		Current healthcare payment systems in the United States are all plagued by the problem of adverse incentivization.

		Capitated systems encourage providers to underutilize resources so as to maximize profit, potentially limiting patients’ access to useful medications or procedures.

		Fee-for-service systems encourage providers to overutilize resources, perversely rewarding poor outcomes that require subsequent costly treatment. 

		A more efficient system, which to my knowledge has not previously been proposed, would be one which leaves billing entities uncertain of how a given patient encounter will be reimbursed.

Current payment systems in United States healthcare are plagued by the problem of adverse incentivization. Capitated systems encourage providers to underutilize resources so as to maximize profit, potentially limiting patients’ access to useful medications or procedures. Fee-for-service systems encourage providers to overutilize resources, straining an already bloated healthcare budget and perversely rewarding poor outcomes that require subsequent costly treatment. A recently proposed reimbursement system—the accountable care organization (ACO)&mdash;seeks to mitigate the magnitude of adverse incentivization by combining system-wide capitation with provider-level quality and safety benchmarks.

While ongoing efforts to simultaneously maximize healthcare quality and minimize cost are to be applauded, early experiences from ACO trial sites have yielded mixed results.

 The ACO model may be equally unable to yield an optimal balance of quality and cost as traditional reimbursement models because, like those systems it is designed to replace, it fails to directly address the economic reasoning of ground-level healthcare providers. Savvy billing entities (providers, provider networks, or entire healthcare systems) with a comprehensive understanding of any reimbursement system will utilize that understanding to maximize profit within that system. It is the a priori knowledge of the reimbursement scheme’s methodology that creates a financial incentive for either underprovision or overprovision of care.

A more efficient system, which to my knowledge has not previously been proposed, would be one which leaves billing entities uncertain of how a given patient encounter will be reimbursed until after that encounter has been concluded. In such a system, a neutral and blinded third-party organization would randomly assign closed patient encounters to either fee-for-service payment or capitated payment. Because providers would not have been cognizant of the reimbursement scheme at the time of care provision, the only incentive for providers at point of contact would have been to provide the best and most judicious possible care to the patient.

Such a system, while purely theoretical, would not be prohibitively difficult to implement, as it could be built upon existing payment schemes. The system would be highly dependent upon the existence of a neutral and trustworthy third-party organization to adjudicate claims, but the underlying relationships between payers and providers would not need to be drastically altered. Until the issue of adverse incentivization is addressed at a more fundamental level, currently envisioned payment schemes may, by design, continue to yield suboptimal outcomes.

Department of Surgery, University of Colorado- Denver, Aurora, CO.

The author reports 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; drafting of the manuscript; critical revision of the manuscript for important intellectual content.

 David Paul Kuwayama, MD, MPA, Department of Surgery, University of Colorado-Denver, 12631 E 17th Ave, Mail Stop C312, Aurora, CO 80045. E-mail: david.kuwayama@ucdenver .edu1. Centers for Medicare & Medicaid Services. Pioneer Accountable Care Organizations succeed in improving care, lowering costs. http://www.cms.gov/Newsroom/MediaReleaseDatabase/Press- Releases/2013-Press-Releases-Items/2013-07-16.html. Published July 16, 2013. Accessed October 21, 2013.

Articles

LETTERS TO THE EDITORS

Current healthcare payment systems in the United States are all plagued by the problem of adverse incentivization. A more efficient system would be one which leaves billing entities uncertain of how a given patient encounter will be reimbursed.

A Novel Approach to Minimizing Adverse Incentivization in Healthcare Payment Systems

To quantify how access to on-patent drugs by tier placement varies by insurance type and therapeutic area.

Retrospective analysis of insurance plan drug coverage data.

 Drug coverage information was collected from the Fingertip Formulary database in May 2011 for 3 drug classes (statins, angiotensin II receptor blockers, and protein-tyrosine kinase inhibitors) across 3 therapeutic areas with varying levels of generic drug availability. A generalized linear model was used to estimate the percentage of available on-patent drugs covered in the formulary tiers with lowest copay requirements (tiers 1 and 2) in different types of healthcare insurance plans in the United States.

There were substantial differences between insurance types in the number of on-patent drugs reimbursed in tiers 1 and 2 (ie, with a low copay). Compared with commercial plans, there were more on-patent drugs reimbursed with a low copay in employer plans, union plans, and with pharmacy benefit management companies, and substantially fewer on-patent drugs with a low copay in Medicare plans (Medicare Advantage, special needs, prescription drug plans) and discount prescription programs. These results were expected, as union plans are known for their generosity and Medicare plans rely heavily on cost containment (eg, cost sharing). For commercial Medicaid and municipal plans, the findings were dependent on the therapeutic class, or were inconclusive. The number of competitors a plan faces did not consistently affect the coverage of on-patent drugs.

The degree of coverage of on-patent drugs in the lowest copay tiers varies dramatically between insurance types, especially for expensive protein-tyrosine kinase inhibitors.

		Insurance plans assign the drugs that they cover (ie, reimburse) to “tiers,” with the copay varying by tier (the lower the tier, the lower the copay).

		For 3 drug classes in distinct therapeutic areas, access to on-patent drugs in the lowest copay tiers (1 and 2) was higher in employer, union, and pharmacy benefit management plans than in commercial plans. However, in Medicare plans, substantially fewer on-patent drugs were included in these tiers, which could create choice restrictions and thus compliance issues.

		Medicare plans should reconsider access to on-patent drugs by tier for high-cost therapeutic areas without affordable alternatives.

In the United States in 2010, approximately 63% of individuals who had health insurance were insured through private plans, 24% by government health programs, and 13% by both types of programs.

 The vast majority (approximately 85%) of privately insured Americans had access to health insurance through private employers.

The main government programs are Medicaid and Medicare, covering 19% and 17% of the insured population, respectively. Medicaid is available to certain low-income individuals and families. Medicare provides health insurance for individuals older than 65 years, those younger than 65 years with certain disabilities, and people with end-stage renal disease. Additional information regarding source of health insurance can be found in 

Health plans usually assign the drugs that they cover to “tiers.” A drug’s tier determines the degree to which the patient will have to contribute a payment for the drug—the lower the tier, the lower the copayment. Tier 1 is typically for generics, tier 2 for preferred brand name drugs, tier 3 for nonpreferred brand name drugs, and tiers 4 (and above) for coinsurance brands. Fixed-sum copayments are required from patients for drugs in tiers 1 through 3 to cover some of the drug costs. Coinsurance copayments are a percentage of a drug's cost and vary from drug to drug. An individual plan may include patent-protected brand name (“on-patent”) drugs, brand name drugs that are no longer patent protected (“off-patent”), and generic bioequivalents. For individuals receiving healthcare coverage via employment, the average 2012 copayment was $10 for first-tier drugs, $29 for second-tier drugs, $51 for third-tier drugs, and $79 for fourth-tier drugs.

 In 2009, the average Medicare Part D copayment was $10 for first-tier drugs, $37 for second-tier drugs, and $75 otherwise.

 The Medicare Part D coinsurance rate on the specialty tier ranged from 25% to as high as 335, with a median of 30%.

