Article Text

Download PDFPDF

Epidural blood patch in post dural puncture headache: a randomised, observer-blind, controlled clinical trial
  1. F van Kooten1,
  2. R Oedit2,
  3. S L M Bakker3,
  4. D W J Dippel1
  1. 1
    Department of Neurology, Erasmus MC University Medical Centre, Rotterdam, The Netherlands
  2. 2
    Department of Neurology, Ruwaard van Putten General Hospital, Spijkenisse, The Netherlands
  3. 3
    Department of Neurology, St Franciscus General Hospital, Rotterdam, The Netherlands
  1. Dr F van Kooten, Department of Neurology, Erasmus Medical Centre, PO Box 2040 3000 CA Rotterdam, The Netherlands; f.vankooten{at}


Objectives: To determine the efficacy of epidural blood patch (EDBP) for the treatment of post dural puncture headache (PDPH).

Methods: We randomised 42 patients who presented with PDPH, lasting 24 h to 1 week, to receive EDBP (n = 19) or conservative treatment (n = 23). The primary end point was any headache at 24 h after the start of treatment. Secondary end points were presence and severity of headache after 1 week. Stratified Mantel–Haenzel analysis was used to adjust for confounders.

Results: Two patients refused to participate directly after randomisation and allocation to conservative treatment. They were excluded from the study. At 24 h after the start of treatment, headache was present in 11 (58%) patients allocated to EDBP and in 19 (90%) patients allocated to conservative treatment (RR 0.64, 95% CI 0.43 to 0.96). At day 7, headache was present in three (16%) patients allocated to EDBP and in 18 (86%) allocated to conservative treatment (RR 0.18, 95% CI 0.06 to 0.53). Headache was mild in all three EDBP patients, but in 10 of 18 conservatively treated patients who had not recovered by day 7 it was classified as moderate or severe. Adjustments for confounders did not affect these results.

Conclusions: EDBP is an effective treatment for PDPH. It offers complete resolution of symptoms in a large proportion of patients. In the remaining patients, it reduces headache severity and allows them to return to their everyday activities.

Statistics from

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.

Headache complicates 10–40% of dural punctures.1 This post dural puncture headache (PDPH) may occur immediately after spinal tap, but it starts within 48 h after the procedure in more than 90% of patients. In 80% of patients, symptoms have been reported to resolve within 7 days or less, but in a minority, they may persist for weeks or even months.2 Patients may be completely incapacitated and confined to bed because of PDPH. Obviously this has financial, social and psychological repercussions.

Several prophylactic measures have been shown to reduce the occurrence of PDPH.36 Despite these prophylactic measures, PDPH may still occur. In such cases, epidural blood patch (EDBP) may be a beneficial therapeutic intervention. EDBP has gained popularity as a therapeutic measure for PDPH, although its effectiveness has remained the subject of debate ever since the method was introduced by Gormly in the 1960s.7 8 Gormly treated six patients suffering from PDPH with EDBP, locating the epidural space with the “hanging drop” or “loss of resistance” method. All six subjects were relieved of their complaints. Many observational studies reported success rates of the EDBP for PDPH of between 70% and 90%.914 Seven controlled trials concerning prophylactic treatment have been published.1521 One of these studies was not blinded,20 three were not randomised1618 and one was only reported as an abstract.19 One other study compared prophylactic EDBP versus no blood patch among obstetric patients and reported a high success rate.15 In this study, adverse effects were not mentioned, which prevents firm conclusions. The most recent study,21 in which prophylactic EDBP was compared with a sham procedure in a double blind setting, no effect on PDPH was shown. The effectiveness of prophylactic EDPB does not seem to have been established firmly.

Only one randomised and blinded trial concerning the therapeutic effect of EDBP has been published.22 In this study, 12 patients suffering from PDPH for more than 4 days despite conservative treatment following lumbar puncture, spinal anaesthesia or myelography were randomly allocated to EDBP or sham treatment. In the placebo group, none of the patients noted complete relief of pain. In the treatment group, five of the six patients gained immediate relief. Subsequently, patients in the placebo group were also treated with an EDBP, resulting in complete relief of PDPH in all patients. The size of the study, the crossover effect and the absence of any documentation regarding the effectiveness of blinding of the observers and patients make it difficult to draw firm conclusions from this study. In a Cochrane review on prophylactic and therapeutic blood patching, it was argued that further randomised trials of epidural blood patching should be carried out before the balance of risks and benefits of this intervention can be properly assessed.23

We conducted a randomised, controlled clinical trial comparing the efficacy of the EDBP with conservative treatment, consisting of 24 h of bed rest and adequate fluid intake for the treatment of PDPH.


