Inappropriate Diagnosis of Pneumonia Among Hospitalized Adults

Importance

Little is known about incidence of, risk factors for, and harms associated with inappropriate diagnosis of community-acquired pneumonia (CAP).

Objective

To characterize inappropriate diagnosis of CAP in hospitalized patients.

Design, Setting, and Participants

This prospective cohort study, including medical record review and patient telephone calls, took place across 48 Michigan hospitals. Trained abstractors retrospectively assessed hospitalized patients treated for CAP between July 1, 2017, and March 31, 2020. Patients were eligible for inclusion if they were adults admitted to general care with a discharge diagnostic code of pneumonia who received antibiotics on day 1 or 2 of hospitalization. Data were analyzed from February to December 2023.

Main Outcomes and Measures

Inappropriate diagnosis of CAP was defined using a National Quality Forum-endorsed metric as CAP-directed antibiotic therapy in patients with fewer than 2 signs or symptoms of CAP or negative chest imaging. Risk factors for inappropriate diagnosis were assessed and, for those inappropriately diagnosed, 30-day composite outcomes (mortality, readmission, emergency department visit, Clostridioides difficile infection, and antibiotic-associated adverse events) were documented and stratified by full course (>3 days) vs brief (≤3 days) antibiotic treatment using generalized estimating equation models adjusting for confounders and propensity for treatment.

Results

Of the 17 290 hospitalized patients treated for CAP, 2079 (12.0%) met criteria for inappropriate diagnosis (median [IQR] age, 71.8 [60.1-82.8] years; 1045 [50.3%] female), of whom 1821 (87.6%) received full antibiotic courses. Compared with patients with CAP, patients inappropriately diagnosed were older (adjusted odds ratio [AOR], 1.08; 95% CI, 1.05-1.11 per decade) and more likely to have dementia (AOR, 1.79; 95% CI, 1.55-2.08) or altered mental status on presentation (AOR, 1.75; 95% CI, 1.39-2.19). Among those inappropriately diagnosed, 30-day composite outcomes for full vs brief treatment did not differ (25.8% vs 25.6%; AOR, 0.98; 95% CI, 0.79-1.23). Full vs brief duration of antibiotic treatment among patients was associated with antibiotic-associated adverse events (31 of 1821 [2.1%] vs 1 of 258 [0.4%]; P = .03).

Conclusions and Relevance

In this cohort study, inappropriate diagnosis of CAP among hospitalized adults was common, particularly among older adults, those with dementia, and those presenting with altered mental status. Full-course antibiotic treatment of those inappropriately diagnosed with CAP may be harmful.

Germline Genetic Testing After Cancer Diagnosis

Importance

Germline genetic testing is recommended by practice guidelines for patients diagnosed with cancer to enable genetically targeted treatment and identify relatives who may benefit from personalized cancer screening and prevention.

Objective

To describe the prevalence of germline genetic testing among patients diagnosed with cancer in California and Georgia between 2013 and 2019.

Design, Setting, and Participants

Observational study including patients aged 20 years or older who had been diagnosed with any type of cancer between January 1, 2013, and March 31, 2019, that was reported to statewide Surveillance, Epidemiology, and End Results registries in California and Georgia. These patients were linked to genetic testing results from 4 laboratories that performed most germline testing for California and Georgia.

Main Outcomes and Measures

The primary outcome was germline genetic testing within 2 years of a cancer diagnosis. Testing trends were analyzed with logistic regression modeling. The results of sequencing each gene, including variants associated with increased cancer risk (pathogenic results) and variants whose cancer risk association was unknown (uncertain results), were evaluated. The genes were categorized according to their primary cancer association, including breast or ovarian, gastrointestinal, and other, and whether practice guidelines recommended germline testing.

Results

Among 1 369 602 patients diagnosed with cancer between 2013 and 2019 in California and Georgia, 93 052 (6.8%) underwent germline testing through March 31, 2021. The proportion of patients tested varied by cancer type: male breast (50%), ovarian (38.6%), female breast (26%), multiple (7.5%), endometrial (6.4%), pancreatic (5.6%), colorectal (5.6%), prostate (1.1%), and lung (0.3%). In a logistic regression model, compared with the 31% (95% CI, 30%-31%) of non-Hispanic White patients with male breast cancer, female breast cancer, or ovarian cancer who underwent testing, patients of other races and ethnicities underwent testing less often: 22% (95% CI, 21%-22%) of Asian patients, 25% (95% CI, 24%-25%) of Black patients, and 23% (95% CI, 23%-23%) of Hispanic patients (P < .001 using the χ2 test). Of all pathogenic results, 67.5% to 94.9% of variants were identified in genes for which practice guidelines recommend testing and 68.3% to 83.8% of variants were identified in genes associated with the diagnosed cancer type.

Conclusions and Relevance

Among patients diagnosed with cancer in California and Georgia between 2013 and 2019, only 6.8% underwent germline genetic testing. Compared with non-Hispanic White patients, rates of testing were lower among Asian, Black, and Hispanic patients.

Adopting Weight-Based Dosing With Pharmacy-Level Stewardship Strategies Could Reduce Cancer Drug Spending By Millions

Immune checkpoint inhibitors, a class of drugs used in approximately forty unique cancer indications, are a sizable component of the economic burden of cancer care in the US. Instead of personalized weight-based dosing, immune checkpoint inhibitors are most commonly administered at "one-size-fits-all" flat doses that are higher than necessary for the vast majority of patients. We hypothesized that personalized weight-based dosing along with common stewardship efforts at the pharmacy level, such as dose rounding and vial sharing, would lead to reductions in immune checkpoint inhibitor use and lower spending. Using data from the Veterans Health Administration (VHA) and Medicare drug prices, we estimated reductions in immune checkpoint inhibitor use and spending that would be associated with pharmacy-level stewardship strategies, in a case-control simulation study of individual patient-level immune checkpoint inhibitor administration events. We identified baseline annual VHA spending for these drugs of approximately $537 million. Combining weight-based dosing, dose rounding, and pharmacy-level vial sharing would generate expected annual VHA health system savings of $74 million (13.7 percent). We conclude that adoption of pharmacologically justified immune checkpoint inhibitor stewardship measures would generate sizable reductions in spending for these drugs. Combining these operational innovations with value-based drug price negotiation enabled by recent policy changes may improve the long-term financial viability of cancer care in the US.

Distinct health care use patterns of patients with chronic gastrointestinal diseases

OBJECTIVES

Patients with complex chronic conditions have varying multidisciplinary care needs and utilization patterns, which limit the effectiveness of initiatives designed to improve continuity of care (COC) and reduce utilization. Our objective was to categorize patients with complex chronic conditions into distinct groups by pattern of outpatient care use and COC to tailor interventions.

STUDY DESIGN

Observational cohort study from 2014 to 2015.

METHODS

We identified patients whose 1-year hospitalization risk was in at least the 90th percentile in 2014 who had a chronic gastrointestinal disease (cirrhosis, inflammatory bowel disease, chronic pancreatitis) as case examples of complex chronic disease. We described frequency of office visits, number of outpatient providers, and 2 COC measures (usual provider of care, Bice-Boxerman COC indices) over 12 months. We used latent profile analysis, a statistical method for identifying distinct subgroups, to categorize patients based on overall, primary care, gastroenterology, and mental health continuity patterns.

RESULTS

The 26,751 veterans in the cohort had a mean (SD) of 13.3 (8.6) office visits and 7.2 (3.8) providers in 2014. Patients were classified into 5 subgroups: (1) high gastroenterology-specific COC with mental health use; (2) high gastroenterology-specific COC without mental health use; (3) high overall utilization with mental health use; (4) low overall COC with mental health use; and (5) low overall COC without mental health use. These groups varied in their sociodemographic characteristics and risk for hospitalization, emergency department use, and mortality.

CONCLUSIONS

Patients at high risk for health care utilization with specialty care needs can be grouped by varying propensity for health care continuity patterns.

Adoption of Extended-Interval Dosing of Single-Agent Pembrolizumab and Comparative Effectiveness vs Standard Dosing in Time-to-Treatment Discontinuation

Importance

Extended-interval dosing of pembrolizumab (400 mg every 6 weeks) was approved by US Food and Drug Administration (FDA) in April 2020 as an alternative to standard-interval dosing (200 mg every 3 weeks). Extended-interval dosing may enhance access, alleviate patient and health system financial toxicity, and improve patient quality of life, particularly during the COVID-19 pandemic. Neither adoption nor effectiveness of extended interval in the US has been adequately described.

Objective

To describe adoption of extended-interval dosing of pembrolizumab since its FDA approval and to measure its preliminary real-world effectiveness compared with standard-interval dosing.

Design, Setting, and Participants

This was a retrospective cohort study that used data from the Veterans Health Administration (VHA), a US-based, nationwide single-payer health system. Participants were veterans who were prescribed single-agent pembrolizumab within the VHA between April 1, 2020, and July 1, 2021. Patients receiving combinations of pembrolizumab and cytotoxic chemotherapy or tyrosine kinase inhibitors were excluded. A subcohort of veterans with non-small cell lung cancer (NSCLC) was also identified using claims-based codes.

Exposures

Single-agent pembrolizumab at extended or standard intervals.

Main Outcomes and Measures

The number and proportion of single-agent pembrolizumab prescriptions that were extended compared with standard interval. Effectiveness was described in terms of time-to-treatment discontinuation (TTD) and extended- to standard-interval pembrolizumab prescriptions were compared using Cox proportional hazards regression.

Results

A total of 835 veterans (mean age [SD], 70.9 [8.7] years; 809 [96.9%] men) began single-agent pembrolizumab during the study period (all-diseases cohort), and of these, 234 (mean [SD] age, 71.6 [7.3] years; 225 [96.2%] men) had NSCLC (NSCLC cohort). Extended-interval adoption reached its steady state plateau of approximately 35% by January 2021; 65% of participants who began standard-interval single-agent pembrolizumab received only standard-interval dosing during the treatment course. In analysis consistent with the intention-to-treat principle, no differences in TTD were observed between standard- and extended-interval dosing in either the all-diseases cohort (HR, 1.00; 95% CI, 1.00-1.00) or the NSCLC cohort (HR, 1.00; 95% CI, 1.00-1.00).

Conclusions and Relevance

This retrospective cohort study found that extended-interval dosing comprised a minority of single-agent pembrolizumab prescriptions despite the FDA approval and its potential health system and public health benefits. The findings support the TTD equivalence of standard- and extended-interval pembrolizumab across indications, complementing clinical pharmacology and single-arm clinical trial data in melanoma. This study provides further support for extended-interval pembrolizumab dosing.

