Cost-effectiveness of emergency contraception options over 1 year

Abstract

Background: The copper intrauterine device is the most effective form of emergency contraception and can also provide long-term contraception. The levonorgestrel intrauterine device has also been studied in combination with oral levonorgestrel for women seeking emergency contraception. However, intrauterine devices have higher up-front costs than oral methods, such as ulipristal acetate and levonorgestrel. Health care payers and decision makers (eg, health care insurers, government programs) with financial constraints must determine if the increased effectiveness of intrauterine device emergency contraception methods are worth the additional costs.

Objective: We sought to compare the cost-effectiveness of 4 emergency contraception strategies-ulipristal acetate, oral levonorgestrel, copper intrauterine device, and oral levonorgestrel plus same-day levonorgestrel intrauterine device-over 1 year from a US payer perspective.

Study design: Costs (2017 US dollars) and pregnancies were estimated over 1 year using a Markov model of 1000 women seeking emergency contraception. Every 28-day cycle, the model estimated the predicted number of pregnancy outcomes (ie, live birth, ectopic pregnancy, spontaneous abortion, or induced abortion) resulting from emergency contraception failure and subsequent contraception use. Model inputs were derived from published literature and national sources. An emergency contraception strategy was considered cost-effective if the incremental cost-effectiveness ratio (ie, the cost to prevent 1 additional pregnancy) was less than the weighted average cost of pregnancy outcomes in the United States ($5167). The incremental cost-effectiveness ratios and probability of being the most cost-effective emergency contraception strategy were calculated from 1000 probabilistic model iterations. One-way sensitivity analyses were used to examine uncertainty in the cost of emergency contraception, subsequent contraception, and pregnancy outcomes as well as the model probabilities.

Results: In 1000 women seeking emergency contraception, the model estimated direct medical costs of $1,228,000 and 137 unintended pregnancies with ulipristal acetate, compared to $1,279,000 and 150 unintended pregnancies with oral levonorgestrel, $1,376,000 and 61 unintended pregnancies with copper intrauterine devices, and $1,558,000 and 63 unintended pregnancies with oral levonorgestrel plus same-day levonorgestrel intrauterine device. The copper intrauterine device was the most cost-effective emergency contraception strategy in the majority (63.9%) of model iterations and, compared to ulipristal acetate, cost $1957 per additional pregnancy prevented. Model estimates were most sensitive to changes in the cost of the copper intrauterine device (with higher copper intrauterine device costs, oral levonorgestrel plus same-day levonorgestrel intrauterine device became the most cost-effective option) and the cost of a live birth (with lower-cost births, ulipristal acetate became the most cost-effective option). When the proportion of obese women in the population increased, the copper intrauterine device became even more most cost-effective.

Conclusion: Over 1 year, the copper intrauterine device is currently the most cost-effective emergency contraception option. Policy makers and health care insurance companies should consider the potential for long-term savings when women seeking emergency contraception can promptly obtain whatever contraceptive best meets their personal preferences and needs; this will require removing barriers and promoting access to intrauterine devices at emergency contraception visits.

Keywords: cost-effectiveness analysis; emergency contraception; incremental cost-effectiveness ratio; intrauterine device.

FIGURE 1. Detailed decision analytic model structure

Blue square represents decision node, or point at which treatment is chosen. Green circles represent chance nodes after which probability is assigned to each event. Purple “M” circles represent Markov nodes after which women transition between health states each menstrual cycle. Red triangles represent terminal nodes, which, in Markov node, indicate state to which women will transition in next cycle. Women accrue costs and effectiveness throughout time horizon based on health states and events that occur during each cycle. Model assumed that women who discontinued contraception would not use contraception for remainder of time horizon. However, if they experienced pregnancy outcome, they may have started contraception. Ectopic pregnancy, induced abortion, and spontaneous abortion health states were tunnel states where patients spent 2–3 menstrual cycles before being forced into nonpregnant health state. Patients who became pregnant and went on to a live birth were assumed to not get pregnant again during time horizon. Tier-1 methods include intrauterine devices (IUDs) and implants; tier-2 methods include injection, pill, patch, ring; and tier 3 includes barrier methods. Cu, copper T380; EC, emergency contraception; LNG, levonorgestrel; UPA, ulipristal acetate. Bellows et al. Cost and emergency contraception. Am J Obstet Gynecol 2018.

FIGURE 2. Cost-effectiveness of emergency contraception (EC) over 1 year

A, Incremental cost-effectiveness scatterplot of each EC strategy vs ulipristal acetate (UPA) in model 1000 iterations. Each point on scatterplot represents mean incremental costs and incremental pregnancies prevented in 1000 women in 1 iteration of model compared to UPA. Larger diamonds represent mean incremental costs and mean incremental pregnancies prevented over all 1000 model iterations. Dashed line represents willingness-to-pay (WTP) threshold of $5167 to prevent pregnancy. On average, oral levonorgestrel (LNG) cost more and was less effective than UPA. In contrast, on average copper T380 (Cu) intrauterine device (IUD) cost more than UPA, but also prevents more pregnancies and does so at acceptable cost. B, Cost-effectiveness acceptability curve shows probability that each EC strategy is most cost-effective across range of WTP values over 1000 model iterations. B, Cu IUD had highest probability of being most cost-effective EC when WTP to prevent pregnancy was above about $3000. Bellows et al. Cost and emergency contraception. Am J Obstet Gynecol 2018.

FIGURE 3. Ten most influential input parameters

Ten most influential variables are shown using INMB framework with UPA as reference group. Horizontal bars represent range of highest INMBs obtained with any emergency contraception (EC) strategy when that variable was varied across range shown at ends of each bar. Dotted and dashed lines represent deterministic INMB for each EC strategy vs UPA. Copper T380 (Cu) intrauterine device (IUD) was preferred strategy (ie, most cost-effective) across nearly all of 1-way sensitivity analyses. Change in preferred EC strategy to UPA (black bars) or oral levonorgestrel (LNG) + LNG IUD (gray bars). In this analysis, change in preferred strategy only occurred at extreme values in 1-way sensitivity analyses. Tier-1 methods include IUDs and implants; tier-2 methods include injection, pill, patch, ring; and tier 3 includes barrier methods. Bellows et al. Cost and emergency contraception. Am J Obstet Gynecol 2018.