Joseph S. Ross, MD, MHS

 To examine differences in the out-of-pocket costs for common generic drugs used to treat chronic conditions when individuals used their Medicare prescription drug plan (PDP) or when purchased through Walmart’s generic drug discount programs (GDDPs) from 2009 to 2017.

 A retrospective analysis of Medicare PDP Formulary files and Walmart’s GDDP retail drug lists from 2009 to 2017.

 We identified all generic drugs used to treat chronic conditions that were on Walmart’s GDDP retail drug list from 2009 to 2017. We then determined the out-of-pocket costs for each drug for each Medicare PDP and compared those costs with Walmart’s GDDP cash price.

 There were 62 and 43 generic medications used to treat common chronic diseases available through Walmart’s GDDP in 2009 and 2017, respectively. Across all PDPs, the median beneficiary out-of-pocket expenditure for a 30-day supply of the GDDP-available medications for chronic diseases decreased from $5.70 (interquartile range [IQR], $2.55-$7.98) in 2009 to $2.00 (IQR, $0.00-$4.00) in 2017 (

 <.001) Approximately three-fifths (60.2%) of PDPs required beneficiaries to pay out-of-pocket costs higher than those of Walmart’s GDDP in 2009, but only one-third (33.4%) did so in 2017.

 Although Medicare beneficiary out-of-pocket costs for commonly used generic drug prescriptions generally decreased over time, Medicare beneficiaries may still be paying more for the same drugs than they would through Walmart’s GDDP. Increased generic drug price transparency, including enforcing bans on gag clauses, is needed to ensure that Medicare beneficiaries obtain drugs using the most affordable options.

Retail generic drug discount programs (GDDPs) improve the affordability of generic medications, especially for individuals without pharmaceutical coverage. We characterized trends in the out-of-pocket costs required by Medicare prescription drug plans for generic drugs used for chronic diseases and compared them with the cash price of the drugs in Walmart’s GDDP from 2009 to 2017.

Out-of-pocket costs for these generic drugs have decreased over time.

Formularies have increasingly classified these drugs into tiers other than preferred generic tiers.

Opportunities to lower out-of-pocket costs still exist, as a substantial proportion of plans still require costs higher than GDDP cash prices for the same medications.

The 2003 Medicare Modernization Act established the Medicare Part D prescription drug benefit to improve access to and affordability of prescription medications for beneficiaries through government subsidies.

 Under the benefit, private plans negotiate with drug manufacturers and compete to enroll Medicare beneficiaries. To control prescription costs, Part D plans steer consumers toward certain drugs through coverage decisions, tier structure (which determines co-payments or coinsurance), and utilization management tools. Since its inception, Part D has grown from offering prescription drug coverage to 27.9 million beneficiaries in 2006 to 43.9 million in 2018,

 and it achieved comparable levels of utilization of generic drugs by enrollees in state Medicaid programs.

 However, billions of dollars in savings could be realized through greater generic substitution of brand-name medications among Medicare beneficiaries.

In 2006, the same year that the Part D benefit went into effect, large wholesale pharmacy chains, such as Walmart, began creating generic drug discount programs (GDDPs). These programs offer common generic medications for low out-of-pocket costs, usually $4 for a 30-day supply and $10 for a 90-day supply. This price can sometimes be lower than what patients would pay through their insurance coverage when purchasing the same medications and, therefore, leads to lower out-of-pocket costs. In 2007, 80% of beneficiaries filled prescriptions for generic medications available through GDDPs, but only 16.3% of those used the $4 GDDP

; the estimated total savings if beneficiaries had used GDDPs instead of their insurance was more than $5.7 billion, demonstrating the significant impact that GDDP utilization could have on prescription medication costs.

 A recent study of 2017 Medicare prescription drug plan (PDP) coverage of generic drugs used to treat cardiovascular disease (CVD) similarly demonstrated potential savings through GDDPs: 21% of PDPs required beneficiary out-of-pocket payments higher than Walmart’s GDDP cash price for a 30-day supply of the same drug.

 However, whether the proportion of PDPs with higher prices than those offered by GDDPs varies across geography, across other medication classes, or over time remains unknown. Accordingly, we used Medicare PDP data to characterize trends in Medicare beneficiary out-of-pocket expenditures for commonly used generic medications that treat any chronic disease and were available through Walmart’s GDDP from 2009 to 2017. We expect our findings to offer important insights for future policy efforts intended to further reduce costs for Medicare beneficiaries and to better clarify the role that GDDPs are likely to play in insurance cost sharing and formulary structure.

&#8203;&#8203;&#8203;&#8203;&#8203;&#8203;

We used data from the second quarters of 2009 through 2017 CMS PDP Formulary and Pricing files. These files provide information submitted by each Part D plan, including the associated formulary, tier assignment for all covered medications, co-payment type (eg, dollar co-payment vs percentage coinsurance), and co-payment amount. Because comprehensive Medicare PDP data to calculate out-of-pocket costs were not available until 2009 for 30-day supplies and 2014 for 90-day supplies, we limited our analysis to annual data from 2009 to 2017. We excluded Special Needs Plans, plans with incomplete or inaccurate data, and plans that operated in states for which Walmart did not guarantee its advertised cash price (generally $4 for a 30-day supply and $10 for a 90-day supply) for all of its GDDP medications through the entire time period of the analysis. Furthermore, we noted whether each plan was a Medicare Advantage (MA) plan, which covers Part A and Part B benefits in addition to prescription drug coverage, or a stand-alone Part D plan, which covers only prescription drugs.

We obtained information on drugs offered through Walmart’s GDDP from 2009 to 2017 directly from Walmart’s pharmacy website.

 We included all oral generic prescription medications used to treat chronic conditions that were not available over the counter. Because drugs could be added or removed from GDDP lists at any time, the drug sample varied from year to year.

Our primary outcome measure was the beneficiary out-of-pocket expenditure for generic prescription drugs for chronic conditions that were available through Walmart’s GDDP from 2009 to 2017. When calculating the median out-of-pocket expenditures, we adjusted for inflation using June 2017 as the index month and year,

 and we estimated the costs incurred by the beneficiary when filling a prescription with coverage for the first time in the year.

As a secondary outcome, we determined the proportion of plans requiring costs greater than Walmart’s GDDP price. For this outcome, we did not adjust for inflation because the GDDP cash price has remained the same since the inception of Walmart’s GDDP. Although nearly all included medications were offered by Walmart’s GDDP at $4 for a 30-day supply and $10 for a 90-day supply, for alendronate sodium, finasteride, and tamoxifen citrate, the 30-day supply cash price was $9 and the 90-day supply cash price was $24.

Other key variables of interest included presence of select care tier designations for certain drugs within plans, therapeutic classes, and geographic distribution of plans. In 2015, CMS increased the weight of adherence to some medications for hypertension, hyperlipidemia, and type 2 diabetes in calculating a plan’s star rating.