Planning and Reporting Effective Web-Based RAND/UCLA Appropriateness Method Panels: Literature Review and Preliminary Recommendations

BACKGROUND

The RAND/UCLA Appropriateness Method (RAM), a variant of the Delphi Method, was developed to synthesize existing evidence and elicit the clinical judgement of medical experts on the appropriate treatment of specific clinical presentations. Technological advances now allow researchers to conduct expert panels on the internet, offering a cost-effective and convenient alternative to the traditional RAM. For example, the Department of Veterans Affairs recently used a web-based RAM to validate clinical recommendations for de-intensifying routine primary care services. A substantial literature describes and tests various aspects of the traditional RAM in health research; yet we know comparatively less about how researchers implement web-based expert panels.

OBJECTIVE

The objectives of this study are twofold: (1) to understand how the web-based RAM process is currently used and reported in health research and (2) to provide preliminary reporting guidance for researchers to improve the transparency and reproducibility of reporting practices.

METHODS

The PubMed database was searched to identify studies published between 2009 and 2019 that used a web-based RAM to measure the appropriateness of medical care. Methodological data from each article were abstracted. The following categories were assessed: composition and characteristics of the web-based expert panels, characteristics of panel procedures, results, and panel satisfaction and engagement.

RESULTS

Of the 12 studies meeting the eligibility criteria and reviewed, only 42% (5/12) implemented the full RAM process with the remaining studies opting for a partial approach. Among those studies reporting, the median number of participants at first rating was 42. While 92% (11/12) of studies involved clinicians, 50% (6/12) involved multiple stakeholder types. Our review revealed that the studies failed to report on critical aspects of the RAM process. For example, no studies reported response rates with the denominator of previous rounds, 42% (5/12) did not provide panelists with feedback between rating periods, 50% (6/12) either did not have or did not report on the panel discussion period, and 25% (3/12) did not report on quality measures to assess aspects of the panel process (eg, satisfaction with the process).

CONCLUSIONS

Conducting web-based RAM panels will continue to be an appealing option for researchers seeking a safe, efficient, and democratic process of expert agreement. Our literature review uncovered inconsistent reporting frameworks and insufficient detail to evaluate study outcomes. We provide preliminary recommendations for reporting that are both timely and important for producing replicable, high-quality findings. The need for reporting standards is especially critical given that more people may prefer to participate in web-based rather than in-person panels due to the ongoing COVID-19 pandemic.

Interpretability, credibility, and usability of hospital-specific template matching versus regression-based hospital performance assessments; a multiple methods study

BACKGROUND

Hospital-specific template matching (HS-TM) is a newer method of hospital performance assessment.

OBJECTIVE

To assess the interpretability, credibility, and usability of HS-TM-based vs. regression-based performance assessments.

RESEARCH DESIGN

We surveyed hospital leaders (January-May 2021) and completed follow-up semi-structured interviews. Surveys included four hypothetical performance assessment vignettes, with method (HS-TM, regression) and hospital mortality randomized.

SUBJECTS

Nationwide Veterans Affairs Chiefs of Staff, Medicine, and Hospital Medicine.

MEASURES

Correct interpretation; self-rated confidence in interpretation; and self-rated trust in assessment (via survey). Concerns about credibility and main uses (via thematic analysis of interview transcripts).

RESULTS

In total, 84 participants completed 295 survey vignettes. Respondents correctly interpreted 81.8% HS-TM vs. 56.5% regression assessments, p < 0.001. Respondents "trusted the results" for 70.9% HS-TM vs. 58.2% regression assessments, p = 0.03. Nine concerns about credibility were identified: inadequate capture of case-mix and/or illness severity; inability to account for specialized programs (e.g., transplant center); comparison to geographically disparate hospitals; equating mortality with quality; lack of criterion standards; low power; comparison to dissimilar hospitals; generation of rankings; and lack of transparency. Five concerns were equally relevant to both methods, one more pertinent to HS-TM, and three more pertinent to regression. Assessments were mainly used to trigger further quality evaluation (a "check oil light") and motivate behavior change.

CONCLUSIONS

HS-TM-based performance assessments were more interpretable and more credible to VA hospital leaders than regression-based assessments. However, leaders had a similar set of concerns related to credibility for both methods and felt both were best used as a screen for further evaluation.

Evaluating alternative methods of comparing antihypertensive treatment intensity

OBJECTIVES

To change blood pressure treatment, clinicians can modify medication count or dose. However, existing studies have measured count modification, which may miss clinically important dose change in the absence of count change. This research demonstrates how dose modification captures more information about management than medication count alone.

STUDY DESIGN

Retrospective cohort study.

METHODS

We included patients 65 years and older with established primary care at the Veterans Health Administration (July 2011-June 2013). We captured medication count and standardized dose change over 90 to 120 days using a validated pharmacy fill algorithm. We determined frequency of dose change without count change (and vice versa), no change in either, change in same direction ("concordant"), and change in opposite direction ("discordant"). We compared change according to systolic blood pressure (SBP) and compared concordance using a minimum threshold definition of dose change of at least 50% (instead of any change) of baseline dose modification.

RESULTS

Among 440,801 patients, 64.2% had dose change; 22.0%, count change; 35.6%, no change in either; 42.4%, dose change without count modification; and 0.2%, count change without dose modification. Discordance occurred in 2.1% of observations. Using the minimum threshold definition of change, 68.7% had no change in either dose or count. Treatment was more frequently changed at SBP greater than 140 mm Hg.

CONCLUSIONS

Measuring change in antihypertensive treatment using medication count frequently missed an isolated dose change in treatment modification and less often misclassified regimen modifications where there was no modification in total dose. In future research, measuring dose modification using our new algorithm would capture change in hypertension treatment intensity more precisely than current methods.

Type 2 Diabetes Management, Control and Outcomes During the COVID-19 Pandemic in Older US Veterans: an Observational Study

BACKGROUND

The COVID-19 pandemic required a change in outpatient care delivery models, including shifting from in-person to virtual visits, which may have impacted care of vulnerable patients.

OBJECTIVE

To describe the changes in management, control, and outcomes in older people with type 2 diabetes (T2D) associated with the shift from in-person to virtual visits.

DESIGN AND PARTICIPANTS

In veterans aged ≥ 65 years with T2D, we assessed the rates of visits (in person, virtual), A1c measurements, antidiabetic deintensification/intensification, ER visits and hospitalizations (for hypoglycemia, hyperglycemia, other causes), and A1c level, in March 2020 and April-November 2020 (pandemic period). We used negative binomial regression to assess change over time (reference: pre-pandemic period, July 2018 to February 2020), by baseline Charlson Comorbidity Index (CCI; > 2 vs. <= 2) and A1c level.

KEY RESULTS

Among 740,602 veterans (mean age 74.2 [SD 6.6] years), there were 55% (95% CI 52-58%) fewer in-person visits, 821% (95% CI 793-856%) more virtual visits, 6% (95% CI 1-11%) fewer A1c measurements, and 14% (95% CI 10-17%) more treatment intensification during the pandemic, relative to baseline. Patients with CCI > 2 had a 14% (95% CI 12-16%) smaller relative increase in virtual visits than those with CCI <= 2. We observed a seasonality of A1c level and treatment modification, but no association of either with the pandemic. After a decrease at the beginning of the pandemic, there was a rebound in other-cause (but not hypo- and hyperglycemia-related) ER visits and hospitalizations from June to November 2020.

CONCLUSION

Despite a shift to virtual visits and a decrease in A1c measurement during the pandemic, we observed no association with A1c level or short-term T2D-related outcomes, providing some reassurance about the adequacy of virtual visits. Further studies should assess the longer-term effects of shifting to virtual visits in different populations to help individualize care, improve efficiency, and maintain appropriate care while reducing overuse.

Association of Family Cancer History With Pathogenic Variants in Specific Breast Cancer Susceptibility Genes

PURPOSE

Family cancer history is an important component of genetic testing guidelines that estimate which patients with breast cancer are most likely to carry a germline pathogenic variant (PV). However, we do not know whether more extensive family history is differentially associated with PVs in specific genes.

METHODS

All women diagnosed with breast cancer in 2013-2017 and reported to statewide SEER registries of Georgia and California were linked to clinical genetic testing results and family history from two laboratories. Family history was defined as strong (suggestive of PVs in high-penetrance genes such as BRCA1/2 or TP53, including male breast, ovarian, pancreatic, sarcoma, or multiple female breast cancers), moderate (any other cancer history), or none. Among established breast cancer susceptibility genes (ATM, BARD1, BRCA1, BRCA2, CDH1, CHEK2, NF1, PALB2, PTEN, RAD51C, RAD51D, and TP53), we evaluated PV prevalence according to family history extent and breast cancer subtype. We used a multivariable model to test for interaction between affected gene and family history extent for ATM, BRCA1/2, CHEK2, and PALB2.

RESULTS

A total of 34,865 women linked to genetic results. Higher PV prevalence with increasing family history extent (P < .001) was observed only with BRCA1 (3.04% with none, 3.22% with moderate, and 4.06% with strong history) and in triple-negative breast cancer with PALB2 (0.75% with none, 2.23% with moderate, and 2.63% with strong history). In a multivariable model adjusted for age and subtype, there was no interaction between family history extent and PV prevalence for any gene except PALB2 (P = .037).

CONCLUSION

Extent of family cancer history is not differentially associated with PVs across established breast cancer susceptibility genes and cannot be used to personalize genes selected for testing.

Hospital-specific Template Matching for Benchmarking Performance in a Diverse Multihospital System

BACKGROUND

Hospital-specific template matching is a newer method of hospital performance measurement that may be fairer than regression-based benchmarking. However, it has been tested in only limited research settings.

OBJECTIVE

The objective of this study was to test the feasibility of hospital-specific template matching assessments in the Veterans Affairs (VA) health care system and determine power to detect greater-than-expected 30-day mortality.

RESEARCH DESIGN

Observational cohort study with hospital-specific template matching assessment. For each VA hospital, the 30-day mortality of a representative subset of hospitalizations was compared with the pooled mortality from matched hospitalizations at a set of comparison VA hospitals treating sufficiently similar patients. The simulation was used to determine power to detect greater-than-expected mortality.

SUBJECTS

A total of 556,266 hospitalizations at 122 VA hospitals in 2017.

MEASURES

A number of comparison hospitals identified per hospital; 30-day mortality.

RESULTS

Each hospital had a median of 38 comparison hospitals (interquartile range: 33, 44) identified, and 116 (95.1%) had at least 20 comparison hospitals. In total, 8 hospitals (6.6%) had a significantly lower 30-day mortality than their benchmark, 5 hospitals (4.1%) had a significantly higher 30-day mortality, and the remaining 109 hospitals (89.3%) were similar to their benchmark. Power to detect a standardized mortality ratio of 2.0 ranged from 72.5% to 79.4% for a hospital with the fewest (6) versus most (64) comparison hospitals.

