Original Contribution

Accurate prediction of the needle depth required for successful lumbar puncture

Sze Yee Chong MBBS, Lee Ai Chong MBBS?, Hany Ariffin MBBS

Department of Pediatrics, University of Malaya, Kuala Lumpur 50603, Malaysia

Received 9 January 2009; accepted 7 February 2009

Abstract

Introduction: The aim of this study is to formulate an accurate estimate of the spinal needle depth for a successful lumbar puncture in pediatric patients.

Methods: This is a prospective study of pediatric oncology patients who had Lumbar punctures in the course of their treatment. The distance from skin entry point to the tip of the spinal needle was measured after lumbar punctures were performed. The relationship between the depth of needle insertion with weight, height, body surface area, body mass index, intervertebral space used, ethnicity, and sex of patient were studied. Predictive statistical models were used for the formulation of the ideal lumbar puncture needle depth.

Results: Two hundred seventy-nine patients who had nontraumatic lumbar punctures were studied. The patient characteristics were as follows: age, 0.5 to 15 years; weight, 7 to 63 kg; and height, 70 to 162 cm. Analysis using multiple regression tests with stepwise approach showed a strong relationship between the lumbar puncture needle depth and weight/height ratio. By using a predictive regression model, ideal depth of needle insertion (cm) = 10 [weight(kg)/height(cm)] + 1, with a regression coefficient r = 0.77.

Conclusion: This formula is accurate and practical with less complex calculations. However, further validation in a prospective study will be needed.

(C) 2010

Introduction

lumbar puncture is a common and important procedure for diagnosis and treatment of a variety of disorders in children. It may be difficult to perform for the inexperienced or in patients who have an unusually large or small body habitus. Multiple attempts to obtain a successful LP will not only cause discomfort to the patient but may also increase the probability of a traumatic LP (TLP), which is

* Corresponding author. Tel.: +60 3 79492065; fax: +60 3 79556114.

E-mail address: [email protected] (L.A. Chong).

defined as cerebrospinal fluid contaminated by 10 or more red blood cells per microliter.

A successful LP is especially important in children with acute lymphoblastic leukemia (ALL) where a TLP at diagnosis may change the overall outcome of the disease. A nontraumatic LP is vital to diagnose central nervous system (CNS) leukemia. Children with CNS leukemia receive more intensive chemotherapy compared with those without and may require cranial irradiation [1]. Gajjar et al

[2] published data demonstrating that initial traumatic diagnostic LP with the presence of blasts (TLP+) negatively affected the Treatment outcome of children with ALL. The event-free survival (EFS) for patients with TLP+ was 60% +-

0735-6757/$ - see front matter (C) 2010 doi:10.1016/j.ajem.2009.02.006

6% compared to 77% +- 2% in those with CNS1 status (nontraumatic LP with b10 red blood cells per microliter and no identifiable leukemic blast cells). The prognosis was worse in patients who had 2 consecutive TLP (TLP++), where the EFS was 46% +- 9%, which is comparable to that of patients with overt CNS leukemia. Similarly, Burger et al [3] from the Berlin-Frankfurt-Munster leukemia study group concluded that ALL patients with TLP+ have an inferior prognosis compared to those without TLP+ (EFS of 73% and 80%, respectively). Therefore, a “perfect” LP is especially important in patients with ALL.

An accurate prediction of the depth of needle insertion before LP may also shorten procedural time. Several studies have shown that prediction of the depth of the LP needle is possible by the use of mathematical formulas incorporating parameters such as the patient’s height, weight, and body surface area [4-8].

In this study, we would like to design a mathematical formula for the ideal depth of needle insertion to attain a successful LP.

Methods

Pediatric oncology patients aged 18 years and younger, treated at the University Malaya Medical Centre from January 1, 2006, to December 31, 2006, were studied with parental consent and approval by the Department of Paediatrics and the Ethics Committee.

Demographic data, depth and the intervertebral space of spinal needle insertion during LP were recorded of all patients. Body surface area (BSA), body mass index (BMI), weight/height ratio (wt/ht), height/weight ratio (ht/wt), and weight x height (wt*ht) were calculated.

All patients were either sedated or had local anesthetic applied to the lumbar area before the LP. All patients underwent LP in a lateral position. The 22G spinal needle (Terumo Corporation, Tokyo, Japan) was inserted in the midline, perpendicular to the skin surface into the intervertebral space either between the third and fourth (L3/L4) or the fourth and fifth (L4/L5) lumbar vertebrae. Entrance into the subarachnoid space was confirmed by good free flow of cerebrospinal fluid. After the LP, the needle was grasped between the thumb and index finger abutting the patient’s back and removed. The depth of insertion was then measured using a standard ruler placed in the room where the LP was performed. The procedure was carried out by a few designated medical staff

with clinical experience at performing LPs in children.