 The drugs assigned to each tier vary with individual healthcare plans, and this is one of the aspects of a plan on which the consumer may base their decision when choosing a plan.

From a patient’s perspective, plans that offer drugs relevant to their condition in the low copay tiers are likely to be more attractive, assuming the monthly premiums are comparable. Another factor that might make a plan more favorable is the degree of choice of drugs within a therapeutic class in the low copay tiers. The ability to choose (doctors, hospitals, and drugs) is seen as very important or extremely important by the vast majority of Americans when selecting healthcare plans.

 Evidence that the level of copayments for a given medication impacts the choice of health plans or pharmacies is limited, however. Zou and Zhang found that only 5% of Medicare Part D beneficiaries chose the cheapest plan available in their region given their medication needs.

 Linton et al found that 22% of Department of Defense beneficiaries older than 65 years used the option with lower copay.

 The ability to choose a drug may have implications for patient adherence to prescribed medication regimens. In particular, having access to, and a choice of, on-patent drugs at a reasonable copay level may be an important factor for patients (ie, for patients who have concerns regarding drug switching or generic substitution). There is evidence that increasing a drug’s copayment can decrease its utilization level, reduce adherence,

The objective of this study was to assess whether the level of on-patent drug coverage varies according to:

		Plan type. Some categories of plans (including some union plans) were expected to offer health insurance packages with generous benefits such as absence of copayments, deductibles, or prior authorizations, even for on-patent drugs for which an alternative generic is available. Those plans can be costly to employers, in which case they would be called “Cadillac plans.”

 In such plans, a range of on-patent drugs might be assigned to lower tiers.

		Therapeutic class. The proportion of on-patent drugs covered by the plan (as a percentage of all on-patent drugs available in the same class) would be expected to be low in therapeutic classes where a large number of drugs are available. For those therapeutic areas, managed care organizations could be effective and aggressive in managing formularies and in influencing physicians’ prescribing behavior.

The implications of the analysis are important for: (1) consumers and employers who, when choosing insurance, want to know whether certain types of plans provide more choice compared with others; (2) lawmakers who want to understand the impact on drug choice of curbing the use of Cadillac plans; (3) public health specialists who want to understand whether the population has affordable access to life-saving medications.

To date, the published literature reporting on drug tiers and formularies in the United States has focused mainly on: the decrease in drug utilization after a copayment increase

; the association between the formulary position (tier) and value of a drug

; and the decision-making process in assigning a drug to a tier.

 For instance, Linton et al observed that the esomeprazole (Nexium) share of the proton pump inhibitor market dropped from 20.0% to 15.7% in the month after TRICARE moved esomeprazole from tier 2 to tier 3 (copay increased from $9 to $22).

Similarly, in the antidepressant market, Hodgkin et al found that drugs that became non-preferred in insurance plans experienced a decrease of 11% in the number of prescriptions filled per enrollee (vs an increase of 5% in the comparison group).

 No articles have reported on the differences in coverage between insurance type and therapeutic classes. By analyzing such differences, this paper contributes to the more limited literature that advises patients on identifying plans that have limited drug formularies

 and whether formularies meet the needs of all patients.

The data set used was the Fingertip Formulary (Fingertip Formulary LLC, Glen Rock, New Jersey), which is a collection of drug coverage data for 1768 health plans in the United States. Fingertip Formulary estimates that between 95% and 98% of the covered lives in the United States are in the database. The data were collected from health plans’ websites or by directly contacting the health plan providers. The information changes regularly; the data used in this paper were downloaded on May 15, 2011. While drug coverage typically changes once a year (before members enroll), it could change during the year if there is a market event (eg, an on-patent drug goes off patent) and/ or if the contract with a manufacturer was not renewed. The author accessed the website on May 15, 2011, and downloaded the information available that day. Since no angiotensin II receptor blockers (ARBs) and no proteintyrosine kinase inhibitors (P-TKIs) lost patent that year, it is likely that the conclusions would not have changed dramatically in 2011. The conclusions on HMG-CoA reductase inhibitors (HMGs) would have remained valid for most of 2011, since Lipitor lost its patent on November 30, 2011.

Fingertip Formulary data include a number of fields that are identical for all drugs: (1) health plan identifier; (2) provider (ie, company providing the health plan) identifier; (3) state(s) where the health plan operated; (4) lives covered (ie, number of enrollees in the plan); (5) plan/ insurance type (commercial, pharmacy benefit manager [PBM], employer, Medicare Advantage [MA], Medicare Part D plan [PDP], special needs plan [SNP], state Medicaid, commercial Medicaid plan [if the Medicaid plan was contracted to private health companies], union, municipal, discount prescription programs). In Fingertip Formulary, employer plans are those offered by organizations that arrange pharmaceutical benefits for workers as part of their total benefit plan and whose data. Fingertip Formulary can source in an ongoing basis. Commercial plans are set by non-Medicare and non-Medicaid organizations that offer pharmaceutical benefits to individuals and/or groups such as businesses and local governments. The database allows for information to be included on the tier status (whether a drug is in tiers 1-6, not reimbursed, or not available) for each drug within each plan. 

 provides the number of plans and the number of insured individuals for each insurance type.

Three different therapeutic classes with different levels of generic competition and disease burden were analyzed in this paper: (1) statins (3-hydroxy-3-methylglutaryl coenzyme A reductase inhibitors; HMGs), used to lower serum cholesterol levels; (2) ARBs, used to reduce hypertension; and (3) P-TKIs, treatments for chronic myeloid leukemia (CML) and gastrointestinal stromal tumor. At the time of the data download, the HMG class comprised 4 off-patent products that had generic equivalents and 8 on-patent products. Six ARBs had patent protection and 1 ARB was available off patent and as a generic equivalent. Azilsartan medoxomil (Edarbi) had only recently been launched and was not included in the analysis, as a number of plans had no information on the drug. All available P-TKIs were still within patent, and no generic versions were available. Coinsurance copays were much higher for P-TKIs than for HMGs and ARBs. For instance, if the coinsurance rate was 30%, the average 2011 copay for a 30-day supply was $1820 for imatinib mesylate Gleevec), $56 for atorvastatin 80 mg (Lipitor) and $48 for valsartan/hydrochlorothiazide 320 mg/25 mg (Diovan HCT).

The objective of the model was to explain how the number of on-patent drugs in a low tier varies by insurance type (eg, commercial, municipal, Medicare) and therapeutic area. The level of drug coverage could be defined in several ways: number of reimbursed drugs, number of reimbursed on-patent drugs, number of drugs with a lowtier copay, number of on-patent drugs with a low-tier copay, etc. Because generics are typically reimbursed and/ or are affordable, the focus was on on-patent products. Therefore, coverage was defined as the number of drugs with patent protection that were in a low copay tier (ie, tiers 1 or 2), and is referred to as “on-patent drug coverage” hereafter. Medicare PDPs, MA, and SN plans were grouped into a Medicare variable.

A logit regression was used to explain drug coverage. Tobit and count models were also considered to ensure that the findings did not depend on the methodology. More details on the models can be found in 

. Plans without coverage information for all products in a therapeutic area were excluded for such analyses. On-patent drugs used in the analysis are shown in 

Generalized Linear Model With Logit Link 

A generalized linear model (GLM) with a logit link function was used to examine the relationship between the percentage of available on-patent drugs in a therapeutic area that are covered in tiers 1 or 2 and the variables of interest: insurance type and therapeutic areas. Hosmer- Lemeshow and Pregibon link tests were used to ensure that the model adequately ï¬t the data.

