PurposePharmacy chains can differ with respect to the characteristics of their patient populations as well as their nonprescription products, services, and practices, and thus may serve as a surrogate for potential unmeasured confounding in observational studies of prescription drugs. This study evaluates whether a single-source drug can have different patient outcomes based on the dispensing pharmacy chain.MethodsSeparate analyses for two anticoagulant drugs, rivaroxaban and apixaban, were conducted using Medicare Fee-for-Service claims evaluating the association between dispensing pharmacy chain and outcomes of acute myocardial infarction, ischemic stroke, intracranial hemorrhage, gastrointestinal (GI) bleeding, all-cause mortality, and major GI bleeding. Inverse probability of treatment weighting (IPTW) was used to balance baseline covariates across pharmacy chain cohorts, and outcome association was assessed with a Cox Proportional Hazards model.ResultsWe observed no differences in outcomes across pharmacy chains for apixaban recipients. Rivaroxaban recipients from pharmacy chain C, however, had lower rates of GI bleeding (adjusted HR 0.83; 95% CI 0.69-1.00) and ischemic stroke (adjusted HR 0.57; 95% CI 0.38-0.87) as compared to chain A in primary analyses with a 3-day grace period. The results moved closer to the null when 14- and 30-day grace periods were implemented.ConclusionsThese results suggest that dispensing pharmacy chains may have the potential to act as a confounder of associations between drug exposure and outcome in some observational studies. Additional studies of potential confounding by pharmacy chain are needed. Further evaluation of potential pharmacy chain effects on safe use would be of value.

Background:Under the Merit-based Incentive Payment System (MIPS), the U.S. Centers for Medicare and Medicaid Services (CMS) evaluate clinicians who manage Medicare patients on the basis of cost and quality outcomes. CMS contractor Acumen, LLC, convened an expert panel to develop a knee arthroplasty episode-based cost measure (EBCM) for use in the MIPS.Methods:A Clinical Subcommittee of 28 clinician experts affiliated with 27 specialty societies provided guidance in developing the knee arthroplasty EBCM. The Clinical Subcommittee specified all aspects of the EBCM including triggering of the episode, services within the episode, risk adjustment, subgrouping, and exclusions. Services were counted only if the Clinical Subcommittee deemed them under the influence of the clinician assigned to the EBCM (selective service assignment; SSA). We assessed the reliability of the EBCM and compared it with an alternative population-based cost measure constructed without SSA.Results:We identified 249,301 knee arthroplasty episodes from June 1, 2016, to May 31, 2017, with 10,681 clinicians having at least 10 attributed episodes. The mean episode cost was $19,321 with a standard deviation of $1,816. SSA increased the reliability score from 0.71 to 0.81 relative to an alternative measure that counted all patient costs. SSA also led to reclassification of 41.8% of clinicians into different quintiles of performance.Conclusions:We found that the use of SSA in the creation of the EBCM substantially reduces random noise (i.e., unrelated medical procedures or costs) and offers a tool for assessing clinicians' costs of management that is focused on care directly related to knee arthroplasty.

BACKGROUND The Vaccine Safety Datalink (VSD) identified a statistical signal for an increased risk of Guillain-Barré syndrome (GBS) in days 1-42 after 2018-2019 high-dose influenza vaccine (IIV3-HD) administration. We evaluated the signal using Medicare. METHODS We conducted early- and end-of-season claims-based self-controlled risk interval analyses among Medicare beneficiaries ages ≥65 years, using days 8-21 and 1-42 postvaccination as risk windows and days 43-84 as control window. The VSD conducted chart-confirmed analyses. RESULTS Among 7 453 690 IIV3-HD vaccinations, we did not detect a statistically significant increased GBS risk for either the 8- to 21-day (odds ratio [OR], 1.85; 95% confidence interval [CI], 0.99-3.44) or 1- to 42-day (OR, 1.31; 95% CI, 0.78-2.18) risk windows. The findings from the end-of-season analyses were fully consistent with the early-season analyses for both the 8- to 21-day (OR, 1.64; 95% CI, 0.92-2.91) and 1- to 42-day (OR, 1.12; 95% CI, 0.70-1.79) risk windows. The VSD's chart-confirmed analysis, involving 646 996 IIV3-HD vaccinations, with 1 case each in the risk and control windows, yielded a relative risk of 1.00 (95% CI, 0.06-15.99). CONCLUSIONS The Medicare analyses did not exclude an association between IIV3-HD and GBS, but it determined that, if such a risk existed, it was similar in magnitude to prior seasons. Chart-confirmed VSD results did not confirm an increased risk of GBS.