Results

A total of 279 patients who had nontraumatic LPs were analyzed. Their age range was 6 months to 15 years 1 month

(mean, 5.9 years; median, 5.2 years). Their body weight ranged from 7.0. to 63.0 kg (mean, 20.5 kg; median, 17.2 kg), and height was 70.0 to 162.0 cm (mean, 110.6 cm; median, 109.0 cm). Our study population consisted of 172 males. The main ethnic groups were Chinese 45.2%, Malays 38.4%, and Indians 13.6%. The main diagnosis of the patients enrolled in the study was ALL (n = 220). The other patients had Acute myeloid leukemia (n = 20), lymphoma (n = 32), and other cancers (n = 7). The LP needle depth ranged from 1.0 to

5.2 cm (mean, 2.8 cm). L3/L4 intervertebral space was used in 73.5% of patients. Most (92%) patients received intravenous sedation for the LP (Table 1).

Results were analyzed using SPSS version 13.0 (SPSS Inc, Chicago, Ill). The main statistical tests used were linear regression and Student t test. Regression coefficient (r N 0.75) indicated a strong relationship between the dependent and independent variables. A P value of less than .05 was considered significant.

A simple Linear regression analysis was performed between the LP needle depth and age, weight, height, BSA, BMI, height/weight ratio, weight/height ratio, and weight x height. Of these variables, weight, BSA, and weight/height ratio had the strongest relationship with the depth of LP needle; r values were 0.76, 0.76, and 0.77, respectively (Table 2). From multiple regression analysis, there was no significant difference in LP needle depth with ethnicity, sex, and intervertebral space used.

To estimate the ideal LP needle depth, manual multiple regression analysis with stepwise approach in a logical way

Table 1 Demographic and procedural data of pediatric oncology patients who underwent an atraumatic LP (N = 279)

Frequency/range

Age (in years and months) Sex

Male Female Ethnicity Malay Chinese India Others Weight (kg) Height (cm)

BSA (kg/cm) BMI (kg/m²) Diagnosis ALL

Acute myeloid leukemia Lymphoma

Other diagnosis

LP needle depth (cm)

LP needle insertion space L3-4

L4-5

6 mo to 15 y 1 mo

172

107

107

126

38

8

7-63

70-162

0.37-1.68

10.74-27.27

220

20

32

7

1.0-5.2

205

74

Independent Variables	Regression coefficient (r) with dependent variable (LP needle depth)	Coefficient of determination (r2)
Age	0.69	0.47
Weight	0.76	0.58
Height	0.68	0.46
BSA	0.76	0.57
BMI	0.60	0.36
Height/weight ratio	-0.74	0.54
Weight/height ratio	0.77	0.59
Weight x height	0.74	0.55

was used to build the best statistical model. We began with a forward selection with one independent variable in the regression model, followed by addition of single variables to the model until all statistically significant variables were included. With each addition of a new variable to the model, all previously entered variables were checked to see whether they maintained their level of significance. Previously entered variables were retained in the model only if their removal would cause a significant reduction in coefficient of determination r².

Table 2 Relationship between independent variables and the depth of LP needle insertion

Among all the models created using stepwise approach, the best model for the depth of the needle was: y = 10.6 (wt/ ht) + 0.93, where y = depth of LP needle in centimeters. This formula had a regression coefficient of r = 0.77.

Discussion

In the past 20 years, there have been several equations published to predict an ideal LP needle depth. Among these formulas are those by Bonadio (1988), Craig (1997), Abe (2005), and Stocker (2005) [4-7]. These are summarized in

Table 3.

From our results, by simple linear regression analysis, variables that were strongly correlated to the LP needle depth were weight, BSA, and weight/height ratio. These findings are consistent with those in similar published studies [4,6,7]

Table 3 Summary of published formulas to predict the LP needle depth

with the exception of Craig et al [5], who only considered height to accurately predict LP needle depth.

Multiple regression analysis showed that the most accurate equation to predict the LP needle depth was y (cm) = 10.6 (wt/ht) + 0.93, where y is the depth of the needle. This equation had an excellent r value (r = 0.77 and r² = 0.59). However, for simplicity and practicality, we modified the equation to y (cm) = 10 (wt/ht) + 1.

To assess the accuracy of our modified formula, the calculated LP needle depth was compared with the observed depth. Student t test was used to compare these 2 means. The P value was .137 (95% confidence interval, -0.146 to 0.105). Hence, our formula maintained its accuracy despite being modified for practicality and has been validated using the holdout sample method as described by Dawson and Trapp [8].