Use of Poisson or negative binomial regressions was also considered, as the number of on-patent drugs with favorable tier coverage was a count variable. However, the Poisson distribution was not deemed relevant because the distributions were overdispersed (

). Therefore, only a negative binomial model was analyzed.

Drug coverage data can be considered right-censored, as the number of covered drugs was, by definition, bound by the number of approved brands. Therefore, applying a censored model (Tobit regression) would be appropriate. Intuitively, a censored model should be more relevant for therapeutic classes with few or no interchangeable options. In this case, one could assume that insurance companies would be willing to cover more drugs if more were available. The high number of plans covering all 4 P-TKIs was consistent with this hypothesis. As the number of approved drugs was specific to each therapeutic area, a Tobit model was run separately for HMGs, ARBs, and P-TKIs. As the assumption of normality of the residuals was rejected (ie, the maximum-likelihood estimator was inconsistent), a bootstrap approach was used (with 1000 replications at 95% CI).

All analyses were performed using the statistical analysis program Stata, version 12.1 (StataCorp LP, College Station, Texas).

shows the average, across all plans, of the number of drugs in each tier. The numbers were calculated by first extracting, from the database, the number of drugs by tier for each plan. Then, for each tier, the average of the number of drugs was calculated across all 1768 plans. The results can be interpreted in terms of how the drugs are split across tiers. On average, 4.7 HMGs and 2.1 ARBs were not reimbursed. Most P-TKIs were reimbursed—an average of only 0.5 drugs out of 4 available drugs (12%) was not reimbursed. HMGs and ARBs were rarely assigned to tiers 4 and above (average 3% of available ARBs and HMGs were in such tiers), but P-TKIs were more frequently found in those tiers (average 31% of available P-TKIs). As expected, onpatent drugs were rarely in tier 1 because it is typically reserved for generics. On average, approximately 30% of on-patent ARBs and HMGs were not reimbursed by health insurance plans. The proportion of on-patent drugs in tier 2 was similar across therapeutic classes (34%-39%), but tier 3 was used much less for on-patent P-TKIs than for on-patent HMGs or ARBs (9% vs 26% and 30%, respectively).

On-patent drug coverage varied greatly by plan type (

). Across therapeutic areas, discount prescription programs and Medicare plans offered the lowest number of on-patent drugs in tiers 1 and 2, especially for P-TKIs. P-TKIs were often in tier 4 or higher in Medicare plans, meaning that a coinsurance was required, likely due to their relatively high cost (eg, in 2011, the average wholesale price of the P-TKI imatinib mesylate [Gleevec] was $202 per day for patients with chronic-phase Philadelphia chromosome— positive CML

). Union and municipal plans had many on-patent HMGs in tier 1 or 2 (ie, the lowest copay tiers), which seems to corroborate the assumption that those plans offer generous drug benefits. Commercial Medicaid plans placed a high proportion of the on-patent drugs in the lowest copay tiers. The results for state Medicaid will be further discussed in the Summary and Discussion section.

 shows summary statistics for the independent variables used in the model specifications. Commercial plans were the most prevalent plans in the database. PBMs were the plans with the most enrollees, on average. The variable capturing the number of competitors is also called “plan competition” in the remainder of the article. Its calculation can be found in online

. The categories of plans with the highest number of competitors were PBMs and Medicare PDP.

The results of the GLM regression are presented in 

. The reference categories for plan and therapeutic dummies were commercial plans and HMGs, respectively. A positive coefficient sign indicates an increased probability of on-patent drug coverage in a low tier. On average, 40.1% of on-patent drugs were in a low tier. All variables were highly significant and therefore had an impact on the probability that an on-patent drug had a low copay requirement. The average marginal effect of P-TKIs represents the incremental probability, relative to HMGs, of on-patent PTKIs to be covered in a low tier. The value was +5.6 points and the branded P-TKIs were predicted to be in a low tier in 44.8% of cases versus 39.0% for HMGs. The marginal effect was —2.4 points for on-patent ARBs, which were predicted to be in a low tier in 36.7% of cases. On average, the percentage of on-patent drugs in tiers 1 or 2 was 20 points higher for union plans and 74 points higher for state Medicaid than for commercial plans. All else being equal, PBMs, commercial Medicaid, employer, and municipal plans also covered more on-patent drugs with a low copay than did commercial plans. However, Medicare on-patent drug coverage in tiers 1 or 2 was substantially lower than that of commercial plans (–23 points). If we do not control for the differences between plans, the conclusion remains that P-TKIs (respective to ARBs) have higher (respective to lower) probability of on-patent drug coverage in a low tier than HMGs (

METHODS

The proportion of available on-patent drugs covered in low copay tiers varied by insurance type, with the lowest proportion being in Medicare plans.

How Does Drug Coverage Vary by Insurance Type? Analysis of Drug Formularies in the United States

 In partnership with a large nonprofit healthcare insurer for the Mid-Atlantic region of the United States, we launched the first cancer clinical pathway in the United States in August 2008. Due to its early success with regard to savings and physician participation and compliance, a second-generation pathways program—the Oncology Medical Home&mdash;was piloted in 2011. This program offered a physician reimbursement model that shifted the source of revenue from drug reimbursement margin to professional charges for cognitive services (evaluation and management codes). We report our observations of the impact of that reimbursement model on physician prescribing behavior.

A select group of practices that participated in the first-generation pathways program were invited to voluntarily participate in the Oncology Medical Home and its cognitive weighted reimbursement design. A matched control group was chosen from the first-generation pathways participants. Comparisons of physician behavior parameters were made pre- and postimplementation and between the Oncology Medical Home practices and the first-generation pathways control group.

Physician behavior was not significantly modified by cognitive weighted reimbursement. No significant change in frequency of office visits for established patients was observed. No change in chemotherapy prescribing was observed. Observed increases in generic regimen use were no different than matched control.

Observations from this oncology medical home pilot program suggest that reimbursement methodology alternatives to the prevailing fee-for-service may have less impact on prescribing behavior than has been conjectured. Future research is ongoing to validate these observations and assess additional influences on prescribing behavior.

Increased use of chemotherapy and more expensive drugs has been correlated with the “buy-and-bill” reimbursement model; therefore, we shifted the primary source of provider revenue from drug reimbursement to professional charges in our payer-sponsored Oncology Medical Home pathways program.

		Analysis showed that this novel reimbursement model did not alter physician prescribing behavior with regard to the type or frequency of chemotherapy administrations, or established and new patient visits.

		These observations suggest that medical oncology treatment selection and cancer care practice patterns may not be influenced by fee-for-service reimbursement.

		Research is ongoing to validate these observations and assess additional influences.

United States direct medical costs associated with cancer are projected to increase exponentially, from $104 billion in 2006 to over $173 billion in 2020.

 Variability in medical practice plays a large role in increasing costs without achieving better patient outcomes.

