This study assessed the association of the Medicare Part D coverage gap with medication adherence among beneficiaries with chronic obstructive pulmonary disease (COPD). 

 Retrospective observational study based on Medicare claims data.

A 5% random sample of Medicare claims data (2006-2010) was used in this study. Beneficiaries diagnosed with COPD and treated with long-acting bronchodilators (LABDs) were assigned to an exposure cohort (at risk of the coverage gap) or a control cohort (otherwise). The exposure and control cohorts were matched using high-dimensional propensity scores. Adherence was defined as ≥80% of the proportion of days covered by LABDs. Logistic regressions controlling for unbalanced covariates post matching were applied to assess the association of the coverage gap with adherence.

 The final matched exposure and control cohorts each included 4147 patient-year observations with about 42% and 46% of them adherent to LABDs, respectively. About 17% of the exposure cohort hit the coverage gap after October 31. Logistic regression showed that, compared with the control cohort, the beneficiaries in the exposure cohort had a significantly lower likelihood of being adherent if they hit the coverage gap later in the year (odds ratio [OR], 0.603; 95% CI, 0.493-0.738), or had a lower likelihood without statistical significance if otherwise (OR, 0.931; 95% CI, 0.846-1.024).

The findings suggest that the Part D coverage gap was associated with lower adherence in patients with COPD, which may serve as evidentiary support for phasing out the

		Medication adherence, in general, is not optimal in beneficiaries with chronic obstructive pulmonary disease (COPD). 

		The Medicare Part D coverage gap is associated with lower adherence to COPD medication, with greater negative impact among the beneficiaries who reach the coverage gap later in the year. 

		The timing of reaching the Part D coverage gap has effects on the level of medication adherence.

Chronic obstructive pulmonary disease (COPD) is characterized by progressive airway narrowing or airflow obstruction, that causes breathing difficulties, reduced exercise capacity, and physical limitation.

 Chronic lower respiratory disease, which primarily includes COPD, has become the third leading cause of death in the United States.

 COPD occurs more often in females and in the elderly, with almost half of all patients with COPD being 65 years or older.

 Pharmacotherapy is a cornerstone of COPD management, and maintenance medications are effective in controlling symptoms, maintaining lung function, and preventing COPD exacerbations.

Previously, access to appropriate pharmacologic therapies to manage COPD was a challenge for Medicare beneficiaries due to the lack of coverage for prescription drugs. However, Part D, the Medicare prescription drug benefit, went into effect on January 1, 2006, and greatly expanded access to pharmacological therapies by subsidizing the cost of prescription drugs and prescription drug insurance premiums for Medicare beneficiaries. Pharmacologic treatments for COPD, such as bronchodilators and inhaled steroids, are now covered by Medicare Part D, with patients responsible for deductibles and co-payments. However, as per policies designed to keep the program financially sustainable, Part D does include a coverage gap for beneficiaries. Specifically, beneficiaries are financially responsible for the full cost of prescriptions once a certain dollar threshold has been reached, up to a maximum amount when catastrophic coverage begins, and then the Part D plan assumes 95% of the cost of prescriptions. These thresholds are established each year by Medicare at CMS.

 approximately 1.5 million and more than 3 million beneficiaries, respectively, were estimated to reach the coverage gap. Multiple studies have assessed the effect of the coverage gap on medication adherence in elderly patients

 and in those with different chronic diseases.

 These studies have shown that the use of brand name medications significantly decreased for most conditions during the coverage gap and substituted by generic drugs. However, there were no generic versions of long-acting bronchodilators (LABDs) available as a long-term maintenance therapy for COPD in the United States. No published studies have evaluated the effect of the Part D coverage gap on medication utilization among Medicare beneficiaries with COPD or explored whether patients in the coverage gap may behave differently due to the lack of generic options. This omission is problematic given the high prevalence of COPD among the elderly

 and the negative health and economic consequences of nonadherence.

 Therefore, the objective of this study was to assess the association between the Part D coverage gap among Medicare beneficiaries diagnosed with COPD and their adherence to COPD maintenance medications, LABDs.

A 5% random sample of Medicare beneficiaries was used for this study. The Medicare administrative claims database is a comprehensive data source covering all beneficiaries who were enrolled in Medicare, capturing information on demographic characteristics, enrollment, prescription drug events, medical encounters in inpatient and outpatient settings, and health services incurred in other facilities, such as hospice or skilled nursing homes. This study was approved by the institutional review board and by the CMS Privacy Board.

 Because the coverage gap thresholds varied by calendar year, patient selection and outcome measures were employed at a calendar-year level. Considering that many Medicare beneficiaries did not have a full-year benefit in 2006, only data files from 2007 to 2010 were used for analysis, and the 2006 data file was used to describe patient baseline characteristics. Beneficiaries who met all of the following inclusion criteria were selected to form a general patient pool: a) had “of age” listed as the reason for Medicare eligibility (ie, age is ≥65 years as of 6 months prior to January 1 of a calendar year), b) had a full year’s eligibility during a respective calendar year and 6 months of eligibility prior to January 1 of the respective calendar year (the 6 months were defined as baseline period), c) had at least 2 outpatient claims with a diagnosis of COPD on different dates or at least 1 emergency department (ED) or inpatient claim with COPD as the primary diagnosis during a respective calendar year, and d) had at least 2 prescriptions for LABDs filled on different dates during a respective calendar year (LABDs are listed in eAppendix Table 1 [

Beneficiaries were excluded from the study if they met at least 1 of the following criteria: a) were enrolled with a Medicare Advantage plan in any month during a respective calendar year; b) had a diagnosis of asthma during a respective calendar year, because some of the LABD medications are also indicated for asthma; c) had a diagnosis of cancer during a respective calendar year, because patients with cancer likely have different medication utilization and spending patterns compared with other Medicare beneficiaries; or d) had a disability or end-stage renal disease during a respective calendar year, because the benefits of such patients can differ substantially from those of other Medicare beneficiaries.

Beneficiaries who met the above selection criteria were divided into 2 study cohorts.

 Beneficiaries were assigned to the control cohort if they fell into 1 of the following categories: a) had Medicare-Medicaid dual eligibility for the whole year, b) qualified for Part D low-income subsidies (LIS) (ie, received LIS for at least 1 month before and after they entered the coverage gap), or c) had additional benefits covering brand and generic drugs during the gap.

Exposure cohort and subgroups in exposure cohort. 

If beneficiaries did not have dual eligibility or low-income subsidies or full benefits to help with the coverage gap during a calendar year, they were assigned to the exposure cohort.