CONCLUSIONS

Hospital-specific template matching may be feasible for assessing hospital performance in the diverse VA health care system, but further refinements are needed to optimize the approach before operational use. Our findings are likely applicable to other large and diverse multihospital systems.

Advancing Patient Safety Through the Clinical Application of a Framework Focused on Communication

OBJECTIVES

The study of communication has evolved from diverse academic disciplines, yet those diverse fields are not well represented in theoretical frameworks that describe communication in health care, narrowing our ability to explain how communication affects patient safety. The purpose of this review article is to describe a conceptual framework of communication drawn from multiple academic disciplines and apply it to health care, specifically for examining communication between providers about the clinical care of their patients.

METHODS

A seminal article in the field of communication that attempted to map the entire field of communication theory inspired our conceptual framework. We adapted these concepts, largely from the social science literature, to find alternative ways of conceptualizing communication and ways to enhance communication in health care.

RESULTS

There are 8 theoretical traditions that informed our conceptual framework: rhetorical, phenomenological, semiotic, cybernetic, sociopsychological, sociocultural, critical, and pragmatic. We provide practical, clinical applications of our conceptual framework, encompassing the interpersonal nature of communication, relationship building and trust, hierarchical differences, and the role of technology in communication. In adopting our conceptual framework, we suggest that researchers and clinicians can choose from any combination of these 8 theoretical traditions to more fully describe and ultimately enhance communication-related phenomena.

CONCLUSIONS

Poor communication remains a stubborn problem in health care in part because of a narrow theoretical and definitional approach to resolving it. Our conceptual framework suggests ways to build relationships and trust, addresses hierarchical differences between communicators, and illuminates the role of technology in communication. It also importantly expands the definition of the value of communication beyond simple information exchange to include creation of new knowledge during communication through the development of shared understanding.

Adding a New Medication Versus Maximizing Dose to Intensify Hypertension Treatment in Older Adults : A Retrospective Observational Study

BACKGROUND

There are 2 approaches to intensifying antihypertensive treatment when target blood pressure is not reached, adding a new medication and maximizing dose. Which strategy is better is unknown.

OBJECTIVE

To assess the frequency of intensification by adding a new medication versus maximizing dose, as well as the association of each method with intensification sustainability and follow-up systolic blood pressure (SBP).

DESIGN

Large-scale, population-based, retrospective cohort study. Observational data were used to emulate a target trial with 2 groups, new medication and maximizing dose, who underwent intensification of their drug regimen.

SETTING

Veterans Health Administration (2011 to 2013).

PATIENTS

Veterans aged 65 years or older with hypertension, an SBP of 130 mm Hg or higher, and at least 1 antihypertensive medication at less than the maximum dose.

MEASUREMENTS

The following 2 intensification approaches were emulated: adding a new medication, defined as a total dose increase with new medication, and maximizing dose, defined as a total dose increase without new medication. Inverse probability weighting was used to assess the observational effectiveness of the intensification approach on sustainability of intensified treatment and follow-up SBP at 3 and 12 months.

RESULTS

Among 178 562 patients, 45 575 (25.5%) had intensification by adding a new medication and 132 987 (74.5%) by maximizing dose. Compared with maximizing dose, adding a new medication was associated with less intensification sustainability (average treatment effect, -15.2% [95% CI, -15.7% to -14.6%] at 3 months and -15.1% [CI, -15.6% to -14.5%] at 12 months) but a slightly larger reduction in mean SBP (-0.8 mm Hg [CI, -1.2 to -0.4 mm Hg] at 3 months and -1.1 mm Hg [CI, -1.6 to -0.6 mm Hg] at 12 months).

LIMITATION

Observational data; largely male population.

CONCLUSION

Adding a new antihypertensive medication was less frequent and was associated with less intensification sustainability but slightly larger reductions in SBP. Trials would provide the most definitive support for our findings.

PRIMARY FUNDING SOURCE

National Institute on Aging and Veterans Health Administration.

A Statewide Collaborative Quality Initiative To Improve Antibiotic Duration And Outcomes Of Patients Hospitalized With Uncomplicated Community-Acquired Pneumonia

Background

Community-acquired pneumonia (CAP) is a common cause for hospitalization and antibiotic overuse. We aimed to improve antibiotic duration for CAP across 41 hospitals participating in the Michigan Hospital Medicine Safety Consortium (HMS).

Methods

This prospective collaborative quality initiative included patients hospitalized with uncomplicated CAP who qualified for a 5-day antibiotic duration. Between 23 February 2017 and 5 February 2020, HMS targeted appropriate 5-day antibiotic treatment through benchmarking, sharing best practices, and pay-for-performance incentives. Changes in outcomes, including appropriate receipt of 5 ± 1–day antibiotic treatment and 30-day postdischarge composite adverse events (ie, deaths, readmissions, urgent visits, and antibiotic-associated adverse events), were assessed over time (per 3-month quarter), using logistic regression and controlling for hospital clustering.

Results

A total of 41 hospitals and 6553 patients were included. The percentage of patients treated with an appropriate 5 ± 1–day duration increased from 22.1% (predicted probability, 20.9% [95% confidence interval: 17.2%–25.0%]) to 45.9% (predicted probability, 43.9% [36.8%–51.2%]; adjusted odds ratio [aOR] per quarter, 1.10 [1.07–1.14]). Thirty-day composite adverse events occurred in 18.5% of patients (1166 of 6319) and decreased over time (aOR per quarter, 0.98 [95% confidence interval: .96–.99]) owing to a decrease in antibiotic-associated adverse events (aOR per quarter, 0.91 [.87–.95]).

Conclusions

Across diverse hospitals, HMS participation was associated with more appropriate use of short-course therapy and fewer adverse events in hospitalized patients with uncomplicated CAP. Establishment of national or regional collaborative quality initiatives with data collection and benchmarking, sharing of best practices, and pay-for-performance incentives may improve antibiotic use and outcomes for patients hospitalized with uncomplicated CAP.

Variation in Provider Connectedness Associates With Outcomes of Inflammatory Bowel Diseases in an Analysis of Data From a National Health System

BACKGROUND & AIMS

Inflammatory bowel diseases (IBD) often require multidisciplinary care with tight coordination among providers. Provider connectedness, a measure of the relationship among providers, is an important aspect of care coordination that has been linked to higher quality care. We aimed to assess variation in provider connectedness among medical centers, and to understand the association between this established measure of care coordination and outcomes of patients with IBD.

METHODS

We conducted a national cohort study of 32,949 IBD patients with IBD from 2005 to 2014. We used network analysis to examine provider connectedness, defined using network properties that measure the strength of the collaborative relationship, team cohesiveness, and between-facility collaborations. We used multilevel modeling to examine variations in provider connectedness and association with patient outcomes.

RESULTS

There was wide variation in provider connectedness among facilities in complexity, rural designation, and volume of patients with IBD. In a multivariable model, patients followed in a facility with team cohesiveness (odds ratio, 0.38; 95% CI, 0.16-0.88) and where providers often collaborated with providers outside their facility (odds ratio, 0.48; 95% CI, 0.31-0.75) were less likely to have clinically active disease, defined by a composite of outpatient flare, inpatient flare, and IBD-related surgery.

CONCLUSIONS

A national study found evidence for heterogeneity in patient-sharing among IBD care teams. Patients with IBD seen at health centers with higher provider connectedness appear to have better outcomes. Understanding provider connectedness is a step toward designing network-based interventions to improve coordination and quality of care.

Agreement among measures examining low-value imaging for low back pain

OBJECTIVES

To quantify the extent of patient-level agreement among 3 published measures of low-value imaging for acute low back pain (LBP).

STUDY DESIGN

In this retrospective cohort study using commercial insurance claims from MarketScan, we assessed 3 published measures of low-value imaging for agreement in identifying LBP diagnoses (denominator), red-flag diagnoses (denominator exclusions), and imaging procedures (numerator).

METHODS

Using a cohort of patients, aged 18 to 64 years, with a diagnosis of LBP in 2014, we assessed agreement surrounding both the overuse event (imaging procedures) and inclusion in the reference population (LBP definition and exclusion diagnoses) using percent agreement and Fleiss κ among 3 overuse measures.

RESULTS

In our cohort of 1,835,620 patients with acute LBP, the 3 measures agreed 100% on the presence of acute LBP and also had excellent agreement (99%; κ = 0.98) in identifying imaging for LBP. However, there was substantial disagreement on whom to exclude for red-flag diagnoses, leading to lower agreement (75%; κ = 0.61) on whom to include in the reference population of acute LBP without red flags, among whom imaging for LBP is considered of low value.

CONCLUSIONS

Our findings demonstrate the need for further consensus surrounding how to translate guideline recommendations to administrative measures that assess overuse of imaging for acute LBP, particularly with respect to defining which patients should be excluded from the measures. This finding is also important for other overuse measures that rely on exclusions.

Clinical outcomes of modifying hypertension treatment intensity in older adults treated to low blood pressure

BACKGROUND/OBJECTIVES

Hypertension treatment reduces cardiovascular events. However, uncertainty remains about benefits and harms of deintensification or further intensification of antihypertensive medication when systolic blood pressure (SBP) is tightly controlled in older multimorbid patients, because of their frequent exclusion in trials. We assessed the association of hypertension treatment deintensification or intensification with clinical outcomes in older adults with tightly controlled SBP.

DESIGN

Longitudinal cohort study (2011-2013) with 9-month follow-up.

SETTING

U.S.-nationwide primary care Veterans Health Administration healthcare system.

PARTICIPANTS

Veterans aged 65 and older with baseline SBP <130 mmHg and ≥1 antihypertensive medication during ≥2 consecutive visits (N = 228,753).

EXPOSURE

Deintensification or intensification, compared with stable treatment.

MAIN OUTCOMES AND MEASURES

Cardiovascular events, syncope, or fall injury, as composite and distinct outcomes, within 9 months after exposure. Adjusted logistic regression and inverse probability of treatment weighting (IPTW, sensitivity analysis).

RESULTS

Among 228,753 patients (mean age 75 [SD 7.5] years), the composite outcome occurred in 11,982/93,793 (12.8%) patients with stable treatment, 14,768/72,672 (20.3%) with deintensification, and 11,821/62,288 (19.0%) with intensification. Adjusted absolute outcome risk (95% confidence interval) was higher for deintensification (18.3% [18.1%-18.6%]) and intensification (18.7% [18.4%-19.0%]), compared with stable treatment (14.8% [14.6%-15.0%]), p < 0.001 for both effects in the multivariable model). Deintensification was associated with fewer cardiovascular events than intensification. At baseline SBP <95 mmHg, cardiovascular event risk was similar for deintensification and stable treatment, and fall risk lower for deintensification than intensification. IPTW yielded similar results. Mean follow-up SBP was 124.1 mmHg for stable treatment, 125.1 mmHg after deintensification (p < 0.001), and 124.0 mmHg after intensification (p < 0.001).

