Measuring cost-effectiveness of surgical procedures.
Our review of CEA of surgical procedures suggests that much of the existing cost analysis literature does not adhere to basic recommended analytic guidelines. However, those authors who specifically planned to perform a CEA analysis met all or nearly all of the methodologic principles (Table 1). Investigators who conduct CEA are strongly encouraged to use the many outstanding methodologic reviews for CEA. An example of threshold analysis was presented by Gray et al in their CEA of laparoscopy versus laparotomy for the treatment of ectopic pregnancy. They calculated that cost per successful treatment would be equal between the two strategies at an initial failure rate of 32% for laparoscopy (compared with their baseline value of 19%). This type of analysis is helpful, in addition to sensitivity analyses, to identify the value of a variable that results in an equal outcome. In the only cost-utility analysis performed on gynecologic surgery, Sculpher studied the trade-offs between a less invasive, less costly procedure (transcervical resection of the endometrium) with a more invasive, more costly, and more effective procedure (abdominal hysterectomy) to treat menorrhagia. Hysterectomy resulted in an incremental cost of 1,500 British pounds per QALY during 2 years of follow-up. This is much less than the range of $30,000 to $100,000 that represents a currently acceptable C/E ratio. Grover et al evaluated the cost-effectiveness of performing a concurrent hysterectomy in women undergoing bilateral salpingo-oophorectomy. They observed that in 45-year-old women, the additional concurrent procedure dominated the alternative strategy of bilateral salpingo-oophorectomy, being both less expensive and increasing average life expectancy. The concurrent hysterectomy strategy also dominated for women aged 55, but both with less cost-savings and gains in life expectancy compared with 45-year-old women. Selecting an appropriate time frame for the analysis is difficult and may dramatically affect the results of the analysis. The time frame should be long enough to measure all clinically relevant costs and benefits. For example, Kung et al compared the cost per cure of stress urinary incontinence of laparoscopic and open Burch procedures. The probability of cure after each procedure was estimated from a retrospective cohort of 62 women with a mean follow-up of 1.2 years for the laparoscopic Burch strategy and 2.7 years in the open Burch strategy. The authors found that the laparoscopic Burch dominated, with lower costs and a higher cure rate. However, the analysis would be more informative with much longer follow-up, because most women who undergo an incontinence procedure have a life expectancy far greater than 1 to 2 years. Ramsey et al performed an economic analysis to assess the long-term costs of behavioral therapy, pharmacotherapy, and surgical therapy used for stress urinary incontinence. They found that in the short-term, behavioral and pharmacotherapy were less costly. However, if life expectancy was equal to or greater than 3.5 years, surgical therapy was least costly. In many articles that evaluate the cost of managing ectopic pregnancy, only short-term costs of the procedures and follow-up visits are considered. Mol et al considered a longer time frame and also included the costs of infertility management based on the future probability of conception correlated with the different management strategies. Selection of an effectiveness measure after surgical intervention is often difficult and controversial. For benign disease, life years or QALYs will be minimally affected by a reasonably safe intervention. In the short-term, utility may be negatively affected by surgery and recovery. In longer-term analyses, these effects will be diluted by time and be negligible. Intermediate measures such as days of hospitalization averted or lives saved are often more appropriate for gynecologic interventions than are longer-term outcomes such as lif