 Clinical care pathways have been cited as solutions to help bend the rising costs of cancer care by reducing unnecessary and costly treatment variation while improving quality of care.

In August 2008, a large nonprofit healthcare insurer for the Mid-Atlantic region of the United States partnered with Cardinal Health to launch the first inclusive provider network cancer care pathway in the United States. This program marked a milestone in the evolution of cancer treatment by demonstrating that an oncology pathway program can be deployed across an entire plan network comprised of disparate providers (single/group, community/academic, independent/affiliated) benefitting all parties by improving consistency and quality of care while reducing costs without reducing provider reimbursement.

We previously reported that high participation and compliance levels in our pathways program led to changes in physician patterns of care that resulted in significant decreases in overall oncology expenditures.8 Similar physician behavior changes were observed in a Blue Cross Blue Shield Michigan pathway program, where aligned stakeholder incentives drove high levels of provider participation and compliance.

 Despite these and other pathway program successes, critics suggest that the observed savings benefits are small and unsustainable.

The fee-for-service payment system has been identified as one of the main drivers of cancer care cost—the more physicians do for patients, the more reimbursement they receive.

In this model, oncologists directly purchase chemotherapy from manufacturers and/or wholesalers (typically below, at, or slightly above average sales price [ASP]) and are reimbursed by the payer at prices usually exceeding ASP by 6% (Medicare presequestration) to 30% or more (commercial payers). Studies have shown that providers’ choice of chemotherapy can be affected by reimbursement, resulting in their prescribing chemotherapy more often and utilizing more costly brand name chemotherapy over less expensive brand or generic alternatives.

Many argue that this “buy-and-bill” model encourages physicians to overprescribe chemotherapy, creates incentives for price inflation, and distorts clinical decision making, thereby driving up the costs of cancer care.

With these factors in mind, we piloted a second-generation pathways program, the Oncology Medical Home, in January 2011. The goal of this program was to decrease cancer costs beyond those observed with the first-generation pathways model, and as such, it would address the concern that chemotherapy prescribing is influenced by a “pay-for-volume” rather than a “pay-for-value” reimbursement methodology. In the Oncology Medical Home, physician reimbursement would shift the drug reimbursement margin to professional charges for cognitive services through a dramatically enhanced professional service evaluation and management (E&M) fee schedule. Other cost-saving measures for the Oncology Medical Home included physician commitment to an intensive continuous quality improvement initiative and an end-of-life program, which have been previously presented and will be addressed in a future manuscript.

This new cognitive weighted reimbursement model provokes several questions. If payment is based on professional charges rather than drug reimbursement, will physicians then evaluate patients more frequently? If chemotherapy reimbursement margins are at or near cost, will the use of chemotherapy decrease? Will there then be a shift away from brand name drugs to less costly brand drugs and/or generic drugs? To answer these questions, we evaluated the impact of the change in reimbursement on business practices and prescribing behavior for physicians participating in the Oncology Medical Home.

The first-generation oncology clinical pathway program was initiated in August 2008 as a collaborative effort between the payer and its contracted providers. Participation in the program was voluntary; however, providers were given financial incentives to participate in the form of an increased brand and generic drug (Jcode) fee schedule as compensation for the additional work flow required to maintain program compliance. A physician steering committee led participating physicians, who jointly and independently developed the content, structure, and implementation of the pathways.

The Oncology Medical Home was initiated in January 2011, with first patients accrued April 1, 2011 (with the delay needed for fee schedule implementation). The mechanism of drug reimbursement was changed for Oncology Medical Home—participating physicians from drug reimbursement margin to cognitive weighted reimbursement (

). A portion of the overall reimbursement to participating physicians was transferred from intravenous drug fees to E&M code reimbursement, while keeping revenue at the same level, when weighted by utilization. Twelve months of claims for intravenous chemotherapy, supportive care, and E&M codes were analyzed. The change in reimbursement between the preprogram rate of 24.5% margin over ASP and the Oncology Medical Home rate of 8% margin over ASP was calculated for each drug for which there were claims. Fourteen generic drugs were kept at higher reimbursement as an incentive for use where the level of reimbursement was the same for both Oncology Medical Home and control practices. In total, 70 drugs were reduced in price. This reduction was transferred as an increase to a select list of 17 E&M and chemotherapy administration codes. This resulted in an aggregate 62% increase in these fees, with an emphasis on new patient consult codes, which were increased by 166%. When weighted by utilization, the increase in revenue to the Oncology Medical Home physicians was equal to the value of the decrease in drug revenue.

The Oncology Medical Home program was offered to a subset of first-generation pathways participants, who had recently organized to form a membership association, Therapeutics and Research in Oncology (TRIO). The TRIO practices operate independently, retaining their individual tax identification numbers, and consist of providers from small and large practices, urban and rural geography, representing the payer’s 2 largest metro areas. The initial membership in TRIO included 42 physicians in 16 practices, all of whom participated in the first-generation pathway program. Thirty-three physicians in 13 practices subsequently chose to participate in the Oncology Medical Home. Reasons for nonparticipation included sale of practice to hospital, retirement, and distrust of program.

Selection of the Study and Control Groups for Comparative Analyses 

Of the 50 practices that participated in the first-generation pathways program, 32 were chosen for initial data evaluation based on the following criteria: (1) consistent data volume throughout the baseline and evaluation periods, defined as April 2010 to March 2012; (2) volume of >100 patients during the baseline period; and (3) location in Virginia, Maryland, or Washington, DC. Ten of these 32 practices were TRIO practices that participated in the Oncology Medical Home. Data from all 32 practices in the baseline year were used to create propensity scores via logistic regression.

K-nearest neighbor analysis of scores and 1-1 matching with replacement resulted in the pairing of 8 Oncology Medical Home practices to 7 first-generation pathways control practices.

 The logistic regression for Oncology Medical Home participation included the covariates of cancer and chemotherapy-treated patient volumes, cancer type, patient age group, extent of treatment, and Charlson comorbidity scores.

Comparisons of physician behavior were made between the Oncology Medical Home participating practices and the first-generation pathways control group for the year prior to (year —1) and the year following (year +1) Oncology Medical Home implementation. Behavioral comparisons included number of patients per practice, number of visits per patient, number of patients receiving chemotherapy, chemotherapy administrations per patient, and use of all generic chemotherapy regimens. Claims data from the insurance network database were collected for year –1 (representing April 1, 2010, to March 31, 2011) and for year +1 (representing April 1, 2011, to March 31, 2012).

To measure extent of treatment, therapy lines were assigned based upon grouping of chemotherapy drugs. Drugs given within 30 days of each other were grouped as a drug combination. Changes that occurred beyond 30 days triggered drug combination reassignment and incremented the line of therapy. Exceptions were made for sequential therapies, which were accounted for by distinct grouping rules. Cormorbidity scores were calculated according to Charlson using Deyo’s mapping of 

International Classification of Diseases, Ninth Revision, Clinical Modification

 A single modification to the mapping was implemented in flagging claims likely to represent solid tumors with metastases. In addition to using claims with ICD-9 codes 196.x through 199.1, patients with solid tumors who received 2 or more lines of chemotherapy were marked as with metastatic solid tumor.