CONCLUSION

Antihypertensive treatment deintensification in older patients with tightly controlled SBP was associated with worse outcomes than continuing same treatment intensity. Given higher mortality among patients with treatment modification, confounding by indication may not have been fully corrected by advanced statistical methods for observational data analysis.

Association of Genetic Testing Results with Mortality Among Women with Breast Cancer or Ovarian Cancer

BACKGROUND

Breast cancer and ovarian cancer patients increasingly undergo germline genetic testing. However, little is known about cancer-specific mortality among carriers of a pathogenic variant (PV) in BRCA1/2 or other genes in a population-based setting.

METHODS

Georgia and California Surveillance Epidemiology and End Results (SEER) registry records were linked to clinical genetic testing results. Women were included who had stages I-IV breast cancer or ovarian cancer diagnosed in 2013-2017; received chemotherapy; and linked to genetic testing results. Multivariable Cox proportional hazard models were used to examine the association of genetic results with cancer-specific mortality.

RESULTS

22,495 breast and 4,320 ovarian cancer patients were analyzed, with a median follow-up of 41 months. PVs were present in 12.7% of breast cancer patients with estrogen and/or progesterone receptor-positive, HER2-negative cancer, 9.8% with HER2-positive cancer, 16.8% with triple-negative breast cancer and 17.2% with ovarian cancer. Among triple-negative breast cancer patients, cancer-specific mortality was lower with BRCA1 (hazard ratio [HR] = 0.49, 95% confidence interval [CI] = 0.35-0.69) and BRCA2 PVs (HR = 0.60, 95% CI = 0.41-0.89), and equivalent with PVs in other genes (HR = 0.65, 95% CI = 0.37-1.13), versus non-carriers. Among ovarian cancer patients, cancer-specific mortality was lower with PVs in BRCA2 (HR = 0.35, 95% CI = 0.25-0.49) and genes other than BRCA1/2 (HR = 0.47, 95% CI = 0.32-0.69). No PV was associated with higher cancer-specific mortality.

CONCLUSIONS

Among breast cancer and ovarian cancer patients treated with chemotherapy in the community, BRCA1/2 and other gene PV carriers had equivalent or lower short-term cancer-specific mortality than non-carriers. These results may reassure newly diagnosed patients and longer follow-up is ongoing.

Factors associated with antihypertensive treatment intensification and deintensification in older outpatients

Background

New hypertension performance measures encourage more intensive treatment in older adults. Treatment intensification includes starting new medications and increasing the dose of old ones. Medication dose is particularly important to older adults, given their vulnerability to dose-related side effects. We previously validated a standardized measure of beneficial doses tested in hypertension trials, Hypertension Daily Dose (HDD).

Aim of the study

To test whether changes in treatment intensity using HDD was associated with systolic blood pressure (SBP) and patient characteristics.

Methods

Longitudinal study of all Veterans aged ≥65 years with a diagnosis of hypertension. We defined 3 groups of risk: 1) cardiovascular risk; 2) geriatric/frail; 3) low-risk (comparator). Using multinomial regression, we assessed the probability of deintensification, intensification, vs. stable treatment, according to SBP and group.

Results

Among 1,331,111 Veterans, 19.9% had deintensification, and 29.6% intensification. Deintensification decreased, while intensification increased, with SBP. Compared to low-risk patients, cardiovascular risk patients had 1.11 (95% CI 1.10-1.13) times the odds of intensifying, and geriatric/frail patients 1.45 (95%CI 1.43-1.47) times the odds of deintensifying.

Discussion

Patient-level HDD change was consistent with an expected association with cardiovascular risk and geriatric/frail conditions, suggesting that HDD can be used longitudinally to assess hypertension treatment modification in large health systems.

Development of a video-observation method for examining doctors' clinical and interpersonal skills in a hospital outpatient clinic in Ibadan, Oyo State, Nigeria

BACKGROUND

Improving the quality of primary healthcare provision is a key goal in low-and middle-income countries (LMICs). However, to develop effective quality improvement interventions, we first need to be able to accurately measure the quality of care. The methods most commonly used to measure the technical quality of care all have some key limitations in LMICs settings. Video-observation is appealing but has not yet been used in this context. We examine preliminary feasibility and acceptability of video-observation for assessing physician quality in a hospital outpatients' department in Nigeria. We also develop measurement procedures and examine measurement characteristics.

METHODS

Cross-sectional study at a large tertiary care hospital in Ibadan, Nigeria. Consecutive physician-patient consultations with adults and children under five seeking outpatient care were video-recorded. We also conducted brief interviews with participating physicians to gain feedback on our approach. Video-recordings were double-coded by two medically trained researchers, independent of the study team and each other, using an explicit checklist of key processes of care that we developed, from which we derived a process quality score. We also elicited a global quality rating from reviewers.

RESULTS

We analysed 142 physician-patient consultations. The median process score given by both coders was 100 %. The modal overall rating category was 'above standard' (or 4 on a scale of 1-5). Coders agreed on which rating to assign only 44 % of the time (weighted Cohen's kappa = 0.26). We found in three-level hierarchical modelling that the majority of variance in process scores was explained by coder disagreement. A very high correlation of 0.90 was found between the global quality rating and process quality score across all encounters. Participating physicians liked our approach, despite initial reservations about being observed.

CONCLUSIONS

Video-observation is feasible and acceptable in this setting, and the quality of consultations was high. However, we found that rater agreement is low but comparable to other modalities that involve expert clinician judgements about quality of care including in-person direct observation and case note review. We suggest ways to improve scoring consistency including careful rater selection and improved design of the measurement procedure for the process score.

Head to head randomized trial of two decision aids for prostate cancer

Background

While many studies have tested the impact of a decision aid (DA) compared to not receiving any DA, far fewer have tested how different types of DAs affect key outcomes such as treatment choice, patient–provider communication, or decision process/satisfaction. This study tested the impact of a complex medical oriented DA compared to a more simplistic decision aid designed to encourage shared decision making in men with clinically localized prostate cancer.

Methods

1028 men at 4 VA hospitals were recruited after a scheduled prostate biopsy. Participants completed baseline measures and were randomized to receive either a simple or complex DA. Participants were men with clinically localized cancer (N = 285) by biopsy and who completed a baseline survey. Survey measures: baseline (biopsy); immediately prior to seeing the physician for biopsy results (pre- encounter); one week following the physician visit (post-encounter). Outcome measures included treatment preference and treatment received, knowledge, preference for shared decision making, decision making process, and patients’ use and satisfaction with the DA.

Results

Participants who received the simple DA had greater interest in shared decision making after reading the DA (p = 0.03), found the DA more helpful (p’s < 0.01) and were more likely to be considering watchful waiting (p = 0.03) compared to those receiving the complex DA at Time 2. While these differences were present before patients saw their urologists, there was no difference between groups in the treatment patients received.

Conclusions

The simple DA led to increased desire for shared decision making and for less aggressive treatment. However, these differences disappeared following the physician visit, which appeared to change patients’ treatment preferences.

Trial registration This trial was pre-registered prior to recruitment of participants.

Supplementary Information

The online version contains supplementary material available at 10.1186/s12911-021-01505-x.

Using multiple imputation to classify potential outcomes subgroups

With medical tests becoming increasingly available, concerns about over-testing, over-treatment and health care cost dramatically increase. Hence, it is important to understand the influence of testing on treatment selection in general practice. Most statistical methods focus on average effects of testing on treatment decisions. However, this may be ill-advised, particularly for patient subgroups that tend not to benefit from such tests. Furthermore, missing data are common, representing large and often unaddressed threats to the validity of most statistical methods. Finally, it is often desirable to conduct analyses that can be interpreted causally. Using the Rubin Causal Model framework, we propose to classify patients into four potential outcomes subgroups, defined by whether or not a patient's treatment selection is changed by the test result and by the direction of how the test result changes treatment selection. This subgroup classification naturally captures the differential influence of medical testing on treatment selections for different patients, which can suggest targets to improve the utilization of medical tests. We can then examine patient characteristics associated with patient potential outcomes subgroup memberships. We used multiple imputation methods to simultaneously impute the missing potential outcomes as well as regular missing values. This approach can also provide estimates of many traditional causal quantities of interest. We find that explicitly incorporating causal inference assumptions into the multiple imputation process can improve the precision for some causal estimates of interest. We also find that bias can occur when the potential outcomes conditional independence assumption is violated; sensitivity analyses are proposed to assess the impact of this violation. We applied the proposed methods to examine the influence of 21-gene assay, the most commonly used genomic test in the United States, on chemotherapy selection among breast cancer patients.

Comparing the use of direct observation, standardized patients and exit interviews in low- and middle-income countries: a systematic review of methods of assessing quality of primary care

Clinical records in primary healthcare settings in low- and middle-income countries (LMIC) are often lacking or of too poor quality to accurately assess what happens during the patient consultation. We examined the most common methods for assessing healthcare workers' clinical behaviour: direct observation, standardized patients and patient/healthcare worker exit interview. The comparative feasibility, acceptability, reliability, validity and practicalities of using these methods in this setting are unclear. We systematically review and synthesize the evidence to compare and contrast the advantages and disadvantages of each method. We include studies in LMICs where methods have been directly compared and systematic and narrative reviews of each method. We searched several electronic databases and focused on real-life (not educational) primary healthcare encounters. The most recent update to the search for direct comparison studies was November 2019. We updated the search for systematic and narrative reviews on the standardized patient method in March 2020 and expanded it to all methods. Search strategies combined indexed terms and keywords. We searched reference lists of eligible articles and sourced additional references from relevant review articles. Titles and abstracts were independently screened by two reviewers and discrepancies resolved through discussion. Data were iteratively coded according to pre-defined categories and synthesized. We included 12 direct comparison studies and eight systematic and narrative reviews. We found that no method was clearly superior to the others-each has pros and cons and may assess different aspects of quality of care provision by healthcare workers. All methods require careful preparation, though the exact domain of quality assessed and ethics and selection and training of personnel are nuanced and the methods were subject to different biases. The differential strengths suggest that individual methods should be used strategically based on the research question or in combination for comprehensive global assessments of quality.

Time Trends in Receipt of Germline Genetic Testing and Results for Women Diagnosed With Breast Cancer or Ovarian Cancer, 2012-2019

PURPOSE

Genetic testing is important for breast and ovarian cancer risk reduction and treatment, yet little is known about its evolving use.

METHODS

SEER records of women of age ≥ 20 years diagnosed with breast or ovarian cancer from 2013 to 2017 in California or Georgia were linked to the results of clinical germline testing through 2019. We measured testing trends, rates of variants of uncertain significance (VUS), and pathogenic variants (PVs).