Escalating United States (US) health care cost is widely recognized to be unsustainable. In 2016, Medicare payments were nearly $700 billion, approximately 15% of the federal budget. If left unchecked, annual expenditures are projected to increase to $1.2 trillion over the coming decade.1Keehan S.P. Stone D.A. Poisal J.A. et al.National health expenditure projections, 2016–25: price increases, aging push sector to 20 percent of economy.Health Affairs. 2017; 36: 553-563Crossref PubMed Scopus (151) Google Scholar,2Congressional Budget OfficeCongress of the United StatesThe budget and economic outlook: 2017 to 2027.2017Google Scholar Congress, in response to rising costs, passed the Medicare Access and Children's Health Insurance Program Reauthorization Act (MACRA) in 2015.3114th Congress. Medicare Access and CHIP Reauthorization Act of 2015. Vol 129. 2015Google Scholar This Act repealed the Sustainable Growth Rate (SGR) and streamlined clinician reporting programs into the Quality Payment Program to promote high-quality cost-efficient care. Clinicians caring for Medicare beneficiaries have 2 options for participation in the Quality Payment Program, the Merit-Based Incentive Payment System (MIPS) or Advanced Alternative Payment Models. In the near term, most gastroenterologists will be reimbursed through MIPS, which maintains fee-for-service reimbursement but adjusts future payment by evaluating clinicians across four dimensions: quality, clinical practice improvement activities (referred to as "improvement activities"), meaningful use of certified electronic health record technology (referred to as "promoting interoperability"), and resource use (referred to as "cost").4Centers for Medicare and Medicaid Services, Department of Health and Human Services. Medicare program; Merit-Based Incentive Payment System (MIPS) and Alternative Payment Model (APM) incentive under the physician fee schedule, and criteria for physician-focused payment models. CMS-5522-FC. Federal Register 2017;82 FR 53568:53568–54229.Google Scholar MACRA mandated that the Centers for Medicare and Medicaid Services (CMS) construct and change episode-based cost-measures (EBCMs) to expand methods to evaluate resource use.3114th Congress. Medicare Access and CHIP Reauthorization Act of 2015. Vol 129. 2015Google Scholar CMS began evaluating clinician resource utilization before MACRA with the Value-Based Payment Modifier program. This program used broad measures that held clinicians accountable for the entirety of health care spending by their patients during a set period of time.5Centers for Medicare and Medicaid ServicesDetailed methodology for the 2018 Value Modifier and the 2016 Quality and Resource Use Report.2017Google Scholar,6Centers for Medicare and Medicaid ServicesValue-Based Payment Modifier.2017Google Scholar For elective procedures that are as routine and well studied as colonoscopy, measures have more meaning if they can discriminate which costs can be influenced by the clinician performing the colonoscopy. Creating episodes of care can focus on the costs of treatment, postoperative care, and complications that are uniquely associated with the procedure and can be influenced by the attributed clinician. This method focuses accountability on areas where clinicians can actually improve value for specific related conditions (eg, decreasing postcolonoscopy bleeding). A group of clinician stakeholders including gastroenterologists, colorectal surgeons, radiologists, nurse practitioners, clinical nurse specialists, anesthesiologists, and internists were convened by the CMS contractor Acumen to create an EBCM for screening/surveillance colonoscopy. Through an iterative transparent process that included public stakeholder feedback, an episode of screening/surveillance colonoscopy that assigned care services to the responsible clinician was developed. The specific goal was to identify a granular set of services that would be clinically related to the procedure and therefore potentially modifiable by the clinician. We exclusively used Medicare claims to construct the episode group to ensure no increase in clinicians' reporting burden. Details regarding the development of this episode are available in the Supplementary Appendix. In brief, a clinical subcommittee of 35 clinician experts affiliated with 23 specialty societies participated in constructing this episode (Figure 1). This group chose to subdivide the population by site of service for the colonoscopy (Hospital Outpatient Department [HOPD], Ambulatory Surgical Center [ASC], or Office) to ensure that costs were compared across homogeneous patient populations and to account for cost differences specifically related to facility type. Diagnostic colonoscopies, inpatient colonoscopies, and patients with inflammatory bowel disease were excluded. An episode was triggered with the use of the primary screening and surveillance Current Procedural Terminology/Healthcare Common Procedure Coding System codes (G0105, G0121) and diagnostic codes (45378, 45380, 45381, 45384, 45385) with a PT modifier to signify the original intention of screening. The group selected a time frame for assessing costs of 14 days after the procedure. The Subcommittee then identified all services within that time frame that were assessed as clinically related to the index colonoscopy, including services related to plausible complications from colonoscopy. Examples of common drivers of costs and services included cardiopulmonary complications, postprocedure lower gastrointestinal