Carolyn Westhoff

Columbia University, New York, New York, United States

Are you Carolyn Westhoff?

Claim your profile

Publications (245)1079.56 Total impact

  • Karla Maguire · Sophie Joslin-Roher · Carolyn L Westhoff · Anne R Davis
    [Show abstract] [Hide abstract]
    ABSTRACT: Objective: To assess baseline dysmenorrhea and insertion-related pain as predictors of intrauterine device (IUD) removal within 1 year following insertion. Study design: System-wide medical record abstraction 1 year after IUD insertion to identify removals and comparison of baseline characteristics (dysmenorrhea, insertion pain) among women with a removal versus women who continued IUD use. Baseline data came from a randomized trial of pain control during insertion. Results: Among 199 insertions, we identified 21 removals and 7 expulsions; a continuation rate of 85.9%. Women with IUD removal had higher median dysmenorrhea scores before insertion than those who continued (42 vs. 25.5, p=0.03). Insertion pain and other characteristics were not associated with removal. Conclusions: Pre-existing dysmenorrhea may predict IUD removal within 1 year.
    Contraception 09/2015; DOI:10.1016/j.contraception.2015.09.005 · 2.34 Impact Factor
  • [Show abstract] [Hide abstract]
    ABSTRACT: Objective: Bleeding data in contraceptive trials areoftencollected using daily diaries, but data quality may vary due to compliance and recall bias. Text messaging is a widespread and promising modality for data collection. Study design: This trial randomized participants 1:1 to use text messages or paper diaries to report on bleeding experiencedduring the 90 days after intrauterine device insertion(IUD). Participants chose either the copper T380A or the 52mg levonorgestrel IUD.Our primary outcome was number of days of reported bleeding data. We hypothesized data gathered with daily text messages would have fewer missing values than paper diaries. Intention to treat analyses used the rank-sum test to compare medians. Results: Two hundred thirty women enrolled, and randomization yielded groups similar in baseline characteristics. Twenty percent of participants provided no bleeding data; of these, 77% were assigned to paper diaries. With 90 days of reporting, approximately twenty percent in each group provided complete bleeding data.The text groupreported amedian of 82 days (interquartile range [IQR] 40-89) and the paper group 36 (IQR 0-88) (p=<0.001).The number of responses received decreased gradually over the 90-day period, but was always higher in the text group. Women who had attained higher levels of education did well regardless of data collection modality, while text messages response rates were greater among those with a high school education or less (p < 0.01). Conclusions: Participants reporting bleeding via text messages provided more complete data than womenusingpaper diaries. Implications: Depending on resources and population of interest, text messages may be a useful modality to improve data collection for patient-reported outcomes.
    Contraception 09/2015; DOI:10.1016/j.contraception.2015.09.004 · 2.34 Impact Factor
  • Carolyn L Westhoff · Lara S Yoon · Rosalind Tang · Vina Pulido · Andrew Eisenberger
    [Show abstract] [Hide abstract]
    ABSTRACT: Venous thromboembolism (VTE) is rare among young women and is often presumed to occur in the setting of a genetic predisposition or during the use of estrogen-containing combined hormonal contraceptive or to have an unknown cause. This study aims to describe the distribution of VTE risk factors among women with a confirmed VTE. We identified all women aged 15-46 years with a VTE diagnosis at Columbia University Medical Center from 2005 to 2012 using medical center databases. We then reviewed all electronic medical records to validate the diagnoses and identify risk factors associated with each confirmed case. We identified 315 cases and confirmed 186 (59%). The proportion of unconfirmed cases increased over time. Forty percent of confirmed cases were associated with hormonal contraceptives or pregnancy. Ninety-five percent of confirmed cases had identifiable major risk factors including a personal history, family history, malignancy or other predisposing illness, recent long-haul travel, trauma, hospitalization, and obesity; many had multiple simultaneous risk factors. None of the confirmed cases was associated with a previously known genetic predisposition, but in 10 confirmed cases a genetic predisposition was identified during evaluation. In only 10 of the 186 confirmed cases could we not identify any acquired risk factor, and only 2 of those 10 women had a genetic predisposition. Many reproductive age women experiencing a VTE have risk factors unique to this group, and most have multiple risk factors, confirming that this is a multifactorial disease. The large proportion of unconfirmed cases suggests the need for great caution in using administrative databases for research due to poor diagnostic specificity and due to lack of information about additional risk factors.
