Post‐dural puncture headache: a prospective study on incidence, risk factors, and clinical characterization of 285 consecutive procedures

Data were analyzed using IBM SPSS software package version 25.0 (Armonk, NY: IBM Corp). Qualitative data were described using number and percent. The Kolmogorov–Smirnov test was used to verify the normality of distribution. Quantitative data were described using range (minimum and maximum), mean, standard deviation, and median. Chi-square test (Fisher or Monte Carlo), and student t-test were used to compare two groups for normally distributed quantitative variables. Mann–Whitney test was used to compare between two groups for not normally distributed quantitative variables. All of the variables that were statistically significant in univariate logistic regression model were entered into multivariate logistic regression model to explore the factors that are independently associated with development of PDPH. The multivariate model contained variables that were associated with p < 0.1 in univariate analysis. Significance of the obtained results was set at p ≤ 0.05 level.

In this study, 29.5% of our patients developed PDPH. The reported incidence of PDPH is variable in the literature, ranging from 10 to 40%, depending on several procedural and non-procedural aspects [9, 10]. Historically, higher incidence was reported in older studies, probably secondary to the use of large gauge, cutting spinal needles. PDPH incidence significantly dropped from 66% in a study from 1898 [11], to around 11% [12], after the development of smaller gauge needles and the modifications of spinal needle tips.

Our findings were similar to the values reported in other regional studies, where PDPH occurred in 27.6% [13], and 28.7% [14] of patients, although cutting needles with different gauge sizes were used in these studies. However, in a similar study using 22G or 24G non-cutting needles, the incidence of PDPH was 21.6% during the immediate post procedural period, and 17.5% at 24-h follow-up [15]. On the other hand, in a 2018 meta-analysis of 110 studies [16], the rate of PDPH with a 20–22 G needle was found to be 11.0% when using traumatic needle group compared to 4.2% with atraumatic needles, signifying a 60% reduction in incidence.

This higher incidence in our study could be understood in the light of the clinical characteristic of our patients, rather than procedural-related factors. The majority of our patients had well-documented risk factors such as; female gender, low BMI, and headaches prior to LP. Moreover, in the aforementioned meta-analysis, participants tended to be older (mean age was 38.6 years), with less female participants (61.7%) than our study. Another reason could be the nature of our prospective study, where we evaluate patients after LP and specifically ask if they have headache. Many PDPH studies are retrospective, thus only patients who report headaches are included.

The majority of headaches (85.7%) started within the first 2 days of the procedure, and lasted for average of 2.4 days. This is similar to findings in the literature, where around 66% of PDPH start within the first 48 h of LP, and nearly 90% within the first 72 h [17]. However, headache duration in our cohort seemed shorter than other studies, in which, only a quarter of patients improve within 2 days of onset, and more than half of patients improve by day 4 [18]. This could be related to the mild/moderate severity of the headaches in the majority of our patients, with good response to conservative medical management. None of our patients developed PDPH after 4 days of LP, which is similar to the literature, as PDPH rarely develops between 5 and 14 days after the procedure [18].

The commonest headache type was dull aching in 58.3%, followed by throbbing in 29.8% of patients. In a review of literature [1], the commonest headache types were reported to be dull aching, throbbing, or burning, similar to our findings. However, they reported that pain in PDPH is usually severe, which contradicts our findings, where the majority in our cohort (76.2%) had headaches of mild to moderate severity, with mean NRS of 4.1 ± 0.9.

Moreover, PDPH was self-limited in the majority of patients, with only 2 patients (0.7%) requiring a blood patch. These values compare favorably to published studies of PDPH. In a study by de Almeida and colleagues [19], PDPH was well-tolerated, and only 0.4% of their cohort needed a blood patch.

The current study supports some earlier findings for PDPH risk factors in the literature, including young age, female gender, low BMI, headache before LP, previous history of PDPH, and number of attempted punctures [4, 20].

In this study, PDPH was statistically significantly higher in patients with ON and CVT, and lower in patients with PN and SMA, although this association was not confirmed in the multivariate regression analysis. This could be attributed to other clinical characteristics in those patients, such as the presence of headache prior to LP, or the predominance of female gender in patients with CVT and ON, in our cohort.

Several studies in the literature reported variable effects of different clinical diagnoses on the development of PDPH. Similar to our findings, dementia [33, 34] and SMA [35] showed low incidence of PDPH in the literature. Interestingly, none of our SMA patients had PDPH, which could be attributed not only to the small number of patients, but might also be related to the removal of small amount of CSF and immediate replacement with the drug during the procedure. Other studies reported higher incidence of PDPH in patients with IIH [36], and MS [37], with an incidence of 23.0% and 57%, respectively.

In our study, CSF opening pressure, closing pressure, CSF volumes or constituents, were not associated with the development of PDPH, similar to several previous reports [19, 38, 39]. Interestingly, higher numbers of RBCs in CSF were found to be inversely correlated with developing PDPH, however, this finding was not confirmed in multivariate analysis. One possible explanation is that the coagulating effect of the blood can act as a patch for the dural tear created during LP, which was observed in few reports [40]. However, other studies failed to find significant correlation of PDPH and RBC counts [19].

Moreover, the lack of association of LP location, position, duration of rest after LP, and post-LP fluids and PDPH in our cohort, was in line with findings in the literature [13, 19, 41]. Furthermore, the experience of the neurologist was not correlated with developing PDPH in our study. In literature, there is no strong evidence that correlates physician’s experience and PDPH. Our findings matched other studies, where there was no effect of operator experience on developing PDPH after 300 diagnostic LPs in one study [21], and after 501 LPs in another [28]. However, this finding contradicts other earlier reports. In one study [42], the incidence of PDPH was 2-times higher with less-experienced anesthesiologists compared to experienced ones (2.5% vs 1.2%). This was also similar to another study [43], where the most junior clinicians had double the incidence of PDPH when compared to consultants, although this finding was not statistically significant (0.2% vs 0.13%, p > 0.05). One can argue that this lack of correlation is attributed to the lesser degrees of variability in the technical performance of the LPs, as all neurologists in the department have more than 3 years of experience, under close supervision and rigorous training.

In literature, more than 85% of PDPH resolves with conservative measures, or without any specific treatment [25]. Despite the lack of evidence on pharmacological management of PDPH [44], the majority of our patients were managed by conservative measures that included bed rest in supine position, oral or intravenous hydration, caffeine supplementation, and simple analgesics, with good response in the majority of them. Only 2 patients had persistent headaches that required epidural blood patch, which is similar to other published values [45, 46].

This study has some limitations. First, the results were obtained from a single-center, and the sample size was not calculated according to a predefined outcome. However, we included all consecutive eligible patients admitted to the largest neurology tertiary hospital in Kuwait, during the study period, which also minimized the possibility of selection bias. Few studies in the literature present a uniform data set regarding LPs and PDPH, as most studies use different techniques and protocols. Moreover, our findings came in line with published data on PDPH, which further validates our sample, and confirms that our cohort is representative to the population of patients undergoing LP in neurology departments.

Secondly, given our department standardized protocol for LPs, we could not evaluate the impact of certain procedural factors such as; needle type, needle size, bevel direction, and non-replacement of the stylet, on the development of PDPH. Moreover, LPs in the study were performed by physicians of varying experience, which could have affected the results. However, all of them have received vigorous training for the procedure, and we aimed to assess this variation on the primary outcome, as this represents the real-life clinical situation in most hospitals.