The BLOPP trial methods have been described elsewhere.24 The study is reported according to the CONSORT recommendations25 (trial registration No ISRCTN 71598245). The study was approved by the local medical ethics committee and review board.

Patients with PDPH were randomised using a computer database which allocated the patient to active (EDBP) or control (conservative) treatment through a random number generator. Written informed consent was obtained from all patients by asking them to return the form by mail or in person, after randomisation by telephone. Follow-up visits at 24 h and 1 week after randomisation, at which outcome was assessed, were carried out by telephone by a research nurse at the trial office. The research nurse was blinded to treatment allocation. At the beginning of the telephone interview the patients received a standard instruction not to inform the observer in any way about the treatment they had received. All patients who underwent a diagnostic lumbar puncture in our hospital were informed of the possibility of developing PDPH and received written information about the study. They were asked to contact their neurologist in the event of headache after lumbar puncture. The neurologists evaluated whether or not the headache was a PDPH. In the case of PDPH, the neurologist informed one of the investigators (RO or FvK). We informed the patient about the study by telephone and asked the patient to participate. To be included in the study, patients should have moderate or severe PDPH for longer than 24 h but not longer than 7 days after a diagnostic spinal tap. Furthermore, patients should be aged 18 years or older. Patients with contraindications for lumbar puncture (ie, haemorrhagic diathesis and space occupying intracranial lesions) and those with a body temperature >38°C were excluded from the study.

Patients allocated to active treatment received EDBP on the day of randomisation. The subject was placed in the lateral position, after which the back was flexed, sterilised and draped. Sterile gloves were used. A needle (Spinocan canule: 0.9×88 mm/206×3.5) was placed in the epidural space, using the loss of resistance technique.26 Subsequently, 20 ml of blood were drawn from the antecubital vein, and 15–20 ml were injected slowly into the epidural space, after which the needle was removed. The subject was held in the supine position for a few minutes, after which there were no further restrictions. Control treatment consisted of the advice to take 24 h bed rest and drink at least 2.0 litres of fluid a day. The use of painkillers was not prohibited. Treatment with EDBP was not an option during the study period of 7 days, not even when conservative treatment failed.

The primary outcome was the presence of headache at 24 h after the start of treatment. Headache was classified on a four point scale: absent, mild (postural headache, slightly restricting daily activities, but the patient is not confined to bed and there are no associated symptoms); moderate (postural headache confining the patient to bed for part of the day, with or without associated symptoms); and severe (postural headache in which the patient is bedridden for the entire day with associated symptoms). Associated symptoms were: nausea, vomiting, dizziness, hearing loss, hyperacusis, tinitus, photophobia, diplopia, stiffness of the neck and scapular pain.2 Secondary outcome measures were the presence and severity of headache at day 7 after the start of treatment, and the number of days until headache subsided. Additionally, we measured general well being after 7 days, expressed as the patient’s rating of his own health status, being either better, the same or worse than his condition at randomisation. Follow-up visits at 24 h and 1 week after randomisation were carried out by telephone by a research nurse at the trial office.

Statistical analysis

Differences in baseline characteristics were measured by t test and Fisher’s exact test where appropriate. The reduction in number of patients with headache after 24 h and 7 days between the two treatment strategies was expressed as a risk ratio (RR) with 95% confidence interval (CI). Adjustments were made for potential confounders such as a history of recurrent headache and the presence of acute headache as indication for the lumbar puncture, with stratification and Mantel–Haenszel weights. Time to recovery was estimated by means of Kaplan–Meier survival analysis. Additionally, the effect of treatment over time was expressed as RR with 95% CI, measured with a proportional hazards model. All analyses were carried out with Stata 8.2 (Statacorp, College Station, Texas, USA). The sample size calculations suggested that 20 patients per treatment group would be needed for 80% power to detect a relative risk reduction of persisting PDPH after 24 h of 50% (RR = 0.5) and a power of 98% to detect a relative risk reduction of 80% (RR = 0.2), at the usual significance level α = 5%, taking into account that the PDPH would have resolved spontaneously after 1 day in 10% of control patients.