 Consequently, these medications (and other common drugs used for chronic conditions) were increasingly allocated into a select care tier, which has low to no out-of-pocket costs, with the goal of improving beneficiary medication adherence and allowing plans to score higher on the star rating measure. We grouped any drug listed in a select care tier with those in tier 1, which typically includes low-cost generic medications. The drugs listed under Walmart’s categorizations were classified into 3 therapeutic classes: cardiovascular, mental health, and endocrine. For geographic distribution, each state is assigned to 1 of 26 total MA regions and 1 of 34 stand-alone regions. All plans operating in MA or stand-alone regions composed of multiple states were assigned to all of those states; beneficiaries can enroll only in plans that operate in their state of residence.

We used descriptive analyses to calculate median out-of-pocket costs across all GDDP-available medications (30- and 90-day supply) offered by all plans each year between 2009 and 2017. We also compared each plan’s out-of-pocket cost for each drug with the Walmart GDDP cash price. Descriptive analyses were also applied to outcomes across geography.

Analyses were stratified by plan type (MA-PDP vs stand-alone PDP) and tier (tier 1 and select care tiers vs all other tiers). The Mann-Whitney 

 test was used to compare MA-PDPs and stand-alone PDPs within the same year. Because plan type and tier stratifications were independent of each other, we did not correct for multiple comparisons. We used the same test for comparing both outcomes in 2009 and 2017. We used the Wilcoxon signed-rank test to examine differences among the 3 therapeutic classes in a pairwise fashion with the Holm-Bonferroni method to correct for multiple comparisons, using a 

 value of .017 as the threshold for statistical significance.

As a sensitivity analysis, we used generalized estimating equations to examine whether within-plan clustering may be driving the results observed regarding median out-of-pocket expenses over time for 30-day supplies of GDDP chronic disease medications from 2009 to 2017. We included Medicaid region and Medicaid expansion status to account for factors that could influence out-of-pocket costs for generic drugs. Results of these analyses were consistent with our main analyses and are presented in the eAppendix material only (

Stata version 15.1 (StataCorp LLC; College Station, Texas) and RStudio version 1.1.423 (RStudio Inc; Boston, Massachusetts) were used for all analyses.

There were 3013 Medicare PDPs in 2009 and 2150 PDPs in 2017 included in our sample (

). Over this period, the composition of Medicare PDP types shifted toward MA-PDPs (51.6% MA-PDPs and 48.4% stand-alone PDPs in 2009; 71.3% MA-PDPs and 28.7% stand-alone PDPs in 2017), primarily because of a reduction in the number of stand-alone PDPs.

The number of medications available through Walmart’s GDDP and eligible for inclusion in our study decreased from 62 generic medications used to treat common chronic diseases in 2009 to 43 in 2017 (

). During this time period, Walmart added only 1 medication (finasteride) but removed 20. Our sample included medications for cardiovascular (32 and 20 medications in 2009 and 2017, respectively), psychiatric (17 and 12 medications in 2009 and 2017), and endocrine (13 and 11 medications in 2009 and 2017) disorders. By 2017, Walmart no longer offered digoxin, anticonvulsants, or selective serotonin reuptake inhibitors as part of the GDDP.

The median proportion of GDDP drugs for common chronic conditions covered in Medicare plans in either tier 1 or a select care tier decreased from 94.8% in 2009 to 60.6% in 2015, and then rose to 75.2% in 2017. By plan type, the median proportion decreased from 96.1% in 2009 to 57.5% in 2015, and then increased to 77.4% in 2017 for MA-PDPs; a similar trend occurred for stand-alone PDPs, from 93.5% in 2009 to 65.4% in 2015 to 69.6% in 2017 (

Across all plans, the median beneficiary out-of-pocket expenditure for a 30-day supply of these GDDP-available medications decreased 64.9%, from $5.70 (interquartile range [IQR], $2.55-$7.98) in 2009 to $2.00 (IQR, $0.00-$4.00) in 2017 (

). By plan type, costs decreased 64.9% for MA-PDPs, from $5.70 (IQR, $1.85-$7.98) in 2009 to $2.00 (IQR, $0.00-$4.00) in 2017 (

<.001), and 82.5% for stand-alone PDPs, from $5.70 (IQR, $2.64-$7.98) in 2009 to $1.00 (IQR, $0.69-$2.39) in 2017 (

Similar declines in median out-of-pocket costs were observed among both MA-PDPs and stand-alone PDPs for drugs in tier 1 or select care tiers and for drugs in all other tiers (

Proportion of Plans Requiring Beneficiary Out-of-Pocket Costs Greater Than GDDP Cash Prices

The median proportion of plans requiring beneficiary out-of-pocket costs greater than the GDDP cash price for a 30-day supply of medications was 60.2% (IQR, 58.5%-63.6%) in 2009 and 33.4% (IQR, 17.9%-55.1%) in 2017 (44.5% decrease; 

 <.001) (Table). By plan type, the proportion for MA-PDPs was 61.2% (IQR, 60.1%-62.9%) in 2009, which decreased to 41.0% (IQR, 22.7%-66.1%) in 2017 (33.0% decrease; 

 <.001). Similarly, the same proportions for stand-alone PDPs were 60.1% (IQR, 56.8%-65.4%) in 2009 and 11.8% (IQR, 3.3%-28.2%) in 2017 (80.4% decrease; 

Declines in proportion of plans requiring beneficiary co-payments more than the GDDP cash price were observed among drugs in tier 1 or select care tiers for both plan types (

 <.001) and among drugs in other tiers for only stand-alone PDPs, not MA-PDPs (

Across geographic areas, there was greater variation in out-of-pocket costs required by MA-PDPs than stand-alone PDPs and greater variation in 2009 than in 2017 (

). In 2009, the IQRs for out-of-pocket costs across states were $5.70 to $6.84 (range, $0-$11.40) for MA-PDPs and $5.70 to $5.70 (range, $5.70-$6.84) for stand-alone PDPs. In 2017, the IQRs were $2.00 to $3.00 (range, $0-$6.00) for MA-PDPs and $1.00 to $2.00 (range, $1.00-$2.00) for stand-alone PDPs.

In 2009, the IQRs for proportions of plans requiring co-payments greater than the GDDP cash price for a 30-day supply across states were 61.9% to 81.0% (range, 19.2%-98.6%) for MA-PDPs and 60.2% to 62.6% (range, 55.1%-66.9%) for stand-alone PDPs. In 2017, the IQRs were 40.4% to 55.8% (range, 27.5%-74.3%) for MA-PDPs and 15.5% to 19.3% (range, 12.1%-22.8%) for stand-alone PDPs.

Across all plans, the differences among the proportions of plans requiring co-payments greater than the GDDP cash price for a 30-day supply were not significantly different by therapeutic class in 2009 (60.2%, 61.9%, and 58.8% for the cardiovascular, mental health, and endocrine classes, respectively; pairwise 

In contrast, in 2017, there were differences by therapeutic class (20.6%, 52.7%, and 28.3% for cardiovascular, mental health, and endocrine classes, respectively; 

 <.017 for cardiovascular compared with mental health class; 

 >.017 for all other pairwise comparisons).

