complications related to regional anesthesia. Anesthesiology 1997;87:479-86.
3. Owens E, Kasten G, Hessel E II. Spinal subarachnoid hematoma after lumbar puncture and heparinization. Anesth Analg 1986;65:1201-7.
4. Lerner S, Gutterman P, Jenkins F. Epidural hematoma and paraplegia after numerous lumbar punctures. Anesthesiology 1973;39:550-1.
5. Renck H. Neurological complications of central nerve blockade. Acta Anaesthesiol Scand 1995;39:859-68.
6. Mallampati SR, Gatt SP, Gugino LD, et al. A clinical sign to predict difficult intubation: a prospective study. Can Anaesth Soc J 1985;32:429-34.
7. Larach MG, Localio AR, Allen GC, et al. A clinical grading scale to predict malignant hyperthermia susceptibility. Anesthesiology 1994;80:771-9.
8. Forrest JB, Rehder K, Kahalan MK, Goldsmith CH. Multicenter study of general anesthesia; predictors of severe postoperative adverse outcomes. Anesthesiology 1992;76:3-15.
9. Horlocker TT, Wedel DJ. Anticoagulants, antiplatelet therapy and neuraxial blockade. Anesth Clin North Am 1992;10:1-11.
10. Kopacz DJ, Neal JM, Pollock JE. The regional anesthesia "learning curve:" what is the minimum number of epidural and spinal blocks to reach consistency? Reg Anesth 1996;21:182-90.
11. Tarkkila P, Huhtala J, Salminen U. Difficulties in spinal needle use: insertion characteristics and failure rates associated with 25-, 27- and 29-gauge Quincke-type spinal needles. Anaesthesia 1994;49:723-5.
12. Harrison DA, Langham BT. A survey of failure rate, postdural puncture headache and patient satisfaction. Anaesthesia 1992;47:902-3.