Determining Compliance of Contraception Guidelines for Women By: - PowerPoint PPT Presentation

Determining Compliance of Contraception Guidelines for Women By: Toyya Pujol-Mitchell Advisor: Dr. Nicoleta Serban

Motivation PREGNANCIES BY INTENTION STATUS Unintended Nearly half of U.S. pregnancies were Pregnancies Lead to unintended in 2011 Worse Outcomes 18% Intended Mistimed Unwanted Poor Outcomes Are 55% 27% Exacerbated by Health Conditions 2

Background • In 2010, Centers for Disease Control and Prevention (CDC) released safety recommendations for contraceptive use • Medical Eligibility Criteria for Contraceptive Use (MEC) • Scope for contraceptives options for multiple medical conditions • Adapted for the US from the World Health Organization MEC • MEC recommended highly effective contraceptives for women with high risk health conditions • 20 conditions include: hypertension, diabetes, stroke, and sickle cell disease • Recommendation: Long-acting reversible contraceptive (LARC) 3

Research Objectives 1. Accurately identify women with these high risk health conditions through administrative claims data 2. Determine if clinicians are following the MEC guidelines 3. Assess if LARC usage of women with chronic health conditions is improving 4. Compare rate to that of the total population 4

Relevant Work • Reeves et al. (2014) • Determined most accurate method to identify sickle cell patients using claims data • Byrd et al. (2015) • Found which states in Medicaid data has most complete data • Dehlendorf et al. (2010) & Russo et al. (2015) • Found healthcare providers’ knowledge of contraceptives and MEC guidelines is low • Massad et al. (2007), Culwell et al. (2009), Lindley et al. (2010) & Sun et al. (2012) • Women with individual medical conditions have high rates of unintended pregnancies • Low levels of highly effective contraceptive provisions 5

Data Overview • Medicaid Analytical Extract medical claims from Centers for Medicare and Medicaid Services (CMS) • 4 billion claims • Inpatient (IP) claims, & other therapy (OT) claims • Focused on 2 years prior (2008 – 2009) and 2 years post (2011-2012) release of guidelines • International Classification of Disease (ICD-9) codes • Current Procedural Terminology (CPT) Codes 6

Identifying Patients Methods & Results

CMS Claims Data Patient Information Provider Service Information Dates Patient Medical Condition CPT ICD-9 Codes Codes 8

Obtaining Study Population • Study population identified as patients who are: 1. Enrolled in Medicaid in Phase 0 (2008 -2009) or Phase 1 (2011-2012) 2. Females within reproductive age 3. Have one or more of the high risk health conditions Study Population: Females Enrolled in Medicaid Phase 1 23.55 Phase 0 20.70 19.00 19.50 20.00 20.50 21.00 21.50 22.00 22.50 23.00 23.50 24.00 Millions Total Females 9

Determining Age Females of reproductive age (10-49 years) determined by difference of year of birth and current year in personal summary table Study Population: Age 400.0 352.8 350.0 Thousands of People 305.8 300.0 252.2 250.0 214.3 200.9 200.0 163.5 150.0 87.8 100.0 79.1 50.0 26.4 22.2 - 10-14 15-24 25-34 35-44 45-49 Phase 0 Count Phase 1 Count 10

Identifying Surgery Patients Surgery patients (Bariatric Surgery and Organ Transplant) were identified if procedure code was in an IP claim within a 2 year time period Study Population: Surgery Patients 9,000 8,140 Number of People 8,000 7,000 6,299 6,000 5,000 4,000 3,000 2,000 779 784 1,000 - Bariatric Surgery Solid Organ Transplant Phase 0 Counts Phase 1 Counts 11

Medical Patient Identification 1. Identify all patients with at least one claim with a medical condition ICD-9 code 2. Obtain all IP and OT claims for patients identified in step 1 3. Identify medical patients by grouping IP and OT claims together by condition 12

Medical Condition Patients PHASE 0 COUNTS PHASE 1 COUNTS Remaining Condition Remaining Condition 6% Stroke 6% Stroke Breast Cancer 2% Breast Cancer 2% 3% Lupus Lupus 3% 3% 3% Ischemic Heart Disease Ischemic Heart Disease 3% Hypertension 3% 41% HIV 3% HIV 4% Epilepsy 8% Epilepsy 8% Hypertension 40% Diabetes Diabetes 31% 31% 13

Final Study Population Five percent of patients identified as having one of the 20 medical conditions Study Population: Health Condition Phase 1 16.26 0.92 Phase 0 14.60 0.78 13.00 14.00 15.00 16.00 17.00 18.00 Millions Total Patient Counts Total Condition Counts 14

Guideline Compliance Methods & Results

Establishing Outcomes • Outcome 1: Family planning ratio • Counseling, insertions, and surveillance for contraceptive methods 1 = 𝑜𝑣𝑛𝑐𝑓𝑠 𝑝𝑔 𝑔𝑏𝑛𝑗𝑚𝑧 𝑞𝑚𝑏𝑜𝑜𝑗𝑜𝑕 𝑞𝑏𝑢𝑗𝑓𝑜𝑢𝑡 𝜈 𝑗𝑘 𝑢𝑝𝑢𝑏𝑚 𝑜𝑣𝑛𝑐𝑓𝑠 𝑝𝑔 𝑞𝑏𝑢𝑗𝑓𝑜𝑢𝑡 • Outcome 2: Highest efficacy method ratio • Intrauterine device (IUD): insertion & surveillance • Implants: insertion 2 = 𝑜𝑣𝑛𝑐𝑓𝑠 𝑝𝑔 𝑀𝐵𝑆𝐷 𝑞𝑏𝑢𝑗𝑓𝑜𝑢𝑡 𝜈 𝑗𝑘 𝑢𝑝𝑢𝑏𝑚 𝑜𝑣𝑛𝑐𝑓𝑠 𝑝𝑔 𝑞𝑏𝑢𝑗𝑓𝑜𝑢𝑡 16

