Original Contribution

Predicting difficult and traumatic Lumbar punctures

Kaushal H. Shah MD^a,*, Daniel McGillicuddy MD^b,

Jeffrey Spear^b, Jonathan A. Edlow MD^b

^aDepartment of Emergency Medicine, University Hospital of Columbia College of Physicians & Surgeons,

St Luke’s-Roosevelt Hospital, New York, NY 10025, USA

^bBeth Israel Deaconess Medical Center, Harvard Medical School, Clinical Center 2, Boston, MA 02215, USA

Received 24 August 2006; revised 13 November 2006; accepted 13 November 2006

Abstract

Objective: The objective of this study is to determine if visual and tactile inspection of the spine is useful in the prediction of a difficult or traumatic lumbar puncture .

Design: This was a prospective, observational, cohort study conducted in the emergency department (ED) on patients who were undergoing an LP. Physicians prospectively completed a structured data form that included information about the patient, number of prior LPs performed, their assessment of the LP difficulty, and the number of needlesticks required. A bdifficultQ LP and a btraumaticQ tap were defined a priori. v², t tests, and regression were used as appropriate; an independent statistician performed the statistical analysis.

Setting: The study was conducted at an urban university teaching hospital with an annual ED census of approximately 48000 patients between November 1, 2002, and June 1, 2003.

Patients: The study population included a convenience sample of patients undergoing LP in the ED. Results: Of the 148 patients enrolled, LP was difficult in 47 (32%) patients and traumatic in 23 (16%) patients. The percentage of patients that did not have a visible spine was significantly higher in the difficult and traumatic groups ( P b .05). Among patients where the physician was unable to visualize the spine, there were significantly more difficult LPs ( P b .05).

Conclusion: It may be possible to predict which patients will have difficult or traumatic LPs before performing the procedure. Simple bedside assessments of spine visibility and palpability may assist in planning the approach to an LP in patients.

D 2007

Introduction

The lumbar puncture is a common and usually relatively simple procedure [1]. However, it is sometimes difficult to obtain the cerebrospinal fluid (CSF)-requiring

* Corresponding author. Tel.: +1 212 265 1770 (Home); +1 646 369

2747 (Cell).

E-mail addresses: [email protected] (K.H. Shah)8 [email protected] (D. McGillicuddy)8 [email protected] (J.A. Edlow).

several needlesticks or another physician. The incidence of a difficult LP is unknown. There is no clear definition as to what constitutes a bdifficultQ LP, nor are there any published data on the average number of needlesticks required to obtain CSF. In addition, the LP is occasionally btraumaticQ (needle-induced bleeding in the subarachnoid space), leading to diagnostic ambiguity. When there are red blood cells in the CSF, the clinician must determine whether they arose from a subarachnoid bleed or inadver- tently from trauma to the venous plexuses. There is no

0735-6757/$ - see front matter D 2007 doi:10.1016/j.ajem.2006.11.025

generally agreed upon number of RBCs that constitute traumatic or true subarachnoid hemorrhage [2]. Diminishing number of RBCs in the series of tubes used to collect CSF is considered a btraumatic tap,Q and the presence of xanthochromia is considered a true SAH [2,3]. The incidence of a traumatic LP (or traumatic tap) ranges from 10% to 20% [1,3,4].

It may be useful for a physician to identify which patients will be more likely to have a difficult or traumatic LP before performing the procedure. For example, in the setting of a high likelihood of a difficult LP, steps may be taken beforehand (eg, a more experienced physician might perform the procedure) to minimize patient discomfort and maximize the chance of a successful LP. Similarly, if there is a high likelihood the LP will be traumatic and there is a suspicion for SAH, one can consider fluoroscopic guidance, as a lower rate of traumatic tap has been demonstrated with this procedure [4].

We found no previously published predictors of a difficult or traumatic LP in any setting. In this study, we evaluate the ability of visual and tactile inspection of the spine to predict a difficult or traumatic LP.

Materials and methods

Study setting and design

This was a prospective, observational, cohort study conducted at an urban university teaching hospital with an annual emergency department (ED) census of approximate- ly 48 000 patients. The study population included a convenience sample of patients undergoing LP in the ED between November 1, 2002, and June 1, 2003. Data forms were available throughout the ED and affixed to stored (unopened) LP kits. All the physicians working in the ED were made aware of the study and were requested to complete the structured LP data form. All physicians in the department were trained on use of the data form, including definitions and difficulty scales. The hospital Institutional Review Board approved the study without need for informed consent of the patients.