The exposure cohort was further categorized into 4 subgroups. Beneficiaries who did not reach the coverage gap in a respective year were identified as “no-reaching-gap subgroup.” This subgroup was not included in the final exposure cohort based on the assumption that beneficiaries who were relatively healthy were much less likely to reach the coverage gap; therefore, their medication-taking behavior was not expected to noticeably change as a result of presence of the coverage gap. Beneficiaries who reached the coverage gap before March 1 were identified as the “early-gap subgroup.” Similar to the “no-reaching-gap subgroup,” the early-gap subgroup was not included in the final exposure cohort based on the assumption that beneficiaries who reached the coverage gap early may have been very sick and wanted to maximize their medication usage during the gap period to enter the catastrophic phase sooner. This group of patients was anticipated to be small and to respond to the coverage gap differently than other subgroups.

Beneficiaries who reached the coverage gap between March 1 and October 31 were identified as the “mid-gap subgroup.” Finally, beneficiaries who reached the coverage gap on or after November 1 were identified as the “late-gap subgroup.” Thus, the final exposure cohort only included the mid-gap and the late-gap subgroups. 

 depicts the subgroup designation within the exposure cohort.

Adherence was measured as the proportion of days covered (PDC). Yearly PDC was defined as the proportion of days covered by LABDs relative to the treatment period during a calendar year. The treatment period was calculated as the duration from the fill date of the first LABD prescription until the end of the year. A dichotomous variable for adherence was constructed as 1 if PDC ≥80%, and 0 otherwise.

The independent variable of interest was a dichotomous indicator of membership in the exposure or the control cohort (1 = exposure cohort, 0 = control cohort). In addition, the following demographic and clinical variables in the baseline period were also assessed: age, gender, ethnicity, residence region, Charlson Comorbidity Index (CCI) score (eAppendix Table 2), presence of major comorbidities (eg, diabetes, heart disease) (specified in eAppendix Table 3), number of unique prescription drugs defined by the first 9 digits of the national drug codes, number of all-cause ED visits, number of all-cause inpatient visits, previous COPD diagnosis, previous use of LABDs, previous supplemental oxygen therapy (eAppendix Table 4), and previous use of oral corticosteroids (eAppendix Table 5).

High-Dimensional Propensity Score Matching

In observational studies, selection bias is an important issue when comparing groups and exposure is not randomly assigned. Propensity score matching (PSM) is commonly used to generate comparable exposure and control cohorts with balanced demographic and clinical characteristics. In a traditional approach to generating propensity scores, a number of relevant confounders or covariates included in a logistic model are defined based on available data, and they are primarily guided by knowledge related to exposure and the study population characteristics. In a typical claims database, important attributes are unavailable (eg, laboratory results, functional status, smoking status, over-the-counter medication); therefore, empirically identifying appropriate proxies for patient health status out of a large number of variables in claims data is a significant challenge. High-dimensional propensity score (HDPS) analysis is an automated algorithm developed by Schneeweiss and colleagues (2009) to set up proxies by assessing diagnosis codes, procedure codes, and prescribed medication codes,

 which helps to overcome the aforementioned challenges in the process of reducing selection bias and controlling for confounding effects.

HDPS analysis was employed in this study with the diagnosis code and the procedure codes in outpatient, ED, and inpatient settings specified as data dimensions. In each data dimension, the 200 most prevalent codes were used, and then the possible amount of confounding was calculated for each variable based on a multiplicative model to sort all variables in a descending order. The top 300 variables were selected to construct a logistic model and generate the propensity score.

 Lastly, PSM was conducted at 1:1 between the exposure cohort and the control cohort using the Greedy 5-to-1 digit technique.

 tests were used to detect differences in patient characteristics between the exposure and the control cohorts for continuous variables (eg, age, CCI score), and the 

 test was used for categorical variables, including demographics (eg, gender, ethnicity) and comorbidities (eg, diabetes, hypertensive disease). After matching, McNemar’s tests were used for categorical variables and the paired 

After matching, a conditional logistic regression model was constructed with adherence (1 if PDC ≥80%, 0 if PDC <80%) as the dependent variable. More than 50% of the beneficiaries had repeated observations for 2 or more years from 2007 to 2010. Therefore, a generalized estimating equation technique was applied in the multivariable models to correct for the correlation between repeated observations of a patient.

 All analyses were performed using SAS version 9.2 (SAS Institute Inc, Cary, North Carolina).

 values less than .05 were considered to be statistically significant.

Application of the patient selection criteria resulted in 5366, 5650, 5991, and 6268 unique beneficiaries diagnosed with COPD and treated with LABDs for the years 2007 to 2010, respectively (

 depicts the patient selection flow and the sample size at different steps. Each year, nearly 20% of those beneficiaries enrolled with Part D benefit were not subject to the coverage gap (ie, assigned into the control cohort), and the remaining beneficiaries were at risk of the coverage gap (ie, assigned into the overall exposure cohort).

From 2007 to 2010, respectively, the final control cohort contained 1011, 1012, 1145, and 1176 beneficiaries, and the final exposure cohort (the mid-gap + the late-gap subgroups) contained 2786, 2746, 2721, and 2751 beneficiaries. Combined across all years, there were 4344 patient-year observations in the control cohort and 11,004 patient-year observations in the exposure cohort before implementation of PSM. After the 1:1 matching, both cohorts included 4147 patient-year observations, which was the final sample for analysis.

Overall, the mean age of the patients was 77.4 years (standard deviation [SD] = 7.6), the majority (71%) was female, and more than 90% were Caucasians (

). Beneficiaries were heavily concentrated in the South, while beneficiaries in the West were underrepresented in the study. The mean CCI score was approximately 2.2. The most common comorbidity in the baseline period was hypertension (>65%), followed by heart disease (>50%) and hyperlipidemia (>45%). More than 70% of the patients had used a LABD, and a large proportion of them received oxygen therapy or oral corticosteroids (about 30%) in the baseline period. Beneficiaries had substantial medication burden in the baseline period, with an average of more than 10 different classes of medications.

Prior to matching, the control and the exposure cohorts were significantly different in almost all of the demographic and baseline characteristics, except for several baseline comorbidities. After matching, the cohorts were generally balanced in demographic and baseline characteristics, with statistical differences observed only for the prevalence of several baseline comorbidities. The standardized differences were generally small, with most of the absolute values less than 10% (except several baseline comorbidity variables), indicative of acceptable balance between the matched cohorts.

After cohort matching, the mean annual PDC in the matched control cohort was 0.70 (SD = 0.25); it was 0.69 (SD = 0.24) in the matched exposure cohort. About 46% of the matched control cohort was adherent versus 42% for the matched exposure cohort.

