Of Competitive Pressures and Care Quality: Infertility Treatment Case Helen Schneider Department of Economics, University of Texas at Austin
Importance of the Study About 11 percent of American women 15 44 years of age have difficulty getting pregnant or carrying a pregnancy to term (CDC). Today, over 1 percent of all infants born in the United States every year are conceived using assisted reproductive technologies (ART). To meet this increased demand for ART, the number of infertility clinics in the United States has increased from 263 in 1995 when CDC started collecting the data to 456 in 2012. Today, in vitro fertilization (ivf) is the most successful infertility treatment but it is costly. On average, nationally, a fresh IVF cycle costs $12,000, before medications, which typically add $3,000 to $5,000. (Forbes 2014) PGD and other tests add extra cost. To address costs, many states passed insurance mandates that require employers to cover - or offer to cover- infertility treatments.
Previous Literature Competition and Quality of Health Outcomes Hospital competition decreases mortality rates (Sari 2002, Kessler & McClellan 2000, Schneider 2008, Gaynor et al. 2011) while Propper et al. 2008 find higher mortality rates in competitive markets Mergers increase readmission rates but do not affect mortality rates (Ho & Hamilton 2000) Mutter et al. 2008 look at 12 different dimensions of inpatient quality. They find that the effect of competition is not unidirectional. Infertility Treatment Markets: The Effect of Competition With more ivf clinics entering the market, many fear that under competitive pressures doctors will pursue aggressive treatments so that the clinics can advertise high success rates. Limits on competition have been proposed. (Kolata 2002, Bergh et al. 1999, Wells 1999) Steiner 2005 measured competition as number of clinics in the area and found that competition did not affect pregnancy rates but decreased high order multiples (triplets and higher). Hamilton & McManus 2005 measure competition with a simple dummy variable (1=monopoly, 0-otherwise). They find that competition does not increase multiple birth rate.
Literature Review (continues) Infertility Treatment Markets: The Effect of Insurance Mandates Universal insurance mandates are associated with greater utilization of ART and other infertility treatments (ovulationinducing drugs and artificial insemination (Henne & Bundorf 2008; Hamilton & McManus 2011; Bitler & Schmidt 2012) Universal insurance mandates decrease multiple births per ART birth (Henne & Bundorf 2008; Hamilton & McManus 2011) Universal insurance mandates increase triplet and higherorder births by 26% (Buckles 2012)
Contribution to Previous Research Recent data allows for a better measure of competitive pressures than what we see from previous research This study calculates competition index (HHI) to measure competitive pressures that ivf clinics face The study examines the relationship between competition and infertility mandates
Empirical Model: Competition Variable We use Herfindahl Hirschman Index (HHI) to measure market competition. The index is constructed based on total ivf cycles performed for each clinic. Increases in the Herfindahl index generally indicate a decrease in competition and an increase of market power, whereas decreases indicate the opposite. The index can vary from zero (perfect competition) to 10,000 (monopoly). We use metropolitan statistical area (MSA) as the relevant market for infertility clinics in our sample.
Empirical Model: Infertility Mandates By 2012 fifteen states passed infertility mandates of which six states (Hawaii, Illinois, New Jersey, Massachusetts, Maryland, and Rhode Island) require all insurance plans to cover ivf. In addition, Arkansas, Montana and Ohio require some plans (all HMO s or all non-hmo s) to cover the costs ivf treatments. We use both definitions of the universal mandate to test the sensitivity of our results. In our definition of mandated infertility benefits, we do not include states like Texas that only require health insurance plants to offer infertility insurance since employers have the right to refuse such coverage. We also exclude states like California that require coverage of all infertility treatments except ivf.
Empirical Model The following empirical model is used in this study: Quality ais 0 1Mandates 2HHIi 3( Mandates HHIi ) 4 Market s ai where the dependent variable measures quality of health outcomes for age cohort a in state s for clinic i. Coefficient β 1 captures the effect of state infertility mandates, coefficient β 2 captures the effect of market competition and β 3 captures the interaction of competition and state mandates. Variable Market s is a vector of controls for variables that vary across states that might also affect the market. These include: median family income (at MSA level), population, female labor force participation rate, and percentage of women educated (women with at least Bachelor degree at state level).