Data collection

Before performing the LP, the following potential risk factors were obtained on a structured data collection form: age; estimated height and weight to calculate body mass index (BMI); history of osteoarthritis, rheumatoid arthritis, back surgery, or scoliosis per patient report; whether the patient is agitated/uncooperative (binary); ability to visual- ize and palpate the spine (binary); and position in which LP will be performed (lying fetal/sitting up). The operators anticipated the level of difficulty subjectively (assessed by both the resident and the attending) on a scale of 1 to 5; and postgraduate year (PGY) and number of previous LPs performed by the physician(s) performing the procedure were obtained. After performing the LP, the number of

needlestick attempts by each physician and the CSF analysis were obtained and documented.

Definitions

We defined a difficult LP a priori as requiring 3 or more needlesticks or requiring a second clinician to attempt the procedure. A traumatic tap was defined a priori as more than 400 RBCs per cubic millimeter in the first CSF tube; although there is no established threshold for number of RBCs to diagnose or exclude the diagnosis of SAH, it is consistent with the cutoff used in previous studies [1,2,5]. Spine visibility was defined as the ability to see the contour of the spinous processes when the patient was in position for the LP. Spine palpability was defined as the ability of the physician to palpate distinct spinous processes. A needlestick to obtain CSF was defined as piercing the skin; simple redirecting of the needle was not considered an additional stick. Given the low incidence of osteoarthritis, rheumatoid arthritis, back surgery, and scoliosis, these 4 variables were combined into the single variable back disease for analysis. The physicians rated the anticipated level of difficulty of the LP on a 5-point scale, and we considered a rating of 4 or 5 to be difficult.

Statistical analysis

An independent statistician performed the statistical analyses using the S-Plus (Insightful Corp, Seattle, WA) and STATA statistical software (Stata Corp, College Stations, TX). Variables were analyzed with univariate analysis, and comparisons between groups were made using Pearson v², Fisher exact, and Student t tests as appropriate. Standard statistical methods were used to calculate means, standard deviations, confidence intervals (CIs), and likeli- hood ratios (LRs). Logistic regression was used to assess the influence of confounders.

Results

There were 148 data forms completed during the study period. Of the 148 LPs enrolled, 75 LPs were performed to rule out meningitis, 60 to rule out SAH, and 11 for some other indication; and 2 data forms were missing an indication for LP. The average age of the patients was 42 F 18 years (range, 16-96 years; 1 missing age). The average BMI was

26.8 F 6 (range, 18-44; 11 missing either height or weight estimate). None of the patients were ultimately diagnosed with SAH. See Table 1 for additional characteristics of patients undergoing LP in the study.

Of the 148 patients enrolled, LP was difficult in 47 (32%) patients and traumatic in 23 (16%) patients. Postgraduate year level 2 physicians performed the most LPs with 41%, followed by interns (PGY level 1) with 33% and by those with PGY 3 and greater level of training with 26%.

We compared the physician’s perception of the spine (visible/palpable) in the difficult group (n = 47) with the

Table 1 Characteristics of patients undergoing LP (N = 148)		Table 3 Comparison	of clinical groups
Characteristic			Difficult LP (%)	Traumatic LP (%)
Mean (SD) {range}			[95% CI]	[95% CI]
Age 42.0 (18) {16-96}		Spine visible	20.0 [10.6-29.4]	8.6 [2.0-15.1]
BMI 26.8 (6) {18-44}		Spine not visible	42.0 [31.3-53.3]	21.8 [12.6-31.0]
n (%) [95% CI]		Spine palpable	26.7 [18.2-35.1]	12.4 [6.1-18.7]
Back disease 34 (23.0) [16-30]		Spine not palpable	44.2 [29.3-59.0]	23.3 [10.6-35.9]
Agitation 22 (14.9) [9-21]		BMI z30	42.1 [26.4-57.8]	18.4 [6.1-30.7]
Spine not visible 78 (52.7) [44.7-60.7]		BMI b30	28.2 [19.8-36.6]	14.6 [8.0-21.1]
Spine not palpable 44 (29.7) [22.4-37.1]		Age z65	42.9 [21.7-64.0]	14.3 [0-29.3]
Fetal position 114 (77.0) [70.2-83.8]		Age b65	29.9 [22.0-37.9]	15.8 [9.4-22.1]
sitting position 34 (23.0) [16.2-29.8]		Lying/fetal position	27.2 [19.0-35.4]	17.5 [10.6-24.5]
Meningitis indication 75 (50.7) [42.6-58.7]		Sitting position	47.1 [30.3-63.8]	8.8 [0-18.4]
SAH indication 60 (40.5) [32.6-48.5]		PGY 1	38.8 [25.1-42.4]	18.4 [7.5-29.2]
Other indication for LP 11 (8.8) [4.2-13.3]		PGY 2	32.8 [21.0-44.6]	16.4 [7.1-25.7]
		PGY 3+	21.1 [8.1-34.0]	10.5 [0.8-20.3]
easy group (n = 101) (Table 2); both residents and