RESULTS

One quarter (25.2%) of 187,535 patients with breast cancer and one third (34.3%) of 14,689 patients with ovarian cancer were tested; annually, testing increased by 2%, whereas the number of genes tested increased by 28%. The prevalence of test results by gene category for breast cancer cases in 2017 were BRCA1/2, PVs 5.2%, and VUS 0.8%; breast cancer-associated genes or ovarian cancer-associated genes (ATM, BARD1, BRIP1, CDH1, CHEK2, EPCAM, MLH1, MSH2, MSH6, NBN, NF1, PALB2, PMS2, PTEN, RAD51C, RAD51D, STK11, and TP53), PVs 3.7%, and VUS 12.0%; other actionable genes (APC, BMPR1A, MEN1, MUTYH, NF2, RB1, RET, SDHAF2, SDHB, SDHC, SDHD, SMAD4, TSC1, TSC2, and VHL) PVs 0.6%, and VUS 0.5%; and other genes, PVs 0.3%, and VUS 2.6%. For ovarian cancer cases in 2017, the prevalence of test results were BRCA1/2, PVs 11.0%, and VUS 0.9%; breast or ovarian genes, PVs 4.0%, and VUS 12.6%; other actionable genes, PVs 0.7%, and VUS 0.4%; and other genes, PVs 0.3%, and VUS 0.6%. VUS rates doubled over time (2013 diagnoses: 11.2%; 2017 diagnoses: 26.8%), particularly for racial or ethnic minorities (47.8% Asian and 46.0% Black, v 24.6% non-Hispanic White patients; P < .001).

CONCLUSION

A testing gap persists for patients with ovarian cancer (34.3% tested v nearly all recommended), whereas adding more genes widened a racial or ethnic gap in VUS results. Most PVs were in 20 breast cancer-associated genes or ovarian cancer-associated genes; testing other genes yielded mostly VUS. Quality improvement should focus on testing indicated patients rather than adding more genes.

A Method to Quantify Mean Hypertension Treatment Daily Dose Intensity Using Health Care System Data

Importance

Simple measures of hypertension treatment, such as achievement of blood pressure (BP) targets, ignore the intensity of treatment once the BP target is met. High-intensity treatment involves increased treatment burden and can be associated with potential adverse effects in older adults. A method was previously developed to identify older patients receiving intense hypertension treatment by low BP and number of BP medications using national Veterans Health Administration and Medicare Part D administrative pharmacy data to evaluate which BP medications a patient is likely taking on any given day.

Objective

To further develop and validate a method to more precisely quantify dose intensity of hypertension treatment using only health system administrative pharmacy fill data.

Design, Setting, and Participants

Observational, cross-sectional study of 319 randomly selected older veterans in the national Veterans Health Administration health care system who were taking multiple BP-lowering medications and had a total of 3625 ambulatory care visits from July 1, 2011, to June 30, 2013. Measure development and medical record review occurred January 1, 2017, through November 30, 2018, and data analysis was conducted from December 1, 2019, to August 31, 2020.

Main Outcomes and Measures

For each BP-lowering medication, a moderate hypertension daily dose (HDD) was defined as half the maximum dose above which no further clinical benefit has been demonstrated by that medication in hypertension trials. Patients' total HDD was calculated using pharmacy data (pharmacy HDDs), accounting for substantial delays in refills (>30 days) when a patient's pill supply was stretched (eg, cutting existing pills in half). As an external comparison, the pharmacy HDDs were correlated with doses manually extracted from clinicians' visit notes (clinically noted HDDs). How well the pharmacy HDDs correlated with clinically noted HDDs was calculated (using C statistics). To facilitate interpretation, HDDs were described in association with the number of medications.

Results

A total of 316 patients (99.1%) were male; the mean (SD) age was 75.6 (7.2) years. Pharmacy HDDs were highly correlated (r = 0.92) with clinically noted HDDs, with a mean (SD) of 2.7 (1.8) for pharmacy HDDs and 2.8 (1.8) for clinically noted HDDs. Pharmacy HDDs correlated with high-intensity, clinically noted HDDs ranging from a C statistic of 92.8% (95% CI, 92.0%-93.7%) for 2 or more clinically noted HDDs to 88.1% (95% CI, 85.5%-90.6%) for 6 or more clinically noted HDDs.

Conclusions and Relevance

This study suggests that health system pharmacy data may be used to accurately quantify hypertension regimen dose intensity. Together with clinic-measured BP, this tool can be used in future health system-based research or quality improvement efforts to fine-tune, manage, and optimize hypertension treatment in older adults.

Focus and features of prescribing indications spanning multiple chronic conditions in older adults: A narrative review

Background:

Inappropriate prescribing is frequent in older adults and associated with adverse outcomes. Prescribing indications aim to optimize prescribing, but little is known about the focus and features of prescribing indications for the most common chronic conditions in older adults. Understanding the conditions, medications, and issues addressed (e.g., patient perspective, drug-disease interaction, adverse drug event) in current prescribing indications may help to identify missing indications and develop standardized measures to improve prescribing quality.

Methods:

We searched Ovid/MEDLINE and EMBASE for articles published between 2015 and 2020 reporting prescribing indications for older adults. Prescribing indication included 1) prescribing “criteria,” or statements that guide prescribing action, and 2) prescribing “measures,” or prescribing actions observed in a population. We categorized their focus by conditions, medications and issues addressed, as well as level of evidence provided.

Results:

Among 16 sets of prescribing indications, we identified 748 criteria and 47 measures. The most common addressed medications were antihypertensives, analgesics/antirheumatics, and antiplatelets/anticoagulants. The most frequently addressed issues were drug-disease interaction, adverse drug event, administration, better therapeutic alternative, and (co-)prescription omission (20.8–36.1%). Age/functioning, drug-drug interaction, monitoring, and efficacy/safety ratio were found in only 9.9–16.5% of indications. Indications rarely focused on the patient perspective or issues with multiple providers.

Conclusion:

Most prescribing indications for chronic conditions in older patients are criteria rather than measures. Indications accounting for patient perspective and multiple providers are limited. The gaps identified in this review may help improve the development of prescribing measures for older adults and ultimately improve quality of care.

Identifying Recommendations for Stopping or Scaling Back Unnecessary Routine Services in Primary Care

IMPORTANCE

Much of health care involves established, routine use of medical services for chronic conditions or prevention. Stopping these services when the evidence changes or if the benefits no longer outweigh the risks is essential. Yet, most guidelines focus on escalating care and provide few explicit recommendations to stop or scale back (ie, deintensify) treatment and testing.

OBJECTIVE

To develop a systematic, transparent, and reproducible approach for identifying, specifying, and validating deintensification recommendations associated with routine adult primary care.

DESIGN, SETTING, AND PARTICIPANTS

A focused review of existing guidelines and recommendations was completed to identify and prioritize potential deintensification indications. Then, 2 modified virtual Delphi expert panels examined the synthesized evidence, suggested ways that the candidate recommendations could be improved, and assessed the validity of the recommendations using the RAND/UCLA Appropriateness Method. Twenty-five physicians from Veterans Affairs and US academic institutions with knowledge in relevant clinical areas (eg, geriatrics, primary care, women's health, cardiology, and endocrinology) served as panel members.

MAIN OUTCOMES AND MEASURES

Validity of the recommendations, defined as high-quality evidence that deintensification is likely to improve patient outcomes, evidence that intense testing and/or treatment could cause harm in some patients, absence of evidence on the benefit of continued or repeated intense treatment or testing, and evidence that deintensification is consistent with high-quality care.

RESULTS

A total of 409 individual recommendations were identified representing 178 unique opportunities to stop or scale back routine services (eg, stopping population-based screening for vitamin D deficiency and decreasing concurrent use of opioids and benzodiazepines). Thirty-seven recommendations were prioritized and forwarded to the expert panels. Panelists reviewed the evidence and suggested modifications, resulting in 44 recommendations being rated. Overall, 37 recommendations (84%) were considered to be valid, as assessed by the RAND/UCLA Appropriateness Method.

CONCLUSIONS AND RELEVANCE

In this study, a total of 178 unique opportunities to deintensify routine primary care services were identified, and 37 of these were validated as high-priority deintensification recommendations. To date, this is the first study to develop a model for identifying, specifying, and validating deintensification recommendations that can be implemented and tracked in clinical practice.

Association of Continuity of Care With Outcomes in US Veterans With Inflammatory Bowel Disease

IMPORTANCE

Health care fragmentation is associated with inefficiency and worse outcomes. Continuity of care (COC) models were developed to address fragmentation.

OBJECTIVE

To examine COC and selected outcomes in US veterans with inflammatory bowel disease (IBD).

DESIGN, SETTING, AND PARTICIPANTS

This retrospective cohort study used the Veterans Health Administration (VHA) Corporate Data Warehouse to identify veterans with IBD who received care in the VHA health care system between January 1, 2002, and December 31, 2014. Included patients were veterans with IBD who had a primary care physician and at least 4 outpatient visits with key physicians (gastroenterologist, primary care physician, and surgeon) within the first year after an index IBD encounter. Data were analyzed from November 2018 to May 2020.

EXPOSURES

Care continuity was measured with the Bice-Boxerman COC index to define care density and dispersion within year 1 after the initial presentation.

MAIN OUTCOMES AND MEASURES

A Cox proportional hazards regression model was used to quantify the association between a low level of COC in year 1 (defined as ≤0.25 on a 0 to 1 scale) and subsequent IBD-related outcomes in years 2 and 3 (outpatient flare, hospitalization, and surgical intervention).

RESULTS

Among the 20 079 veterans with IBD who met the inclusion criteria, 18 632 were men (92.8%) and the median (interquartile range [IQR]) age was 59 (48-66) years. In the first year of follow-up, substantial variation in the dispersion of care was observed (median [IQR] COC, 0.24 [0.13-0.46]). In a Cox proportional hazards regression model, a low level of COC was associated with a higher likelihood of outpatient flares that required corticosteroids (adjusted hazard ratio [aHR], 1.11; 95% CI, 1.01-1.22), hospitalizations (aHR, 1.25; 95% CI, 1.06-1.47), and surgical interventions (aHR, 1.72; 95% CI, 1.43-2.07).

CONCLUSIONS AND RELEVANCE

Results of this cohort study showed a wide variation in dispersion of IBD care and an association between a lower level of COC and active IBD and worse outcomes. The findings suggest that investigating the barriers to COC in integrated systems that have invested in care coordination is key to not only better understanding COC but also identifying opportunities to improve care fragmentation.