All statistical analyses were conducted using SPSS 19 statistical software (IBM, Armonk, New York) with the exception of k-nearest neighbor analyses, which were performed using SPSS Modeler 14.2. Due to the large sample sizes, χ² analyses were not used for categorical comparisons. Instead, individual measures were calculated at the practice level and group means were compared by independent 

 tests. All chemotherapy evaluations were based on intravenous chemotherapy claim code 96413. New and established patient visit claim codes 99201-99205, 99241- 99245, and 99211-99215 were used in these analyses.

A total of 33 physicians from 13 TRIO practices within the insurance network who participated in the first-generation pathways project chose to join the Oncology Medical Home program. Propensity score analysis identified 8 Oncology Medical Home practices and 7 firstgeneration pathways practices that were well matched for use in these analyses (

). The matching variables were chosen as a reflection of practice disease focus, treatment preferences, and overall patient volume. After matching, baseline demographics and characteristics were similar in the 2 groups (

). The first-generation pathways control group treated 4847 patients at baseline versus 7213 for the Oncology Medical Home. The ages of patients were similarly distributed between the 2 groups. Approximately 13.5% of patients were under age 50 years, 35% were aged 50 to 64 years, approximately 25% were aged 65 to 74 years, and 25% were aged 75 years or older. Both study groups treated primarily solid tumors (85%). Other tumor types were hematologic (11%) and gynecologic (5%). Seventy percent of patients recieved first-line chemotherapy, 22% percent recieved second-line chemotherapy, and approximatley 9% recieved third-line or higher chemotherapy.

There was minimal difference in behavior change between the Oncology Medical Home and first-generation pathways providers and between study years among providers (

). The number of established patient visits in the Oncology Medical Home group remained stable from year —1 to year +1, with a mean of 3.7 and 3.8 patient visits, respectively. The same was true for the firstgeneration pathways control group, with a mean of 4.6 and 5 patient visits, respectively. New patient visits increased 2% from year –1 to year +1 for Oncology Medical Home providers compared with a decrease of 2% for the first-generation pathways control group. The percentage of chemotherapy administrations per patient remained stable among study years for both groups, with approximately 8 chemotherapy administrations per patient in each group for each year. The percentage of patients who received chemotherapy remained stable for each group among study years at approximately 15%.

There was no difference in the percentage of patients who received all-generic chemotherapy regimens between the study groups, though there was a trend for increased use of generics in year +1 for both groups; Oncology Medical Home providers increased the use of generic-only regimens by 43% in year +1 compared with first-generation pathways providers, who increased use of generics by 42%. The migration of broadly used agents gemcitabine, docetaxel, irinotecan, and oxaliplatin (transiently) from brand to generic status during the study period was largely responsible for the increase in generic-only regimen prescribing among both groups. The generic regimens lacking any of these 4 drugs accounted for 25% and 23% of control and Oncology Medical Home patients, respectively, in the program year. In the preprogram year, all-generic regimens accounted for roughly 25% of control and 28% of Oncology Medical Home patients.

We found that, surprisingly, moving financial incentives from drug administration toward cognitive services did not alter physician behavior with regard to type or frequency of chemotherapy administrations. We say surprisingly, as expectations based on prevailing wisdom suggest physicians behave in their economic best interest as long as patient outcomes are not jeopardized. This has been demonstrated in previously published reports showing that decreased reimbursement schedules correlate with increased rates of chemotherapy administration and use of more costly chemotherapeutic agents.

 In fact, a recent New York Times editorial cosigned by 20 leading academics cited fee-for-service reimbursement as the primary driver for the spiraling cost of cancer care in this country.

 Our results are quite inconsistent with this idea. Whether this pattern of care was pathway-influenced or is the result of National Comprehensive Cancer Network and other guidelines, brand name prescription drug detailing, cognitive dissonance, our culture of medicine, or other factors, is speculative without more information.

Additionally, the results from this study indicate that, despite a nearly 3-fold increase in E&M code reimbursement, no significant change in established or new patient visits was observed. This was contrary to expectation and could be related to external influences on physician practice behavior, including the historically lower contribution of E&M reimbursement to revenue, standardized and established practice patterns, and maximized throughput within office flow. If so, then the speculated impact of reimbursement reform may be overestimated.

The reimbursement level for generics did not differ between the control and the Oncology Medical Home participants. We acknowledge the potential impact due to the patent expiry of 4 drugs in the program year, which may have minimized pressure to increase reimbursement from E&M claims. However, this is a separate issue from the main question of whether physicians behave to maximize financial gain. The data we provided suggest that given the opportunity to maximize revenue by increasing select cognitive services, physicians remained unchanged in their behavior.

The nature of this observational study may raise questions over bias, and ultimately, conclusions. The methodology of conducting research in such circumstances is difficult; by definition, selection bias exists when programs are voluntary and financially incentivized. However, any selection bias incurred impacted both control and experimental cohorts, which were then matched by propensity scoring. To account for differences in disease focus, diagnosis mix was considered. To account for differences in heavily treated versus newly treated patients, and in earlystage versus later-stage treatment, the distribution of chemotherapy lines (“extent of treatment”) was considered. To account for overall treated patient burden, comorbidity index was considered. Taken together, we believe these measures yielded propensity scores indicative of case mix. However, we acknowledge that claims data, and nearly all secondary data sources, are not sufficient to account for more diverse disease stratifications. For example, race, socioeconomic status, biomarker status, disease stage, or disease histology are not sufficiently represented in claims data. Despite the absence these factors, we believe these cohorts are appropriately matched and the comparisons are valid.

We initiated the Oncology Medical Home program with the intent of further removing financial impact by offering a “white bag” drug delivery system where patient specific medications and supplies are delivered directly to practices from a dedicated specialty pharmacy. Alternatively, practices could continue their legacy buy-and-bill arrangement at a reduced fee schedule, 8% margin over ASP predetermined to be adjusted annually, where they would assume risk for price increases in brand drugs not mitigated by price equilibration for those drugs recently converted to generic. The 8% margin was selected for 2 reasons. First, this was the lowest bid by specialty pharmacy suppliers, as they are unable to purchase drugs at the same discounted prices as oncologists. Second and more importantly, modeling suggested that the 8% margin over ASP, drug utilization from the baseline control period, and the most recent Medicare fee schedule would create no net profit. All practices in the Oncology Medical Home chose to retain their current buy-and-bill practice for drugs. Acceptance of white bagging would have made cohort comparisons less complicated, but the continued buy-and-bill preference by participants, despite profit neutrality and assumed risk, represents an interesting behavioral observation. We recognize that the Oncology Medical Home program was limited to a single payer, which may not have accounted for sufficient per practice volumes to impact behavior. However, that payer represented approximately one-third of the payers of first-generation pathway providers and more than 50% of their profit, making the program financially relevant to participating practices. Although Medicare may contribute the majority of patients to an oncologist’s practice, its contribution to profit is much less significant, making commercial payers increasingly relevant to a practice’s financial integrity.

Finally, we recognize that Oncology Medical Home providers, being mature pathways participants of nearly 3 years, may have had established patterns of care that limited variance, thereby reducing the influence of reimbursement. If this is true, then pathways programs such as the first-generation program described herein may be a more palatable provider solution to an unsustainable cost curve than radical reimbursement reform.