    Journal of Women's Health 08/2015; DOI:10.1089/jwh.2015.5259 · 2.05 Impact Factor
  • Contraception 08/2015; DOI:10.1016/j.contraception.2015.08.005 · 2.34 Impact Factor
  • Dr. Carolyn L. Westhoff
    Contraception 07/2015; 92(1). DOI:10.1016/j.contraception.2015.05.001 · 2.34 Impact Factor
  • [Show abstract] [Hide abstract]
    ABSTRACT: To examine the relationship of ovulation-stimulating drugs to risk of cancers other than breast and gynecologic malignancies. Retrospective cohort study, with additional follow-up since initial report. Five US reproductive endocrinology practices. Among a cohort of 12,193 women evaluated for infertility between 1965 and 1988, a total of 9,892 women (81.1% of the eligible population) were followed through 2010, via passive and active (questionnaire) approaches. None. Hazard ratios (HRs) and 95% confidence intervals (CIs) for various fertility treatment parameters for select cancers. During 30.0 median years of follow-up (285,332 person-years), 91 colorectal cancers, 84 lung cancers, 55 thyroid cancers, and 70 melanomas were diagnosed among study subjects. Clomiphene citrate (CC), used by 38.1% of patients, was not associated with colorectal or lung cancer risks, but was related significantly to melanoma (HR = 1.95; 95% CI: 1.18-3.22), and non-significantly to thyroid cancer risks (HR = 1.57; 95% CI: 0.89-2.75). The highest melanoma risks were seen among those with the lowest drug exposure levels, but thyroid cancer risk was greatest among the heavily exposed patients (HR = 1.96; 95% CI: 0.92-4.17) for those receiving >2,250 mg. Clomiphene citrate-associated risks for thyroid cancer were somewhat higher among nulligravid, compared with gravid, women, but did not differ according to distinct causes of infertility. Gonadotropins, used by only 9.7% of subjects, were not related to risk of any of the assessed cancers. Our results provide support for continued monitoring of both melanoma and thyroid cancer risk among patients receiving fertility drugs. Published by Elsevier Inc.
    Fertility and sterility 07/2015; DOI:10.1016/j.fertnstert.2015.06.045 · 4.59 Impact Factor
  • Beverly Winikoff · Carolyn Westhoff
    Contraception 07/2015; 92(3). DOI:10.1016/j.contraception.2015.06.019 · 2.34 Impact Factor
  • [Show abstract] [Hide abstract]
    ABSTRACT: Unintended pregnancy remains a serious public health challenge in the USA. We assessed the effects of an intervention to increase patients' access to long-acting reversible contraceptives (LARCs) on pregnancy rates. We did a cluster randomised trial in 40 reproductive health clinics across the USA in 2011-13. 20 clinics were randomly assigned to receive evidence-based training on providing counselling and insertion of intrauterine devices (IUDs) or progestin implants and 20 to provide standard care. Usual costs for contraception were maintained at all sites. We recruited women aged 18-25 years attending family planning or abortion care visits and not desiring pregnancy in the next 12 months. The primary outcome was selection of an IUD or implant at the clinic visit and secondary outcome was pregnancy within 12 months. We used generalised estimating equations for clustered data to measure the intervention effect on contraceptive selection, and used survival analysis to assess pregnancy rates. Of 1500 women enrolled, more at intervention than control sites reported receiving counselling on IUDs or implants (565 [71%] of 797 vs 271 [39%] of 693, odds ratio 3·8, 95% CI 2·8-5·2) and more selected LARCs during the clinic visit (224 [28%] vs 117 [17%], 1·9, 1·3-2·8). The pregnancy rate was lower in intervention group than in the control group after family planning visits (7·9 vs 15·4 per 100 person-years), but not after abortion visits (26·5 vs 22·3 per 100 person-years). We found a significant intervention effect on pregnancy rates in women attending family planning visits (hazard ratio 0·54, 95% CI 0·34-0·85). The pregnancy rate can be reduced by provision of counselling on long-term reversible contraception and access to devices during family planning counselling visits. William and Flora Hewlett Foundation. Copyright © 2015 Elsevier Ltd. All rights reserved.