Unadjusted results showed the matched exposure cohort had lower adherence rates than the matched control cohort. After adjusting for age, gender, and the unbalanced covariates, beneficiaries who reached the coverage gap had lower odds of adherence compared with beneficiaries who were not exposed to the coverage gap. Specifically, beneficiaries in the late-gap subgroup had nearly 40% lower odds of adherence than beneficiaries in the control cohort (OR, 0.603; 95% CI, 0.493-0.738). Beneficiaries in the mid-gap subgroup also had lower odds of adherence, although this relationship was not statistically significant (OR, 0.931; 95% CI, 0.846-1.024). In addition, hyperlipidemia, depression, and diseases of the musculoskeletal system and connective tissues were found to be associated with lower likelihood of adherence (

The study results suggest that reaching the Part D coverage gap may be negatively associated with medication adherence among Medicare patients with COPD, and the association was stronger among the beneficiaries who reach the coverage gap later (ie, on or after November 1). One explanation for these findings may be that when Medicare beneficiaries enter the coverage gap, they bear a higher economic burden to obtain their medications and they may be more likely to choose nonadherence to more expensive brand name drugs. In addition, we assessed patients’ use of short-acting bronchodilators (SABDs) before and after they hit the coverage gap and did not observe a remarkable shift (results not reported here). One possible reason is that SABDs are a class of medications that are usually used to relieve acute symptoms, not a substitute for long-term maintenance therapy.

The results are consistent with previous studies finding that the Part D coverage gap is associated with reduced medication adherence.

 For example, Fung et al (2010) found that the odds of adherence among patients with diabetes with the Part D coverage gap decreased by 17% compared with those patients without the Part D coverage gap.

 Likewise, Stuart and colleagues (2013) found that the PDC was 7.8% lower for statins, 7.0% lower for clopidogrel, or 5.9% lower for beta-blockers for beneficiaries exposed to the coverage gap compared with those not exposed.

Articles

WEB EXCLUSIVE

Using longitudinal Medicare claims data, this study quantified the association of the Medicare Part D coverage gap with medication adherence among beneficiaries with chronic obstructive pulmonary disease.

Association of Part D Coverage Gap With COPD Medication Adherence

Although team-based care can improve coronary heart disease (CHD) risk factors and is considered cost-effective from a healthcare system perspective, little is known about the financial impact of team-based primary care for secondary prevention of CHD. The purpose of this study was to define the impact of team-based care for CHD on utilization, costs, and revenue of a private primary care practice.

 Between March 1, 2010, and March 31, 2013, we assisted a private medical practice, comprising 5 primary care clinic sites, to organize and deliver team-based care for patients with CHD. We used billing records and the registered nurse care manager’s diary to calculate the cost of team-based care, differences in the average number of visits per patient, and revenue per patient before and after the implementation of team-based care.

The net cost of team-based primary care was $291 per patient over the 1-year period of observation.

The findings from this study are consistent with other economic analyses of team-based care and suggest that payment for care must be restructured if patients are expected to enjoy the benefits of team-based primary care.

		The Community Preventive Services Task Force has concluded that although it generates costs, team-based care is both effective and cost-effective in controlling coronary heart disease (CHD) risk factors. 

		We also found that team-based care has a negative financial impact on primary care under a fee-for-service payment system. 

		However, team-based care for patients with CHD has the potential to be cost-saving for accountable care organizations if team-based care reduces overall costs by as little as 2%.

Minimizing levels of risk factors for individuals who have coronary heart disease (CHD) could prevent or postpone 71% of deaths among Americans with heart disease who are aged 30 to 84 years,

 and a team-based approach has been shown to improve control of blood pressure and serum cholesterol levels.

 Although the Community Preventive Services Task Force has concluded that a team-based approach to blood pressure control is cost-effective, they have also concluded that it generates a net cost.

We hypothesized that, because patients with CHD are at a much higher risk of another cardiac event in the near term, it might be possible to implement a team-based care system that would be revenue-neutral or revenue-positive for primary care. In this report, we describe the results of our attempt to design and implement such a system in partnership with a private practice that comprised 5 clinic sites.

The HealthPartners Institute Institutional Review Board approved the study as protocol #09-132.

The 5 clinic sites that participated in the trial were located 30 to 50 miles from a large Midwestern urban center and cared both for patients who worked for large companies in the metropolitan area and patients who were self-employed in farming or worked in light manufacturing in small towns. Nearly all of the patients served by these clinics identify as non-Hispanic white and have some sort of insurance.

We based our intervention on the principles of complex adaptive systems and diffusion of innovation

 and were able to improve low-density lipoprotein cholesterol (LDL-C) control and increase the rate at which taking aspirin was documented in the medical record.

 We also documented the activities that the clinics undertook to implement and maintain their care systems.

Our only firm requests to the participating clinics were that they use the grant resources to improve their Minnesota Health Scores vascular disease quality scores and hire a registered nurse (RN) care manager to coordinate their team-based care program. Although the National Institutes of Health grant paid her salary, the RN care manager was a clinic employee who reported to one of the clinical services managers and a lead primary care physician; she did not report to the investigators. The clinics were otherwise free to use existing care processes or develop new ones as they chose. The design of the team-based care system and the tasks undertaken by the staff are described more fully in previous publications.

We identified 663 patients who were treated for CHD—but may also have had diabetes&mdash;from March 1, 2009, to February 28, 2010; this was the pre-intervention (baseline) period during which there was no assistance. We also identified a reference population of 1643 patients without CHD who were treated for diabetes during the same period. We collected baseline data from randomly selected samples of these 2 groups of patients (n = 551 for the sample with CHD and n = 485 for the sample with diabetes). The implementation period was 18 months in each clinic site. During the postimplementation evaluation period of December 1, 2011, to March 31, 2013, the clinics treated 1012 patients for CHD and treated 1828 patients without CHD for diabetes. We collected postimplementation data on independently identified random samples of these patients (n = 529 for patients with CHD and n = 509 for patients with diabetes). The postimplementation period was longer than 1 year because implementation was staggered in the 5 clinic sites.

To capture the tasks that she addressed, the RN care manager kept a diary of her activities on randomly selected days of each month during the intervention. She recorded both the type of activity and with whom, if anyone, she was meeting.

We calculated the costs of team-based care by applying an estimate of employee compensation to employee hours. We estimated employee hours from study time logs, and we estimated employee salaries from national median hourly earnings, reported in Occupational Employment Statistics by the US Bureau of Labor Statistics for occupational codes 11-1000, 11-3030, 11-3120, 13-000, 13-2000, and 29-1141 for administration; codes 11-3021 and 15-1120 for information technology; and 29-1141 and 29-9090 for patient care.

