Sherman C. Stein, Mark G. Burnett and Seema S. Sonnad
The average 65-year-old patient with moderate dementia can look forward to only 1.4 quality-adjusted life years (QALYs), that is, longevity times quality of life. Some of these patients suffer from normal-pressure hydrocephalus (NPH) and respond dramatically to shunt insertion. Currently, however, NPH cannot be diagnosed with certainty. The authors constructed a Markov decision analysis model to predict the outcome in patients with NPH treated with and without shunts.
Transition probabilities and health utilities were obtained from a review of the literature. A sensitivity analysis and Monte Carlo simulation were applied to test outcomes over a wide range of parameters. Using shunt response and complication rates from the literature, the average patient receiving a shunt would gain an additional 1.7 QALYs as a result of automatic shunt insertion. Even if 50% of patients receiving a shunt have complications, the shunt response rate would need to be less than 5% for empirical shunt insertion to do more harm than good. Authors of most studies have reported far better statistics.
In summary, many more patients with suspected NPH should be considered for shunt insertion.
Mark G. Burnett, Seema S. Sonnad and Sherman C. Stein
Many tests have been proposed to help choose candidates for shunt insertion in cases of suspected normal-pressure hydrocephalus (NPH). It is unclear what sensitivity and specificity a prospective test must have to improve outcomes, compared with the results of automatic shunt insertion.
The authors adapted the decision analysis model used in a companion article to allow for application of a screening test. Using the reported sensitivities and specificities of several such tests, they evaluated the effects such tests would have on the expected outcome of an average 65-year-old patient with moderate dementia. They also evaluated the cost-effectiveness of a theoretical screening test with superior sensitivity and specificity.
Although external lumbar drainage comes quite close, none of the screening tests reported to date have sufficient sensitivity and specificity to improve expected outcome in an average candidate, compared with the results of automatic shunt placement in cases of suspected NPH. In addition, even a theoretically improved test would need to be considerably less expensive than prolonged lumbar drainage to be cost-effective in clinical practice.
Bradley C. Lega, Shabbar F. Danish, Neil R. Malhotra, Seema S. Sonnad and Sherman C. Stein
Chronic subdural hematoma (CSDH), a condition much more common in the elderly, presents an increasing challenge as the population ages. Treatment strategies for CSDH include bur-hole craniostomy (BHC), twist-drill craniostomy (TDC), and craniotomy. Decision analysis was used to organize existing data and develop recommendations for effective treatment.
A Medline search was used to identify articles about treatment of CSDH. Direct assessment by health care professionals of the relative health impact of common complications and recurrences was used to generate utility values for treatment outcomes. Monte Carlo simulation and sensitivity analyses allowed comparisons across treatment strategies. A second simulation examined whether intraoperative irrigation or postoperative drainage affect the outcomes following BHC.
On a scale from 0 to 1, the utility of BHC was found to be 0.9608, compared with 0.9202 for TDC (p = 0.001) and 0.9169 for craniotomy (p = 0.006). Sensitivity analysis confirmed the robustness of these values. Craniotomy yielded fewer recurrences, but more frequent and more serious complications than did BHC. There were no significant differences for BHC with or without irrigation or postoperative drainage.
Bur-hole craniostomy is the most efficient choice for surgical drainage of uncomplicated CSDH. Bur-hole craniostomy balances a low recurrence rate with a low incidence of highly morbid complications. Decision analysis provides statistical and empirical guidance in the absence of well-controlled large trials and despite a confusing range of previously reported morbidity and recurrence.
John Douglas Pickard, David Spiegelhalter and Marek Czosnyka