bleeding, pathology services, perforation or peritonitis, and repeated colonoscopy or flexible sigmoidoscopy. Admissions and treatments for medical care deemed to be unrelated to the index colonoscopy were not included.Figure 2Changes in total cost of anesthesia/anesthesia complications and pathology clinical themes across performance quintiles.View Large Image Figure ViewerDownload Hi-res image Download (PPT) The Subcommittee also defined risk adjustors to adjust episode costs for factors outside a clinician's control. The risk-adjustment model incorporated patient age, reason for Medicare eligibility (age, disability, or end-stage renal disease), long-term care residence, and comorbid condition variables from CMS' Hierarchical Condition Category (HCC) model (2016 v22) in addition to specific clinical variables (Supplementary Appendix).7Evans M.A. Pope G.C. Kauttner J. et al.Evaluation of the CMS-HCC risk adjustment model.2011Google Scholar,8Centers for Medicare and Medicaid ServicesDetails for title: 2018 Model Software/ICD-10 Mappings.2018Google Scholar Once constructed, these episode specifications were then field tested with the use of Medicare claims data from June 1, 2016, to May 31, 2017.9Centers for Medicare and Medicaid ServicesMerit-Based Incentive Payment System (MIPS): Episode-Based Cost Measure field test reports fact sheet.2018Google Scholar The Clinical Subcommittee further refined the measure after considering public comments and results from field testing. To assess the value of this measure, we compared it with a hypothetical measure without these clinically selected service assignments to simulate an alternative cost measure method. In the field test period, 780,025 screening/surveillance colonoscopy episodes from 776,718 beneficiaries were identified. The average age for patients receiving these colonoscopies across each of the settings was 70 years. There were 4,192 clinicians or clinician groups defined by Tax Identification Number (TIN) with at least 10 episodes included in our analysis where many practiced in multiples sites of service. Not surprisingly, patients with more comorbidities had their colonoscopies at HOPDs (average number of HCC comorbidities 0.97, followed by ASCs [0.76] and then Offices [0.74]). We calculated a statistic of measure reliability to assess how well the measure captures differences across clinicians rather than differences in patient case mix or random noise. The mean overall reliability increased from 0.30 to 0.93 with the use of selective service assignments at the TIN level of analysis. The average risk-adjusted episode cost for all providers was $962 (SD $88). After service assignments, this episode produced a narrower and lower cost distribution. The unadjusted average provider costs for episodes that included service assignments for HOPDs, ASCs, and Offices were $1,129, $841, and $611, respectively. These were comparatively less than an episode that excluded service assignment (HOPDs $1,403, ASCs $1,032, and Offices $830). Furthermore, there were decreased inclusions of high-cost services unrelated to the colonoscopy. While the distribution is tighter, the variation is more meaningful because of the careful construction of service assignment rules to focus only on clinically related services. This is reflected by the large improvement in reliability after service assignment, which measures the actual variation across providers. The application of selective service assignments in this EBCM ensures that a substantial amount of unrelated costs are not included. To further elaborate on the difference between using this selective EBCM vs a hypothetical nonselective EBCM, clinicians (by TIN) were ranked by quintiles in each measure, with quintile 1 being the lowest-cost group (ie, observed over expected costs) and quintile 5 being the highest-cost group. Overall, a total of 52.4% of TINs would be reclassified into a different quintile with the use of this selective EBCM. Among TINs who incurred the highest costs (quintiles 4–5) in the nonselective EBCM, 26% would be reclassified as lower-cost TINs (quintiles 1–3) with the use of the selective EBCM. Among clinicians who incurred lower costs (quintiles 1–3) in the nonselective EBCM, 17.4% would be reclassified as high-cost TINs (quintiles 4–5) with the use of the selective EBCM. When comparing distribution of costs across EBCM quintiles by clinical theme, the difference between low- and high-cost providers are not due to frequency of gastrointestinal or cardiopulmonary complications from the colonoscopy nor due to need for repeated colonoscopy. The primary driver of cost differences between quintiles appeared to be related to anesthesia utilization and pathology costs (Figure 2). A high-cost clinician may have relatively higher pathology costs due to increased detection of polyps generating more pathology services. In this specific case, higher costs may translate into higher quality, but may not be appreciated in the 2-week episode window as currently designed because interval cancer development would take years to measure. This highlights the need to not view this measure in isolation, because it does not provide a comprehensive view of clinician care. Rather, it will need to be assessed in the context of performance in other MIPS categories and within the MIPS overall scoring framework. The Screening/Surveillance Colonoscopy EBCM was endorsed by the National Quality Forum in 2019 and has been in use in the MIPS cost performance category since then. The development of