GDDP-Available Medications in 2009 Versus 2017

Twenty generic drugs were available through GDDPs in 2009 that were not available in 2017; the median out-of-pocket costs for these drugs, and the proportion of plans offering them for more than the GDDP cash price, were consistently higher when compared with the 43 generic drugs available through GDDPs only in 2017 (

The trends in median out-of-pocket costs for a 90-day supply of GDDP-available medications were similar to those of a 30-day supply for both outcomes (eAppendix Table 4).

In the decade following the creation of the Medicare PDP benefit and GDDPs, there was a significant decrease in beneficiary out-of-pocket costs for generic medications treating common chronic conditions that were available through Walmart’s GDDP. Similarly, a smaller proportion of plans required beneficiary out-of-pocket costs greater than the GDDP cash price for both a 30-day and a 90-day medication supply in 2017 than in 2009. Not only were these reductions more pronounced for stand-alone PDPs than for MA-PDPs, stand-alone PDPs also had less geographic variation across states compared with their MA-PDP counterparts. Despite reductions across all medication classes from 2009 to 2017, changes were uneven by therapeutic class: For example, in 2017, out-of-pocket costs for CVD medications were significantly less than those for mental health medications. Nevertheless, median beneficiary out-of-pocket costs and the proportion of plans requiring co-payments greater than the GDDP cash price decreased for all plan type and tier stratifications.

The availability of cheaper medications through GDDPs may have led plans to lower out-of-pocket costs for these drugs for beneficiaries, because a cheaper nonplan option is available. Despite this progress, it is still necessary to find opportunities to lower costs further, particularly given concerns about the ability of older adults to afford medications and the number of beneficiaries taking GDDP-available medications.

 In 2017, $3.8 billion was spent by Medicare on the 43 medications included in our drug sample and available through the Walmart GDDP.

 Although the proportion of plans requiring out-of-pocket co-payments greater than GDDP cash prices decreased from three-fifths of plans in 2009 to one-third in 2017, alternative options, such as utilization of GDDPs, may be crucial for saving costs month after month if insurance plans remain more expensive for these medications.

A 2014 study estimated that only one-third of adults 65 years or older utilized GDDPs, showing that significant room for potential savings remains.

 One reason that plans may continue to require higher out-of-pocket costs is the increasing use of mechanisms that steer consumers toward particular drugs. The proportion of PDPs listing generic drugs in tiers other than tier 1 or a select care tier grew over time from 2009 to 2017, a result consistent with others’ findings that Part D plans have placed an increasing number of generic drugs in nonpreferred drug tiers over time.

 This means patients are incentivized to choose certain medications from preferred tiers (ie, tier 1 and select care tier); this limits patient choice and is contingent on clinicians prescribing a generic medication listed in a preferred tier for patients’ respective plans.

 However, at the same time, Walmart’s GDDP similarly narrowed the selection of drugs made available, from 62 to 43 medications for chronic diseases from 2009 to 2017.

Our finding of cost differences to beneficiaries, depending on whether the prescription is obtained using their pharmacy benefit plan or a GDDP, even if obtained from the same pharmacy, is a concern. Inconsistencies may result in instances when a drug’s co-payment is higher than the cost to the insurance or pharmacy benefit manager (ie, “clawbacks”). These instances have garnered increasing national attention and are in part due to gag clauses, which are contractual terms that prevent pharmacists from disclosing list prices and informing patients about other potentially cheaper options for the same medication.

 Although recent legislation has formally banned these clauses,

 the enforcement and effects of these bans remain unknown. Ideally, all stakeholders would have readily available access to data and would be able to inform patients of the best way to lower out-of-pocket costs for individual medications and simultaneously minimize the inconvenience for patients who may achieve the lowest total cost for their medications by filling their generic medications using different modalities (eg, using a combination of insurance coverage and accessing GDDPs). Opportunities include informing clinicians at the time of ordering prescriptions, patients when choosing the pharmacy with the most affordable option(s), and pharmacists at the time of filling prescriptions.

Articles

Out-of-pocket costs required by Medicare prescription drug plans for drugs available through Walmart’s generic drug discount program have decreased from 2009 to 2017.

Medicare Beneficiaries’ Out-of-Pocket Costs for Commonly Used Generic Drugs, 2009-2017

To assess whether Medicare formularies restrict access to drugs receiving new FDA black box warnings for which safer drug alternatives are available.

A retrospective analysis using Medicare Prescription Drug Plan Formulary files to determine formulary changes for drugs receiving FDA black box warnings between 2007 and 2013.

We identified all FDA-approved medications available in tablet or capsule formulation that received a black box warning between 2007 and 2013 related to death and/or cardiovascular risk. We then determined formulary coverage of these drugs pre-black box warning, 1 year after, and 2 years after. For each formulary, we identified formulary restrictiveness, defined as: unrestrictive coverage (no prior authorization or step therapy), restrictive coverage (prior authorization or step therapy required), or no coverage.

Nine drugs with at least 1 FDA-approved safer drug alternative received 10 new black box warnings for death and/or cardiovascular risk between 2007 and 2013. In response to FDA black box warnings, overall formulary restrictiveness increased for 40% (n = 4) of drugs at 1 year, and for 50% (n = 5) at 2 years. However, for the majority of drugs (n = 7),  most formularies remained unrestrictive 2 years after a new black box warning.

Medicare formularies became more restrictive for half of the drugs that recently received new FDA black box warnings for death and/or cardiovascular risk and for which safer drug alternatives are available. However, a substantial proportion of formularies remained unrestrictive, suggesting inconsistent responses to new safety information to curtail the use of these medications.

The impact of FDA black box warnings on formulary restrictiveness is not well known. This study investigated whether Medicare formulary coverage changed after oral drugs with safe available drug alternatives received new FDA black box warnings related to death and/or cardiovascular risk. 

Some formularies either dropped coverage or began requiring prior authorization to restrict the prescribing of unsafe drugs. 

A substantial proportion of formularies made no changes to restrict coverage. 

There are opportunities to improve consistency in formulary responses to FDA black box warnings to curtail utilization of drugs with known safety risks and for which safer drug alternatives are available.

The FDA requires that drugs with serious or life-threatening risks have a boxed warning (commonly referred to as a “black box warning” [BBW]) on their label. The BBW is one of the FDA’s strongest actions short of drug recall or withdrawal, and is intended to inform patients and prescribers about serious safety concerns that must be considered in assessing the risks and benefits of a drug or that the FDA allows use of the drug only with restrictions to ensure safe use. However, BBWs have not been consistently associated with reduced prescribing

 for several possible reasons. First, prescribers may be unaware that the FDA had issued the warnings.