Surprisingly, these findings are not revelatory, as related research into patterns of cancer care has resulted in similar observations. Morden et al found that hospice referral rates, hospitalizations, intensive care unit admissions, and chemotherapy use in the last weeks of life were remarkably uniform regardless of institution; profit versus nonprofit, academic versus community hospital, or small versus large facility.

 These direct observations imply that salaried physicians practice similarly to those reimbursed by fee-for-service methodology, despite vastly different economic incentives.

Observations from Morden and this study suggest that, contrary to prevailing dogma, medical oncology treatment selection and cancer care practice patterns may not be influenced by fee-for-service reimbursement as is often ascribed. Such observations warrant careful consideration as reimbursement methodology is modeled as part of healthcare reform. Research is ongoing to validate these observations and assess additional influences. Other value measures of our Oncology Medical Home that were believed to be both cost-saving and qualityenhancing, included physician commitment to an intensive continuous quality improvement initiative and their participation in an end-of-life care coordination program. The observations related to these components of this Oncology Medical Home pilot were separately reported at the American Society of Clinical Oncology Annual Meeting in 2013 and will be published in the near future.

 Cardinal Health Specialty Solutions, Dublin, OH (BF, SM, TO, JS); CareFirst BlueCross Blue Shield, Baltimore, MD (WW, DW); Greater Washington Oncology Associates, Rockville, MD (RT).

The Pathways Programs, data from which are used in this study, are a business line of Cardinal Health Specialty Solutions (CHSS). Several of the study authors are employed by and own stock in CHSS (BF, SM, TO, JS). The other authors (WW, DW, RT) report no relationship or financial interest with any entity that would pose a conflict of interest with the subject matter of this article.

MANAGERIAL

Shifting physician revenue from drug sales to cognitive services for mature pathways providers did not affect practice behavior in this payer-sponsored Oncology Medical Home.

Physician Behavior Impact When Revenue Shifted From Drugs to Services

Reliable measures of emergency department (ED) use are important for studying ED utilization and access to care. We assessed the association of emergent classification of an ED visit based on the New York University ED Algorithm (EDA) with hospital mortality and hospital admission.

Using diagnosis codes, we applied the EDA to classify ED visits into emergent, intermediate, and nonemergent categories and studied associations of emergent status with hospital mortality and hospital admissions.

 We used a nationally representative sample of patients with visits to hospital-based EDs from repeated cross sections of the National Hospital Ambulatory Medical Care Survey from 2006 to 2009. We performed survey-weighted logistic regression analyses, adjusting for year and patient demographic and socioeconomic characteristics, to estimate the association of emergent ED visits with the probability of hospital mortality or hospital admission.

 The EDA measure of emergent visits was significantly and positively associated with mortality (odds ratio [OR]: 3.79, 95% confidence interval [CI]: 2.50-5.75) and hospital admission (OR: 5.28, 95% CI, 4.93-5.66).

 This analysis assessed the NYU algorithm in measuring emergent and nonemergent ED use in the general population. Emergent classification based on the algorithm was strongly and significantly positively associated with hospitalization and death in a nationally representative population. The algorithm can be useful in studying ED utilization and evaluating policies that aim to change it.

The New York University Emergency Department (ED) Algorithm is a powerful tool that can be used to classify ED visits.

		Researchers are in need of methods to categorize ED visits as emergent and nonemergent.

		Public and private decision makers may use the algorithm to evaluate the impact of policies that alter ED utilization when information on diagnosis codes is available.

Emergency departments (EDs) are a critical component of the healthcare system, but face growing demand and often have inadequate capacity.

 Between 2000 and 2008, hospital-based ED visits grew by 15% to 124 million visits, while the total number of hospital-based EDs declined.

Most of this increase in ED visits can be attributed to increases in visits by publicly insured patients.

 Further, EDs often serve as primary care providers despite the fact that they are not optimally designed to provide this type of care.

 In response, payers and policy makers have attempted to deter nonemergent ED use and increase efficiency using financial incentives. For example, some health insurance plans and state Medicaid programs have implemented or increased ED copayments in recent years.

Efforts to improve access to primary care may also reduce nonemergent ED use. Researchers require reliable tools for classifying emergent and nonemergent ED use in order to evaluate policies and initiatives designed to improve access to primary care and reduce nonemergent use. Application of these methods to assess changes in access and reliance on EDs will become even more salient after implementation of the Patient Protection and Affordable Care Act in 2014, when much of the population will gain health insurance coverage and the number of publicly insured individuals will increase dramatically.

To this end, we assessed a tool for categorizing ED visits as emergent, by applying each visit to a nationally representative sample of ED visits and testing its strength as a predictor of hospital admission and mortality. Specifically, we used the New York University ED Algorithm (EDA). The EDA is based on a full chart review, using information on patient complaints, symptoms, vital signs, diagnoses, procedures, and ED resource use in order to determine the emergent nature of ED visits.

 The EDA could be a useful tool to evaluate ED use, but has only been tested in settings with more limited generalizability and has not previously been validated using a nationally representative sample of ED visits.

A previous study evaluated the EDA using data on commercial and Medicare patients from an integrated delivery system from 1999 to 2001, and found that patients classified as having an emergent visit under the EDA were more likely to be hospitalized within 1 day or to die within 30 days than were patients with visits classified as nonemergent.

Because the EDA may perform differently among groups of patients with limited access to care who are more likely to use the ED for nonemergent reasons, such as uninsured or Medicaid populations,

results from that study may not generalize to other populations. Recent research has also shown significant changes in the demographics of ED patients over the past decade, so previous results may be outdated.

This study improves on previous research in a number of ways. We applied the EDA to a national sample of hospital-based ED visits. We used the same criteria to assess the EDA as the previous study, but improve on it by using it in a nationally representative sample of all insurance groups in order to examine its generalizability to the US population, while previous research focused on privately insured and Medicare populations in a managed care context. Finally, we used several years of more recent data—from 2006 to 2009&mdash;while the previous study used data from 1999 to 2001. Similar to previous work, we examined the association between EDA categorization and hospital mortality or hospital admission, controlling for observed patient demographic and socioeconomic factors.

We used a nationally representative sample of visits to hospital-based EDs at noninstitutional, general, and short-stay hospitals (excluding Federal, military, and Veterans Administration hospitals) located in the 50 states and the District of Columbia, from repeated cross sections of the 2006 to 2009 National Hospital Ambulatory Medical Care Survey (NHAMCS). EDs within hospitals were randomly selected using a 4-stage probability design and surveyed over a randomly selected 4-week period each year. The sampling unit was the visit, with a target sample of 100 visits per ED sampled. Visit-level information was reported by hospital staff.

 Our sample included 117,477 ED visits with nonmissing primary diagnosis codes, representing a population of about 420 million. We used patient age, sex, and race, and socioeconomic indicators for the patient’s zip code, including median household income and urban-rural classification, as controls. We recoded the expected source of payment into private insurance, Medicare, Medicaid/State Children’s Health Insurance Program (SCHIP), self-pay/charity, and other insurance categories.

