Intracytoplasmic sperm injection use in states with and without insurance coverage mandates for infertility treatment, United States, 2000-2015.

A total of 1,356,377 fresh IVF cycles were included from 2000 to 2015; of these, 25.8% (n = 350,344) and 74.2% (n = 1,006,033) were performed in clinics located within states with and without a mandate, respectively. Overall, ICSI use was higher for cycles with male-factor infertility diagnosis compared with non–male-factor diagnoses, regardless of the presence of insurance mandates ( Fig. 1 ). During this time period, the increase in ICSI use among non–male-factor infertility cycles was greater among nonmandate states (34.6% to 73.9%) compared with mandate states (39.5% to 63.5%; P <.0001; Fig. 1 ). Among male-factor infertility cycles, the increase in ICSI use among nonmandate states was smaller compared with mandate states (mandate: 85.4% to 92.9%, a 7.5% percentage point increase; nonmandate: 87.7% to 94.8%, a 7.1% percentage point increase; P <.0001). Limiting our analyses to 2011–2015 yielded 445,684 fresh IVF cycles. Overall, patient and treatment characteristics varied by state mandate status ( Table 1 ). The proportions of IVF cycles for women aged ≥35 years and those of non-Hispanic black or non-Hispanic Asian race/ethnicity were higher in states with an insurance mandate for IVF compared with states without a mandate. Despite a high proportion of missing values for the race/ethnicity variable (35.0%), which may limit reliability, IVF cycles among non-Hispanic white women were of the highest proportions compared with other race/ethnicities, regardless of mandate status. For the following characteristics, the proportion of IVF cycles was lower in states with an insurance mandate: no previous live birth, all infertility diagnoses except for unexplained infertility and “other,” and Hispanic and “other” race. The overall percentage of ICSI use was higher for nonmandate states (79.4%) compared with mandate states (70.8%; P <.05). During 2011–2015, ICSI use was significantly higher in nonmandate states than in states with an insurance mandate for each of the selected non–male-factor infertility indications except PGD/PGS ( Fig. 2 ). The difference was most pronounced for low oocyte yield and unexplained infertility. Multivariable analyses for years 2011–2015 revealed that after adjustments for age, non–male-factor infertility diagnoses, and clustering by clinic, ICSI use was lower in mandate states compared with nonmandate states (91.5% vs. 94.5%; aRR 0.97, 95% confidence interval [CI] 0.94–0.999; P =.04) among male-factor infertility cycles ( Table 2 ). Similarly, for non–male-factor infertility cycles, ICSI use was lower in mandate states compared with nonmandate states (60.3% vs. 70.9%; aRR 0.85, 95% CI 0.74–0.985; P =.03). Results from the sensitivity analysis, in which the state mandate group was limited to states that specifically mention ICSI in their mandates, were similar: male-factor infertility cycles: aRR 0.96 (95% CI 0.92–0.99; P =.02); versus non–male-factor infertility cycles: aRR 0.83 (95% CI 0.68–0.999; P =.049). Clinic size, estimated by the average number of fresh cycles performed per clinic, was not significantly different by quartile of ICSI use, regardless of mandate status ( Supplemental Table 2 ). The characteristics that reached statistical significance by ICSI quartile were average proportion of cycles with two or more previous ART cycles and no live birth ( P <.05 in both mandate and nonmandate states) and average use of PGD/PGS in nonmandate states ( P <.0001), which ranged from 3.6% in clinics in the lowest quartile of ICSI use to 9.7% for clinics in the highest quartile of ICSI use.

This population-based cross-sectional study used data from the National ART Surveillance System (NASS), a web-based reporting system that allows the Centers for Disease Control and Prevention (CDC) to monitor the safety and effectiveness of ART procedures conducted in the U.S. and its territories ( 14 ). NASS was established in response to the Fertility Clinic Success Rate and Certification Act of 1992, which requires that U.S. fertility clinics report annual data on ART procedures and outcomes to the CDC. NASS data contains cycle-level information pertaining to patient demographics, reproductive history, ART procedure type, and pregnancy outcomes. NASS also contains information about male-factor infertility diagnosis, but it has very limited details on semen parameters or the severity of male-factor infertility for the years included in the present study. This analysis focused on fresh-embryo cycles performed during 2000–2015. Because information on ICSI procedures is not consistently collected for frozen-embryo transfers, frozen cycles were excluded. Additional exclusion criteria included: IVF cycles that were cancelled before egg retrieval; cycles from U.S. territories, non-U.S. residents, and gestational carriers; gamete intrafallopian transfer; and zygote intrafallopian transfer cycles. States were classified into two groups: those with an explicit infertility insurance mandate to cover IVF (n = 8; AR, CT, HI, IL, MA, M.D., NJ, and RI) and those without such mandates (n = 43; the remaining 42 states and DC). Because ICSI is performed with IVF, we assumed that the bulk of the ICSI procedure cost would be covered by the IVF mandate. In addition, we performed a sensitivity analysis of just those states that specifically mention ICSI in their coverage details. States were classified as having a mandate only in the years that a mandate was present; because NJ and CT mandates were enacted during the observation period, in 2001 and 2005, respectively, these states were classified as mandate states only for the years following the legislation (2002–2015 and 2006–2015, respectively). To assess differences in linear trends in ICSI use by mandate status over the study period, we used linear regression models, with ICSI use as the dependent variable and year of cycle start, mandate status, and the interaction of these terms included as independent