    The Lancet 06/2015; 386(9993). DOI:10.1016/S0140-6736(14)62460-0 · 45.22 Impact Factor
  • [Show abstract] [Hide abstract]
    ABSTRACT: Evaluate the levonorgestrel pharmacokinetics in nonobese and obese women after placement of LNG20 intrauterine contraceptive at 3 years of use. In a clinical trial to evaluate the LNG20 intrauterine contraceptive for contraception, levonorgestrel plasma concentrations were determined in 279 participants at the 36-month visit. Blood samples were collected from 166 nonobese LNG20, 77 obese LNG20, 23 nonobese Mirena, and 13 obese Mirena participants. Levonorgestrel plasma levels between nonobese and obese participants were compared using a Wilcoxon Mann-Whitney test. The mean plasma levonorgestrel concentrations at month 36 was 135±51 pg/mL for LNG20 overall; 145.8 ±47.7 pg/mL and 111.3±50.4 pg/mL for LNG20 nonobese and obese participants and 132.1±62.6 pg/mL for Mirena overall; 148.5±71.5 pg/mL and 103.2±27.5 pg/mL for Mirena nonobese and obese participants, respectively. Obese participants using LNG20 and Mirena had lower plasma levels than nonobese participants (P<.5). No difference in efficacy in obese and nonobese LNG20 participants was observed for the first 3 years of clinical use. Levonorgestrel plasma concentration at month 36 did not differ between LNG20 and Mirena nonobese and between LNG20 and Mirena obese participants. Obese women demonstrate significantly lower levonorgestrel concentrations than nonobese women in LNG20 with comparable efficacy in both groups, highlighting the predominantly local mechanism of action of the intrauterine contraceptive.
    Obstetrics and Gynecology 05/2015; 125 Suppl 1:84S-85S. DOI:10.1097/ · 5.18 Impact Factor
  • [Show abstract] [Hide abstract]
    ABSTRACT: The pelvic examination is under scrutiny given revised cervical cancer screening guidelines. Recent studies have examined physician attitudes and practices, but data are scarce on women's preferences. General mistrust of medical care is higher among Hispanics and blacks, but we know little about attitudes toward the pelvic examination by race and ethnicity. We examined variation in pelvic examination attitudes with baseline data from a cluster randomized trial on contraception. Women ages 18-25 years (n=1,500) were enrolled at 40 Planned Parenthood clinics across the United States. This secondary analysis used multivariable logistic regression with generalized estimating equations for clustering to examine variation in pelvic examination attitudes by race and ethnicity, controlling for individual variables (age, parity, prior contraceptive use, health insurance), site characteristics, and intervention. Hispanic women (67%) were less likely than black (84%) and white (85%) women to have had a pelvic examination (P=.001). However, Hispanic (61%) and black (69%) women were more likely than white women (43%) to report they would want a pelvic examination even if not required for contraception (Hispanic: adjusted odds ratio [OR] 2.2, 95% confidence interval [CI] 1.8-2.7; black: adjusted OR 3.2, 95% CI 2.3-4.4). Hispanic (8%) and black (10%) women were less likely than white women (17%) to have delayed going to a clinic for contraception to avoid a pelvic examination (Hispanic: adjusted OR 0.5, 95% CI 0.4-0.7; black: adjusted OR 0.6, 95% CI 0.4-0.9). The majority of young women in this contraceptive study had a pelvic examination, and perspectives on the examination varied by race and ethnicity. Women should be given the opportunity to avoid the examination when not medically necessary, as in provision of contraceptives.
    Obstetrics and gynecology; 05/2015
  • Source
    [Show abstract] [Hide abstract]
    ABSTRACT: To assess 3-year data on the efficacy and safety of a new 52-mg levonorgestrel intrauterine contraceptive (LNG20) designed for up to 7 years use. Nulliparous and parous women aged 16-45 years at enrollment with regular menstrual cycles and requesting contraception were enrolled in an open-label, partially randomized trial to evaluate LNG20. The primary outcome was pregnancy rate for women aged 16-35 years calculated as the Pearl Index. Women aged 36-45 years received LNG20 for safety evaluation only. All participants had in-person or phone follow-up approximately every 3 months during the study. A total of 1600 women aged 16-35 years and 151 women aged 36-45 years agreed to LNG20 placement, including 1011 (57.7%) nulliparous and 438 (25.1%) obese women. Successful placement occurred in 1714 (97.9%) women. Six pregnancies occurred, four of which were ectopic. The Pearl Index for LNG20 was 0.15 (95% CI 0.02-0.55) through Year 1, 0.26 (95% CI 0.10-0.57) through Year 2, and 0.22 (95% CI 0.08-0.49) through Year 3. The cumulative life-table pregnancy rate was 0.55 (95% CI 0.24-1.23) through 3 years. Expulsion was reported in 62 (3.5%) participants, most (50 [80.6%]) during the first year of use. Of women who discontinued LNG20 and desired pregnancy, 86.8% conceived spontaneously within 12 months. Pelvic infection was diagnosed in 10 (0.6%) women. Only 26 (1.5%) LNG20 users discontinued due to bleeding complaints. The LNG20 intrauterine system is highly effective and safe over 3 years of use in nulliparous and parous women. A new 52-mg levonorgestrel-releasing intrauterine system is effective and safe for nulliparous and parous women for at least 3 years. Copyright © 2015. Published by Elsevier Inc.