 We used average benefits rates from the National Compensation Survey calculator to estimate the benefit rate associated with the paid wages.

 We excluded from our estimates all time spent on research evaluation activities and initial planning and implementation.

We calculated revenue to primary care during the postimplementation period using billing data for the 534 patients in the sample for whom they were available; data were not available for 17 patients. Revenue included total payments received by the medical group from both insurance plans and patients, but excluded unpaid bills. To better understand the source of revenue change, we also estimated average revenue per visit, average revenue per procedure, the number of office visits for any reason, the number of visits with a care provider associated with an evaluation and management (E&M) code, the average number of total procedures, and the average number of laboratory tests per patient.

We used a generalized linear model with a gamma distribution with a log-link function to estimate revenue measures and a negative binomial distribution and a log-link function for count data. We estimated the statistical model on the full data set of pre- and postintervention observations for both the 2 samples of patients with CHD and the 2 reference samples of patients with diabetes using a difference-in-difference variable specification. We applied the results to the pre-intervention CHD sample in order to predict what their costs would be in the postintervention period. Doing so allowed us to control for certain population characteristics (age, sex, insurance status) while comparing paid amounts pre- and post intervention.

The pre-implementation CHD random sample comprised 635 patients and the postimplementation sample comprised 534 patients (

). Fewer than 5% of patients were insured by Medicaid or lacked insurance altogether. Although we do not have data on supplemental insurance for patients who were insured by Medicare, national surveys suggest that about 90% would have had supplemental insurance.

 The demographic characteristics of the diabetes samples used to statistically model the financial impact of the team-based care program are presented in eAppendix Table A (

During the 1-year intervention period, the staff spent 6856 hours providing team-based care to patients with CHD; 91% of these hours were devoted to direct patient care activities (

). Applied to the 1012 patients treated for CHD, the cost is $291 per patient.

When her time allocation was defined by activity, the RN care manager spent 25% of her time updating the registry; 25% charting; and 18% attending, preparing for, and traveling to planning and patient care meetings (eAppendix Table B). The other activities that required her attention—discussing patient care, placing orders, requesting records, reviewing charts, and activities otherwise undefined&mdash;each consumed less that 10% of her time.

When defined by interaction, the RN spent 38% of her time in activities conducted alone; further, she spent an approximately equal proportion of time with care coordinators (17%) and patients (16%), and she spent 12% of her time with healthcare providers (physicians and other licensed providers), as well as 3% of her time with clinic support staff. She spent 14% of her time with individuals not otherwise described.

Pre-implementation, the average number of total visits per patient was 9.2, the average number of visits with a provider (ie, a visit associated with an E&M code) was 4.4, and the average revenue per patient was $878 (eAppendix Table C). Post implementation, the average number of total visits declined 5% to 8.7, the average number of visits with a provider declined 8% to 4.0, and the average revenue per patient declined 11% to $779. In addition, the average revenue per office visit declined 4%, from $109 to $104; the average number of procedures declined 6%, from 21.8 to 20.6; the average revenue per procedure declined 3%, from $44 to $43; and the average number of laboratory tests per patient declined 17%, from 9.7 to 8.1.

Based on statistical modeling with comparison to patients with diabetes, team-based care for patients with CHD would be expected to produce a 5.7% increase in revenue per total visits and a 1.4% increase in revenue per procedure. However, team-based care would also be expected to produce a decline in total visits (—4.3%), visits with a provider (–1.4%), total procedures (–2.8%), and laboratory tests (–5.6%). The net result would be a 2.5% decline in revenue per patient (eAppendix Table D).

In this trial of team-based primary care in 5 clinic sites, we were able to demonstrate an improvement in LDL-C control and an increase in the rate at which aspirin use was documented.

 However, we were unable to create a system that would be revenue-neutral or revenue-positive for the primary care practice that is providing the service. Using a pair of diabetes patient samples as a comparator, statistical modeling indicated that although the revenue per visit increased (5.7%), the net effect of the team-based care system was to reduce the revenues earned per patient (—2.5%) because the number of visits was reduced (–4.3%).

The annual cost of $291 per patient is very close to the median of 20 studies of team-based care for hypertension control considered by the Community Guide (median intervention cost per patient per year [PPPY], $284; interquartile interval [IQI], $153-$670)

 and a randomized trial to improve care for patients with hypertension, diabetes, or heart failure by adding laypersons to the care team ($286 PPPY).

Based on the 20 studies it analyzed, the Community Guide estimated that team-based care for hypertension generates an additional $65 (IQI, —$235 to $318) PPPY for outpatient and inpatient care, emergency department visits, and medications.

 If we assume that this same cost applies to the patients in our study, the annual total cost of team-based care (intervention costs + healthcare costs) for patients in this study would be $356 ($291 + $65).

However, patients who have CHD have a near-term risk of an event that is much higher than the risk for patients who have hypertension alone. This offers a number of near-term cost-reduction opportunities that are not available when caring for patients with hypertension. Total annual healthcare costs for patients with CHD have been reported to be $18,953 in 2008 dollars

—$6000 greater than the costs for patients who simply have hypertension.

 The cost of implementing team-based care that we computed earlier, $356 PPPY, is 1.9% of this total cost. It is indeed possible that team-based primary care could reduce costs for an accountable care organization to this extent by reducing acute and emergent events and low-value and no-value care.

 These savings could then be shared with primary care providers; however, until savings are shared, the additional cost of team-based care will be a barrier to the adoption of the strategy by primary care practices. 

Our study has a number of important limitations. Some might consider a design that compared the intervention clinics with a set of parallel reference clinics to be a stronger design, but we were concerned that with such a design, unrecognized differences between patients in the intervention and control clinics might confound the results. Another weakness is the fact that the study was conducted in a single primary care practice with only 5 clinic sites and, because the team-based care model implemented to manage patients with CHD was new to the practice, it is possible that costs might decline as efficiencies developed over time.  The trial is also limited by the short time horizon.

The average annual cost of implementing team-based care in our trial for a patient with CHD could be recovered by an accountable care organization if team-based care reduced the total cost of treating a patient who has CHD by as little as 2%. In the current payment system, however, all of the costs, and none of the savings from team-based care, are borne by primary care. This suggests that primary care practices will not adopt team-based care and their patients will not experience the benefits until new payment models are developed and implemented.

 HealthPartners Institute (TEK, MVM, JAH, EDP, JOT), Bloomington, MN; Hudson Physicians (PM, JMN), Hudson, WI.

 NIH grant R18HL096563 was the sole source of funding for the study.