variables. We restricted the study population to the most recent 5-year period (2011–2015) and used chi-square tests to compare the distribution of patient and treatment characteristics for cycles in states with and without a mandate. We considered the following variables for each patient: age, race/ethnicity, infertility diagnosis, number of previous live births, number of previous spontaneous abortions, number of previous ART cycles, oocyte/embryo source, number of oocytes retrieved, number of embryos transferred, embryo stage at transfer (days 2–3 or days 5–6), number of embryos cryopreserved, use of assisted hatching, and preimplantation genetic testing. We also compared ICSI use for selected non–male-factor indications by mandate status, including: use of preimplantation genetic diagnosis (PGD)/preimplantation genetic screening (PGS), maternal age ≥38 years, low oocyte yield (<5 oocytes retrieved), unexplained infertility, and having had two or more previous ART cycles and no history of live births (proxy for previous unsuccessful cycles). Data were missing for <2% of patient and treatment characteristics except race/ethnicity (35.0%) and variables not collected for cycles cancelled between oocyte retrieval and embryo transfer (number of embryos transferred, embryo stage, and use of assisted hatching). We also used log binomial regression to calculate adjusted risk ratios (aRRs) for the association between ICSI use and mandate status for 2011–2015. Generalized estimating equations (GEEs) were used to account for correlation of outcomes from the same clinic. The variables used in the multivariable analysis were based on a priori knowledge of potential confounders ( 15 ). We controlled for female patient age, and non–male-factor infertility diagnoses, including tubal factor, endometriosis, uterine factor, ovulatory disorder, and diminished ovarian reserve. We also included the interaction between male-factor infertility and state mandate status (the interaction term was included in the regression model); we report the relationship between ICSI and state mandate status stratified by male-factor infertility diagnosis. To investigate possible differences in underlying patient populations that may explain clinic-level variations in ICSI use in the context of non–male-factor infertility, we categorized clinics as being located in a state with or without a mandate, aggregated the data by clinic, and calculated the proportion of ICSI use in all fresh cycles in each clinic, averaged across the years 2011–2015. We then classified clinics according to quartiles of average ICSI use in non–male-factor cycles. Analysis of variance was used to compare the average number of fresh cycles performed per clinic (a proxy for clinic size), male-factor infertility cycles, and the five non–male-factor indications as previously described across the quartiles. Clinic averages for mandate and nonmandate state clinics were derived from all cycles (including male-factor and non–male-factor infertility diagnoses). The following SAS version 9.3 procedures were used to analyze the data: FREQ (descriptive statistics, categoric data analysis), GENMOD (log binomial regression with GEE), GLM (trend analysis), ANOVA (for clinic tables), and MEANS for quartiles estimates. P values of <.05 were considered to be statistically significant. This study was approved by the CDC Institutional Review Board.

Infertility insurance coverage mandates were associated with lower use of ICSI for non–male-factor infertility cycles, especially for cycles with unexplained infertility and low oocyte yield. These findings suggest there may be impacts on clinical practice from insurance coverage for infertility treatment.

ICSI use in the United States increased from 2000 to 2015, with greater increases observed for non–male-factor infertility cycles performed for residents of states without an IVF insurance coverage mandate versus those in states with a mandate. In recent years (2011–2015), residency in a state with an IVF insurance coverage mandate was associated with lower rates of ICSI use when a male-factor infertility diagnosis was not present, even after accounting for factors such as patient age and female infertility diagnoses. Furthermore, ICSI use was higher in states without an IVF insurance mandate than in states with an insurance mandate for nearly all of the non–male-factor infertility indications, especially low oocyte yield and unexplained infertility. Unlike earlier ASRM guidelines on ICSI use ( 4 , 16 ), guidelines released in 2012 ( 1 ) offer a more thorough review and conclude that current evidence does not support routine ICSI use for non–male-factor infertility indications. Nevertheless, there are several possible reasons why ICSI may be used in the absence of male-factor infertility. First, in an effort to optimize reproductive outcomes (e.g., pregnancy, live birth) for their patients, providers may use ICSI to prevent failed fertilization and overcome the intrinsic etiology of infertility (i.e., genetic problems preventing viability of the egg, low oocyte yield) or unexplained infertility, especially because many patients may not be able to afford multiple cycles of ART ( 17 ). Second, potential neonatal health risks of ICSI ( 3 – 7 ) may not be widely known, acknowledged, or considered to be important by providers and/or patients. Finally, despite the existence of national guidelines from ASRM, there remains the possibility of a financial conflict of interest in a fee-for-service environment where pressures to achieve a successful outcome need to be balanced with cost of success in a for-profit setting ( 18 ). Just as some studies have shown increased IVF use with insurance coverage mandates as overall access to ART increases in such states ( 19 ), the presence of a mandate may increase the number of cycles requiring ICSI over time. Jain and Gupta previously concluded that “there was more use of ICSI for infertility not attributed to male-factor conditions in states with mandated health insurance coverage for IVF services than in