    Contraception 04/2015; 92(1). DOI:10.1016/j.contraception.2015.04.006 · 2.34 Impact Factor
  • [Show abstract] [Hide abstract]
    ABSTRACT: Female adolescents at high pregnancy risk frequently visit the emergency department (ED) and lack primary providers. It is unclear if current methods of ED referral are successful. The objective of this pilot study was to assess the potential effect and feasibility of a standardized, enhanced method of referral of sexually active females from an ED for family planning (FP). We conducted an ED-based intervention study using an enhanced referral process, which included a wallet card advertising a walk-in, adolescent-friendly FP clinic and a standardized ED physician monologue. ED physicians were instructed to disseminate the enhanced process to sexually active females ages 12-19 years (intervention group). Patients receiving the intervention were prospectively enrolled. The control group consisted of patients who, on retrospective review, came to the ED during the enrollment period and were eligible but were not enrolled. We used the electronic medical record (EMR) to identify and randomly select missed eligible patients (control group) and accumulated one control for each case. The primary outcome was FP follow-up within 2 months, measured by EMR review. The intervention (n=101) and control groups (n=101) were similar in age, prior ED visits (54% versus 56%), and previous FP visits (28% versus 28%). The absolute difference in follow-up to a FP clinic between the enhanced referral group (7%; 7/101) compared with the nonenhanced referral group (5%; 5/101) was only 2% (95% confidence interval -5% to 9%). Feasibility of the intervention was modest, with a best-case scenario of 59/160 (37%) of eligible patients captured. An enhanced referral initiative relying on physician participation did not substantially increase follow-up rates to a FP clinic and showed modest feasibility. More research is required to identify effective means of ED-based referral for preventive reproductive care.
    Journal of Women's Health 04/2015; 24(4):324-328. DOI:10.1089/jwh.2014.4994 · 2.05 Impact Factor
  • [Show abstract] [Hide abstract]
    ABSTRACT: This study was conducted to compare single-dose pharmacokinetics of ethinyl estradiol in an oral contraceptive to steady-state values, and to assess whether any simpler measures could provide an adequate proxy of the 'gold standard' 24-hour steady-state area-under-the-curve. Identifying a simple, less expensive, measure of systemic ethinyl estradiol exposure would be useful for larger studies designed to assess the relationship between an individual's ethinyl estradiol exposure and her side effects. We conducted a 13 samples over 24 hours pharmacokinetic analysis on day 1 and day 21 of the first cycle of a monophasic oral contraceptive containing 30 mcg ethinyl estradiol and 150 mcg levonorgestrel in 17 non-obese healthy white women. We also conducted an abbreviated single dose 9-sample pharmacokinetic analysis after a month washout. Ethinyl estradiol was measured by liquid chromatography-tandem mass spectrometry. We compared results of full 13-sample steady-state pharmacokinetic analysis with results calculated using fewer samples (9 or 5) and following the single doses. We calculated Pearson correlation coefficients to evaluate the relationships between these estimates of systemic ethinyl estradiol exposure. The area-under-the-curve, maximum (Cmax), and 24-hour (C24) values were similar following the two single oral contraceptive doses (area-under-the-curve, r = 0.92). The steady-state 13-sample 24-hour area-under-the-curve was highly correlated with the average 9-sample area-under-the-curve after the two single doses (r = 0.81, p = 0.0002). This correlation remained the same if the number of samples was reduced to 4, taken at time 1, 2.5, 4 and 24 hours. The C24 at steady-state was highly correlated with the 24-hour steady-state area-under-the-curve (r = 0.92, p < 0.0001). The average of the C24 values following the two single doses was also quite highly correlated with the steady-state area-under-the-curve (r = 0.72, p = 0.0026). Limited blood sampling, including results from two single doses, gave highly correlated estimates of an oral contraceptive user's steady-state ethinyl estradiol exposure. Copyright © 2014 Elsevier Inc. All rights reserved.