attendings were able to predict which patients were going to be difficult LPs ( P b .05, LR+ 5.0). Inability to visualize and inability to palpate the spine were significantly higher in the difficult group compared with those in the easy group (visibility: 70% vs 45%, P = .004, LR = 8.7; palpability:

40% vs 24%, P = .033, LR = 6.9). The percentage of difficult LPs was twice as great if performed in the sitting position. In addition, there was found to be no statistical difference in age, BMI, back disease, agitation, or indication for LP.

We also compared the physician’s perception of the spine (visible/palpable) in the traumatic group (n = 23) with the nontraumatic group (n = 125) (Table 2). Spine palpability demonstrated no statistically significant association, but spine visibility was significantly less in the traumatic group ( P b .05 and LR+ 5.12). The spine was not visible in approximately 75% of patients who had a traumatic LP. The traumatic group did not require more needlesticks than the nontraumatic group ( P = .4). The incidence of traumatic LPs was 50% less if performed in the sitting position but almost twice as likely to be difficult (Table 3). Of note, the average attending physician difficulty prediction was significantly higher in the group that had an LP in the sitting position compared with the fetal position (41% vs 16%, P = .01).

Table 2 Statistically significant risk factors

Risk factors	Difficult vs easy group		Traumatic vs nontraumatic group
	P	LR+		P	LR+
Spine not visible	.004	8.67	.027			5.12
Spine not palpable	.033	6.87	.239			2.85
Sitting position	.029	4.57	.218			1.68
Resident predicted	.022	5.04	.072			2.98
difficult
Attending predicted	.016	5.54	.612			0.25
difficult

The P values and the likelihood (LR+) ratios for the predictors of interest (spine visibility, spine palpability, and physician’s prediction) are listed in Table 2.

Table 3 compares the clinical groups (eg, physician determines spine visible or not visible at bedside) with relation to incidence of difficult and traumatic LPs. In patients whose spines were visible, only 20% of LPs were difficult and 9% were traumatic; if the spine was not visible, the difficulty and traumatic rates doubled: 42.0% and 22%, respectively. Similarly, in patients whose spines were palpable compared with those that were not, the rate of difficult LPs increased from 26% to 44% and the rate of traumatic tap increased from 13% to 23%.

The average PGY level was only slightly lower in the difficult group (1.8) compared with that in the easy group (2.1) ( P = .05). More noteworthy, the incidence of difficult and traumatic LPs was lowest among LPs performed by a PGY 3+ (21% and 11%, respectively), followed by those performed by a PGY 2 (33% and 16%, respectively), and was highest among those performed by a PGY 1 (39% and 18%, respectively). These differences were not found to be significant statistically, and CIs overlapped.

Discussion

The LP is most often performed in the ED to diagnose 2 Serious diseases, meningitis and SAH [6]. In the case of meningitis, the LP and analysis of the CSF are the critical Diagnostic tests. In the case of SAH, interpretation of the LP results is far more difficult if the procedure is traumatic [2].

Our data suggest that spine visibility and palpability can help predict a difficult or traumatic LP. Among the variables examined here, spine visibility is clearly the best predictor of difficult and traumatic LPs. The fact that spine palpability is not as good a predictor as spine visibility makes sense intuitively; accurate palpation requires more skill and may

be less accurate than simply visualizing the contours of the spinous processes.