The study ran for 139 weeks, from August 2002 to May 2005, and had a steady accrual (fig 1). We estimated that 285 patients underwent a diagnostic lumbar puncture at our outpatient clinic during the study period (fig 2). We expected 20% of patients to have had PDPH. In fact, 55 patients were diagnosed by their neurologist with PDPH. Of these, 13 (24%) refused to participate and 42 (76%) were included in the trial. Two of the patients who were allocated to conservative treatment insisted on being treated with EDBP. When this request was not honoured, these patients refused all further participation and withdrew consent. Eventually, 40 patients were included in the study protocol of whom 19 received EDBP and 21 received conservative treatment. None of these patients was lost to follow-up. Table 1 shows the baseline characteristics. Patients allocated to conservative treatment had a history of recurrent headaches more often than those allocated to EDBP (p = 0.02).

Figure 1 Inclusion rate of patients in the trial.
Figure 2 Trial profile. *Number of diagnostic lumbar punctures performed in our department during the study period was estimated. †Estimated number of patients with post dural puncture headache, 20% of the estimated number of patients with a diagnostic lumbar puncture.
Table 1 Baseline characteristics

Table 2 shows the main results of the study. The primary outcome, headache at 24 h after randomisation, occurred in 11 (58%) compared with 19 (90%) patients allocated active and conservative treatment, respectively (p = 0.03, RR 0.64 (0.43 to 0.96)). Seven days after randomisation, headache was still present in 3 (16%) versus 18 (86%) patients in the active and conservative groups, respectively (p<0.001, RR 0.18 (0.06 to 0.53)). Moreover, if headache was present at 24 h, it was mild in all but one patient in the active treatment group, and moderate to severe in all patients who were treated conservatively. The cumulative probability of complete recovery was 84% (95% CI 65% to 96%) after 1 week in patients allocated to EDBP and 14% (95% CI 5% to 38%) in patients allocated to conservative treatment (fig 3). The difference in recovery rates between EDBP and conservative treatment was highly significant (p (log rank) <0.0001). Remarkably, in the majority of conservatively treated patients, headache was not completely resolved, even after 7 days. The headache was severe in 24%, moderate in 24% and mild in 38% of the conservatively treated patients. Seven days after treatment with EDBP, headache was present in only three patients (16%), and in all cases this residual headache was mild.

Figure 3 Recovery from post dural puncture headache.
Table 2 Headache, back pain and general well being in patients who received active treatment and in controls

Back pain at 24 h was mostly mild and not more common in the EDBP treated group than in controls. Other complications of the blood patch were not reported by patients. More patients in the active treatment group felt better at 24 h compared with the conservative group (17 (89%) vs 4 (19%); p<0.001). Although it became smaller, this effect was still present after 7 days. After adjustment for imbalances in the frequency of a history of recurrent headache, and the presence of acute headache as indication for the diagnostic lumbar puncture between the two treatment groups, the effects of treatment remained approximately the same (table 3).

Table 3 Treatment effects

In addition, we studied the effect of an alternative dichotomisation of the primary outcome: we grouped patients without headache together with patients with mild headache as favourable outcome with the argument that moderate and severe headache is incapacitating and mild headache is not. After grouping, the effect of EDBP on headache at 24 h after treatment was even stronger (RR 0.06, 95% CI 0.01 to 0.39). The effect did not change after adjustment for potential confounders.