    American Journal of Obstetrics and Gynecology 12/2014; 212(5). DOI:10.1016/j.ajog.2014.12.007 · 4.70 Impact Factor
  • [Show abstract] [Hide abstract]
    ABSTRACT: To evaluate skin irritation and patch adhesiveness of a new weekly low-dose levonorgestrel (LNG) and ethinyl estradiol (EE) contraceptive patch (LNG/EE patch). This analysis was part of an open-label, parallel-group, multicenter, phase 3 study that randomized healthy women to the LNG/EE patch (one patch weekly for three consecutive weeks, followed by a patch-free week for 13 cycles) or to an oral contraceptive for six cycles followed by seven LNG/EE patch cycles. Participants selected patch application sites of abdomen, buttock or upper torso. Investigators rated patch adhesiveness and skin irritation using standardized scales. Participants rated skin irritation and itching daily using standardized scales and recorded patch fall-off on daily diary cards. A total of 32,508 patches were applied (n=1273). At the five clinic visits in which investigators rated the patches, they rated adhesiveness=0 (no lift) for ≥84% of participants and skin irritation=absent/mild for 97% of patches. Participants reported that 2-3.7% of patches fell off and rated skin irritation as absent or mild for 92- 95% of patches, according to site. Investigator- and participant-rated assessments of LNG/EE patch adhesiveness and irritation demonstrated a low incidence of patch detachment, skin irritation and pruritus. This secondary analysis of a phase 3 clinical trial of a new weekly low-dose LNG and EE contraceptive patch, which used assessment by both investigators and participants, observed a low incidence of skin irritation, pruritus and patch detachment. Copyright © 2014 Elsevier Inc. All rights reserved.
    Contraception 11/2014; 91(3). DOI:10.1016/j.contraception.2014.11.013 · 2.34 Impact Factor
  • [Show abstract] [Hide abstract]
    ABSTRACT: Pregnancy rates in US contraceptive clinical trials are increasing due to decreased treatment compliance. This study compared compliance with a new low-dose levonorgestrel (LNG) and ethinyl estradiol (EE) contraceptive patch (CP, Twirla™) with that of a low-dose combination oral contraceptive (COC) in a demographically diverse population. This analysis was part of an open-label, parallel-group, multicenter phase 3 study that randomized healthy sexually active women (17-40years) to 13cycles of LNG/EE CP or 6cycles of COC, then 7cycles of LNG/EE CP. We defined self-reported compliance as cycles that, according to diaries, show 21days of patch wear without missed days or any patch worn >7days or 21days of pill-taking without missed pill days. We verified compliance by detectable plasma presence of LNG and EE at cycles 2, 6, and 13. Of the intention-to-treat population with diary information (N=1328, mean age 26.4years, 46% minorities, 33% obese), 10.0% of the CP (n=998) versus 21.2% of the COC group (n=330) self-reported noncompliance after 6cycles (p<.001). Laboratory assessments verified 10-14% of participants in both groups as noncompliant. Self-reported perfect use did not vary between obese [body mass index (BMI) ≥30kg/m(2)] versus nonobese (BMI <30kg/m(2)) participants in both groups or when stratified by age, education, or race/ethnicity in the CP group. Self-reported compliance was significantly greater in the CP than COC group and did not vary by obesity status. Discrepancies between self-reported and verified compliance question reliability of patient diaries. This paper, based on an analysis of a phase 3 trial, shows that compliance was significantly greater with a new weekly transdermal CP than with a once-daily COC in obese as well as nonobese participants. Discrepancies between self-reported compliance and laboratory-verified compliance raise questions regarding the reliability of patient diaries. Copyright © 2014 Elsevier Inc. All rights reserved.