 Dr McGinnis was compensated for his time through the grant for this study. The remaining 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 (TEK, MVM, JAH, PM, JMN); acquisition of data (TEK, PM, JMN, EDP, JOT); analysis and interpretation of data (TEK, MVM, PM, JMN, EDP, JOT); drafting of the manuscript (TEK, MVM, EDP); critical revision of the manuscript for important intellectual content (JAH, TEK, MVM, PM, EDP); statistical analysis (MVM, EDP); obtaining funding (JAH, TEK); administrative, technical, or logistic support (JAH, JOT); and supervision (TEK).

Although team-based care improved cardiovascular disease risk factors, it had a negative financial impact on a primary care practice.

The Financial Impact of Team-Based Care on Primary Care

 To determine the opinions of US hospital leadership on the Hospital Readmissions Reduction Program (HRRP), a national mandatory penalty-for-performance program.

 We developed a survey about federal readmission policies. We used a stratified sampling design to oversample hospitals in the highest and lowest quintile of performance on readmissions, and hospitals serving a high proportion of minority patients. 

 We surveyed leadership at 1600 US acute care hospitals that were subject to the HRRP, and achieved a 62% response rate. Results were stratified by the size of the HRRP penalty that hospitals received in 2013, and adjusted for nonresponse and sampling strategy.

 Compared with 36.1% for public reporting of readmission rates and 23.7% for public reporting of discharge processes, 65.8% of respondents reported that the HRRP had a “great impact” on efforts to reduce readmissions. The most common critique of the HRRP penalty was that it did not adequately account for differences in socioeconomic status between hospitals (75.8% “agree” or “agree strongly”); other concerns included that the penalties were “much too large” (67.7%), and hospitals’ inability to impact patient adherence (64.1%). These sentiments were each more common in leaders of hospitals with higher HRRP penalties.

 The HRRP has had a major impact on hospital leaders’ efforts to reduce readmission rates, which has implications for the design of future quality improvement programs. However, leaders are concerned about the size of the penalties, lack of adjustment for socioeconomic and clinical factors, and hospitals’ inability to impact patient adherence and postacute care. These concerns may have implications as policy makers consider changes to the HRRP, as well as to other Medicare value-based payment programs that contain similar readmission metrics.

	Despite the fact that the Hospital Readmissions Reduction Program (HRRP) has increased efforts to reduce readmissions, hospital leaders identified important issues with the program. Our findings from a national survey of hospital leaders indicate that: 

		Leaders are concerned about the size of the penalties and the lack of adjustment for socioeconomic and clinical factors. 

		Currently, the HRRP remains a lower priority for leaders than other areas of quality improvement, such as patient safety and adherence to guidelines. 

		Federal policy makers may want to address these issues as they consider future changes to the program and seek to maximize its impact.

Reducing hospital readmissions has the potential to simultaneously improve patient outcomes and reduce healthcare spending and, as such, has become a major target for US policy makers. In an effort to spur a reduction in readmissions, Medicare began publicly reporting on hospitals’ discharge planning in 2007 and, in 2009, added public reporting on readmission rates for acute myocardial infarction (AMI), heart failure (HF), and pneumonia. Despite these efforts, 30-day readmission rates remained stable near 20% during this time frame.

 Consequently, with the passage of the Affordable Care Act in 2010, Congress included legislation establishing the Hospital Readmissions Reduction Program (HRRP).

 Under the HRRP, CMS penalizes hospitals with higher than expected readmission rates for Medicare patients; it has been in effect since the beginning of fiscal year (FY) 2013.

 In the HRRP’s third year, hospitals performing poorly may lose up to 3% of their base Medicare diagnosis-related group (DRG) payments—a substantial amount given that many hospitals have negative Medicare inpatient margins at baseline.

However, the HRRP has been controversial. Initial reports suggested that the program was more likely to penalize large, teaching, and safety net hospitals.

 Multiple organizations have argued that the program’s methodology should take sociodemographic factors into account and exclude readmissions unrelated to the initial reason for hospitalization,

 and at least 2 bills have been proposed in Congress to address these concerns and others.

 On the other hand, early data show that readmission rates have fallen by 1% to 2% since the implementation of the HRRP, suggesting that this program may have had a positive impact on this outcome, although causality cannot be established.

The HRRP is one of a number of value-based payment models within Medicare, and the US Secretary of HHS recently announced a goal to have 85% of Medicare fee-for-service payments tied to quality or value by 2016.

 Many of these new payment programs are closely related to the HRRP; for example, the forthcoming Skilled Nursing Facility Value-Based Payment program is similarly based on a single readmission measure: 30-day readmission following a hospitalization.

 Readmissions metrics similar to the one used in the HRRP are also now included in quality measures for the Medicare Shared Savings Program

 and the Physician Value-Based Modifier Program,

 and will be included in payment programs in additional settings, such as dialysis facilities, in future years.

Given the importance of the HRRP as a model for future value-based payment programs, its controversy, and its initial success, it is crucial to understand how hospital leaders have responded to the program and closely examine their concerns about its methodology. Therefore, we surveyed hospital leadership—including chief executive officers (CEOs), chief medical officers (CMOs), and chief quality officers (CQOs)&mdash;at approximately 1600 hospitals, stratified by whether their hospitals received a penalty under the HRRP. We aimed to answer 3 key questions: first, how has the HRRP impacted hospitals’ readmission reduction efforts, particularly compared with prior readmissions policies such as public reporting? Second, how have leaders prioritized the HRRP in the context of multiple other federal quality improvement initiatives that they face simultaneously? Third, what are the opinions of hospital leaders on the program’s methodology and implementation?

Our first step in survey development was to conduct a set of case studies examining hospitals’ efforts to reduce readmission rates; this work has been described previously.

 Based on this work, we developed a survey instrument that was tested with hospital leaders, hospital personnel, and survey experts, and revised accordingly.

We began in mid-2012 with a list of all acute care hospitals that were eligible for the HRRP, excluding Critical Access Hospitals and other facilities not paid under the Inpatient Prospective Payment System and, thus, ineligible for participation. 

Based on calculations performed prior to survey administration, we anticipated needing 1000 survey responses to have adequate power to address our hypotheses. We anticipated a response rate of 60% to 65%; thus, our final sample consisted of 1600 hospitals. We also designed our survey sample to enable us to pursue secondary analyses that focused on differences between hospitals that care for a large proportion of black patients (who have previously been shown to have particularly high readmission rates

 and are also more likely to face unique challenges

) versus other hospitals; and differences between hospitals that had high, average, or low 30-day readmission rates. Thus, we calculated the overall proportion of Medicare patients at each hospital that self-identified as black. We then calculated 30-day risk-adjusted readmission rates for AMI, HF, and pneumonia, from 2008 to 2010 (the years used to assign hospital penalties during the first year of the HRRP) for each hospital, using methods that have been described previously.

