Currently Viewing:
The American Journal of Managed Care June 2018
Prevalence and Predictors of Hypoglycemia in South Korea
Sun-Young Park, PhD; Eun Jin Jang, PhD; Ju-Young Shin, PhD; Min-Young Lee, PhD; Donguk Kim, PhD; and Eui-Kyung Lee, PhD
Initial Results of a Lung Cancer Screening Demonstration Project: A Local Program Evaluation
Angela E. Fabbrini, MPH; Sarah E. Lillie, PhD, MPH; Melissa R. Partin, PhD; Steven S. Fu, MD, MSCE; Barbara A. Clothier, MS, MA; Ann K. Bangerter, BS; David B. Nelson, PhD; Elizabeth A. Doro, BS; Brian J. Bell, MD; and Kathryn L. Rice, MD
A Longitudinal Examination of the Asthma Medication Ratio in Children
Annie Lintzenich Andrews, MD, MSCR; Daniel Brinton, MHA, MAR; Kit N. Simpson, DrPH; and Annie N. Simpson, PhD
Physician Practice Variation Under Orthopedic Bundled Payment
Joshua M. Liao, MD, MSc; Ezekiel J. Emanuel, MD, PhD; Gary L. Whittington, BSBA; Dylan S. Small, PhD; Andrea B. Troxel, ScD; Jingsan Zhu, MS, MBA; Wenjun Zhong, PhD; and Amol S. Navathe, MD, PhD
Simply Delivered Meals: A Tale of Collaboration
Sarah L. Martin, PhD; Nancy Connelly, MBA; Cassandra Parsons, PharmD; and Katlyn Blackstone, MS, LSW
Currently Reading
Placement of Selected New FDA-Approved Drugs in Medicare Part D Formularies, 2009-2013
Bruce C. Stuart, PhD; Sarah E. Tom, PhD; Michelle Choi, PharmD; Abree Johnson, MS; Kai Sun, MS; Danya Qato, PhD; Engels N. Obi, PhD; Christopher Zacker, PhD; Yujin Park, PharmD; and Steve Arcona, PhD
Assessing Markers From Ambulatory Laboratory Tests for Predicting High-Risk Patients
Klaus W. Lemke, PhD; Kimberly A. Gudzune, MD, MPH; Hadi Kharrazi, MD, PhD, MHI; and Jonathan P. Weiner, DrPH
Satisfaction With Care After Reducing Opioids for Chronic Pain
Adam L. Sharp, MD, MS; Ernest Shen, PhD; Yi-Lin Wu, MS; Adeline Wong, MPH; Michael Menchine, MD, MS; Michael H. Kanter, MD; and Michael K. Gould, MD, MS
Cost Sharing for Antiepileptic Drugs: Medication Utilization and Health Plan Costs
Nina R. Joyce, PhD; Jesse Fishman, PharmD; Sarah Green, BA; David M. Labiner, MD; Imane Wild, PhD, MBA; and David C. Grabowski, PhD

Placement of Selected New FDA-Approved Drugs in Medicare Part D Formularies, 2009-2013

Bruce C. Stuart, PhD; Sarah E. Tom, PhD; Michelle Choi, PharmD; Abree Johnson, MS; Kai Sun, MS; Danya Qato, PhD; Engels N. Obi, PhD; Christopher Zacker, PhD; Yujin Park, PharmD; and Steve Arcona, PhD
There is significant heterogeneity in formulary placement and restrictions on new drug approvals in the Part D marketplace.
Data Sources, Sample Selection, and Study Variables

Primary data for the study were obtained from CMS monthly formulary files for all Part D plans from 2009 through 2014. The formulary files were supplemented with annual Part D plan characteristics from CMS Part D landscape files. Data on drug approvals and National Drug Code (NDC) assignments were obtained from the FDA.7

To ensure consistent tracking of all time-related variables, we restricted the sample to 863 Part D plans with continuous CMS contracts over the study period (from 3728 plans in 2009 decreasing to 2660 in 2013). Continuity of contracts was determined based on unique identification (ID) numbers assigned to each Part D plan. Some plans had such short CMS contracts that we could not assess duration of formulary placement for most drugs of interest. Other contracts had gaps in coverage or changed sponsor-level names, making it impossible to accurately align formulary coverage and plan characteristics over time. The selected plans had a total Part D enrollment of 10 million in December 2009. By December 2013, enrollment in these plans reached 16.6 million, representing approximately 50% of total Medicare Part D enrollment that year.