Classification of Emergent ED Visits Using the EDA 

We applied the EDA to the NHAMCS data to categorize ED visits as emergent or nonemergent. The EDA used the primary International Classification of Diseases, Ninth Revision diagnosis code at discharge for each visit and assigned the probability of the visit being: (1) nonemergent (NE); (2) emergent but treatable in a primary care setting (E-PCT); (3) emergent/ED care required, but preventable or avoidable if appropriate ambulatory care had been received (E-PA); and (4) emergent/ED care required and not preventable or avoidable (E-NPA). The EDA also separately assigned primary diagnosis codes to categories for injury, mental health, alcohol, and drug-related diagnoses. For about 16% of diagnosis codes, the EDA was not able to assign a visit to any classification category due to missing or uncommon primary diagnosis codes.

Following Ballard et al, we categorized ED visits into 3 levels: emergent, intermediate, and nonemergent.

We classified a visit as emergent if the sum of the probabilities of E-NPA and E-PA were greater than 50%. We classified a visit as nonemergent if the sum of the probabilities of NE and E-PCT were greater than 50%. Visits where the sum of the probabilities of E-NPA and E-PA or the sum of the probabilities of NE and E-PCT equaled exactly 50% were categorized as intermediate. We also conducted a sensitivity analysis of probability thresholds at 75% and 90%, following other work using the EDA.

To estimate the association of emergent and intermediate ED visit classification based on the EDA with mortality and hospitalization, we estimated logistic regression models, using NHAMCS complex survey weights. The 2 main outcome variables of interest were dichotomous indicators for whether an individual died (either on arrival to the ED, in the ED, or after being hospitalized) and whether the patient was admitted to the hospital from the ED. All models controlled for individual characteristics, including age, gender, race (white, Black, other race); payment source (private, Medicare, Medicaid/SCHIP, self-pay/charity, other); socioeconomic indicators at the patient zip code level, including low median household income and urban location; and survey year. The covariates in the model were selected because they are known to be associated with hospitalization and health outcomes.

 presents the breakdown of ED visits as classified by the EDA. We were able to classify 73,076 of 111,970 total ED visits into the emergent, intermediate, and nonemergent categories. This represented a population of about 260 million ED visits. The remaining ED visits (38,894) were classified into injury (32,149), drug (304), alcohol (1718), and mental health (4723). We were not able to classify about 16% of ED visits using the EDA, which is similar to the percentage of diagnosis codes that are uncommon and could not be classified as reported by the developers of the EDA.

 We limited our analysis sample to patients with diagnosis codes that could be classified into emergent, intermediate, and nonemergent. We hypothesized that patients with ED visits classified as emergent would have a higher probability of mortality and hospitalization. Like Ballard et al, we excluded visits categorized as drug, alcohol, or mental health, since they may have been emergent but were less clearly associated with mortality or hospitalization.

 In addition, we excluded visits categorized as injury since they may have been justifiably emergent, but had a low probability of mortality or hospitalization. Within the sample that could be classified into 1 of the 3 levels, 24.1% of visits were categorized as emergent, 1.9% as intermediate, and 74.0% as nonemergent, using the 50% probability threshold.

presents summary statistics for the analysis sample. Patients with ED visits in the sample had an average age of 37 years and were more likely to be female, white, have Medicaid or SCHIP as a primary payer, and live in an urban area. About 15% of patients were either admitted or died on arrival to the ED, in the ED, or in the hospital after an ED-associated visit.

Analysis of Emergent ED Visits Using the EDA 

 presents logistic regression results estimating the association between EDA visit classification and the outcome variables, mortality, and hospital admission. We found that emergent ED visit classification was positively and significantly associated with mortality (odds ratio [OR]: 3.79, 

 <.01) and with ED-related hospitalization (OR: 5.28, 

 <.01), relative to nonemergent classification. Intermediate classification was also positively associated with mortality (OR: 2.16, 

 = .12) and with ED-related hospitalization (OR: 4.14, 

 <.01), relative to nonemergent status. The results on the control variables were as expected. Older patients, males, and patients with public insurance or self-pay/charity as the primary payer (vs private insurance) were associated with significantly higher probabilities of ED-associated death and hospitalization. The findings regarding publicly insured and self-pay ED visits were consistent with other research that has shown these patient populations have poorer health and higher inpatient mortality than the privately insured.

We performed 2 specification checks to test the sensitivity of our results (full results available upon request). First, we estimated conditional logit models with year-specific ED fixed effects in order to control for any unobserved time-invariant ED-specific differences (such as staffing and resources) and to examine within-ED differences in hospitalization and mortality. Results from these models were very similar to our baseline models for both mortality (OR: 3.52, 95% confidence interval [CI]: 2.58-4.80) and hospital admission (OR: 5.22, 95% CI, 4.96-5.50). Second, following Ballard et al, we repeated our analyses using different probability thresholds (75% and 90%) for the ED measures in order to check the sensitivity of our results.

 In these models, visits were categorized as intermediate if the probability they were emergent was between 25% (10%) and 75% (90%). The results for our emergent variable represent the comparison of visits with a predicted probability of being emergent of 75% (90%) or greater with visits with a predicted probability of being emergent of 25% (10%) or less, thus we expect the ORs to increase, but be in the same direction as our baseline results. We found that this is generally the case, with ORs for our mortality outcome of 6.11 (95% CI, 3.73-10.02) and 4.90 (95% CI, 2.16-11.15) for the 75% and 90% cutoffs, respectively. Similarly, ORs for our admission outcome are 8.35 (95% CI, 7.60-9.17) and 11.26 (95% CI, 9.83-12.92) for the 75% and 90% cutoffs, respectively. All specification checks support our main finding that emergent visit status is significantly associated with hospital admission and death.

This analysis evaluated methods for classifying ED visits as emergent, using the EDA in a nationally representative sample of ED visits, and found that measures based on the EDA were significantly associated with mortality and EDassociated hospitalizations. The magnitude of association of emergent classification with hospitalization and mortality were similar to, but larger than, those from Ballard et al, who used data from privately insured individuals in a single integrated delivery system.

 The difference may be due to the fact that our sample included both the Medicaid and uninsured populations, and was not restricted to privately insured and Medicare patients in an integrated delivery system, who are likely to have better access to care. In addition, our sample excluded visits categorized as injury, which may also explain our larger result. Further, our results suggest that the EDA can be used to categorize visits as emergent and nonemergent by researchers when information is available on diagnosis codes but not on triage time (eg, claims data) or when a full chart review is not possible.

There were a few limitations to our approach. In categorizing visits using the EDA, there may have been some measurement error in determining if a visit is truly emergent. For example, some diagnoses may have been appropriately categorized as emergent, but were not associated with death or hospital admission (eg, a broken leg). We attempted to address this possible issue by excluding ED admissions for injury. Another limitation to our research was that we were only able to observe mortality and hospital admission as a direct admit from the ED, but not able to observe subsequent mortality or hospital admission within a limited time frame after the ED visit. Further, we used a blunt measure of hospital admission in order to directly compare our results with Ballard et al,

 which may not necessarily reflect severity in cases where hospitalizations resulted in shorter lengths of stay or were less acute. Finally, thresholds for hospital admission may have differed by payer type or other subjective patient characteristics. EDs may serve as a gateway through which to admit patients who do not have access to care through other channels due to their insurance status or other socioeconomic factors. This question is beyond the scope of this paper and requires further research.