 We selected all of the top 900 hospitals in terms of the highest proportion of black patients hospitalized with either AMI, HF, or pneumonia for inclusion in our sample. We divided the remaining hospitals into 3 groups based on performance on readmissions from 2008 to 2010, which was determined by ranking hospitals with the mean risk-adjusted readmission rates for the 3 target conditions into quintiles: top (best) quintile, middle 3 quintiles, and bottom quintile. We selected 266 hospitals from each of these groups using random number generation. There were a small number of hospitals in our sample that had closed, merged with other hospitals, or become Critical Access Hospitals or long-term care facilities; we replaced these using random selection from the same group.

To identify clinical leaders, we obtained the hospital leadership list of CMOs from the American Hospital Association. Study staff called each hospital leader to verify contact information, and once a recipient was verified, his or her hospital was moved into the active fielding stage. The survey was then fielded in 2 phases. The first phase (June 2013 to June 2014) was conducted by DataStat Inc, of Ann Arbor, MI. A hard copy of the survey was mailed to hospitals, along with a cover letter explaining the intent of the survey and the consent process. This was followed by follow-up phone calls and a second mailing. If requested, recipients were sent a version of the survey as a PDF file. The second phase (June to December 2014) was conducted by research staff at the Harvard T.H. Chan  School of Public Health, and followed a similar protocol—a second mailing and follow-up phone calls&mdash;but also gave hospital leaders the option of completing a Web-based version of the survey instrument. The second phase was instituted to ensure an adequately high response rate. Throughout the survey, although the initial point of contact at the hospitals was the office of the CMO, we encouraged that individual to reach out to other leaders within the hospital who were best equipped to help to either provide assistance or actually complete the survey.

We computed summary statistics both overall and stratified by HRRP penalty amount. We stratified the hospitals into 3 groups based on their penalty in FY2013, the main time frame in which the survey was in the field. Penalty statuses included a) no penalty, b) minor penalty (greater than 0 but less than the median penalty of 0.32% of base DRG payments), and c) major penalty (equal to or greater than the median penalty). Responses were tabulated for each question. For multiple choice or Likert scale questions, responses were summed within groups as they were defined on the survey (ie, “not important,” “somewhat important,” “very important,” and “extremely important”; or “disagree strongly,” “disagree,” “neither agree nor disagree,” “agree,” or “agree strongly”). For open-ended questions, we created a taxonomy based on the frequency of similar responses, and grouped responses accordingly.

Survey responses were adjusted for both sampling strategy and nonresponse to better reflect a national representation of US hospitals. To adjust for sampling strategy, we assigned sample weights to each group. To adjust for nonresponse, we constructed a logistic regression model, in which, returning the survey was the primary outcome, and hospital characteristics—including size, teaching status, ownership, and urban location&mdash;were predictors, as has been done previously.

 Each hospital received a likelihood of response based on this model; responses were then weighted with the inverse of this likelihood. Finally, we conducted additional regression analyses, in which, we further adjusted responses for the hospital characteristics listed above, as well as for the safety net status of the hospitals—those hospitals in the top quintile of Disproportionate Share Hospital Index were considered to be in the safety net

—and the proportion of black patients at each hospital.

All responses were de-identified before analysis. Informed consent was obtained within the survey itself and the introductory page to the survey included detailed information about privacy and data de-identification, as well as consent, stating, “Completion of this survey implies informed consent.” The study was approved by the Office of Human Research Administration at the Harvard T.H. Chan School of Public Health.

Of the 1600 hospitals contacted, we received completed surveys from 992 (62% response rate). Of that group, 951 were eligible for HRRP penalties in FY2013 and, thus, comprise our analytic sample. The other hospitals, mainly those in Maryland, or hospitals that no longer had enough cases to qualify for the program, were not eligible for HRRP. Hospital characteristics differed significantly by penalty receipt, with nonteaching, public, and Northeastern hospitals more likely to be in the “major penalty” group, and higher proportions of black and Medicaid patients in the highly penalized hospitals. Readmission rates were, as expected, higher in the highly penalized hospitals (

Compared with nonrespondents, respondents were more often leaders from large, nonprofit, or teaching hospitals; respondents were also more likely to represent urban hospitals and those located in the Northeast and Midwest (eAppendix Table 1 [

Of the respondents, 29.6% identified themselves as directors of case management or equivalent, 27.1% as CQOs or equivalent, 26.3% as CMOs or chiefs of staff, 4.6% as chief nursing officers, 2.5% as CEOs, and 9.8% as “other,” including vice president for medical affairs and chief operating officer.

Impact of the HRRP on Efforts to Reduce Readmissions

Nearly two-thirds (65.8%) of hospital leaders reported that the HRRP had a “significant” or “great” impact on increasing their hospital’s efforts to reduce readmissions compared with the 2 readmissions policies that preceded the HRRP: public reporting of readmission rates (36.1%) and public reporting of discharge planning (23.7%). When we examined these opinions, stratified by receipt of a penalty, we found that leaders at hospitals receiving a penalty were much more likely to report that each of the federal policies had impacted their efforts to reduce admissions than at those hospitals not receiving a penalty (

). Adjusting these results for hospital characteristics and socioeconomic status (SES) factors yielded similar results (eAppendix Figure 1).

In terms of potential responses to the HRRP, 26.6% of leaders reported that it was more than moderately likely that hospitals would increase the use of observation status to improve their perceived performance on readmissions, and 15.1% felt it was more than moderately likely that hospitals would increasingly avoid high-risk patients. These responses were similar across penalty strata (

 = .14 for differences in response, respectively).

Prioritization of Readmissions Reduction in the Context of Other Federal Programs

When asked to prioritize readmissions reduction among other current federal quality improvement initiatives, only 44.1% of leaders reported that it was of “highest priority” compared with 79.2% for improving patient safety, 76.6% for improving patient experience, 75.2% for reducing hospital-acquired infections, 65% for meeting Meaningful Use requirements, and 44.4% for improving compliance with guideline-based care. The biggest gap in prioritization between leaders at hospitals with readmission penalties versus without readmission penalties was in prioritizing readmissions. However, leaders at highly penalized hospitals still rated all 6 of the competing priorities more highly than those at nonpenalized hospitals (

). Results adjusted for hospital characteristics and SES factors were similar (eAppendix Figure 2).