Validation of a Health System Measure to Capture Intensive Medication Treatment of Hypertension in the Veterans Health Administration

IMPORTANCE

Blood pressure (BP) targets are the main measure of high-quality hypertension care in health systems. However, BP alone does not reflect intensity of pharmacological treatment, which should be carefully managed in older patients.

OBJECTIVES

To develop and validate an electronic health record (EHR) data-only algorithm using pharmacy and BP data to capture intensive hypertension care (IHC), defined as 3 or more BP medications and BP less than 120 mm Hg, and to identify conditions associated with greater IHC, either through greater algorithm false-positive IHC, or by contributing clinically to delivering more IHC.

DESIGN, SETTING, AND PARTICIPANTS

This cross-sectional study was conducted among 319 randomly selected patients aged 65 years or older receiving IHC from the Veterans Health Administration (VHA) from July 1, 2011, to June 30, 2013. Data were collected from a total of 3625 primary care visits. Data were analyzed from January 2017 to March 2020.

EXPOSURES

Calibration and measurement of the algorithm for IHC (algorithm IHC).

MAIN OUTCOMES AND MEASURES

For each primary care visit, the reference standard, clinical IHC, was determined by detailed review of free-text clinical notes. The correlation in BP medication count between the EHR-only algorithm vs the reference standard and the sensitivity and specificity of the algorithm IHC were calculated. In addition, presence vs absence of contributing conditions acting in combination with hypertension management were measured to examine incidence of IHC associated with contributing conditions, including an acute condition that lowered BP (eg, dehydration), another condition requiring a BP target lower than the standard 140 mm Hg (eg, diabetes), or the patient needing a BP-lowering medication for a nonhypertension condition (eg, β-blocker for atrial fibrillation) resulting in low BP.

RESULTS

Among 319 patients with 3625 visits (mean [SD] age, 75.6 [7.2] years; 3592 [99.1%] men), 911 visits (25.1%) had clinical IHC by the reference standard. The algorithm for determining medication count was highly correlated with the reference standard (r = 0.84). Sensitivity of detecting clinical IHC was 92.2% (95% CI, 89.3%-95.1%), and specificity was 97.2% (95% CI, 96.1%-98.3%), suggesting that clinical IHC can be identified from routinely collected data. Only 75 visits (2.1%) were algorithm IHC false positives, 55 visits (1.5%) involved IHC with contributing conditions, and 125 visits (3.5%) involved either false-positive or IHC with contributing conditions. Among select contributing conditions, congestive heart failure (37 patients [5.2%]) was most associated with a prespecified combined false-positive or IHC with contributing conditions rate higher than 5%.

CONCLUSIONS AND RELEVANCE

These findings suggest that health system data can be used reliably to estimate IHC.

Developing a template matching algorithm for benchmarking hospital performance in a diverse, integrated healthcare system

Template matching is a proposed approach for hospital benchmarking, which measures performance based on matching a subset of comparable patient hospitalizations from each hospital. We assessed the ability to create the required matched samples and thus the feasibility of template matching to benchmark hospital performance in a diverse healthcare system.Nationwide Veterans Affairs (VA) hospitals, 2017.Observational cohort study.We used administrative and clinical data from 668,592 hospitalizations at 134 VA hospitals in 2017. A standardized template of 300 hospitalizations was selected, and then 300 hospitalizations were matched to the template from each hospital.There was substantial case-mix variation across VA hospitals, which persisted after excluding small hospitals, hospitals with primarily psychiatric admissions, and hospitalizations for rare diagnoses. Median age ranged from 57 to 75 years across hospitals; percent surgical admissions ranged from 0.0% to 21.0%; percent of admissions through the emergency department, 0.1% to 98.7%; and percent Hispanic patients, 0.2% to 93.3%. Characteristics for which there was substantial variation across hospitals could not be balanced with any matching algorithm tested. Although most other variables could be balanced, we were unable to identify a matching algorithm that balanced more than ∼20 variables simultaneously.We were unable to identify a template matching approach that could balance hospitals on all measured characteristics potentially important to benchmarking. Given the magnitude of case-mix variation across VA hospitals, a single template is likely not feasible for general hospital benchmarking.

Measures to Improve Diagnostic Safety in Clinical Practice

Timely and accurate diagnosis is foundational to good clinical practice and an essential first step to achieving optimal patient outcomes. However, a recent Institute of Medicine report concluded that most of us will experience at least one diagnostic error in our lifetime. The report argues for efforts to improve the reliability of the diagnostic process through better measurement of diagnostic performance. The diagnostic process is a dynamic team-based activity that involves uncertainty, plays out over time, and requires effective communication and collaboration among multiple clinicians, diagnostic services, and the patient. Thus, it poses special challenges for measurement. In this paper, we discuss how the need to develop measures to improve diagnostic performance could move forward at a time when the scientific foundation needed to inform measurement is still evolving. We highlight challenges and opportunities for developing potential measures of "diagnostic safety" related to clinical diagnostic errors and associated preventable diagnostic harm. In doing so, we propose a starter set of measurement concepts for initial consideration that seem reasonably related to diagnostic safety and call for these to be studied and further refined. This would enable safe diagnosis to become an organizational priority and facilitate quality improvement. Health-care systems should consider measurement and evaluation of diagnostic performance as essential to timely and accurate diagnosis and to the reduction of preventable diagnostic harm.

Developing Strategies to Reduce Unnecessary Services in Primary Care: Protocol for User-Centered Design Charrettes

Background

Overtreatment and overtesting expose patients to unnecessary, wasteful, and potentially harmful care. Reducing overtreatment or overtesting that has become ingrained in current clinical practices and is being delivered on a routine basis will require solutions that incorporate a deep understanding of multiple perspectives, particularly those on the front lines of clinical care: patients and their clinicians. Design approaches are a promising and innovative way to incorporate stakeholder needs, desires, and challenges to develop solutions to complex problems.

Objective

This study aimed (1) to engage patients in a design process to develop high-level deintensification strategies for primary care (ie, strategies for scaling back or stopping routine medical services that more recent evidence reveals are not beneficial) and (2) to engage both patients and primary care providers in further co-design to develop and refine the broad deintensification strategies identified in phase 1.

Methods

We engaged stakeholders in design charrettes—intensive workshops in which key stakeholders are brought together to develop creative solutions to a specific problem—focused on deintensification of routine overuse in primary care. We conducted the study in 2 phases: a 6.5-hour design charrette with 2 different groups of patients (phase 1) and a subsequent 4-hour charrette with clinicians and a subgroup of phase 1 patients (phase 2). Both phases included surveys and educational presentations related to deintensification. Phase 1 involved several design activities (mind mapping, business origami, and empathy mapping) to help patients gain a deeper understanding of the individuals involved in deintensification. Following that, we asked participants to review hypothetical scenarios where patients, clinicians, or the broader health system context posed a barrier to deintensification and then to brainstorm solutions. The deintensification themes identified in phase 1 were used to guide phase 2. This second phase primarily involved 1 design activity (WhoDo). In this activity, patients and clinicians worked together to develop concrete actions that specific stakeholders could take to support deintensification efforts. This activity included identifying barriers to the actions and approaches to overcoming those barriers.

Results

A total of 35 patients participated in phase 1, and 9 patients and 7 clinicians participated in phase 2. The analysis of the deintensification strategies and survey data is currently underway. The results are expected to be submitted for publication in early 2020.

Conclusions

Health care interventions are frequently developed without input from the people who are most affected. The exclusion of these stakeholders in the design process often influences and limits the impact of the intervention. This study employed design charrettes, guided by a flexible user-centered design model, to bring clinicians and patients with differing backgrounds and with different expectations together to cocreate real-world solutions to the complex issue of deintensifying medical services.

International Registered Report Identifier (IRRID)

RR1-10.2196/15618

Association of Attending Surgeon With Variation in the Receipt of Genetic Testing After Diagnosis of Breast Cancer

Importance

Genetic testing after diagnosis of breast cancer is common, but little is known about the influence of the surgeon on the variation in testing.

Objectives

To quantify and explain the association of attending surgeon with rates of genetic testing after diagnosis of breast cancer.

Design, Setting, and Participants

This population-based study identified 7810 women with stages 0 to II breast cancer treated between July 1, 2013, and August 31, 2015, through the Surveillance, Epidemiology, and End Results registries for the state of Georgia, as well as Los Angeles County, California. Surveys were sent approximately 2 months after surgery. Also surveyed were 488 attending surgeons identified by the patients.

Main Outcomes and Measures

The study examined the association of surgeon with variation in the receipt of genetic testing using information from patient and surgeon surveys merged to Surveillance, Epidemiology, and End Results and genetic testing data obtained from 4 laboratories.

Results

In total, 5080 women (69.6%) of 7303 who were eligible (mean [SD] age, 61.4 [0.8] years) and 377 surgeons (77.3%) of 488 (mean [SD] age, 53.8 [10.7] years) responded to the survey. Approximately one-third (34.5% [1350 of 3910] of patients had an elevated risk of mutation carriage, and 27.0% (1056 of 3910) overall had genetic testing. Surgeons had practiced a mean (SE) of 20.9 (0.6) years, and 28.9% (107 of 370) treated more than 50 cases of new breast cancer per year. The odds of a patient receiving genetic testing increased more than 2-fold (odds ratio, 2.48; 95% CI, 1.85-3.31) if she saw a surgeon with an approach 1 SD above that of a surgeon with the mean test rate. Approximately one-third (34.1%) of the surgeon variation was explained by patient volume and surgeon attitudes about genetic testing and counseling. If a patient with higher pretest risk saw a surgeon at the 5th percentile of the surgeon distribution, she would have a 26.3% (95% CI, 21.9%-31.2%) probability of testing compared with 72.3% (95% CI, 66.7%-77.2%) if she saw a surgeon at the 95th percentile.

Conclusions and Relevance

In this study, the attending surgeon was associated with the receipt of genetic testing after a breast cancer diagnosis. Variation in surgeon attitudes about genetic testing and counseling may explain a substantial amount of this association.

Template matching for benchmarking hospital performance in the veterans affairs healthcare system

Comparing hospital performance in a health system is traditionally done with multilevel regression models that adjust for differences in hospitals' patient case-mix. In contrast, "template matching" compares outcomes of similar patients at different hospitals but has been used only in limited patient settings.Our objective was to test a basic template matching approach in the nationwide Veterans Affairs healthcare system (VA), compared with a more standard regression approach.We performed various simulations using observational data from VA electronic health records whereby we randomly assigned patients to "pseudo hospitals," eliminating true hospital level effects. We randomly selected a representative template of 240 patients and matched 240 patients on demographic and physiological factors from each pseudo hospital to the template. We varied hospital performance for different simulations such that some pseudo hospitals negatively impacted patient mortality.Electronic health record data of 460,213 hospitalizations at 111 VA hospitals across the United States in 2015.We assessed 30-day mortality at each pseudo hospital and identified lowest quintile hospitals by template matching and regression. The regression model adjusted for predicted 30-day mortality (as a measure of illness severity).Regression identified the lowest quintile hospitals with 100% accuracy compared with 80.3% to 82.0% for template matching when systematic differences in 30-day mortality existed.The current standard practice of risk-adjusted regression incorporating patient-level illness severity was better able to identify lower-performing hospitals than the simplistic template matching algorithm.