The BLOPP study compared conservative treatment consisting of bed rest and sufficient fluid intake with EDBP for the treatment of patients with PDPH between 24 h and 7 days after a diagnostic spinal tap. Conservative treatment consisted of the advise to take bed rest for 24 h and have a fluid intake of at least 2 litres. We chose to compare the blood patch with this strategy because it is the most frequently used treatment in general practise and there is no evidence of other effective treatment strategies. We observed that EDBP was superior by far to conservative treatment in reducing the number of patients with headache, duration of headache, as well as the severity of residual headache after the start of treatment. Moreover, the effect was not compromised by adverse effects of the EDBP such as back pain or other unexpected side effects. This finding may have been expected, in view of the results of former observational studies914 and one small randomised and blinded trial of 12 patients.22 On the other hand, evidence from observational studies is always insufficient, and evidence from randomised controlled trials was scarce when we started our study.22 Since then, one more randomised study of EDBP in PDPH was published.27 In this study, 32 patients with either PDPH after a diagnostic procedure or after epidural anaesthesia were allocated within 24 h from onset to EDBP or conservative treatment. Outcomes were patients’ self assessments of headache on a visual analogue scale ranging from 0 to 10 at 2 and 24 h after EDBP. The difference in mean scores on the visual analogue scale was 7.1 (95% CI 6.5 to 7.7). No other outcomes were reported. The method of randomisation was not described in this study, the follow-up was short and outcome assessment was apparently not blind to treatment allocation.

It may be argued that as a result of the self-limiting character of the PDPH, large numbers of patients are needed to show the potential effectiveness of EDBP. Part of the favourable effect of the blood patch can be attributed to the low recovery rate in the conservatively treated patients. We estimated beforehand that approximately 10% of the conservatively treated patients would recover spontaneously within 24 h after randomisation. In fact, only one additional patient recovered completely in the remaining 6 days of follow-up. As many as 10 (48%) patients who were treated conservatively had moderate to severe headache at day 7 after randomisation, confining them to bed for at least part of the day. This was more than we expected on the basis of previous reports.2 Because we were not convinced of the positive effect of EDBP at the time of the study, we did not administer EDBP to patients in whom conservative treatment had failed after a week. This made it possible, although it was not a part of the study protocol, to globally evaluate additional recovery in the post follow-up period based on information from the medical charts. PDPH had disappeared in all patients who were available for follow-up within a month after randomisation. This indicates that PDPH is indeed mostly self-limiting, but the process takes longer than might have been expected. Consequently, the benefit of intervention by EDBP is larger than expected.

It may be argued that the results of our study are influenced by a placebo effect because we did not apply a sham procedure in the conservative group. However, we chose to directly compare two realistic treatment strategies, avoiding adverse effects by inserting a needle in the lumbar musculature in patients who were treated conservatively. We do not believe that the placebo effect, which is undoubtedly present as part of every intervention, can entirely explain the robust decrease in the presence and severity of headache in EDBP treated patients. Concerns may rise about the blinding of the observer. We accomplished blinding of the observer by instructing patients not to give information about their treatment prior to the start of the questionnaire. Our results confirm strongly the efficacy of EDBP as treatment for PDPH. Whether EDBP should be used as a prophylactic procedure when a lumbar puncture is performed remains a subject for debate. This question was not addressed in our study.

Considering that some of our patients immediately responded to treatment, while in others the effect was noted only after a few days, we propose two pathophysiological mechanisms to explain the treatment effects. Firstly, the introduced blood clot may compress the CSF space, reduce its volume and increase the CSF pressure, which leads to prompt relief. Later, the dural tear may heal which may explain the remote effect that we found in some of our patients. Too small volumes of autologous blood may lead to failure of the first mechanism, and relapses may indicate that the second mechanism has failed. This suggests that repeating the procedure in patients who do not respond after a first blood patch is a sensible action. Some suggest that even larger volumes of blood, up to 20 or 30 ml, are needed to induce prompt relief of post punctional headache. However, our data do not allow us to draw conclusions concerning the amount of blood injected.

We conclude that EDBP is an effective treatment for PDPH. It offers complete resolution of symptoms in a large proportion of patients. In the remaining patients, it reduces the severity allowing them to return to their everyday activities within 24 h after administration. We recommend the EDBP as the treatment of choice for PDPH.


We thank Ron Meijer, neurologist, for training us in delivery of the blood patch and Naziha El Ghannouti, research assistant, for her contribution in the follow-up of patients.

FvK participated in data acquisition, coordination of the study and drafted the manuscript. RO participated in the study design, acquisition of data and helped to draft the manuscript. SLMB participated in the conception of the study and in data acquisition. DWJD participated in the conception of the study, data acquisition, data analysis and drafting of the manuscript.



  • Funding: This study was supported by The Netherlands Headache Society.

  • Competing interests: None.

  • Ethics approval: The study was approved by the local medical ethics committee and review board.