Our data suggest that inability to visualize or palpate the spine when the patient is in position for an LP increases the likelihood of a difficult and traumatic tap. As with any anticipated potentially difficult procedure, steps should be taken to maximize success of the procedure before starting, that is, paying close attention to patient position, having an experienced physician perform the procedure or at least be present, and having a backup plan in the event of failure.

Although performing the LP in the fetal position was less likely to be difficult and more likely to be traumatic compared with that in the sitting position, we found that physicians tended to start with the sitting position in patients predicted to be more difficult; thus, the results are likely confounding and prohibiting any definitive conclusions. To determine which position yields less difficulty and a lower incidence of traumatic LPs, further studies are necessary.

The attendings likely used weight and age in predict- ing which patients would be difficult LPs. The overweight (BMI N30) and elderly (age z65 years) patients were predicted by the attendings to be far more difficult than those patients with a BMI less than 30 and age less than 65 (both P b .001). In this study, although not statistically significant, the incidence of a difficult LP was higher in the elderly group (43% vs 30%) and the overweight group (42% vs 28%); it is possible that with a greater number of enrollments, a statistical difference would have been realized.

The existing literature on the impact of physician experience on the incidence of a difficult or traumatic LP is conflicting, with one study concluding no difference [5] and another demonstrating a possible difference [1]. Although not statistically significant, in this study, there was a clear inverse relationship between the likelihood of a difficult or traumatic LP and PGY training levels. With a greater number of enrollments, it is possible these differ- ences would have also been statistically significant.

Limitations

Although this was a prospective study and data forms were completed anonymously on the day of the LP, there was no independent observer to verify the information provided by the physician performing the LP. It is possible physicians underestimated the number of needlesticks required.

This was a convenience sample of patients, allowing for potential selection bias in our patient recruitment. The ED billed for 296 LPs during the study period; therefore, we enrolled approximately half of the LPs performed. Although many steps were taken to maximize enrollment of patients, it is possible that patients (especially those that were very difficult or traumatic) were not entered because of physician embarrassment; however, we suspect the major cause to be perceived lack of time in a busy ED environment to enroll patients in the research study.

Given that the study was performed at a teaching hospital with residents of all levels, we did not control for variability in the skill level of the persons performing the LP.

Future studies

When the contours of the spinous processes are not easily visualized and the spine is difficult to palpate, we believe the LP is more likely to be difficult. The best location to insert the needle becomes unclear. Fluoroscopic guidance has been demonstrated to be successful [4]; however, this may not be readily available. Many EDs now have ultrasound machines. We are presently conducting a randomized trial to study the benefit of ultrasound to identify the spinous processes in patients in whom the spine is not clearly visible. If the ultrasound can bseeQ the spine as well as the naked eye, the number of difficult and traumatic LPs among this subgroup of patients could be cut in half.

Conclusions

We found an inability to either visualize or palpate the spinous processes to be predictive of a difficult LP, whereas lack of ability to visualize the spinous processes was associated with a traumatic LP. Although further assessment of potential confounders and controlling for provider ability need to be performed in future validation studies, our re- sults are quite suggestive that these bedside assessments are useful and may assist in preparing and planning the approach to a successful LP.

Acknowledgment

The authors thank Drs David Newman and Nathan Shapiro for their significant help in the editing of the manuscript.

References

1. Shah KH, Richard KM, Nicholas S, Edlow JA. Incidence of traumatic lumbar puncture. Acad Emerg Med 2003;10(2):151 - 4.
2. Shah KH, Edlow JA. Distinguishing traumatic lumbar puncture from true subarachnoid hemorrhage. J Emerg Med 2002;23(1):67 - 74.
3. Marton KI, Vender MI. The lumbar puncture. Patterns of use in clinical practice. Med Decis Making 1981;1(4):331 - 44.
4. Eskey CJ, Ogilvy CS. Fluoroscopy-guided lumbar puncture: decreased frequency of traumatic tap and implications for the assessment of CT- negative acute subarachnoid hemorrhage. AJNR Am J Neuroradiol 2001;22(3):571 - 6.
5. Williams C, Fost N. Doctors’ experience and traumatic lumbar punctures. Lancet 1994;344(8929):1086 - 7.
6. Straus SE, Thorpe KE, Holroyd-Leduc J. How do I perform a lumbar puncture and analyze the results to diagnose Bacterial meningitis? JAMA 2006;296(16):2012 - 22.