this EBCM was deeply informed by public stakeholder feedback and a panel of technical experts. Principles guiding measure development included minimizing reporting burden by constructing the measure on claims data alone, establishing clear rules to determine when an episode is attributed to a clinician, and including only clinically relevant services. This EBCM on screening/surveillance colonoscopy removed costs that were not clinically relevant but maintained variation in overall costs. The measure may also sync with important quality indicators such as reducing repeated examinations due to poor bowel preparation or inability to reach the cecum and their associated increased costs.10Manes G. Repici A. Hassan C. et al.Randomized controlled trial comparing efficacy and acceptability of split- and standard-dose sodium picosulfate plus magnesium citrate for bowel cleansing prior to colonoscopy.Endoscopy. 2014; 46: 662-669Crossref PubMed Scopus (23) Google Scholar,11Martel M. Barkun A.N. Menard C. et al.Split-dose preparations are superior to day-before bowel cleansing regimens: a meta-analysis.Gastroenterology. 2015; 149: 79-88Abstract Full Text Full Text PDF PubMed Scopus (136) Google Scholar This measure complements 2 population-based measures, Total Per Capita Cost (TPCC) and Medicare Spending per Beneficiary (MSPB), which were the only 2 measures in the 2017 and 2018 MIPS cost category. TPCC assesses the global costs of care, and MSPB looks at cost of care associated with a hospitalization.5US Government Accountability OfficeMedicare: Per Capita Method can be Used to Profile Physicians and Provide Feedback on Resource Use.https://www.gao.gov/products/GAO-09–802Date: 2009Google Scholar As population-based measures that apply to a large and diverse number of clinicians in MIPS, they help provide comparability between clinician cost performance. By including more episodes on which to evaluate a clinician than the narrower EBCMs, the population-based measures will provide more opportunities for attributed clinicians to receive feedback on the overall cost of care. In addition, they align with measures across other Medicare programs, such as the MSPB Hospital measure, to align incentives to provide optimal patient care. In conclusion, an EBCM for screening/surveillance colonoscopy was constructed with extensive, detailed input from the clinician community to be used in the MIPS program to serve as a reliable measure of cost. Though developed with certain limitations associated with using a claims-based system, there is no additional reporting burden for the clinician. This measure compares only costs that can be reasonably influenced by the clinician. Further development of such measures will lead to broader acceptance by clinicians of the validity and value of cost performance measures within the MIPS program. Unlisted authors: Nirmal Choradia,1 Colleen Schmitt,2 Caroll Koscheski,3 Joyce Lam,1 Sam Bounds,1 Rose Do,1 Laurie Feinberg,1 Daniel Vail,1 Sriniketh Nagavarapu,1 and Jay Bhattacharya4,5; from 1Acumen, Burlingame, California; 2Galen Medical Group, Chattanooga, Tennessee; 3Gastroenterology Associates, Hickory, North Carolina; 4Division of Gastroenterology and Hepatology, Department of Medicine, Stanford University School of Medicine, Stanford, California; and 5Stanford Health Policy, Stanford University School of Medicine, Stanford, California. The main purpose of episode construction is to define discrete health care events that constitute the entirety of medical care that a patient receives. Prior versions of episode construction focused on the patient's experience of illness within the health care system, with cost assignment and episode attribution occurring after completion of the episode.1Keehan S.P. Stone D.A. Poisal J.A. et al.National health expenditure projections, 2016–25: price increases, aging push sector to 20 percent of economy.Health Affairs. 2017; 36: 553-563Crossref PubMed Scopus (151) Google Scholar,2Congressional Budget OfficeCongress of the United StatesThe budget and economic outlook: 2017 to 2027.2017Google Scholar Assessing resource use through the patient's experience of illness is relatively straightforward when evaluating care that is bundled or delivered within a single comprehensive health care system. However, in the context of a Merit-Based Incentive Payment System (MIPS) cost measure evaluating the effects on care by physicians in a fee-for-service system, a patient-centric framework can be problematic. Multiple clinicians are often involved in the care of a single condition. Patients also usually have multiple conditions that might qualify as separate episodes. Previous attempts have failed to coherently attribute services that are provided for multiple conditions and to clearly delineate the responsible provider when multiple clinicians are involved with a single condition. This shortcoming has made it difficult for clinicians to identify areas of improvement in patient care, one of the putative goals of the Centers for Medicare and Medicaid Services' (CMS) in the promotion of cost-efficient high-value care. To overcome these limitations, Acumen (a contractor of CMS) and CMS worked with clinical stakeholders to construct episodes focused on the clinician's role in caring for a patient's illness. This approach allows for identifying the services and outcomes that a clinician can credibly influence and should therefore be held accountable for. Consequently, a clinician can identify areas in which they can both decrease cost and improve