This study demonstrated that the EDA can be used to identify ED visits associated with mortality and hospitalization. Classifying ED visits as emergent or nonemergent has been a shortcoming of the literature on ED use. The EDA is increasingly being used at the state and local levels

and, despite its limitations, the EDA has the potential to be a useful tool for understanding patterns of use and assessing the effects of policies and programs aimed at reducing nonemergent ED use. While the developers of the EDA have cautioned that the algorithm would not be appropriate to use for making individual reimbursementbased decisions, and recent research has supported this assessment,

 it can be applied to assess overall trends in ED use and to study how interventions and policies may affect these trends. For example, it has been used by researchers studying how new programs providing primary care for the uninsured affected ED use among these particular patient populations.

 In such contexts, where administrative data are available to assess how a program or policy change affected utilization, the EDA can be useful.

With the implementation of health reform, there will be major changes in the number of uninsured, the distribution of insurance coverage types, the payment and organization of healthcare providers, and other aspects of the healthcare system that could affect the way various patient populations utilize the ED. It will be important to have tools to classify ED visits and to test the success of interventions and policies designed to alter ED utilization and improve access to alternative sources of care. We have shown that the conclusions of earlier research validating the EDA in the context of managed care also hold when a nationally representative sample of ED visits is examined, suggesting that the EDA is a useful tool for health services and policy researchers.

 RTI International, Washington, DC (SOG); Department of Healthcare Policy and Research, Virginia Commonwealth University School of Medicine (LS).

The authors 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 (SOG, LS); acquisition of data (SOG, LS); analysis and interpretation of data (SOG, LS); drafting of the manuscript (SOG, LS); critical revision of the manuscript for important intellectual content (SOG, LS); statistical analysis (SOG, LS).

 Sabina Ohri Gandhi, PhD, RTI International, 701 13th St NW, Ste 750, Washington, DC 20005-3967. E-mail: sgandhi@rti.org.1. Institute of Medicine. Hospital-Based Emergency Care: At the Breaking Point. Washington, DC: National Academies Press; 2007.

2. National Center for Health Statistics; US Department of Health and Human Services. Health, United States, 2010: with special feature on death and dying. http://www.cdc.gov/nchs/data/hus/hus10.pdf#089. Published 2011. Accessed February 21, 2012.

3. Hsia RY, Kellerman AL, Shen YC. Factors associated with closures of emergency departments in the United States. JAMA. 2011;305(19): 1978-1985.

4. Tang N, Stein J, Hsia RY, Maselli JH, Gonzales RJ. Trends and characteristics of US emergency department visits, 1997-2007. JAMA. 2010;304(6):664-670.

5. Hsu J, Price M, Brand R, et al.Cost-sharing for emergency care and unfavorable clinical events: findings from the safety and financial ramifications of ED copayments study. Health Serv Res. 2006;41(5): 1801-1820.

6. Mortensen K. Copayments did not reduce Medicaid enrollees’ nonemergency use of emergency departments. Health Aff (Millwood). 2010;29(9):1643-1650.

7. Congressional Budget Office. Selected CBO publications reltaed to healthcare legislation, 2009-2010. Washington, DC: Congress of the United States; 2010.

8. Billings J, Parikh N, Mijanovich T. Emergency room use: the New York story. Commonwealth Fund issue brief 2000. http://wagner.nyu .edu/chpsr/index.html?p=25. Published 2000. Accessed October 31, 2011.

9. Ballard DW, Price M, Fung V, et al. Validation of an algorithm for categorizing the severity of hospital emergency department visits. Med Care. 2010;48(1):58-63.

10. Cunningham PJ, May J. Insured Americans drive surge in emergency department visits: issue brief for the Center for Studying Health System Change. 2003;(70):1-6.

11. Zuckerman S, Shen YC. Characteristics of occasional and frequent emergency department users: do insurance coverage and access to care matter? Med Care. 2004;42(2):176-182.

12. Inter-University Consortium for Political and Social Research. National Hospital Ambulatory Medical Care Survey series. http://www .icpsr.umich.edu/icpsrweb/ICPSR/series/42. Published 2011. Accessed February 21, 2012.

POLICY

This article assesses a classification tool for categorizing emergency department visits as emergent and nonemergent.

Emergency Department Visit Classification Using the NYU Algorithm

 Unhealthy behaviors are responsible for most chronic disease, ample healthcare costs, and around 40% of deaths. This study assessed patient satisfaction and behavioral outcomes for a program that enables healthcare settings to deliver evidence-based, cost-saving behavioral screening and intervention (BSI) services, as recommended by the US Preventive Services Task Force and others.

Thirty-one diverse primary care clinics and 2 other sites participated. Over 110,000 patients completed multibehavioral screening questionnaires. Those with positive alcohol or drug screens met with dedicated, on-site health educators for further assessment, and then received either reinforcement, brief intervention, or referral. Over 300 patients completed satisfaction questionnaires during year 3 of the program. A pseudo-randomly selected group of 675 patients participated in a 6-month follow-up telephone interview. In addition, for a short time in a pilot project at 3 clinics, 29 patients received depression screening, collaborative care, and behavioral activation, and 22 completed a 3-month follow-up telephone interview.

Mean patient satisfaction scores for all services exceeded 4.2 on a 5-point scale. Over 6 months, binge drinking episodes declined by over 20% for most subgroups. Recent marijuana use decreased by 15%. Depression symptom scores decreased by 55%.

With intensive training and ongoing support, cost-efficient paraprofessionals can deliver effective alcohol, drug, and depression screening and intervention services in busy healthcare settings. The approach holds promise for systematically addressing on a population-wide basis a variety of important behavioral health determinants and reducing related healthcare costs.

This article describes a program that successfully delivers evidence-based, costsaving alcohol, drug, and depression screening and intervention services in primary care settings.

Ample research has documented the effectiveness of these services and their return on investment, yet they are seldom delivered, largely because of provider and staff time pressures.

Trained and supported, cost-efficient paraprofessionals can deliver these services, elicit high patient satisfaction, and attain substantial declines in risky drinking, illicit drug use, and depressive symptoms.

Over time, clinical settings were able to modify work flow so that most patients received recommended services.

Expanding primary care teams with trained and supported paraprofessionals enables systematic delivery of widely recommended, evidence-based, cost-saving alcohol, drug, and depression screening and intervention services.

April 2014

A Novel Approach to Minimizing Adverse Incentivization in Healthcare Payment Systems

How Does Drug Coverage Vary by Insurance Type? Analysis of Drug Formularies in the United States

Physician Behavior Impact When Revenue Shifted From Drugs to Services

Emergency Department Visit Classification Using the NYU Algorithm

A Team Approach to Systematic Behavioral Screening and Intervention

Hepatitis B Virus Infection Testing and Prevalence Among Asian and Pacific Islanders

Adherence to Statins and LDL-Cholesterol Goal Attainment

Are Obese Patients Assisted in Losing Weight?

Toward a Better Understanding of Patient-Reported Outcomes in Clinical Practice

Consumer Perspectives on Personal Health Records: A 4-Community Study