Mixed messages to consumers from Medicare: Hospital Compare grades versus value-based payment penalty

OBJECTIVES

To (1) compare the 2015 hospital grades reported on Medicare's Hospital Compare website for heart failure (HF) and acute myocardial infarction (AMI) readmissions with the HF- and AMI-specific scores for excess readmissions used to assess Medicare readmission penalties and (2) assess how often hospitals were penalized for excess readmissions in only 1 or 2 conditions, given that hospitals received a penalty impacting all Medicare payments based on an overall readmission score calculated from 5 conditions (HF, AMI, pneumonia, chronic obstructive pulmonary disease, and total hip/knee arthroplasty).

STUDY DESIGN

Retrospective secondary data analysis.

METHODS

Descriptive analyses of hospital-specific, condition-specific grades and excess readmission scores and hospital-level penalties downloaded from Hospital Compare.

RESULTS

Of the 2956 hospitals that had publicly reported HF grades on Hospital Compare, 91.9% (2717) were graded as "no different" than the national rate for HF readmissions, which included 48.6% that were scored as having excessive HF admissions, and 87% received an overall readmission penalty. Of 120 (4.1%) hospitals graded as "better" than the national rate for HF, none were scored as having excessive HF readmissions and 50% were penalized. AMI data yielded similar results. Among 2591 hospitals penalized for overall readmissions, 26.6% had only 1 condition with excess readmissions and 27.5% had 2 conditions.

CONCLUSIONS

Many hospitals with an HF and AMI readmission grade of "no different" than the national rate on Hospital Compare received penalties for excessive readmissions under the Hospital Readmissions Reduction Program. The value signal to consumers and hospitals communicated by grades and penalties is therefore weakened because the methods applied to the same hospital data produce conflicting messages of "average grades" yet "bad enough for penalty."

Recent Trends in Chemotherapy Use and Oncologists' Treatment Recommendations for Early-Stage Breast Cancer

Background

There is growing concern about overtreatment of breast cancer as outcomes have improved over time. However, little is known about how chemotherapy use and oncologists' recommendations have changed in recent years.

Methods

We surveyed 5080 women (70% response rate) diagnosed with breast cancer between 2013 and 2015 and accrued through two Surveillance, Epidemiology, and End Results registries (Georgia and Los Angeles) about chemotherapy receipt and their oncologists' chemotherapy recommendations. We surveyed 504 attending oncologists (60.3% response rate ) about chemotherapy recommendations in node-negative and node-positive case scenarios. We conducted descriptive statistics of chemotherapy use and patients' report of oncologists' recommendations and used a generalized linear mixed model of chemotherapy use according to time and clinical factors. All statistical tests were two-sided.

Results

The analytic sample was 2926 patients with stage I-II, estrogen receptor-positive, human epidermal growth factor receptor 2-negative breast cancer. From 2013 to 2015, keeping other factors constant, chemotherapy use was estimated to decline from 34.5% (95% confidence interval [CI] = 30.8% to 38.3%) to 21.3% (95% CI = 19.0% to 23.7%, P < .001). Estimated decline in chemotherapy use was from 26.6% (95% CI = 23.0% to 30.7%) to 14.1% (95% CI = 12.0% to 16.3%) for node-negative/micrometastasis patients and from 81.1% (95% CI = 76.6% to 85.0%) to 64.2% (95% CI = 58.6% to 69.6%) for node-positive patients. Use of the 21-gene recurrence score (RS) did not change among node-negative/micrometastasis patients, and increasing RS use in node-positive patients accounted for one-third of the chemotherapy decline. Patients' report of oncologists' recommendations for chemotherapy declined from 44.9% (95% CI = 40.2% to 49.7%) to 31.6% (95% CI = 25.9% to 37.9%), controlling for other factors. Oncologists were much more likely to order RS if patient preferences were discordant with their recommendations (67.4%, 95% CI = 61.7% to 73.0%, vs 17.5%, 95% CI = 13.1% to 22.0%, concordant), and they adjusted recommendations based on patient preferences and RS results.

Conclusions

For both node-negative/micrometastasis and node-positive patients, chemotherapy receipt and oncologists' recommendations for chemotherapy declined markedly over time, without substantial change in practice guidelines. Results of ongoing trials will be essential to confirm the quality of this approach to breast cancer care.

Abstract 102: Impact of Using Older Data on the Accuracy of Cardiovascular Risk Scores

Background: Recent scores to predict atherosclerotic cardiovascular disease, ASCVD, have been found inaccurate, with some concern that risk scores become inaccurate with time, as changing demographics may also change ASCVD risk. A partial solution to the timeliness problem could be creating risk scores using electronic health records (EHR). EHR-based risk scores can be easily updated and tested for changes over time. They can also use more variables than traditional risk scores. This could capture clinical change that would otherwise be missed.

We hypothesized that ASCVD risk scores change over time, but this could be minimized with more robust risk scores. To test this, we looked at change of two EHR-based risk score over three follow-up periods and using different statistical techniques to design the risk scores.

Methods: Data sources: VA national data linked to Medicare and the National Death Index.

Population: 3 overlapping cohorts from 2002, 2006, and 2009. Each consisted of all active VA patients aged 45-80 who had no documented history of CVD, clinical heart failure or loop diuretic use at baseline.

Prediction models

1. VARS-ASCVD: uses the same variables as traditional risk scores, but all variables were re-calibrated to our population.

2. VARS-EHR: Uses 41 predictor variables and more interaction effects.

Outcome variables: First occurrence of fatal or nonfatal ASCVD during 5 years of follow-up.

Analysis: We looked at the accuracy of risk scores developed in 2002 on patients in 2006 and 2009. The discrimination of the risk scores (the ability to distinguish between those who do and do not develop an event), was evaluated with C-statistisic. The calibration (how closely the predicted probabilities reflect true risk) was evaluated with the Hosmer-Lemeshow Goodness of Fit statistic (GoF).

Results: Each cohort had at least 1.4 million participants. Between the 3 cohorts the rate of diabetes mellitus increased from 201% to 27% and statin use increased from 25% to 45% of the population.

The VARS-ASCVD risk scores for men developed in 2002 had the same discrimination of 0.67 in 2006 and 2009, but in women fell from 0.77 to 0.72 then increased to 0.74. The goodness of fit worsened. Using the VARS-EHR model, discrimination stayed similar in men and women. The GOF worsened, but by substantially less.

Conclusions: ASCVD risk prediction tools become poorly calibrated over fairly short time periods. For effective use, they must be updated regularly.

An Examination of Deintensification Recommendations in Clinical Practice Guidelines: Stepping Up or Scaling Back?

In light of initiatives to decrease use of unnecessary services, this article examines whether current guidelines for diabetes and cardiovascular disease preferentially recommend intensification rather than deintensification of care.

Improving Breast Cancer Surgical Treatment Decision Making: The iCanDecide Randomized Clinical Trial

Purpose This study was conducted to determine the effect of iCanDecide, an interactive and tailored breast cancer treatment decision tool, on the rate of high-quality patient decisions-both informed and values concordant-regarding locoregional breast cancer treatment and on patient appraisal of decision making. Methods We conducted a randomized clinical trial of newly diagnosed patients with early-stage breast cancer making locoregional treatment decisions. From 22 surgical practices, 537 patients were recruited and randomly assigned online to the iCanDecide interactive and tailored Web site (intervention) or the iCanDecide static Web site (control). Participants completed a baseline survey and were mailed a follow-up survey 4 to 5 weeks after enrollment to assess the primary outcome of a high-quality decision, which consisted of two components, high knowledge and values-concordant treatment, and secondary outcomes (decision preparation, deliberation, and subjective decision quality). Results Patients in the intervention arm had higher odds of making a high-quality decision than did those in the control arm (odds ratio, 2.00; 95% CI, 1.37 to 2.92; P = .0004), which was driven primarily by differences in the rates of high knowledge between groups. The majority of patients in both arms made values-concordant treatment decisions (78.6% in the intervention arm and 81.4% in the control arm). More patients in the intervention arm had high decision preparation (estimate, 0.18; 95% CI, 0.02 to 0.34; P = .027), but there were no significant differences in the other decision appraisal outcomes. The effect of the intervention was similar for women who were leaning strongly toward a treatment option at enrollment compared with those who were not. Conclusion The tailored and interactive iCanDecide Web site, which focused on knowledge building and values clarification, positively affected high-quality decisions largely by improving knowledge compared with static online information. To be effective, future patient-facing decision tools should be integrated into the clinical workflow to improve decision making.

Surgeon Influence on Variation in Receipt of Contralateral Prophylactic Mastectomy for Women With Breast Cancer

Importance

Rates of contralateral prophylactic mastectomy (CPM) have markedly increased but we know little about the influence of surgeons on variability of the procedure in the community.

Objective

To quantify the influence of the attending surgeon on rates of CPM and clinician attitudes that explained it.

Design, Setting, and Participants

In this population-based survey study, we identified 7810 women with stages 0 to II breast cancer treated in 2013 to 2015 through the Surveillance, Epidemiology, and End Results registries of Georgia and Los Angeles County. Surveys were sent approximately 2 months after surgery. Surveys were also sent to 488 attending surgeons identified by the patients.

Main Outcomes and Measures

We conducted multilevel analyses to examine the impact of surgeon influence on variations in patient receipt of CPM using information from patient and surgeon surveys merged to Surveillance, Epidemiology, and End Results data.

Results

A total of 5080 women responded to the survey (70% response rate), and 377 surgeons responded (77% response rate). The mean (SD) age of responding women was 61.9 (11) years; 28% had an increased risk of second primary cancer, and 16% received CPM. Half of surgeons (52%) practiced for more than 20 years and 30% treated more than 50 new patients with breast cancer annually. Attending surgeon explained a large amount (20%) of the variation in CPM, controlling for patient factors. The odds of a patient receiving CPM increased almost 3-fold (odds ratio, 2.8; 95% CI, 2.1-3.4) if she saw a surgeon with a practice approach 1 SD above a surgeon with the mean CPM rate (independent of age, diagnosis date, BRCA status, and risk of second primary). One-quarter (25%) of the surgeon influence was explained by attending attitudes about initial recommendations for surgery and responses to patient requests for CPM. The estimated rate of CPM was 34% for surgeons who least favored initial breast conservation and were least reluctant to perform CPM vs 4% for surgeons who most favored initial breast conservation and were most reluctant to perform CPM.