Comparing Rates Use Poisson means method 𝑙 − 𝜌 0 𝑙 − 𝜌 1 𝑙 = 𝜈 1𝑘 𝑙 𝐼 0 : 𝜈 0𝑘 𝑙 − 𝜌 0 𝑙 < 𝜈 1𝑘 𝑙 − 𝜌 1 𝑙 𝐼 1 : 𝜈 0𝑘 𝑙 ∶ 𝑞𝑠𝑝𝑞𝑝𝑠𝑢𝑗𝑝𝑜 𝑝𝑔 𝑞𝑏𝑢𝑗𝑓𝑜𝑢𝑡 𝑔𝑝𝑠 𝑞ℎ𝑏𝑡𝑓 𝑗 𝑏𝑜𝑒 𝑛𝑓𝑒𝑗𝑑𝑏𝑚 𝑑𝑝𝑜𝑒𝑗𝑢𝑗𝑝𝑜 𝑘 𝑝𝑔 𝑡𝑢𝑣𝑒𝑧 𝑞𝑝𝑞𝑣𝑚𝑏𝑢𝑗𝑝𝑜 𝜈 𝑗𝑘 𝑙 : 𝑞𝑠𝑝𝑞𝑝𝑠𝑢𝑗𝑝𝑜 𝑝𝑔 𝑞𝑏𝑢𝑗𝑓𝑜𝑢𝑡 𝑔𝑝𝑠 𝑞ℎ𝑏𝑡𝑓 𝑗 𝑝𝑔 𝑑𝑝𝑜𝑢𝑠𝑝𝑚 𝑞𝑝𝑞𝑣𝑚𝑏𝑢𝑗𝑝𝑜 𝜌 𝑗 𝑞ℎ𝑏𝑡𝑓 𝑗 ∈ (0,1) ℎ𝑓𝑏𝑚𝑢ℎ 𝑑𝑝𝑜𝑒𝑗𝑢𝑗𝑝𝑜 𝑘 ∈ 1,2, … 20 𝑝𝑣𝑢𝑑𝑝𝑛𝑓 𝑙 ∈ 1,2 17

Outcome Adjustment Factors NATIONAL LARC USAGE Outcome 1: Overall contraceptive use has remained constant between the two phases (62%) Outcome 2: Increase in LARC usage in most recent years (4% to 7.2%) 18

Increase in Both Outcomes Outcome (Patients/Health Patients) Rate Before Rate After 11.96% 13.16% Outcome 1: Family Planning Outcome 2: LARC 1.13% 2.37% Based on the Poisson means test: The change of the family planning patient rate and LARC usage between phases is statistically significant 19

Outcome 1: Inconsistent Increase • Outcome 1 significant medical Percentage of Family Planning Patients by Condition conditions: 45.0% • Breast Cancer 40.0% 35.0% • Diabetes 30.0% 25.0% • Epilepsy 20.0% • HIV 15.0% 10.0% • Ischemic Heart Disease 5.0% • Lupus 0.0% • Sickle Cell Disease • Thrombogenic Mutation • Valvular Heart Disease Phase 0 Phase 1 20

Outcome 2: Consistent Increase Outcome 2 significant medical Percentage of LARC Patients by Condition 10.00% conditions: • All conditions significant 8.00% 6.00% 4.00% 2.00% 0.00% Phase 0 Phase 1 21

Conclusion • The proportion of women who receive family planning has increased • These women are also choosing LARC methods more frequently • LARC usage improved at both the aggregated and condition level • Family planning improved for some health conditions • Intervention should focus on conditions which have not seen increase in family planning 22

Future Work • Update adjustment factor to reflect derived Medicaid patient rates • Add sterilization to analysis • Make recommendations across other factors, such as age, state and provider types 23

Thank you Toyya Pujol-Mitchell Georgia Institute of Technology Industrial & Systems Engineering Health Analytics Center pujol@gatech.edu

APPENDIX 25

20 Chronic Health Conditions CDC Identified High-Risk Conditions for Unplanned Pregnancy Peripartum Bariatric surgery HIV Stroke Cardiomyopathy Schistosomiasis with Liver Systemic Lupus Breast Cancer Hypertension Fibrosis* Erythematosus Diabetes Ischemic heart Disease Severe Cirrhosis Thrombogenic Mutations Endometrial & Ovarian Malignant Gestational Sickle Cell Disease Tuberculosis Cancer Trophoblastic Disease* Epilepsy Malignant Liver Tumors* Solid Organ Transplant* Valvular Heart Disease *Dropped from analysis due to low numbers 26

States Under Consideration 14 states were included in analysis: • Alabama (AL) • Minnesota (MN) • Arkansas (AR) • North Carolina (NC) • California (CA) • New York (NY) • Florida (FL) • Pennsylvania (PA) • Georgia (GA) • South Carolina (SC) • Louisiana (LA) • Tennessee (TN) • Mississippi (MS) • Texas (TX) 27