    Contraception 11/2014; 91(3). DOI:10.1016/j.contraception.2014.11.011 · 2.34 Impact Factor
  • Carolyn L. Westhoff · Beverly Winikoff
    Contraception 10/2014; 90(4):353. DOI:10.1016/j.contraception.2014.08.008 · 2.34 Impact Factor
  • Contraception 09/2014; 90(3):349. DOI:10.1016/j.contraception.2014.05.194 · 2.34 Impact Factor
  • Carolyn L. Westhoff · Isabel Reinecke · Keith Bangerter · Martin Merz
    [Show abstract] [Hide abstract]
    ABSTRACT: Background Body mass index (BMI) may influence ovulation inhibition resulting from transdermal hormone delivery. Investigation of this effect is important given the high prevalence of obesity in the US. Study Design This open-label, uncontrolled, Phase 2b trial stratified 173 women (18–35 years) according to three BMI groups (Group 1, n = 56, ≤ 30 kg/m2; Group 2, n = 55, > 30 kg/m2 and ≤ 35 kg/m2; and Group 3, n = 47, > 35 kg/m2). Women used a contraceptive patch containing 0.55 mg ethinyl estradiol (EE) and 2.1 mg gestodene (GSD). The EE/GSD patch was used weekly for three 28-day cycles (one patch per week for 3 consecutive weeks followed by a 7-day, patch-free interval) and its effect on ovulation was assessed by the Hoogland score, a composite score that comprises transvaginal ultrasound and estradiol (E2) and progesterone levels every 3 days in Cycles 2 and 3. Evaluation of PK parameters was a secondary aim of the study, and blood samples for analytic determination of EE, GSD and SHBG were taken during the pre-treatment cycle, Cycle 2 and Cycle 3. Compliance was assessed using diary information and serum drug levels. Results In the per-protocol set, there were only six ovulations during the study, and no participant ovulated in both study cycles. One ovulation occurred in Group 1, three in Group 2, and two in Group 3. Ovulation inhibition was unaffected by BMI; in all groups, most participants had Hoogland scores of 1 or 2 (i.e. follicle-like structures < 13 mm (Group 1, ≤ 30 kg/m2, 80.0% in Cycle 2, 85.7% in Cycle 3; Group 2, > 30 kg/m2 and ≤ 35 kg/m2, 61.4% in Cycle 2, 75.0% in Cycle 3; Group 3, > 35 kg/m2, 78.0% in Cycle 2, 72.5% in Cycle 3). Serum levels of follicle-stimulating hormone, luteinizing hormone, E2 and progesterone were similar between groups. Body weight had a limited effect on EE clearance that was unlikely to be clinically relevant. Conclusion The EE/GSD patch provided effective ovulation inhibition, even in women with higher BMI. Implications This is the largest-to-date study of physiologic endpoints and found no clinically important differences in ovarian suppression among obese and normal-weight users of the EE/GSD contraceptive patch, thus providing reassurance that obese women can achieve the same high level of contraceptive protection as normal-weight users.
    Contraception 09/2014; 90(3). DOI:10.1016/j.contraception.2014.04.018 · 2.34 Impact Factor
  • K.M. Morrell · S. Cremers · C. Westhoff · A.R. Davis
    Contraception 09/2014; 90(3):292–293. DOI:10.1016/j.contraception.2014.05.204 · 2.34 Impact Factor
  • Contraception 09/2014; 90(3):331-332. DOI:10.1016/j.contraception.2014.05.135 · 2.34 Impact Factor

Publication Stats

3k Citations
1,079.56 Total Impact Points


  • 1994–2015
    • Columbia University
      • • Department of Obstetrics and Gynecology
      • • Department of Epidemiology
      • • College of Physicians and Surgeons
      New York, New York, United States
  • 1999–2013
    • New York Presbyterian Hospital
      • • Department of Obstetrics and Gynecology
      • • Department of Urology
      New York, New York, United States
  • 2012
    • University of Campinas
      Conceição de Campinas, São Paulo, Brazil
  • 2011–2012
    • Princeton University
      • Office of Population Research
      Princeton, New Jersey, United States
  • 1999–2009
    • CUNY Graduate Center
      New York, New York, United States
  • 2008
    • Boston University
      Boston, Massachusetts, United States
    • Harbor-UCLA Medical Center
      Torrance, California, United States
  • 2004
    • National Institute of Child Health and Human Development
      베서스다, Maryland, United States
  • 1996
    • Icahn School of Medicine at Mount Sinai
      Borough of Manhattan, New York, United States