quality.3National Quality ForumMeasure evaluation criteria and guidance for evaluating measures for endorsement.http://www.qualityforum.org/Measuring_Performance/Submitting_Standards/2016_Measure_Evaluation_Criteria.aspxDate: 2016Date accessed: May 23, 2018Google Scholar This paradigm allows services to be shared across multiple episodes and can be attributed to multiple clinicians simultaneously. Because episodes are compared only with similar episodes, services are not inappropriately double-counted. A key advantage to this approach focuses the episode around the set of services that can be reasonably influenced by the attributed clinician. When developing new episodes for potential use in MIPS to evaluate clinician resource use, Acumen applied criteria published by the National Quality Forum (NQF) Resource Use Measure (RUM)3National Quality ForumMeasure evaluation criteria and guidance for evaluating measures for endorsement.http://www.qualityforum.org/Measuring_Performance/Submitting_Standards/2016_Measure_Evaluation_Criteria.aspxDate: 2016Date accessed: May 23, 2018Google Scholar to evaluate previous episode constructions and guide its episode design. Applying the organizing principles outlined by the NQF to the development of episode-based RUMs for use in MIPS led to the following goals: 1) clinical face validity, 2) clear attribution, 3) clinical coherence, 4) transparency and comprehensibility, 5) utility and actionability, and 6) no increased reporting burden. Clinical face validity refers to the clinical community's trust in the source and accurate application of clinical input, and its importance has been stressed across many different programs, public comments, and technical expert panels. Clinical guidance should not be limited in its scope and should inform every facet of episode group development. Acumen allowed for the selection of peers whose clinical knowledge is directly applicable to the clinician's role in the episode being developed. A public nomination period was held to gather subcommittees from the clinical community who address these episodes and understand the inherent concerns facing their colleagues. The next goal in episode construction methodology is clearly attributing an episode to the correct clinician. As indicated above, previous approaches of episode development did not attribute episodes until after their completion, making it difficult for managing clinicians to know whether they would be attributed the episode. This dramatically reduces the incentive for clinicians to meaningfully affect resource use. In addition, clear and early attribution allows us to focus on the clinician's role during the episode and to evaluate providers only on services that they can reasonably influence. This difficulty is magnified when a single service is used to treat multiple conditions (such as an outpatient follow-up visit). Previous efforts struggled with developing consistent rules in assigning costly services to the "correct" provider and have either lumped the entire service cost into a single episode or split service costs (often arbitrarily) across multiple episodes. By focusing on clear attribution and service assignment early in an episode's course, we ensure that clinicians can accurately identify the moment an episode begins, so that they can change behavior in real time. Once the episode is attributed to the correct provider, the population and the assigned costs must be clinically coherent, so that a clinically homogenous population is represented and clinicians are assessed only on services that they can influence. The goal is to maximize variations in cost that are due to a clinician's care and to minimize variations that are outside a clinician's influence. An often-expressed example is tertiary care centers or highly specialized clinicians being inordinately affected due to caring for the sickest populations. Although risk-adjustment can reduce some of the variation outside a clinician's influence, robust clinical input on service assignment often dramatically improves the performance of these risk-adjusted models. One strategy for improving clinical homogeneity is to develop episode subgroups, so that episodes can be further stratified into comparable units. The goals of transparency and comprehensibility derive from the guiding principle of "usability." As with the concerns for attribution, the costs assigned for an episode should be clearly defined and consistent and should be determined by the clinical community that deals directly with the disease process and its consequences. This allows the affected clinicians to know which costs they must focus on to improve their resource use. Episodes must be easily understood by clinicians, and a clinician's performance must be easily evaluated through feedback to clinicians. Our iterative process of collaboration with clinical stakeholders along with public reporting has fostered greater transparency and comprehensibility of these episodes. All episode construction algorithms have been made publicly available, so that clinicians can easily reconstruct episodes and identify areas of excess cost. Actionability refers to a measure's usefulness for decision making and its ability to influence changes in behavior. The actionability of an episode characterizes how well it presents feedback that clinicians can act on to reduce the cost of the care they provide. Existing literature has generally considered actionability in the context of the type of information and level of detail included in