Second, prescribers may think that, despite BBWs, the drugs have a superior benefit/risk ratio to alternative therapies or that the safety concern is not as severe as suggested, and opt to continue prescribing while using strategies to mitigate risk, such as closer patient monitoring.

 Third, BBWs may apply only to limited patient populations, and prescribers may not believe the risk applies more generally.

Another possibility is that BBWs are not taken into account within insurance plan drug formularies. Formularies limit prescribing of unsafe drugs through 2 strategies: formulary exclusion (ie, no coverage) or utilization management (eg, prior authorization or step therapy). One study found no change to state Medicaid plans’ prior authorization requirements following the 2005 BBW for atypical antipsychotic use among the elderly,

 and another found that only 2 state Medicaid programs restricted use after the 2007 BBW for rosiglitazone.

 To better understand the potential role that insurance plan drug formularies may play in restricting the use of potentially unsafe drugs that have safer available alternatives, we characterized changes to Medicare formularies for oral drugs that received new BBWs between 2007 and 2013 for death and/or cardiovascular risk. As these are among the most severe adverse events, we hypothesized that they would have the greatest likelihood of resulting in formulary changes to restrict use. Because each BBW is specific to the drug’s indications and risk, we focused on a limited sample to provide an illustrative summary of formulary changes.

We conducted a retrospective analysis of Medicare formularies using data from the CMS Prescription Drug Plan Formulary Files. We obtained data on both standalone prescription drug plans and Medicare Advantage prescription drug plan formularies for Medicare Part D, as of each June and December from 2006 through 2015.

We used the FDA’s MedWatch website to identify all FDA-approved tablet or capsule formulation medications used in an outpatient setting that received BBWs between 2007 and 2013 related to death and/or cardiovascular risk. MedWatch is the agency’s safety information and adverse event reporting program. We examined all BBWs and first excluded those for drugs used in-hospital only or not relevant to Medicare beneficiaries. We also excluded BBWs for combination drugs, since the primary drug would already be included in our sample. Next, we excluded BBWs for cancer, HIV, or immunosuppressants as these are protected classes treating diseases that require specialized pathways of care. We also excluded warnings for a narrow subset of patients and warnings related to drug-drug interactions. Lastly, we limited our study to drugs with at least 1 drug alternative without a BBW approved for use for the same primary condition, based on the US Pharmacopeial Convention Category & Class versions 5.0 and 6.0. When a drug had multiple BBWs, each warning was considered a unique event.

Our main outcome was formulary restrictiveness based on formulary exclusion, step therapy, or prior authorization. For each drug’s lowest available dose, we classified all Medicare formularies as unrestrictive, restrictive, or no coverage at 3 time points: immediately preceding, at least 1 year after, and at least 2 years after the BBW. An unrestrictive formulary covered the drug without prior authorization or step therapy requirements. A restrictive formulary covered the drug but required either prior authorization and/or step therapy. A formulary that did not explicitly provide coverage did not include the drug on its formulary, which means patients are responsible for the entire drug cost except in very rare circumstances.

We used descriptive statistics to characterize the proportion of formularies that were unrestrictive, restrictive, and did not provide coverage for each drug at all 3 time points. We used χ

 or Fischer’s Exact tests to examine differences in formulary restrictiveness for each drug, comparing restrictiveness pre-BBW to restrictiveness at both 1 year and 2 years afterward. All statistical tests were 2-tailed with a type 1 error rate of 0.025 to account for multiple comparisons for each drug.

Nine medications in a tablet or capsule formulation received a total of 10 new BBWs related to death and/or cardiovascular risk between 2007 and 2013 and had at least 1 drug alternative without a BBW. Four of the 9 drugs (44%) had existing BBWs before the FDA issued BBWs between 2007 and 2013. Seven (70%) of the 10 BBWs were related to death and 5 (50%) to cardiovascular risk (2 related to both cardiovascular risk and death). Two BBWs, those for propylthiouracil and ketoconazole, specifically advised against use of the given medication except in very unique circumstances.

Two years after each of the 10 new BBWs, for 5 (50%) drugs, Medicare formularies became more restrictive, on average, by removing coverage or adding prior authorization requirements. There was no overall increase in step therapy requirements. For 1 (10%) drug, formularies became less restrictive on average and for 4 (40%) drugs there was no overall change in average formulary restrictiveness. When restricting to the 4 drugs with prior BBWs, results were overall similar: Medicare formularies became more restrictive for 2 (50%) drugs at both 1 and 2 years.

Rosiglitazone is approved to improve glycemic control for type 2 diabetes. In August 2007, thiazolidinediones, a medication class that includes rosiglitazone and pioglitazone, received a BBW for congestive heart failure. Prior to this BBW, 78% (n = 255) of formularies provided unrestrictive coverage, 19% (n = 62) provided restrictive coverage, and 3% (n = 10) provided no coverage. There was no significant change in formulary restrictiveness at 1 or 2 years (Figure).

Prior to its strengthened February 2011 BBW for a significantly increased risk of myocardial infarction, 57% (n = 148) of formularies provided unrestrictive coverage, 36% (n = 94) provided restrictive coverage, and 7% (n = 18) provided no coverage. At 1 year, rates were 32%, 36%, and 32%, respectively (

 <.001) (Figure), suggesting that formularies had become significantly more restrictive. Changes in restrictiveness were due to fewer formularies providing coverage and covering formularies requiring prior authorization (Table).

Pioglitazone is approved to improve glycemic control for type 2 diabetes. Prior to its August 2007 BBW for congestive heart failure, 67% (n = 220) of formularies provided unrestrictive coverage, 19% (n = 62) provided restrictive coverage, and 14% (n = 45) provided no coverage. At 1 year, rates were 73%, 27%, and 0%, respectively (

<.001) (Figure), suggesting that formularies had become significantly less restrictive. Changes in restrictiveness were predominantly due to more formularies providing coverage.

Zyban is approved for smoking cessation. Prior to its July 2009 BBW for serious neuropsychiatric events, including suicide, 24% (n = 75) of formularies provided unrestrictive coverage, 5% (n = 11) provided restrictive coverage, and 73% (n = 232) provided no coverage. There was no significant change in formulary restrictiveness at 1 year (Figure). At 2 years, rates were 19%, 0%, and 80%, respectively (

 = .006), suggesting that formularies had become significantly more restrictive. Changes in restrictiveness were predominantly due to fewer formularies providing coverage.

Varenicline is approved for smoking cessation. Prior to its July 2009 BBW for serious neuropsychiatric events including suicide, 60% (n = 191) of formularies provided unrestrictive coverage, 21% (n = 68) provided restrictive coverage, and 19% (n = 59) provided no coverage. At 1 year, rates were 51%, 49%, and 0%, respectively (

 <.001) (Figure). Changes in restrictiveness were predominantly due to more formularies requiring prior authorization (Table).

