Assessment: Prevention of post–lumbar puncture headaches

Abstract

Headache (HA) is a common sequel to lumbar puncture (LP), whether performed for diagnosis or anesthesia. 1-6 In their monograph summarizing the world literature through 1960, Tourtellotte et al. 1 considered separately three principal patient populations: 1) patients undergoing diagnostic LPs (excluding myelography, pneumoencephalography, and cisternal puncture), excluding also patients whose condition might reduce the reliability to report HA; 2) patients undergoing nonobstetric spinal anesthesia; and 3) patients undergoing obstetric spinal anesthesia. They reported several observations. 1. The average frequency of post-LP HAs (PLPHA) in patients after diagnostic LP (excluding myelography, pneumoencephalography, and cisternal puncture), excluding also patients whose condition might reduce the reliability to report HA, was 32%. For nonobstetric spinal anesthesia, the average frequency was 13%. For obstetric spinal anesthesia, the average frequency was 18%. 2. In reports in which patients received special measures to prevent PLPHA, the average frequencies were 6% for diagnostic LPs, 5.5% for nonobstetric spinal anesthesia, and 6.2% for obstetric spinal anesthesia. The actual frequencies in individual series ranged from 0 to 18%. 3. The frequency of PLPHA was 36% in their own series of 105 normal individuals, 30% in 317 patients with diagnostic LPs, and 2% definite and 2% probable in 100 patients undergoing spinal anesthesia (but 30 patients with HAs of other types were excluded from the latter count). In analyzing risk factors for PLPHA, they concluded that the evidence, including their own prospective series, was convincing to consider younger age and female gender as definite risk factors. They attributed the difference of PLPHAs in obstetric and nonobstetric patients undergoing spinal anesthesia at least in part to these factors. They further considered the data fairly convincing that the smaller the needle size, the lower the frequency of PLPHA, but were unable to show this in their prospective series. They commented that the great variability of HA frequency for the same needle size between authors may reduce the reliability of this observation. With regard to all other risk factors, they concluded that the evidence was inconclusive. With regard to all preventive or therapeutic measures, they commented that proponents of a particular treatment, in general, found it to be beneficial; however, some of the reports were uncontrolled and some results could not be replicated by others. Today, we would consider the latter findings a result of publication bias-if one first tried a new approach to reducing the incidence of PLPHA, one would be less likely to publish a failure than a success. They commented on the different frequency of PLPHA in patients undergoing diagnostic LP compared with those undergoing spinal anesthesia, and considered the following factors in their series: age, gender, needle size, fasting, hydration, premedication and postoperative medication, minimal amount of trauma to the meninges, duration of recumbency, and the amount of CSF removed. Even after controlling for age and gender, the frequency of PLPHA in the spinal anesthesia group was low. They speculated that "if patients undergoing an LP for diagnostic purposes were treated like patients undergoing spinal anesthesia the incidence ['frequency'] of PLPHAs could be markedly reduced." 1 The work of Tourtellotte et al. 1 demonstrated the large variability in the frequency of PLPHA in different settings and in different series, and the apparent ability to reduce this frequency, based on uncontrolled reports. The average frequency that they reported in their monograph has been replicated or even exceeded in more recent experience (e.g., 37% in the From the