feedback reports.3National Quality ForumMeasure evaluation criteria and guidance for evaluating measures for endorsement.http://www.qualityforum.org/Measuring_Performance/Submitting_Standards/2016_Measure_Evaluation_Criteria.aspxDate: 2016Date accessed: May 23, 2018Google Scholar, 4Medicare Payment Advisory CommissionReport to the Congress: Improving incentives in the Medicare Program.http://www.medpac.gov/docs/default-source/reports/Jun09_EntireReport.pdfDate: 2009Date accessed: May 24, 2018Google Scholar, 5US Government Accountability OfficeMedicare: Per Capita Method can be Used to Profile Physicians and Provide Feedback on Resource Use.https://www.gao.gov/products/GAO-09–802Date: 2009Google Scholar However, it is also necessary that the clinical information reported in measures is designed to be actionable. The NQF has argued that episode-based resource use measurement provides more focused and actionable results than per-capita measurement because it gives clinicians information about their management of specific conditions.3National Quality ForumMeasure evaluation criteria and guidance for evaluating measures for endorsement.http://www.qualityforum.org/Measuring_Performance/Submitting_Standards/2016_Measure_Evaluation_Criteria.aspxDate: 2016Date accessed: May 23, 2018Google Scholar To achieve these goals, 8 initial episode-based cost measures were developed in 2017, drawing on extensive input from technical expert panels, public comment postings, and condition-specific clinical subcommittees that nominated clinician stakeholders from relevant specialties and qualified professional society nominees. One of these 8 measures was screening/surveillance colonoscopy, which forms the topic of the present paper. Episode group. Patients undergoing screening or surveillance colonoscopy for colon cancer screening. Trigger rules. Part B claims (G0105, G0121) including the following with a PT modifier to signify screening intention (45378,45380, 45381, 45384, 45385). Subgroups.1.Hospital Outpatient Department2.Ambulatory Surgical Center3.Office Exclusion rules.1.Procedure performed in inpatient setting2.Diagnostic colonoscopy (non-G code without PT modifier)3.Colonoscopies where upper GI endoscopy performed in same setting4.Procedures with endoscopic mucosal resection (HCPCS code 45390)5.Patients with inflammatory bowel disease6.Death within episode window7.Primary payer other than Medicare within previous 120 days8.Not enrolled in Medicare Part A and B for previous 120 days9.Enrolled in Medicare Part C Attribution rules. TIN or TIN-NPI with Part B claim. Episode window. Day 0 (procedure day) to day 14 after procedure. Service grouping rules.1.Services billed in the following settings: Emergency Room, Inpatient care (including inpatient rehabilitation facilities), Outpatient care, and Long-term care.2.Services billed within episode window (varied depending on the clinical theme and claim)3.Each clinical code was vetted by the Subcommittee and was categorized in the following clinical themes: cardiopulmonary complications, postprocedure lower gastrointestinal bleeding, pathology services, perforation or peritonitis, repeated procedure, and uncategorized. Risk-adjustment rules. The risk-adjustment model incorporated patient age, reason for Medicare eligibility, and comorbid conditions variables from the CMS Hierarchical Condition Model.6Centers for Medicare and Medicaid ServicesValue-Based Payment Modifier.2017Google Scholar,7Evans M.A. Pope G.C. Kauttner J. et al.Evaluation of the CMS-HCC risk adjustment model.2011Google Scholar The Subcommittee also incorporated the following variables: anticoagulant use, presence of acute ischemic cerebrovascular disease, asthma/obstructive sleep apnea, history of anesthesia difficulties, valvular disease and hypertrophic cardiomyopathy, home oxygen or respiratory failure, poorly controlled hypertension, and pulmonary hypertension. The goal of episode development is to produce clinician performance scores that allow comparison of average episode costs between clinicians. These scores must incorporate risk-adjustment and also must account for differences in payments due to systematic differences such as geography. Here, we outline the procedure for taking the set of services assigned to an episode and computing clinician performance scores. Once the episodes have been defined (as the set of assigned services), there are four steps to calculate payment-standardized, risk-adjusted cost measure scores: (1) exclude episodes, (2) calculate standardized observed episode costs, (3) calculate expected episode costs through risk adjustment, and (4) calculate measure scores. Clinicians or group of clinicians (if group reporting) with fewer than 10 attributed episodes were excluded. First, claim payments are standardized to account for differences in Medicare payments for the same service(s) across Medicare providers. We eliminate variation in payments due to adjustments for geographic differences in wage levels or policy-driven payment adjustments with the use of the CMS payment standardization methodology.8Centers for Medicare and Medicaid ServicesMedicare Spending Per Beneficiary (MSPB) measure methodology.https://qualitynet.cms.gov/inpatient/measures/mspb/methodologyDate accessed: May 24, 2018Google Scholar We then calculate standardized observed episode costs by taking the total standardized amount paid to Medicare providers for all services assigned to each episode. This step calculates expected costs for each episode, as predicted through a risk