Quinine is approved for malaria treatment, although it was widely used for nocturnal leg cramps. Prior to its November 2009 BBW explaining that use for nocturnal leg cramps could result in life-threatening hematological reactions, 60% (n = 159) of formularies provided unrestrictive coverage, 18% (n = 56) provided restrictive coverage, and 32% (n = 103) provided no coverage. There was no significant change in formulary restrictiveness at 1 year (Figure). At 2 years, rates were 41%, 27%, and 31%, respectively (

 = .012), suggesting that formularies had become significantly more restrictive. Changes in restrictiveness were predominantly due to more formularies requiring prior authorization (Table).

Propylthiouracil is approved for treatment of hyperthyroidism. Prior to its April 2010 BBW for potentially fatal liver failure, stating propylthiouracil was only indicated in patients intolerant of methimazole, 100% (n = 292) of formularies provided unrestrictive coverage. There was no significant change in formulary restrictiveness at 1 or 2 years (Figure).

Dronedarone is approved for treatment of paroxysmal or persistent atrial fibrillation or atrial flutter but included a BBW contraindicating its use in patients with advanced heart failure. Prior to its additional December 2011 BBW contraindicating its use in patients with permanent atrial fibrillation because of cardiovascular risk, 82% (n = 199) of formularies provided unrestrictive coverage, 9% (n = 22) provided restrictive coverage, and 9% (n = 23) provided no coverage. At 1 year, rates were 72%, 10%, and 18%, respectively (

 <.001) (Figure), suggesting that formularies had become significantly more restrictive. Changes in restrictiveness were predominantly due to fewer formularies providing coverage. At 2 years, rates were 80%, 8%, and 12%, respectively (

 = .63), suggesting that changes in formulary restrictiveness had reversed, as more formularies began covering the drug again.

Dantrolene is approved for treatment of chronic spasticity. In July 2012, its prior BBW on fatal and nonfatal hepatotoxicity was updated to note that a higher proportion of hepatic events had fatal outcomes in elderly patients, although the majority of these cases were complicated by confounding factors. Prior to the addition to this BBW, 88% (n = 242) of formularies provided unrestrictive coverage, 0% (n = 11) provided restrictive coverage, and 12% (n = 34) provided no coverage. There was no significant change in formulary restrictiveness at 1 or 2 years (Figure).

Ketoconazole is approved as an antifungal therapy. Previous BBWs had been issued for QT prolongation and hepatotoxicity. Prior to the July 2013 BBW stating that the drug should only be used when other effective antifungal therapy was not available or tolerated, 100% (n = 314) of formularies provided unrestrictive coverage. At 1 year, 84% of formularies provided unrestrictive coverage, 16% provided restrictive coverage, and 1% provided no coverage (

 <.001) (Figure), suggesting that formularies had become significantly more restrictive. Changes in restrictiveness were predominantly due to more formularies requiring prior authorization (Table).

After 10 new BBWs were issued by FDA between 2007 and 2013 for death and/or cardiovascular risk among 9 tablet or capsule drugs with safer available drug alternatives, we observed that Medicare formularies became somewhat, but not universally, more restrictive at 1 and 2 years afterward. Formularies sometimes removed coverage and/or implemented prior authorization requirements to deter the use of drugs with new safety warnings. However, a substantial number of formularies continued to provide unrestrictive coverage for these drugs after the FDA’s strong safety warning. Providing unrestrictive coverage when other effective but safer drug alternatives are available suggests that formularies are not taking advantage of existing utilization management strategies to promote the use of safer available alternative medications, thus unnecessarily exposing some Medicare beneficiaries to risky medications. There were no consistent patterns in the drugs for which BBWs were not associated with increased formulary restrictiveness.

CMS instructs its contracting Part D sponsor formularies to consider drug safety and efficacy when making decisions about which drugs to include on formularies.

 Further, although changes are generally constrained once the coverage year begins, CMS specifically allows formularies to add utilization management tools at any time after a new FDA BBW and allows these unsafe drugs to be removed without meeting advance notice requirements.

First, our conclusions are limited to BBWs related to death and/or cardiovascular risk. Second, our analysis excluded drugs in a non-tablet or capsule formulation and those without safer available drug alternatives. Third, we did not examine changes in Medicare formulary tier, which is a strategy most likely to be used to impose cost-sharing requirements and limit the use of expensive, as opposed to unsafe, therapies. Fourth, our data may not be generalizable to formularies of insurers other than Medicare. Nevertheless, our findings provide important insights into the role of formulary policy because Medicare beneficiaries represent the largest proportion of insured patients using prescription medications. Finally, we did not examine the impact of formulary changes at the plan level or on prescriptions dispensed. Ultimately, linking formulary restrictiveness after BBWs to patient outcomes would help inform future formulary management decisions.

Medicare formularies increased restrictions to some drugs with new FDA BBWs for death and/or cardiovascular risk. Although this indicates important action, a substantial portion of formularies provided unrestrictive drug coverage 1 and 2 years afterward. Increasing the use of formulary exclusion or restrictiveness for drugs that receive BBWs could help curtail prescriptions of unsafe medications when safer drug alternatives exist.

Medicare formularies were inconsistent in increasing restrictiveness to drugs that received FDA black box warnings for death and/or cardiovascular risk with safer available drug alternatives.

Association Between FDA Black Box Warnings and Medicare Formulary Coverage Changes

 Medicare coverage of skilled nursing facility (SNF) care requires that beneficiaries have a 3-night inpatient stay in the prior 30 days to be eligible. Time spent by beneficiaries receiving hospital-based observation services does not count toward this requirement.

 To examine the frequency of Medicare beneficiary discharge from hospital-based observation services to SNFs and its impact on Medicare coverage.

 In 2010, 195,068 community-dwelling beneficiaries received hospital-based observation services. Beneficiaries were overwhelmingly (96.5%) discharged back to the community without home health services. Only 1.2% (2319) were discharged to non-covered SNFs, while 0.6% (1196) were discharged to covered SNFs. Patients discharged to SNFs experienced longer lengths of stay (LOS) than those discharged back to the community (34.9 hours vs 25.5 hours; P <.01). Approximately one-fourth of beneficiaries discharged to SNFs had an observation LOS of 48 hours or more.

While only a small minority of community-dwelling Medicare beneficiaries who received hospital-based observation services in 2010 were discharged to an SNF not covered by Medicare, the implications for these patients and the associated costs deserve attention. These findings have important implications for Medicare’s observation service and 2-midnight policies.

	Medicare coverage of skilled nursing facility (SNF) care requires that beneficiaries have a 3-night inpatient stay in the prior 30 days to be eligible—time spent in hospital—based observation does not count toward this requirement. In 2010, only 1.2% of community-dwelling Medicare beneficiaries who received hospital-based observation services were discharged to an SNF not covered by Medicare. Beneficiaries were overwhelmingly (96.5%) discharged back to the community.

		This study addresses concerns that have been raised about potential unintended consequences of increased use of observation services, including whether Medicare beneficiaries may be left responsible for costly, post discharge nursing care following an observation encounter—the subject of a lawsuit against the HHS.