Opinions on the Methodology and Impact of the HRRP

A majority (67.5%) of leaders felt that the HRRP penalties were “much too large”; this was more common among leaders at hospitals receiving major penalties (74.7%) than those hospitals without penalties (65.2%: 

 <.001) but was still a majority in all groups. The most commonly endorsed critique of the HRRP penalty was that it did not adequately account for differences in SES between hospitals (76.2% “agree” or “agree strongly”). Other common concerns included an inadequate account of medical complexity by the penalty (75.9%), and hospitals' limited ability to impact patient adherence (64.1%) (

). Each concern was expressed more often among leaders of hospitals receiving major or minor penalties than among leaders of hospitals without penalties (Table 2); results adjusted for hospital characteristics and SES factors were similar, although the differences between groups narrowed somewhat (eAppendix Table 2).

Only a minority of study hospitals were participating in bundled payment programs or accountable care organizations (ACOs), and just over half of hospitals were participating in private pay-for-performance programs (

, top panels). When asked whether these value-based payment programs were likely to improve quality, 42.5% of leaders responded affirmatively about the HRRP compared with 32% for bundled payment programs, 45.6% for ACOs, and 52.6% for pay-for-performance (Table 3, bottom panels).

Response patterns were generally similar when stratified by receipt of a penalty, but leaders at hospitals receiving penalties were less likely to respond that the HRRP was likely to improve care (35.7% for hospitals with major penalties vs 45% for minor penalties vs 48.4% for no penalties [Table 3]) and response patterns were also similar when adjusting for hospital characteristics and SES factors (eAppendix Table 3). The highest proportion of respondents (54.8%) felt that the HRRP was likely to reduce costs compared with the other programs (Table 3); responses were similar across penalty strata and after adjustment (eAppendix Table 3).

In a large, national survey, hospital leaders reported that the HRRP has had a sizable impact on their hospitals’ efforts to reduce readmissions. However, despite paying more attention to readmissions than previously done, hospital leaders continue to prioritize other quality improvement efforts, such as improving patient safety, improving patent experience, and adhering to clinical guidelines. Hospital leaders also reported critiques of the policy—largely centered around risk adjustment for SES and clinical factors&mdash;and the ability of hospitals to impact patient adherence, as well as postacute, ambulatory, and institutional care. Leaders at hospitals that were receiving penalties under the HRRP tended to have more negative opinions about the program than leaders at hospitals without penalties.

According to hospital leaders, the HRRP has had a significantly greater impact on their own efforts to reduce readmission rates than its policy predecessors—namely, public reporting of discharge planning and public reporting of readmission rates. This observation, that financial incentives alter behavior to a greater degree than public reporting alone, is consistent with prior observations

 and may explain why the HRRP has been associated with improvements in readmission rates

 whereas public reporting was not. This experience may suggest that policy makers should move more rapidly to financially reward or penalize hospitals for desired outcomes rather than merely reporting them publicly.

 Of course, the fact that over one-fourth of respondents suggested that hospital leaders might increase the use of observation status to improve performance on readmissions, and 15% thought hospitals might avoid high-risk patients, could serve as cautionary counterpoints to the enthusiasm for financial incentives.

Nevertheless, despite the reported impact of the HRRP, nearly every other mandatory federal quality improvement program was rated higher in terms of its importance in our survey, although the study design did not allow us to determine why this was the case. One possibility is that readmissions were seen by many as being outside the control of the hospital; in the setting of competing priorities, perceiving a lack of ability to change an outcome could cause hospital leadership to focus on other areas for intervention. As programs, such as ACOs, increasingly bridge the inpatient and outpatient settings, it is possible that increasing integration could alter these perceptions. Additionally, as the number of Medicare programs that reward readmissions as components of performance continues to grow—not only in ACOs,

 but also including the Physician Value-Based Modifier

 and the coming pay-for-performance programs in the dialysis

—it is feasible that inpatient facilities may prioritize readmission reduction more highly.

In terms of methodology, the frequently cited critiques of the HRRP included its lack of adjustment for SES or patient adherence and concern about its adjustment for medical complexity. Although these findings were not necessarily surprising given prior publications

 our survey allows us, for the first time, to quantify the degree to which these are concerns for hospital leaders. Given that more than 3 of 4 hospital leaders reported SES as a critical issue, it is clear that the concerns are not just among those who disproportionately care for the poor. On the other hand, a sizable proportion of leaders did not feel that SES adjustment was necessary, suggesting that support is not unanimous among the hospital community. There is a great deal of current activity around SES and readmission policy: the National Quality Forum recently released recommendations in this area,

 and is currently undertaking analyses to determine if adjustment for SES is appropriate for certain measures, including many having to do with readmissions.

 Two bills that were recently proposed in Congress aimed to incorporate measures of poverty, income, and education into risk adjustment for the HRRP,

  but neither legislation has moved forward. The Medicare Payment Advisory Commission has argued that the HRRP should stratify hospitals into groups based on SES,

 which is one promising strategy. Congress also passed the Improving Medicare Post-Acute Care Transformation (IMPACT) Act in October 2014, which calls on the HHS to study the relationship between SES and performance in Medicare’s incentive-based programs, and to suggest changes in these programs that might be warranted.

As is the case with all surveys, it is possible that nonrespondents were different than those who responded to our survey. Although we used appropriate techniques to deal with nonresponse, these statistical techniques are imperfect. Further, we believe that hospital leaders answered these questions to the best of their ability, but, nevertheless, there may be differing opinions within hospitals, such that our results reflect only the individual who filled out the instrument. Therefore, it is possible that sampling different individuals within these hospitals would have yielded different results. Finally, we surveyed hospitals during the first 2 years of the HRRP, and leaders’ opinions may change over time.

In a national survey of hospital leaders, we found that the HRRP has had a major impact on hospital leaders’ efforts to reduce readmissions; however, the HRRP currently remains a lower priority for leaders than other areas of quality improvement, such as patient safety, patient experience, and adherence to guidelines. Further, concerns remain about its manner of accounting for social and medical risk factors and whether hospitals, by themselves, can impact patient adherence or the transitional and postacute care that helps determine whether a patient is readmitted. These findings may be useful for policy makers contemplating future iterations of the HRRP and other programs using readmissions as a quality metric that may have a synergistic effect on improving patient care. 

 Department of Health Policy and Management (KEJ, JFF, AKJ) and Department of Biostatistics (EJO), Harvard T.H. Chan School of Public Health, Boston, MA; Division of General Internal Medicine (JFF, EJO, AKJ) and Division of Cardiovascular Medicine (KEJ), Department of Medicine, Brigham and Women’s Hospital, Boston, MA.

 This research was funded by NHLBI grant number 1R01HL113567-01.

