Adjusting Risk Adjustment - Accounting for Variation in Diagnostic Intensity

Management Practices and Quality of Care: Evidence from the Private Health Care Sector in Tanzania

We measure the adoption of management practices in over 220 private for-profit and non-profit health facilities in 64 districts across Tanzania and link these data to process quality-of-care metrics, assessed using undercover standardised patients and clinical observations. We find that better managed health facilities are more likely to provide correct treatment in accordance with national treatment guidelines, adhere to a checklist of essential questions and examinations, and comply with infection prevention and control practices. Moving from the 10th to the 90th percentile in the management practice score is associated with a 48% increase in correct treatment. We then leverage a large-scale field experiment of an internationally recognised management support intervention in which health facilities are assessed against comprehensive standards, given an individually tailored quality improvement plan and supported through training and mentoring visits. We find zero to small effects on management scores, suggesting that improving management practices in this setting may be challenging.

Essential Concepts for Reducing Bias in Observational Studies

Randomized controlled trials (RCTs) are the gold standard study design for clinical research, as prospective randomization, at least in theory, balances any differences that can exist between groups (including any differences not measured as part of the study) and isolates the studied treatment effect. Any remaining imbalances after randomization are attributable to chance. However, there are many barriers to conducting RCTs within pediatric populations, including lower disease prevalence, high costs, inadequate funding, and additional regulatory requirements. Researchers thus frequently use observational study designs to address many research questions. Observational studies, whether prospective or retrospective, do not involve randomization and thus have more potential for bias when compared with RCTs because of imbalances that can exist between comparison groups. If these imbalances are associated with both the exposure of interest and the outcome, then failure to account for these imbalances may result in a biased conclusion. Understanding and addressing differences in sociodemographic and/or clinical characteristics within observational studies are thus necessary to reduce bias. Within this Method/ology submission we describe techniques to minimize bias by controlling for important measurable covariates within observational studies and discuss the challenges and opportunities in addressing specific variables.

Elevated end-of-life spending: A new measure of potentially wasteful health care spending at the end of life

OBJECTIVE

To construct a new measure of end-of-life (EoL) spending-the elevated EoL spending-and examine its associations with measures of quality of care and patient and physician preferences in comparison with the commonly used total Medicare EoL spending measures.

DATA SOURCES AND STUDY SETTING

Medicare claims data for a 20% random sample of Medicare fee-for-service (FFS) patients, from the health care quality data for 2015-2016, from the Hospital Compare and the Medicare Geographic Variation public use file, and survey data about patient and physician preferences.

STUDY DESIGN

We constructed the elevated EoL spending measure as the differential monthly spending between decedents and survivors with the same one-year mortality risk, where the risk was predicted using machine learning models. We then examined the associations of the hospital referral region (HRR)-level elevated EoL spending with various health care quality measures and with the survey-elicited patient and provider preferences. We also examined analogous associations for monthly total EoL spending on decedents.

DATA EXTRACTION METHODS

Medicare FFS patients who were continuously enrolled in Medicare Parts A & B in 2015 and were alive as of January 1, 2016.

PRINCIPAL FINDINGS

We found a large variation in the elevated EoL spending across HRRs in the United States. There was no evidence of an association between HRR-level elevated EoL spending and established health care quality measures, including those specific to EoL care, whereas total EoL spending was positively associated with certain quality of care measures. We also found no evidence that elevated EoL spending was associated with patient preferences for EoL care. However, elevated EoL spending was positively and significantly associated with physician preferences for treatment intensity.

CONCLUSIONS

Our findings suggested that elevated EoL spending captures different resource use from conventional measures of EoL spending and may be more valuable in identifying potentially wasteful spending.

Team Relationships and Performance: Evidence from Healthcare Referral Networks

We examine the teams that emerge when a primary care physician (PCP) refers patients to specialists. When PCPs concentrate their specialist referrals-for instance, by sending their cardiology patients to fewer distinct cardiologists-repeat interactions between PCPs and specialists are encouraged. Repeated interactions provide more opportunities and incentives to develop productive team relationships. Using data from the Massachusetts All Payer Claims Database, we construct a new measure of PCP team referral concentration and document that it varies widely across PCPs, even among PCPs in the same organization. Chronically ill patients treated by PCPs with a one standard deviation higher team referral concentration have 4% lower health care utilization on average, with no discernible reduction in quality. We corroborate this finding using a national sample of Medicare claims and show that it holds under various identification strategies that account for observed and unobserved patient and physician characteristics. The results suggest that repeated PCP-specialist interactions improve team performance.

Association between socioeconomic status and presenting characteristics and extent of disease in patients with surgically resected nonfunctioning pituitary adenoma

OBJECTIVE

The aim of this study was to evaluate the association between zip code-level socioeconomic status (SES) and presenting characteristics and short-term clinical outcomes in patients with nonfunctioning pituitary adenoma (NFPA).

METHODS

A retrospective review of prospectively collected data from the University of Southern California Pituitary Center was conducted to identify all patients undergoing surgery for pituitary adenoma (PA) from 2000 to 2021 and included all patients with NFPA with recorded zip codes at the time of surgery. A normalized socioeconomic metric by zip code was then constructed using data from the American Community Survey estimates, which was categorized into tertiles. Multiple imputation was used for missing data, and multivariable linear and logistic regression models were constructed to estimate mean differences and multivariable-adjusted odds ratios for the association between zip code-level SES and presenting characteristics and outcomes.