adjustment model that uses a linear regression. To account for the limitations of risk adjustment, episodes predicted to have expected costs that are substantially different from observed costs are excluded as outliers. This step is performed separately for each subgroup. The linear regression model includes a common set of independent variables such as those from the CMS Hierarchical Condition Categories (HCCs) as well as variables that were recommended by the Clinical Subcommittee for this measure (as defined above). We dropped risk adjustors that were defined for fewer than 15 episodes nationally, including Inflammatory Bowel Disease (HCC 35), Cystic Fibrosis (HCC 110), Severe Head Injury (HCC 166), and Respiratory Arrest (HCC 83), across all subgroups, including certain variables within subgroups to avoid perfect estimation of cost for episodes with that variable. We categorized beneficiaries into age ranges by means of their date of birth information in the Medicare beneficiary enrollment database. If an age range had a cell count less than 15, we collapsed it with the next adjacent higher age range category. We used an ordinary least squares (OLS) regression to estimate the relationship between the independent variables and the dependent variable, standardized observed episode costs, and used a separate regression for each episode subgroup. To limit the effects of extreme values on expected costs, we truncated outliers through winsorization, a statistical transformation that limits extreme values. Winsorization of the lower end of the distribution (ie, bottom coding) involved setting extremely low predicted values below a predetermined limit to be equal to that predetermined limit. Specifically, for expected episode costs below the 0.5th percentile, we assigned the value of the 0.5th percentile. We then renormalized values by multiplying each episode's winsorized expected cost by the subgroup's average expected cost and dividing the result by the subgroup's average winsorized expected cost. This process ensures that the winsorized average is equal to the original average. We then excluded episodes with outlier residual values to maintain a consistent average episode cost level. We performed this step separately for each subgroup. After calculating each episode's residual as the difference between the renormalized winsorized expected cost and the observed cost, we excluded episodes with residuals below the 1st percentile or above the 99th percentile of the residual distribution. We renormalized the resultant expected cost values by multiplying by the subgroup's average observed cost and dividing by the subgroup's average. We produced the clinician's measure score by obtaining the average ratio of observed cost to expected episode cost across a provider's episodes, multiplied by the national average observed episode cost. We did this by using episodes from all subgroups. Mathematically, the risk-adjusted cost for clinician group practice j is:MeasureScorej=(1nj∑i∈{Ij}YijYijˆ)(1n∑j∑i∈{Ij}Yij)where: Yij is the attributed standardized payment for episode i and clinician (or clinician group practice) j. Yijˆ is the expected standardized payment for episode i and clinician (or clinician group practice) j, as predicted from risk adjustment. nj is the number of episodes for clinician (or clinician group practice) j. n is the total number of episodes nationally. i∈{Ij} is all episodes i in the set of episodes attributed to clinician (or clinician group practice) j. We identified important sources of cost variation for the Screening/Surveillance Colonoscopy Cost Measure from discussion with the Subcommittee members as well as from an extensive literature search. A limit of 5 clinical themes for each episode was set in preparation for the field test with the understanding that all relevant services may not be included within these themes. Once decided on, Acumen clinicians mapped service to clinical themes to include all relevant services for diagnosis, treatment, and aftercare. In addition, a number of other categories were reported in field test reports based on previous CMS categories of classification. Because these cost measures were risk adjusted on an episode level as opposed to an individual service level, the clinical themes and other categories were compared with clinicians in the same risk deciles, which represented similar patient case mixes. Calculation of clinical theme costs could not be directly risk adjusted for comparison due to risk adjustment on an episode level, because such clinical theme costs were compared with the national average as well as across risk-score deciles. The steps for calculating the clinical themes were to 1) sum total cost across Medicare Parts A and B for services within the clinical theme, 2) average total cost across number of episodes, 3) assess average risk score for the provider based on patient case mix, 4) calculate average cost for clinical theme across episodes within risk-score decile inclusive of the specific clinician's episodes, and 5) compare clinical theme cost of clinician with that of their risk-score decile. This methodology provides a suggestive comparison of costs for clinicians who treat similarly risky patient populations, but it does suffer from the drawback of possibly having different comorbidities leading to similar risk scores. For this paper, clinical themes were adjusted to more fully represent costs during the episode as well as to show the effect of the clinical themes on episodes where they were present.