Data We use 2012 ART Fertility Clinic Success Rates Report. The data is publicly available by Center of Disease Control and Prevention (CDC). The unit of analysis is a clinic performing ART (no patient level data is available). Clinic-level data was collected on 176,247 ART cycles at 456 reporting clinics in the United States during 2012, resulting in 51,267 live births (deliveries of one or more living infants) and 65,160 live born infants (CDC). Market area characteristics came from publicly available state and MSA-level data. Female labor force participation for year 2012 was collected by the Bureau of Labor Statistics (BLS) at the state level. Percentage of educated women variable is based on National Center for Education Statistics report. This data is collected at the state level and captures percent of women with at least a bachelor s degree. MSAlevel income per variable data came from the US Census Bureau. Data on state infertility mandates came from the American Society for Reproductive Medicine.
Descriptive statistics for selected variables Mean (st. deviation) Ivf cycles for women aged under 92.01 35 (123.27) Ivf cycles for women aged 35-39 89.21 (145.63) Ivf cycles for women aged 40 and 38.35 above (78.22) Average embryos transferred for 1.97 women aged under 35 (0.30) Average embryos transferred for 2.30 women aged under 35-39 (0.36) Average embryos transferred for 2.74 women aged 40 and above (0.76) Mandate 0.15 (0.36) HHI 4138.93 (3289.28) Multiples rate for women aged 31.2 under 35 (17.2) Multiples rate for women aged 26.9 35-39 (20.5) Min Max 0 1463 0 1694 0 884 1 4.3 1 3.8 1 6 0 1 576.52 10000 0 1 0 1
Descriptive statistics continued Multiples rate for women aged 40+ Mean (st. deviation) 29.2 (20.1) ICSI 70.17 (19.48) PGD 5.904 (11.26) Women s education Income per capita 28.245 (4.66) 47,698.83 (8975.783) Min Max 0 1 0 100 0 100 17.4 48.6 22,400 81,068
Preliminary Results Number of ivf cycles HHI does not increase ivf cycles Health insurance mandates have a positive and significant effect on ivf cycles (p< 0.01 for women under 35 and p<0.05 for women between 35 and 39) Other significant variables include women s education and SART membership. Both significantly increase average number of cycles for all women. Average embryos transferred For women younger than 35 none of the policy variables are significant For women over 35 years of age competition decreases number of embryos transferred (P<0.1); mandates decrease number of embryos transferred although the effect is not statistically significant; interaction variable is positive and significant (p<0.1 for women 35-40 and P<0.05 for women >40) Other significant variables are women s education, pgd rate (negative and statistically significant effect), SART membership (significant for women under 35)
Preliminary Results: Multiple gestations women 35-39 women 35-39 women >40 women> 40 Mandate -2.383 (3.28) -1.057 (2.537) -8.62 (12.18) -16.107* (8.84) HHI 0.000565** (0.000314) 0.000459* (0.000250) 0.00297* (0.00178) 0.00973* (0.00567) Volume 0.00572 (0.00515) -0.00495 (0.00520) 0.0270 (0.0448) 0.0310 (0.0457) Income per capita 0.000286** (0.000132) 0.000314** (0.000131) 0.00108 (0.000678) 0.00119* (.000650) Women s education -0.533** (0.266) -0.487* (0.253) -2.39* (1.43) -1.20 (1.39) ICSI -0.0608 (0.0531) -0.0640 (0.0523) -0.448 (0.275) -0.208 (-.259) PGD -0.0502 (0.0666) -0.0483 (0.0683) -1.029** (0.444) -1.096*** (0.391) SART membership -11.874*** (3.378) -11.854*** (3.389) 0.207 (16.92) -2.633 (15.338) Interaction 0.00158* (0.0009) 0.0105** (0.00511) N 406 406 404 404 F 2.53** 2.56*** 2.33** 2.34** R-squared 0.0748 0.0848 0.0994 0.0997
Future research Estimate the effect of changes in HHI on multiple gestations Re-estimate the model for high order multiples Sensitivity analyses: use a different definition of market
Policy Implications Additional competition can significantly increase the number of patients while decreasing multiple birth rates especially for older patients. Competition in medical industry can also lower prices and cost of care (Dranove & White 1994, Gaynor & Haas-Wilson 1999, Keeler et al. 1999) Mounting empirical evidence shows that competitive pressures can decrease costs, prices while improving quality of care, thus improving patients welfare. Antitrust policies as well as technological improvements can play a role in determining improvements in health care quality. Antitrust analysis should incorporate the potential effects of procompetitive policies on health outcomes.