Conclusions and Relevance

In this study, attending surgeons exerted influence on the likelihood of receipt of CPM after a breast cancer diagnosis. Variations in surgeon attitudes about recommendations for surgery and response to patient requests for CPM explain a substantial amount of this influence.

Interobserver Reliability of the Berlin ARDS Definition and Strategies to Improve the Reliability of ARDS Diagnosis

BACKGROUND

Failure to reliably diagnose ARDS may be a major driver of negative clinical trials and underrecognition and treatment in clinical practice. We sought to examine the interobserver reliability of the Berlin ARDS definition and examine strategies for improving the reliability of ARDS diagnosis.

METHODS

Two hundred five patients with hypoxic respiratory failure from four ICUs were reviewed independently by three clinicians, who evaluated whether patients had ARDS, the diagnostic confidence of the reviewers, whether patients met individual ARDS criteria, and the time when criteria were met.

RESULTS

Interobserver reliability of an ARDS diagnosis was "moderate" (kappa = 0.50; 95% CI, 0.40-0.59). Sixty-seven percent of diagnostic disagreements between clinicians reviewing the same patient was explained by differences in how chest imaging studies were interpreted, with other ARDS criteria contributing less (identification of ARDS risk factor, 15%; cardiac edema/volume overload exclusion, 7%). Combining the independent reviews of three clinicians can increase reliability to "substantial" (kappa = 0.75; 95% CI, 0.68-0.80). When a clinician diagnosed ARDS with "high confidence," all other clinicians agreed with the diagnosis in 72% of reviews. There was close agreement between clinicians about the time when a patient met all ARDS criteria if ARDS developed within the first 48 hours of hospitalization (median difference, 5 hours).

CONCLUSIONS

The reliability of the Berlin ARDS definition is moderate, driven primarily by differences in chest imaging interpretation. Combining independent reviews by multiple clinicians or improving methods to identify bilateral infiltrates on chest imaging are important strategies for improving the reliability of ARDS diagnosis.

Effects of Guideline and Formulary Changes on Statin Prescribing in the Veterans Affairs

OBJECTIVE

To compare the effects of two sequential policy changes-the addition of a high-potency statin to the Department of Veterans Affairs (VA) formulary and the release of the American College of Cardiology/American Heart Association (ACC/AHA) cholesterol guidelines-on VA provider prescribing.

DATA SOURCES/STUDY SETTING

Retrospective analysis of 1,100,682 VA patients, 2011-2016.

STUDY DESIGN

Interrupted time-series analysis of changes in prescribing of moderate-to-high-intensity statins among high-risk patients and across high-risk subgroups. We also assessed changes in prescribing of atorvastatin and other statin drugs. We estimated marginal effects (ME) of formulary and guideline changes by comparing predicted and observed statin use.

DATA COLLECTION/EXTRACTION METHODS

Data from VA Corporate Data Warehouse.

PRINCIPAL FINDINGS

The use of moderate-to-high-intensity statins increased by 2 percentage points following the formulary change (ME, 2.4, 95% confidence interval [CI], 2.2 to 2.6) and less than 1 percentage point following the guideline change (ME, 0.8, 95% CI, 0.6 to 0.9). The formulary change led to approximately a 12 percentage-point increase in the use of moderate-to-high-intensity atorvastatin (ME, 11.5, 95% CI, 11.3 to 11.6). The relatively greater provider response to the formulary change occurred across all patient subgroups.

CONCLUSIONS

Addition of a high-potency statin to formulary affected provider prescribing more than the ACC/AHA guidelines.

Trends in Reoperation After Initial Lumpectomy for Breast Cancer: Addressing Overtreatment in Surgical Management

Importance

Surgery after initial lumpectomy to obtain more widely clear margins is common and may lead to mastectomy.

Objective

To describe surgeons' approach to surgical margins for invasive breast cancer, and changes in postlumpectomy surgery rates, and final surgical treatment following a 2014 consensus statement endorsing a margin of "no ink on tumor."

Design, Setting, and Participants

This was a population-based cohort survey study of 7303 eligible women ages 20 to 79 years with stage I and II breast cancer diagnosed in 2013 to 2015 and identified from the Georgia and Los Angeles County, California, Surveillance, Epidemiology, and End Results registries. A total of 5080 (70%) returned a survey. Those with bilateral disease, missing stage or treatment data, and with ductal carcinoma in situ were excluded, leaving 3729 patients in the analytic sample; 98% of these identified their attending surgeon. Between April 2015 and May 2016, 488 surgeons were surveyed regarding lumpectomy margins; 342 (70%) responded completely. Pathology reports of all patients having a second surgery and a 30% sample of those with 1 surgery were reviewed. Time trends were analyzed with multinomial regression models.

Main Outcomes and Measures

Rates of final surgical procedure (lumpectomy, unilateral mastectomy, bilateral mastectomy) and rates of additional surgery after initial lumpectomy over time, and surgeon attitudes toward an adequate lumpectomy margin.

Results

The 67% rate of initial lumpectomy in the 3729 patient analytic sample was unchanged during the study. The rate of final lumpectomy increased by 13% from 2013 to 2015, accompanied by a decrease in unilateral and bilateral mastectomy (P = .002). Surgery after initial lumpectomy declined by 16% (P < .001). Pathology review documented no significant association between date of treatment and positive margins. Of 342 responding surgeons, 69% endorsed a margin of no ink on tumor to avoid reexcision in estrogen receptor-positive progesterone receptor-positive cancer and 63% for estrogen receptor-negative progesterone- receptor-negative cancer. Surgeons treating more than 50 breast cancers annually were significantly more likely to report this margin as adequate (85%; n = 105) compared with those treating 20 cases or fewer (55%; n = 131) (P < .001).

Conclusions and Relevance

Additional surgery after initial lumpectomy decreased markedly from 2013 to 2015 concomitant with dissemination of clinical guidelines endorsing a minimal negative margin. These findings suggest that surgeon-led initiatives to address potential overtreatment can reduce the burden of surgical management in patients with cancer.

Oncologists' influence on receipt of adjuvant chemotherapy: does it matter whom you see for treatment of curable breast cancer?

PURPOSE

We know little about whether it matters which oncologist a breast cancer patient sees with regard to receipt of chemotherapy. We examined oncologists' influence on use of recurrence score (RS) testing and chemotherapy in the community.

METHODS

We identified 7810 women with stages 0-II breast cancer treated in 2013-15 through the SEER registries of Georgia and Los Angeles County. Surveys were sent 2 months post-surgery, (70% response rate, n = 5080). Patients identified their oncologists (n = 504) of whom 304 responded to surveys (60%). We conducted multi-level analyses on patients with ER-positive HER2-negative invasive disease (N = 2973) to examine oncologists' influence on variation in RS testing and chemotherapy receipt, using patient and oncologist survey responses merged to SEER data.

RESULTS

Half of patients (52.8%) received RS testing and 27.7% chemotherapy. One-third (35.9%) of oncologists treated >50 new breast cancer patients annually; mean years in practice was 15.8. Oncologists explained 17% of the variation in RS testing but little of the variation in chemotherapy receipt (3%) controlling for clinical factors. Patients seeing an oncologist who was one standard deviation above the mean use of RS testing had over two-times higher odds of receiving RS (2.47, 95% CI 1.47-4.15), but a parallel estimate of the association of oncologist with the odds of receiving chemotherapy was much smaller (1.39, CI 1.03-1.88).

CONCLUSIONS

Clinical algorithms have markedly reduced variation in chemotherapy use across oncologists. Oncologists' large influence on variation in RS use suggests that they variably seek tumor profiling to inform treatment decisions.

Treatment-associated toxicities reported by patients with early-stage invasive breast cancer

BACKGROUND

Patient-reported toxicities help to appraise the breast cancer treatment experience. Yet extant data come from clinical trials and health care claims, which may be biased. Using patient surveys, the authors sought to quantify the frequency, severity, and burden of treatment-associated toxicities.

METHODS

Between 2013 and 2014, the iCanCare study surveyed a population-based sample of women residing in Los Angeles County and Georgia with early-stage, invasive breast cancer. The authors assessed the frequency and severity of toxicities; correlated toxicity severity with unscheduled health care use (clinic visits, emergency department visits/hospitalizations) and physical health; and examined patient, tumor, and treatment factors associated with reporting increased toxicity severity.

RESULTS

The overall survey response rate was 71%. From the analyzed cohort of 1945 women, 866 (45%) reported at least 1 toxicity that was severe/very severe, 9% reported unscheduled clinic visits for toxicity management, and 5% visited an emergency department or hospital. Factors associated with reporting higher toxicity severity included receipt of chemotherapy (odds ratio [OR], 2.2; 95% confidence interval [95% CI], 2.0-2.5), receipt of both chemotherapy and radiotherapy (OR, 1.3; 95% CI, 1.0-1.7), and Latina ethnicity (OR vs whites: 1.3; 95% CI, 1.1-1.5). A nonsignificant increase in at least 1 severe/very severe toxicity report was observed for bilateral mastectomy recipients (OR, 1.2; 95% CI, 1.0-1.4).

CONCLUSIONS

Women with early-stage invasive breast cancer report substantial treatment-associated toxicities and related burden. Clinicians should collect toxicity data routinely and offer early intervention. Toxicity differences observed by treatment modality may inform decision making. Cancer 2017;123:1925-1934. © 2017 American Cancer Society.

Corrected ROC analysis for misclassified binary outcomes

Creating accurate risk prediction models from Big Data resources such as Electronic Health Records (EHRs) is a critical step toward achieving precision medicine. A major challenge in developing these tools is accounting for imperfect aspects of EHR data, particularly the potential for misclassified outcomes. Misclassification, the swapping of case and control outcome labels, is well known to bias effect size estimates for regression prediction models. In this paper, we study the effect of misclassification on accuracy assessment for risk prediction models and find that it leads to bias in the area under the curve (AUC) metric from standard ROC analysis. The extent of the bias is determined by the false positive and false negative misclassification rates as well as disease prevalence. Notably, we show that simply correcting for misclassification while building the prediction model is not sufficient to remove the bias in AUC. We therefore introduce an intuitive misclassification-adjusted ROC procedure that accounts for uncertainty in observed outcomes and produces bias-corrected estimates of the true AUC. The method requires that misclassification rates are either known or can be estimated, quantities typically required for the modeling step. The computational simplicity of our method is a key advantage, making it ideal for efficiently comparing multiple prediction models on very large datasets. Finally, we apply the correction method to a hospitalization prediction model from a cohort of over 1 million patients from the Veterans Health Administrations EHR. Implementations of the ROC correction are provided for Stata and R. Published 2017. This article is a U.S. Government work and is in the public domain in the USA.