Purpose The US Food and Drug Administration monitors the risk of Guillain-Barre syndrome (GBS) following influenza vaccination using several data sources including Medicare. In the 2017 to 2018 season, we transitioned our near real-time surveillance in Medicare to more effectively detect large GBS risk increases early in the season while avoiding false positives. Methods We conducted a simulation study examining the ability of the updating sequential probability ratio test (USPRT) to detect substantially elevated GBS risk in the 8- to 21-day postvaccination versus 5x to 30x the historical rate. We varied the first testing week (weeks 5-8) and the null rate (1x-3x) and evaluated power. We estimated signal probability and the risk ratio (RR) after signaling when high-risk seasons were rare. Results Applying fixed alternatives, we found >80% power to detect a risk 30x the historical rate in week 5 for the 1x null and in week 6 for the 1.5x to 3x nulls. Nearly all testing schedules had >80% power for a 5x risk by week 11. To test the robustness of USPRT, we further simulated seasons where 1% were true high-risk seasons. Using a 1x null led to 10% of seasons signaling by week 11 (median RR approximately 1.4), which decreased to approximately 1% with the >= 2.5x null (median RR approximately 16.0). Conclusions On the basis of the results from this simulation and subsequent consultations with experts and stakeholders, we specified USPRT to test continuously from weeks 7 to 11 using the null hypothesis that the observed GBS rate was 2.5x the historical rate. This helped improve the ability of USPRT to provide early detection of GBS risk following influenza vaccination as part of a multilayered system of surveillance.

BACKGROUND:Institutionalized adults are at increased risk of morbidity and mortality from influenza and pneumococcal infection. Influenza and pneumococcal vaccination have been shown to be effective in reducing hospitalization and deaths due to pneumonia and influenza in this population. OBJECTIVE:To assess trends in influenza vaccination coverage among US nursing home residents from the 2005-2006 through 2014-2015 influenza seasons and trends in pneumococcal vaccination coverage from 2006 to 2014 among US nursing home residents, by state and demographic characteristics. METHODS:Data were analyzed from the Centers for Medicare and Medicaid Services' (CMS's) Minimum Data Set (MDS). Influenza and pneumococcal vaccination status were assessed for all residents of CMS-certified nursing homes using data reported to the MDS by all certified facilities. RESULTS:Influenza vaccination coverage increased from 71.4% in the 2005-2006 influenza season to 75.7% in the 2014-2015 influenza season and pneumococcal vaccination coverage increased from 67.4% in 2006 to 78.4% in 2014. Vaccination coverage varied by state, with influenza vaccination coverage ranging from 50.0% to 89.7% in the 2014-2015 influenza season and pneumococcal vaccination coverage ranging from 55.0% to 89.7% in 2014. Non-Hispanic black and Hispanic residents had lower coverage compared with non-Hispanic white residents for both vaccines, and these differences persisted over time. CONCLUSION:Influenza and pneumococcal vaccination among US nursing home residents remains suboptimal. Nursing home staff can employ strategies such as provider reminders and standing orders to facilitate offering vaccination to all residents along with culturally appropriate vaccine promotion to increase vaccination coverage among this vulnerable population.

Background:Tens of millions of seniors are at risk of herpes zoster (HZ) and its complications. Live attenuated herpes zoster vaccine (HZV) reduces that risk, although questions regarding effectiveness and durability of protection in routine clinical practice remain. We used Medicare data to investigate HZV effectiveness (VE) and its durability.Methods:This retrospective cohort study included beneficiaries ages ≥65 years during January 2007 through July 2014. Multiple adjustments to account for potential bias were made. HZV-vaccinated beneficiaries were matched to unvaccinated beneficiaries (primary analysis) and to HZV-unvaccinated beneficiaries who had received pneumococcal vaccination (secondary analysis). HZ outcomes in community and hospital settings were analyzed, including ophthalmic zoster (OZ) and postherpetic neuralgia (PHN).Results:Among eligible beneficiaries (average age 77 years), the primary analysis found VE for community HZ of 33% (95% CI: 32%-35%) and 19% (95% CI: 17%-22%), for the first 3, and subsequent 4+ years postvaccination, respectively. In the secondary analysis, VE was, respectively, 37% (95% CI: 36%-39%) and 22% (95% CI: 20%-25%). In the primary analysis, VE for PHN was 57% (95% CI: 52%-61%) and 45% (95% CI: 36%-53%) in the first 3 and subsequent 4+ years, respectively; VE for hospitalized HZ was, respectively, 74% (95% CI: 67%-79%) and 55% (95% CI: 39%-67%). Differences in VE by age group were not significant.Conclusions:In both the primary and secondary analyses, HZV provided protection against HZ across all ages, but effectiveness declined over time. VE was higher and better preserved over time for PHN and HZ-associated hospitalizations than for community HZ.