states without mandated insurance coverage.” ( 10 ) However, their study occurred at a time when fewer states with insurance mandates existed and when evidence-based clinical practice and guidelines regarding ICSI use were changing. Our study provides an updated analysis of trends in ICSI use, including eight states with current IVF mandates. We found that, despite potentially improving affordability and access to ART, the presence of an infertility insurance mandate was associated with decreased use of ICSI for non–male-factor infertility in recent years. One possible explanation for this finding is that the presence of a mandate may allow flexibility for providers and patients to consider conventional IVF or to undergo multiple IVF cycles without feeling pressured to use additional interventions, such as ICSI, to reduce the chance of fertilization failure. Another possibility is that insurance companies may be limiting coverage for ICSI when there is no demonstrated benefit compared with IVF alone. Although this premise has yet to be asserted in the literature for ICSI use specifically, literature in other areas (e.g., single-embryo transfer practices [13] and infant immunization [20]) demonstrates how shifts in insurance coverage based on evidence-based findings or guidelines can lead to improved practices and outcomes. Our clinic-level analysis did not reveal significant differences for most characteristics across varying quartiles of ICSI use, regardless of whether the clinic was located in a state with or without a mandate. Because there were few apparent differences in underlying patient populations for clinics with low versus high rates of ICSI use in cycles with non–male-factor infertility, variations in ICSI among clinics may reflect clinic-level treatment protocols rather than the characteristics of the patients undergoing treatment. Notably, we found that among nonmandate states, there was a higher average percentage of cycles with the PGD/PGS procedure performed at higher-volume clinics. Because ICSI is commonly used during PGD/PGS to ensure that the embryo being tested has no other “potential paternal contamination” ( 1 ), clinics with high rates of PGD/PGS will likely have high rates of ICSI use regardless of male-factor infertility diagnosis. Interestingly, although clinical practice guidelines do not explicitly discourage PGD/PGS use with ICSI, ASRM did not find sufficient evidence to promote routine ICSI use with PGD/PGS and without male-factor infertility diagnosis ( 1 ). The present study has several strengths. We had a large sample size accumulated over multiple years that was representative of fresh ART cycles initiated in the U.S. All fertility clinics are mandated to report to NASS, which helps to ensure completeness and representativeness of data. NASS includes more than 98% of ART cycles performed annually ( 21 ). Our study expands on the initial work performed by Jain and Gupta on data from 1995–2004 ( 10 ). At the time of their analysis, complete and comprehensive mandates were present in three states—Illinois, Massachusetts, and Rhode Island; all other states were nonmandate states ( 10 ). Since that study, ART technology has evolved, ICSI use has increased substantially, and more states have issued mandates, allowing for a more comprehensive comparison of mandate and nonmandate states. Finally, because the present analysis used individual cycle-specific rather than aggregate data, we were able to control for the effect of many potential confounding factors. Despite these strengths, our analysis has limitations. Within any state, patients may have differing access to private and public insurance ( 22 ). In addition, living in a state with a mandate does not ensure access to ICSI; only four out of the eight mandate states explicitly mentioned ICSI. Treatment access can vary also by race, income, or other factors ( 23 , 24 ) and could have affected our interpretation of the impact that mandates have on ICSI use. Another limitation is that our assumptions about the presence of a mandate do not apply to those individuals who work for self-insured employers, because state mandates do not apply to those plans ( 25 ). In addition, there may be temporal changes in state-level characteristics (besides insurance coverage) affecting ICSI use that we were unable to control for, such as population demographics or state economic features. Due to the high percentage of missing data on race/ethnicity (35.0%), we were unable to include this variable as a potential confounder in our models. Finally, the NASS does not collect detailed information on potentially confounding male partner characteristics such as semen parameters (potentially affecting accurate diagnosis of male-factor infertility) or paternal age. The current male-factor infertility diagnosis variable is validated annually in the NASS, and in 2014 it had a low discrepancy rate of 3.1%; this validation helps to ensure accuracy of reported data from the clinic to CDC ( 21 ). A few studies have shown an influence of paternal age on pregnancy outcomes from ART, though with some maternal age bias ( 26 ). Male characteristics (semen parameters, paternal age) are potential confounders that may be important to adjust for in examining the relationship between ICSI use and insurance mandate coverage. The results of this study highlight how the presence of mandated insurance coverage for IVF could affect ICSI use. We observed decreased use of ICSI in cycles without male-factor infertility among residents of states with insurance coverage mandates compared with states without such mandates, suggesting that mandates may complement evidence-based medical practices and help to limit nonindicated use of ICSI. Examining ICSI use in states before and after enactment of an insurance coverage mandate may be useful to further demonstrate the effect of insurance coverage mandates for infertility on ICSI use. In addition, clinic-level policies may also warrant special consideration, because they may affect the provision of safe and evidence-based care.