 Dr Joynt is currently employed in the United States Department of Health and Human Services, Office of the Assistant Secretary for Planning and Evaluation, where Dr Orav also serves as an advisor. The work described here was conducted when the authors were employees of Harvard University. The views expressed herein are those of the authors alone, and do not represent the official position of the federal government. The remaining authors report no relationship or financial interest with any entity that would pose a conflict of interest with the subject matter of this article. 

The Hospital Readmissions Reduction Program has had a major impact on hospital leaders’ efforts to reduce readmission rates; however, important concerns about the program remain.

Opinions on the Hospital Readmission Reduction Program: Results of a National Survey of Hospital Leaders

		Choosing Wisely is a program aimed at reducing harm and costs associated with unnecessary testing. 

		Spinal pain is common, but few patients with spinal pain require advanced imaging (ie, radionuclide bone, computed tomography, or magnetic resonance imaging scans). 

		There are various red flags that can aide the physician in determining who does not need advanced imaging. 

		Using red flags as the threshold for ordering advanced imaging yields a very low risk of missing any serious cases of spinal pain.

Robert Ferrari, MD, MSc (Med), FRCPC, FACP

 confirm in their February 2016 article, in 

, that there is a high percentage of wasteful magnetic resonance imaging (MRI) scans being ordered for patients with low back pain. Their study protocol gives the ordering physician considerable leeway in determining what is inappropriate, and it is likely that the wasteful testing is even higher than the 31% found, given that failure to improve with conservative therapy is not a reason in itself for ordering an MRI. Regardless of how one defines “wasteful” or “inappropriate” testing, there is no doubt that this behavior exists.

Gidwani et al quite rightly point out that efforts to reduce inappropriate ordering cannot rely on financial incentives alone; as such, they suggest that interventions target specific providers, as the researchers found a high concentration of inappropriate ordering among a few providers. Although this seems logical, and there would not seem to be a need to target “nonoffenders,” a more universal solution is even more appropriate. An intervention that universally discourages wasteful testing can be applied to future and new physicians, or physicians whose practices or practice styles change. It also reduces the approach of “blame”—that is, identifying certain physicians as culprits.

One approach may be to change the requisition form used for ordering the MRI. It has been shown that if MRI scanning is used only when red flags are present in patients with subacute or chronic spinal pain, there is a very low risk of missing any cases of serious spinal disorders.

 In the aforementioned study, subjects with any region of spinal pain for a period of 6 to 52 weeks were assessed to determine if radiological studies beyond x-rays were indicated, including MRI, computed tomography (CT), or radionuclide bone scans. An a priori threshold (the presence of at least 1 red flag) was set before an MRI, CT, or bone scan would be considered. Those who did not have one of these 3 scans ordered for them were followed for at least 1 year to determine if any went on to have a diagnosis of a more serious form of spinal disorder (eg, infection, fracture, spondylitis, tumor, neurological compression). Of the 1003 subjects, 110 met an a priori threshold for ordering at least 1 MRI, CT, or bone scan. The remaining 893 subjects were followed for at least 1 year, and none developed evidence of a nonbenign cause for their spinal pain. Thus, in the evaluation of nonspecific spinal pain and symptoms, setting an a priori threshold for ordering MRI, CT, or bone scans with the spirit of the current Choosing Wisely recommendations for spinal imaging results in a very low risk of missing a case of a serious cause of back pain.  This approach can be universally applied to a wide array of practitioner types.

In our article, we suggest that targeted, rather than universal, quality improvement efforts are better suited to reduce inappropriate order of lumbar spine magnetic resonance imaging (LS MRI), based on our results indicating 11% of providers ordered 50% of inappropriate scans and that 24% of providers ordered 74% of inappropriate scans.

In his letter to the editor, Dr Ferrari suggests that a universal quality improvement approach is still preferred, as it applies to all physicians and reduces the approach of blaming physicians. He also proposes a universal requisition form to be used for ordering an MRI, limiting the order only to patients with “red flag” conditions and subacute or chronic spinal pain.

We respectfully disagree with this approach and believe the targeted quality improvement approach is preferred for 3 reasons. First, universal application of a requisition form to justify imaging orders is unlikely to reduce inappropriate scans. Previous attempts to apply this strategy have had small effect.

 By the time the order is written, the provider has already made a decision to order an MRI, and providers easily learn the answers that will get their order approved

Second, it is not clear that allowing only patients with red flags—which Dr Ferrari includes as those with cancer, fever, diabetes, muscle weakness, age greater than 65, among other factors&mdash;to undergo spinal imaging results in a low risk of missing cases of serious spinal disorder. The study Dr Ferrari references is based on a cohort of patients he saw in clinical practice, and, like all observational studies, suffers from selection bias.

 He found that younger, healthier patients without red flag conditions had fewer cases of nonbenign back pain relative to older, sicker patients with red flag conditions. Given the systematic differences in these patients at baseline, these results are not surprising. The generalizability of this study is also compromised by this selection issue, as the healthiest and youngest patients with low back pain are systematically different than the population of other patients who present at physicians’ offices with low back pain. 

Third, when implementing quality improvement programs, one must also be cognizant of the burden of the (unnecessary) quality improvement intervention on the providers who do not order inappropriately. When the majority of inappropriate behavior is concentrated in a few individuals, as our study found, universal programs have the effect of penalizing—in the form of wasted time and undue cognitive burden&mdash;the providers who are acting appropriately. 

A more effective approach must address the root causes of the inappropriate orders; however, these are not yet fully understood, and may vary in different settings and health plans. Multiple surveys have found that failure to implement guidelines is not only due to lack of familiarity. Providers may not believe the guidelines are correct, or structural factors may make their implementation difficult.

A letter on reducing unnecessary spinal imaging through the introduction of an a priori threshold before an MRI, CT, or bone scan would be considered in patients with spinal pain, with a reply from Dr Gidwani, of the original February 2016 article.

August 2016

Association of Part D Coverage Gap With COPD Medication Adherence

The Financial Impact of Team-Based Care on Primary Care

Opinions on the Hospital Readmission Reduction Program: Results of a National Survey of Hospital Leaders

Choosing Wisely: Spine Imaging

The Impact of Formulary Drug Exclusion Policies on Patients and Healthcare Costs

Socioeconomic Disparities in Adoption of Personal Health Records Over Time

The Impact of Patient-Centered Medical Homes on Safety Net Clinics

Variation in US Outpatient Antibiotic Prescribing Quality Measures According to Health Plan and Geography

Aligning Payment Reform and Delivery Innovation in Emergency Care

Association Among Change in Medical Costs, Level of Comorbidity, and Change in Adherence Behavior