The CMS monthly formulary files assign other unique ID numbers to each formulary. These numbers change every time there is a modification in formulary coverage, such as placing a newly approved drug. This means that, over time, all Part D plans have many formulary IDs. It was for this reason that we used the plan rather than the formulary as our unit of observation, recognizing that multiple plans may use the same formulary at any given time.

Study variables relating to timing of formulary placement included the FDA approval date, NDC assignment date, and calendar month in which each drug of interest first appeared on formulary among the 863 plans. In most cases, NDC assignment dates followed FDA approval dates by a few days, but in a few cases, the delay was 6 months or more. Because plans cannot place a drug on formulary without an NDC, we tracked Part D plan placement trends from the NDC assignment month. We measured the difference in months between the month the first Part D plan was observed to place the drug (“first adopter”) and the adoption month by each subsequent plan. We used first adopter date as the baseline, rather than FDA approval or NDC assignment date, because time prior to first adoption by any Part D plan was—by definition—the same for every adopting plan. The time horizon for measuring months to formulary adoption varied from a minimum of 12 months for drugs approved at the end of 2013 up to 5 years for drugs approved in early 2009.

Drug characteristics included drug type—new chemical entity (NCE), line extension (LE), or combination product (CP)—and timing of drug approval relative to other agents in the same pharmacologic class—first in class, second in class, or third or later in class. To determine place in class, we first identified the FDA Established Pharmacologic Class (EPC) for each drug of interest. We then searched Micromedex to identify all drugs within each EPC. Next, we identified FDA approval dates from the Drugs@FDA database.6 For plans that did place each drug on formulary, we captured ST and PA restrictions in the initial placement month from the monthly CMS formulary files.

Finally, we captured essential characteristics of plan structure and performance. On the structural side, we classified plans as either PDPs or MAPDs and determined whether each plan provided basic or enhanced benefits at time of NDC assignment. We also classified PDPs according to whether they offered premiums at or below regional benchmarks. We assessed each plan’s performance using CMS star ratings, with 5 stars representing the highest quality. We hypothesized that plans with enhanced benefits and higher star ratings would have earlier and higher formulary placement rates than basic benefit plans, benchmark plans, and those with lower star ratings. Unfortunately, we did not have information on manufacturer rebate offers and so could not model net cost as a variable in formulary placement decisions.

Statistical Analysis

Our descriptive analyses consisted of drug-by-drug tabulations of FDA approval and NDC assignment dates, months to first Part D plan formulary placement, proportion of plans adopting each drug 6 months and 12 months following first placement, and ST and PA restrictions at time of initial formulary adoption.

We employed 2-part regression models to estimate the impact of drug and plan characteristics on formulary placement (part 1) and we identified factors expected to be associated with placement (months to formulary adoption and ST and PA restrictions) for plans that added the drugs to their formularies (part 2).

To assess factors associated with formulary placement (part 1), we created a dataset of 28,479 observations (863 plans × 33 drugs), with values of 1 for plans that placed a given drug on their formularies during the study period and 0 for plans that did not. We employed a similar model structure to estimate the part 2 models. Here, plan/drug observations were restricted to plans whose formularies adopted the drugs of interest. The same strategy was used to estimate the effects of drug and plan characteristics on ST and PA restrictions. All models were estimated using ordinary least squares regression so that the magnitude of the estimated effects could be readily compared across the various models.

We note 2 reasons why the effective sample sizes for these regression models are actually smaller than the nominal samples. First, multiple plans used the same formulary in any given month. The minimum number of unique formularies among the 863 plans was 91 in March 2011; the maximum was 133 in November 2012. Second is the potential for coordinated behavior in formulary adoptions within the same Part D plan. We corrected for the effect of clustering on standard errors using the Robust command in Stata (StataCorp; College Station, Texas).


 
Copyright AJMC 2006-2019 Clinical Care Targeted Communications Group, LLC. All Rights Reserved.
x
Welcome the the new and improved AJMC.com, the premier managed market network. Tell us about yourself so that we can serve you better.
Sign Up