		These findings have important implications for Medicare’s observation service and 2-midnight policies for a covered inpatient stay.

Hospital-based observation services allow short-term evaluation, treatment, and assessment of patients as an alternative to inpatient admission. Recently, there has been a steady increase in the number of observation units and in the utilization and duration of observation services.

 In limited studies, observation encounters have been found to be associated with decreased Medicare and hospital costs, decreased overall hospital length of stay (LOS), and increased patient satisfaction.

More recently, however, concerns have been raised about the potential unintended consequences of increased use of observation services, including whether Medicare beneficiaries may be left responsible for costly post discharge nursing care following an observation encounter—the subject of a recent lawsuit against the federal HHS.

 Under current Medicare regulations, observation services are billed as outpatient treatment, not inpatient care. One of the main requirements for Medicare coverage of post acute care in a skilled nursing facility (SNF) is a 3-day hospital admission. Time spent by patients in observation is not counted toward the required 3-day inpatient stay,

 and consequently, beneficiaries may bear prohibitively high out-of-pocket costs.

The frequency of discharge from observation services to SNFs and the characteristics of beneficiaries who receive SNF care after observation services are currently unknown. We used a nationally representative sample of community-dwelling Medicare fee-for-service (FFS) beneficiaries to evaluate patterns and predictors of care after observation services, and to estimate the potential financial impact on beneficiaries; such data are necessary to assess the impact of the 3-day rule on beneficiaries and the potential costs related to the use of observation services.

We performed a retrospective cohort study of FFS Medicare beneficiaries receiving observation services in 2010 using a 20% nationally representative sample from the CMS Chronic Conditions Data Warehouse. Beneficiaries were included if they had at least 1 observation service encounter.

Observation encounters were linked to an administrative Medicare file that specifies beneficiaries’ daily location of care: hospital, SNF, community with home health services, community without services, and deceased. This timeline file is created from Medicare inpatient claims, SNF claims, home health claims, the Long-Term Care Minimum Data Set (MDS), and the Outcome and Assessment Information Set.

The study cohort was constructed by identifying beneficiaries who met Medicare’s criteria for payment for observation services.

 In the 20% sample from 2010, 241,929 beneficiaries received observation services. We limited our analysis to beneficiaries who were community dwelling prior to their observation service (n = 209,613) and for beneficiaries with multiple encounters, we utilized the first service. We further limited our analysis to beneficiaries who were enrolled in Medicare Parts A and B for the full study period or covered until their date of death (n = 195,143) and discharged alive from their index encounter (n = 195,068). After exclusions, 195,068 beneficiaries remained in our cohort.

We assessed the rate of SNF utilization directly following hospital-based observation services. Post observation discharge destinations included return to the community without home health services, return to the community with home health services, or discharge to an SNF. Discharge to an SNF outcomes were further separated to delineate stays that were covered by Medicare and stays that were not covered by Medicare.

Demographic and clinical variables from administrative files, including age, gender, race, dual status, and presence of end-stage renal disease, were used to describe the sample. Beneficiaries’ chronic conditions were identified using Chronic Conditions Data Warehouse condition flags,8 and the following 14 conditions were used: dementia (Alzheimer’s disease, related disorders, or senile dementia), atrial fibrillation, cancer (breast, colorectal, lung, prostate), chronic kidney disease, chronic obstructive pulmonary disease, depression, diabetes, glaucoma, congestive heart failure, ischemic heart disease, osteoporosis, arthritis (rheumatoid or osteoarthritis), hip fracture, and cerebrovascular disease (stroke or transient ischemic attack).

Medicare Part A deductible and coinsurance amounts and the beneficiary deductible liability amounts were obtained from the 20% Medicare SNF Standard Analytic File. The presence or absence of supplemental Medicaid coverage was obtained from the Medicare Master Beneficiary Summary File.

We described the characteristics of the sample and evaluated unadjusted differences in beneficiary characteristics, characteristics of the stay, and discharge settings using 

 tests. For beneficiaries discharged to Medicare-covered SNFs following index observation stay—and eligible for Medicare coverage due to the fact that they previously met Medicare eligibility requirements for coverage&mdash;beneficiary liability was calculated by combining the MedicarePart A deductible and coinsurance amounts with the beneficiary blood deductible liability amount.

Because available data only capture Medicare costs, the estimated financial impacts of current policy on non—Medicare-covered SNF costs following observation stays were calculated by classifying beneficiaries discharged to non–Medicare-covered SNF stays following index observation stay, into 3 subgroups: 1) those with full Medicaid coverage, 2) those with partial Medicaid coverage, and 3) those with no Medicaid coverage. We were unable to accurately estimate the financial impact of Medicaid coverage due to the lack of availability of 2010 Medicaid financial data. For the purposes of this analysis, those with full Medicaid coverage were assumed to have no financial liability, and for those without Medicaid coverage, we assumed beneficiaries might have been liable for up to 100% of the costs for care. As such, cost estimates for this group were derived using average SNF LOS for this beneficiary group and reported average cost per day for a semi-private SNF room in 2010.

 For the small group of beneficiaries with partial Medicaid coverage, we did not have sufficient information to make any financial estimates.

All analyses were performed using SAS version 9.1 (SAS Institute, Cary, North Carolina). This study was considered exempt from review by the Yale University Human Investigations Committee.

In 2010, 195,068 community-dwelling beneficiaries received observation services and met study criteria, representing 3.2% of community-dwelling Medicare FFS beneficiaries. The mean age was 72.5 years (SD = 13.3) with beneficiaries distributed across all age groups (

). The majority of beneficiaries were white (84.9%), male (58.9%), and of non-dual status (77.1%). Overall, the average time spent in observation status averaged 25.7 hours, but varied substantially (SD = 17.2) (

). Those discharged to SNFs experienced longer LOS in observation than those discharged back to the community (34.9 hours vs 25.5 hours; 

<.01). Approximately one-fourth of beneficiaries discharged to SNFs had observation LOS of 48 hours or more.

Disposition Following Observation Encounters

Following the index observation encounter, beneficiaries were overwhelmingly (96.5%) discharged back to the community without home health services (Table 1); less than 2% (1.7%) of beneficiaries were discharged back to the community with home health services. The remaining beneficiaries in our sample—only 1.8%&mdash;were discharged to an SNF. Less than 1% (0.6%) of the sample were discharged to an SNF with the stay covered by Medicare, while 1.2% were discharged to an SNF with their stay not covered by Medicare.

Beneficiaries discharged to a non-covered SNF tended to be older compared with beneficiaries discharged to a covered facility (82.5 years vs 80.1 years; 

<.01). Beneficiaries discharged to a covered SNF also tended to have more comorbid conditions (

<.01), with 60.8% having at least 5 comorbid conditions compared with those discharged to a non-covered SNF—of whom 40.8% had at least 5 comorbid conditions.

