2018 Section 5 - Rhinology and Allergic Disorders

Otolaryngology–Head and Neck Surgery 156(1)

Figure 3. Forest plot illustrating the odds ratios (ORs) for successful endoscopic cerebrospinal fluid leak repair with and without lumbar drain (LD). ‘‘Events’’ represent number of successful repairs. CI, confidence interval.

Figure 4. Forest plot illustrating the odds ratios (ORs) for successful endoscopic cerebrospinal fluid leak repair with and without lumbar drain (LD) among patients whose origin of leak was anterior skull base resection. ‘‘Events’’ represent number of successful repairs. CI, confi- dence interval.

CSF leak repair must be carefully weighed against the asso- ciated risks, such as headache, retained catheter tip, celluli- tis at the puncture site, and sequelae related to overdrainage (ie, tension pneumocephalus, subdural hemorrhage, and uncal herniation). 7,10 Although LDs have been implicated in creating an increased risk of meningitis, the rates of meningitis between the aggregate treatment arms in our study were similar (4.7% in patients with LDs vs 3.9% in patients without LDs). 10,45 Headaches, how- ever, occurred with higher frequency in patients with LDs (21.0%) versus those without LDs (8.9%). Despite the risks associated with LDs, we do, however, agree with other authors who have suggested that a particular subpo- pulation likely stands to benefit from CSF diversion: patients at high risk of CSF leak recurrence, such as those with high-flow defects and risk factors that deter healing of skull base repairs (history of irradiation, intracranial hypertension, etc). 10,13,34 Given the limited resources of our health care system, one must also consider the length of hospitalization and cost associated with perioperative lumbar drainage in the setting of endoscopic CSF leak repair. Zuckerman and DelGaudio 46 found that length of hospitalization for patients with LDs averaged 5.5 days, compared with only 1.7 days in patients without LDs. They also demonstrated that placement of an

LD added, on average, $3124 per patient to the overall cost of hospitalization. Our study has several limitations. Given the lack of level 1 and 2 evidence on the subject matter, 11 of the 12 studies that met the inclusion criteria were retrospective observa- tional studies and largely composed of case series. As such, selection bias and reporting bias are expected to influence the data reported. Although publication bias was not detected among studies, such bias for the overall data set is a concern, as authors with favorable results are typically more inclined to publish their findings. Only 2 of the 12 studies (Albu et al 22 and Caballero et al 23 ) had a robust number of subjects in each arm and thus were assigned much more weight than other studies in the meta-analysis, possibly skewing the overall results. Additionally, the popu- lations among the studies were highly variable, as were other factors, such as techniques of repair, materials used for repairs, and pre- and postoperative modality of CSF leak determination. Conclusions There is insufficient evidence to suggest that adjunctive lumbar drainage significantly reduces the rate of postopera- tive CSF leak recurrence in patients undergoing endoscopic CSF leak repair. Subgroup analysis examining only those

270