To compare our formula with other previously published formulas, these formulas were applied to our study population. Student t test was used to compare the means of the LP needle depth generated from the published formulas and that observed in this study. Only Bonadio’s and Stocker’s formulas as well as ours could accurately predict the LP needle depth (Table 4). Craig’s and Abe’s formulas predicted the depth 0.5 and 1.2 cm longer than the actual depth, respectively.

Our formula is more practical with less complex calculations than formulas by Bonadio et al [4] and Stocker and Bonsu’s [7]. Stocker’s formula is very accurate (r = 0.85), but his study sample size was small (n = 54), and this may reduce the statistical power and validity of his formula. Our formula is also more accurate especially when compared to the widely quoted Craig’s formula [5]. The increased accuracy of our formula is because we retained the constant 1, which was omitted by Craig. The constant in a regression model is important as it represents other factors that have not been studied but may affect the outcome.

In comparison to published formulas, our formula takes into consideration both weight and height, which is

Table 4 Comparison of the mean depth of LP needle calculated from the modified formulas of this study and published formulas with that observed from this study

SD

95% Confidence Interval of

the Difference

t

P

Lower Upper

Formula (y = depth of LP needle insertion)
Bonadio’s formula	y (cm) = 2.56 x BSA + 0.77 (BSA in m2)
(1988)
Craig’s formula	y (cm) = 0.03 x height (cm)
(1997)
Abe’s formula	y (cm) = 17 wt/ht + 1 (weight in kg,
(2005)	height in cm)
Stocker’s formula	y (mm) = 0.5 x weight (kg) + 18
(2005)

Modified	0.51022	-0.01463	0.10563	1.490	.137
formulas
Bonadio’s	0.51975	-0.01584	0.10666	1.459	.146
Stocker’s	0.52061	-0.06832	0.05439	-0.223	.823
Craig’s	0.58282	-0.57109	-0.43371	-14.399	.000
Abe’s	0.62447	-1.26715	-1.11996	-31.925	.000
P value of less than .05 is considered to be significant.

important in children of varying body habitus. This may be especially so in children who are either malnourished or have steroid toxicity where weight or height alone may not reliably estimate the depth of LP needle insertion.

One caveat to the use of our formula is that the LP should be performed with the needle perpendicular to the skin at point of entry. One study has shown that the LP needle depth, if insertion is 30? to the skin surface, may be increased by up to 1.5 mm for every 10-mm perpendicular distance [9]. A paramedian approach as described by Armitage would also increase the needle insertion distance by varying degrees depending on the inclination of the needle [10].

Conclusion

Analysis of 279 successful nontraumatic LPs in pediatric oncology patients yielded a strong correlation between the LP needle depth and weight/height ratio, allowing us to predict the ideal LP needle depth. By multiple regression analysis, the best and simplest equation to accurately predict the ideal LP needle depth is y (cm) = 10 [weight (kg)/height (cm)] + 1. However, this formula will need further validation in a prospective study.

References

1. Pui CH, Howard SC. Current management and challenges of Malignant disease in the CNS in paediatric leukaemia. Lancet Oncol 2008;9:257-68.
2. Gajjar A, Harrison PL, Sandlund JT, Rivera GK, Ribeiro RC, Rubnitz JE, et al. Traumatic lumbar puncture at diagnosis adversely affects outcome in children with acute lymphoblastic leukaemia. Blood 2000;96:3381-4.
3. Burger B, Zimmermann M, Mann G, et al. Diagnostic cerebrospinal fluid examination in children with acute lymphoblastic leukaemia: significance of low leukocyte counts with blasts or traumatic lumbar puncture. J Clin Oncol 2003;21(2):184-8.
4. Bonadio WA, Smith DS, Metrou M, Dewitz B. Estimating lumbar puncture depth in children. N Engl J Med 1988;319(14):952-3.
5. Craig F, Stroobant J, Winrow A, Davies H. Depth of insertion of a lumbar puncture needle. Arch Dis Child 1997;77:450.
6. Abe KK, Yamamoto LG, Itoman EM, et al. Lumbar puncture length determination. Am J Emerg Med 2005;23(6):742-6.
7. Stocker DM, Bonsu B. A rule based on body weight for predicting the optimum depth of spinal needle insertion for lumbar puncture in children. Acad Emerg Med 2005;5(1):105-6.
8. Dawson B, Trapp RG. Basic and clinical biostatistics. 4th edition. Lange Medical Books/McGraw-Hill; 2004. p. 256.
9. Bosenberg AT. Skin-epidural distance in children. Anaesthesia 1995;5:895-7.
10. Armitage EN. Regional anaesthesia. In: Summer E, Hatch DJ, editors. Textbook of Paediatric Anaesthetic Practice. London: Bailiere Tindall; 1989. p. 213-33.