Of the 1196 beneficiaries in our sample who were discharged to a Medicare-covered SNF, LOS and Medicare payment records for services provided in 2010 were obtained for 1188 beneficiaries. These beneficiaries had an average LOS in observation of 35.3 hours (SD = 29.6 hours), and their average LOS in the SNF was 19.0 days (SD = 15.4 days) in 2010 for the stay immediately following their observation service encounter. On average, beneficiaries were liable for $1414.84 (SD = $827.11) beyond their Medicare SNF coverage.

Of the 2319 beneficiaries who received observation services and were discharged to a non—Medicare-covered SNF, 989 beneficiaries had full Medicaid coverage and 41 beneficiaries had partial Medicaid coverage. No financial analysis was done for these Medicaid subgroups. For the 1289 beneficiaries discharged to a non–Medicare-covered SNF without Medicaid coverage, the average length of observation services was 34.1 hours (SD = 21.0 hours). Beneficiaries had an average SNF LOS of 44.1 days (SD = 63.9 days) following their observation service encounter. Given that the average cost per day for a semiprivate SNF room in 2010 was $205,9 these beneficiaries may have been liable for up to 100% of the $9040.50 estimated payments for care.

In this sample of community-dwelling Medicare beneficiaries receiving observation services in 2010, 1.2% were discharged from observation services to an SNF that was not covered by Medicare; this represents 0.04% of the community-dwelling Medicare population. Less than 2% of those receiving observation services were discharged to home with home health, suggesting that this is a rare substitution.

There was considerable variation in time spent in observation. Beneficiaries who were discharged to SNFs had a LOS nearly 10 hours longer than those who were discharged to the community, and were more likely to have observation services exceeding 48 hours. Because beneficiaries discharged to an SNF tend to be older, their observation encounters may involve the integration of diverse elements of health and social care services. This combination of factors is likely to result in longer LOS at a time when LOS in observation services and inpatient encounters are increasingly under scrutiny.

While CMS currently recommends that observation LOS not exceed 48 hours, current policy does not prohibit a longer stay. Responding to the growth in long-stay observation cases (those greater than 48 hours) which have increased from 8% of all observation cases in 2006 to 12% in 2008, and concerns regarding the protection of beneficiaries as well as hospital requests for guidance, CMS introduced a policy change in fiscal year 2014.

 In the Acute Inpatient Prospective Payment System rule, CMS revised its guidance on inpatient admissions by stating that an inpatient admission is generally appropriate if the physician anticipates that the stay requires a duration of at least 2 midnights (meanwhile, the Medicare coverage of SNF stays still requires a prior 3-day inpatient stay). This rule change will likely result in more short-stay observation admissions and decrease the number of long-stay observation cases. In light of our finding that nearly

a quarter of the 2319 beneficiaries in our sample that were discharged to a non—Medicare-covered SNF have observation stays of 48 hours or more, limiting the number of long-stay observation cases may decrease the number of beneficiaries adversely affected by the 3-day rule required for SNF coverage.

Although our study finds that only a small percentage of beneficiaries are affected by Medicare’s SNF coverage rules, for those affected, the costs may be substantial. Beneficiaries discharged to an SNF not covered by Medicare have a longer LOS than those beneficiaries discharged to a Medicare-covered SNF. This may indicate that the former group of beneficiaries are more similar to long-term care beneficiaries than they are to those beneficiaries receiving SNF services. That being said, the daily costs of care can quickly accumulate, particularly for those without supplemental insurance. Future research should more rigorously assess the receiving of observation services by Medicare beneficiaries, including their clinical needs and the financial ramifications of those needs.

The benefits and consequences of an observation encounter will vary across beneficiaries and their individual needs. Therefore, it is necessary to provide beneficiaries with adequate information to make informed choices. Medicare provides publicly available guidance to beneficiaries, and state already requires hospitals to provide notice to patients placed in observation services and to explain the implications thereof.

 As observation service utilization continues to increase, it will become increasingly imperative to include patients in the decision-making process.

The findings from this analysis should be viewed in the context of a number of limitations. First, our sample did not include beneficiaries who were initially placed in observation but later converted to an inpatient stay. Second, identification of beneficiaries discharged to a noncovered SNF relied upon an MDS assessment submitted to Medicare by nursing facilities. We applied a generous 5-day window following discharge from observation to identify these MDS assessments; while MDS assessments are required on admission for all residents in certified nursing homes, regardless of source of payment, it is possible that some facilities may not have completed forms for all beneficiaries, thereby underestimating our results. Third, SNF LOS data were limited to 2010, and consequently, there may be some beneficiaries with a truncated LOS, thereby also underestimating time spent in the SNF and associated costs. Finally, we are unable to definitively determine the level of financial liability for those with full or partial Medicaid coverage due to the lack of Medicaid data availability and the possibility of retrospective coverage. Our financial analysis also does not account for any supplemental private insurance. As such, our financial impact analysis should be viewed as composed

of estimates only, and may not reflect actual beneficiary out-of-pocket expenses.

To date, individual and anecdotal data presented by the public media about negative potential unintended consequences of observation encounters have lacked evidence regarding the number of Medicare beneficiaries who are actually impacted. These findings quantify the potential adverse impact of being placed in observation and then subsequently being discharged to SNFs. 

Center for Innovation to Implementation, VA Palo Alto Healthcare System (AAV), Palo Alto, CA; Rapid-Cycle Evaluation Group, Centers for Medicare and Medicaid Innovation, CMS (SGC), Baltimore, MD; Office of Information Products and Data Analytics, CMS (CAP), Baltimore, MD; Section of General Internal Medicine, Yale University School of Medicine (JSR), New Haven, CT; Center for Outcomes Research and Evaluation, Yale-New Haven Hospital (JSR), New Haven, CT; Department of Medicine, Brigham and Women’s Hospital and Harvard Medical School (WHS), Boston, MA.

This work was not supported by any external grants or funds. At the time this work was conducted, Dr Vashi was a fellow in the Clinical Scholars program at Yale University sponsored by the Robert Wood Johnson Foundation. Dr Ross is supported by the National Institute on Aging (K08 AG032886) and by the American Federation for Aging Research through the Paul B. Beeson Career Development Award Program.

Dr Ross is a member of a scientific advisory board for FAIR Health, Inc. Dr Ross also receives research funding through Yale University from CMS to develop and maintain performance measures that are used for public reporting; from Medtronic, Inc, to develop methods of clinical trial data sharing; and from the Pew Charitable Trusts to examine regulatory issues at the FDA. At the time of this study, Dr Shrank received grant funding from CVS Caremark, where he is now employed. Dr Shrank has also received research grants from Lilly, Teva, and the National Association of Chain Drug Stores. Drs Vashi and Powers and Ms Cafardi report no relationship or financial interest with any entity that would pose a conflict of interest with the subject matter of this article.

Concept and design (AV, SGC, JSR, WHS); acquisition of data (SGC, CAP, WHS); analysis and interpretation of data (AV, SGC, CAP, JSR, WHS); drafting of the manuscript (); critical revision of the manuscript for important intellectual content (AV, SGC, CAP, JSR, WHS); statistical analysis (AV, SGC, CAP); and supervision (JSR).

Anita A. Vashi, MD, MPH, MHS, 16 Laguna St, Apt 303, San Francisco, CA 94102. E-mail: anitavashi@gmail.com.

Only 1.2% of community-dwelling Medicare beneficiaries who received hospital-based observation services in 2010 were discharged to an SNF not covered by Medicare.

Observation Encounters and Subsequent Nursing Facility Stays

To determine the prevalence of physician incentives for quality and to test the hypothesis that the quality of ambulatory medical care is better when provided by physicians with

Cross-sectional study using data from the National Ambulatory Medical Care Survey.

We examined the association between 12 measures of high-quality ambulatory care and physician compensation based on quality, physician compensation based on satisfaction, and public reporting of quality measures.

 Overall, 20.8% of visits were to physicians whose compensation was partially based on quality, 17.7% of visits were to physicians whose compensation was partially based on patient satisfaction, and 10.0% of visits were to physicians who publicly reported quality measures. Quality of ambulatory care varied: weight reduction counseling occurred in 12.0% of preventive care visits by obese patients, whereas no urinalysis in patients with no indication was achieved in 93.0% of preventive care visits. In multivariable analyses, there were no statistically significant associations between compensation for quality and delivery of any of the 12 measures, nor between compensation for satisfaction and 11 of the 12 measures; the exception was body mass index screening in preventive visits (47.8% vs 56.2%, adjusted 

 = .004). There was also no statistically significant association between public reporting and delivery of 11 of 12 measures; the exception was weight reduction counseling for overweight patients (10.0% vs 25.5%, adjusted 

We found no consistent association between incentives for quality and 12 measures of high-quality ambulatory care.

Using data from the National Ambulatory Medical Care Survey, we examined the association between 3 incentives—financial compensation for quality and patient satisfaction, and public reporting of quality measures&mdash;and 12 ambulatory care quality measures.

		 We found fewer than one-fifth of ambulatory visits were to physicians receiving these incentives.

		 There were no consistent associations between financial incentives or public reporting and 12 measures of ambulatory quality.

Variation in quality is a problem of the US healthcare system.

 Pay for performance and public reporting of quality measures are 2 incentives that health insurers and payers use to promote high-quality medical care.

 Almost half of commercial health maintenance organizaions (HMOs) use some form of pay for performance, according to a 2006 national survey.

 In addition, the Deficit Reduction Act of 2005 and the Tax Relief and Health Care Act of 2006 required the Centers for Medicare & Medicaid Services (CMS) to establish plans for value-based payments and physician quality reporting mechanisms.

 In 2007, CMS instituted a voluntary physician reporting program—the Physician Quality Reporting System&mdash;as part of its valuebased purchasing program.

 CMS also has an ongoing demonstration project, the Physician Group Practice project, to study the impact of pay for performance and other payment models on quality of care.

Patient Protection and Affordable Care Act of 2010 has allocated funding for models such as accountable care organizations and the Patient- Centered Medical Home, which tie compensation to quality.

Despite increasing interest, the impact of pay for performance and public reporting on the quality of ambulatory medical care is unclear.

 A 2006 systematic review of the literature found that very few studies had assessed the effect of pay for performance on quality.

 Of these studies, the impact of pay for performance varied: several programs showed improvements in quality and others showed little or none. Even less is known about the effect of physician public reporting of quality measures on ambulatory quality of care.

 Furthermore, evaluations of physician-level incentives to improve quality have focused on a small number of health plans or systems, or programs outside the United States.

 To our knowledge, no studies have examined the association between physician incentives and quality of ambulatory care on a national level.

Using a nationally representative survey of ambulatory visits in the United States, we sought to determine the prevalence of physician incentives for quality and to test the hypothesis that quality of ambulatory medical care is better in the context of these physician incentives. We looked specifically at financial incentives that reward higher quality care and financial incentives that reward better patient satisfaction with the hypothesis that these methods of compensation will be associated with higher quality care. We also examined the association between quality and public reporting of quality measures with the assumption that physicians are motivated to improve performance on publicly reported quality measures to maintain or improve their professionalreputation and patient volume.

We performed a cross-sectional analysis using data from the 2006 and 2007 National Ambulatory Medical Care Survey (NAMCS). NAMCS is a nationally representative survey administered by the Centers for Disease Control’s National Center for Health Statistics. NAMCS contains information about patient visits to non—federally funded, non–hospitalbased offices throughout the United States. Physicians in the fields of anesthesiology, radiology, and pathology are excluded from the survey. Physicians who participate in the survey cannot participate again for at least 3 years.

NAMCS uses a 3-stage sampling design. The first stage is based on geographic location, the second stage identifies offices in each geographic location, and the third stage samples visits within each office. The visits sampled take place during a 1-week period that is randomly assigned for each practice. Between 20% and 100% of the visits that week are sampled, depending on the size of the practice. The National Center for Health Statistics weighs each visit so that the data can be used for national estimates. Each visit weight accounts for selection probability, adjusts for nonresponse, and accounts for other factors so that the national estimates properly reflect the scope of ambulatory visits in the United States.

The survey collects patient and office demographics, and visit-specific clinical information. The information from each visit is recorded on a standardized survey form by the physician, office staff, or a US Census Bureau representative. Clinical characteristics include up to 3 reasons for the visit coded using a reason-for-visit classification, up to 3 diagnoses coded using the International Classification of Diseases, Ninth Revision, Clinical Modification (ICD-9-CM), and up to 8 drugs coded using the Lexicon Plus classification, a proprietary database of Cerner Multum, Inc. Because this study used publicly available NAMCS data without respondent identifiers, the Mount Sinai Institutional Review Board exempted it from review.

We limited our sample to visits by nonpregnant patients who were >18 years of age. Because this study focused on rates of delivery of high-quality ambulatory care, we further limited our sample to visits to primary care physicians (ie, general internists, family practitioners, gynecologists) and internal medicine subspecialists (eg, cardiologists, endocrinologists). We excluded visits to surgical specialists and nonmedical specialists (eg, dermatologists, psychiatrists, occupational medicine) because the measures of high-quality care of interest did not pertain to their areas of practice. Finally, we excluded visits in which a physician was not seen.

We found no consistent associations between physician incentives for quality improvement and 12 measures of ambulatory quality of care.

Joseph S. Ross, MD, MHS

Articles

Medicare Beneficiaries’ Out-of-Pocket Costs for Commonly Used Generic Drugs, 2009-2017

Association Between FDA Black Box Warnings and Medicare Formulary Coverage Changes

Observation Encounters and Subsequent Nursing Facility Stays

Association Between Physician Quality Improvement Incentives and Ambulatory Quality Measures

Scheduled and Unscheduled Hospital Readmissions Among Patients With Diabetes