RESULTS

A total of 637 patients were included in the overall analysis. Compared with patients in the lowest SES tertile, those in the highest tertile were more likely to be treated at a private (rather than safety net) hospital, and were less likely to present with headache, vision loss, and apoplexy. After multivariable adjustment for age, sex, and prior surgery, SES in the highest compared with lowest tertile was inversely associated with tumor size at diagnosis (-4.9 mm, 95% CI -7.2 to -2.6 mm, p < 0.001) and was positively associated with incidental diagnosis (multivariable-adjusted OR 1.72, 95% CI 1.02-2.91). Adjustment for hospital (private vs safety net) attenuated the observed associations, but disparities by SES remained statistically significant for tumor size. Despite substantial differences at presentation, there were no significant differences in length of stay or odds of an uncomplicated procedure by zip code-level SES. Patients from lower-SES zip codes were more likely to require postoperative steroid replacement and less likely to achieve gross-total resection.

CONCLUSIONS

In this series, lower zip code-level SES was associated with more severe disease at the time of diagnosis for NFPA patients, including larger tumor size and lower rates of incidental diagnosis. Despite these differences at presentation, no significant differences were observed in short-term postoperative complications, although patients with higher zip code-level SES had higher rates of gross-total resection.

Racial and Ethnic Disparities in the Management of Postdural Puncture Headache With Epidural Blood Patch for Obstetric Patients in New York State

IMPORTANCE

Characterizing and addressing racial and ethnic disparities in peripartum pain assessment and treatment is a national priority.

OBJECTIVE

To evaluate the association of race and ethnicity with the provision and timing of an epidural blood patch (EBP) for management of postdural puncture headache in obstetric patients.

DESIGN, SETTING, AND PARTICIPANTS

This cross-sectional study used New York State hospital discharge records from January 1, 1998, to December 31, 2016, from mothers 15 to 49 years of age with a postdural puncture headache after neuraxial analgesia or anesthesia for childbirth. Statistical analysis was performed from February 2020 to February 2022.

EXPOSURES

Patients' race and ethnicity (reported as provided by each participating hospital; the method of determining race and ethnicity [ie, self-reported or not] cannot be determined from the data) were categorized into non-Hispanic White (reference group), non-Hispanic Black, Hispanic, and other race and ethnicity (including Asian and Pacific Islander, American Indian, Alaskan Native, and other).

MAIN OUTCOMES AND MEASURES

The primary outcome was the rate of EBP use. The secondary outcome was the interval (days) between hospital admission and provision of EBP. Odds ratios (ORs) and 95% CIs of EBP use associated with race and ethnicity were estimated using mixed-effect logistic regression models, adjusting for patient and hospital characteristics.

RESULTS

During the study period, 8921 patients (mean [SD] age, 30 [6] years; 1028 [11.5%] Black; 1301 [14.6%] Hispanic; 4960 [55.6%] White; and 1359 [15.2%] other race and ethnicity) with postdural puncture headache were identified among 1.9 million deliveries with a neuraxial procedure. Of these 8921 patients, 4196 (47.0%; 95% CI, 46.0%-48.1%) were managed with an EBP. A total of 2650 White patients (53.4%; 95% CI, 52.0%-54.8%) used an EBP; this rate was significantly higher than that among Hispanic patients (41.7% [543]; 95% CI, 39.9%-44.5%), Black patients (35.7% [367]; 95% CI, 32.8%-38.7%), or patients of other race and ethnicity (35.2% [478]; 95% CI, 32.6%-37.8%). Timing of EBP was at a median of 2 days (IQR, 2-3 days) after hospital admission for White patients compared with a median of 3 days (IQR, 2-4 days) for Hispanic patients, Black patients, and patients of other race and ethnicity (P < .001 for the comparison with White patients). After adjustment for patient and hospital characteristics, the EBP rate was not different between White and Hispanic patients (adjusted OR, 1.11; 95% CI, 0.94-1.30). It was significantly lower for Black patients (adjusted OR, 0.80; 95% CI, 0.67-0.94) and patients of other races and ethnicities (adjusted OR, 0.85; 95% CI, 0.73-0.99).

CONCLUSIONS AND RELEVANCE

In this study, significant racial and ethnic disparities in the management of postdural puncture headache with EBP were observed, with both lower rates and delayed timing, which may be associated with long-term adverse effects.

Use of past care markers in risk-adjustment: accounting for systematic differences across providers

Risk-adjustment models are used to predict the cost of care for patients based on their observable characteristics, and to derive efficient and equitable budgets based on weighted capitation. Markers based on past care contacts can improve model fit, but their coefficients may be affected by provider variations in diagnostic, treatment and reporting quality. This is problematic when distinguishing need and supply influences on costs is required.We examine the extent of this bias in the national formula for mental health care using administrative records for 43.7 million adults registered with 7746 GP practices in England in 2015. We also illustrate a method to control for provider effects.A linear regression containing a rich set of individual, GP practice and area characteristics, and fixed effects for local health organisations, had goodness-of-fit equal to R² = 0.007 at person level and R² = 0.720 at GP practice level. The addition of past care markers changed substantially the coefficients on the other variables and increased the goodness-of-fit to R² = 0.275 at person level and R² = 0.815 at GP practice level. The further inclusion of provider effects affected the coefficients on GP practice and area variables and on local health organisation fixed effects, increasing goodness-of-fit at GP practice level to R² = 0.848.With adequate supply controls, it is possible to estimate coefficients on past care markers that are stable and unbiased. Nonetheless, inconsistent reporting may affect need predictions and penalise populations served by underreporting providers.

Place-Based Drivers of Mortality: Evidence from Migration

We estimate the effect of current location on elderly mortality by analyzing outcomes of movers in the Medicare population. We control for movers' origin locations as well as a rich vector of pre-move health measures. We also develop a novel strategy to adjust for remaining unobservables, using the correlation of residual mortality with movers' origins to gauge the importance of omitted variables. We estimate substantial effects of current location. Moving from a 10th to a 90th percentile location would increase life expectancy at age 65 by 1.1 years, and equalizing location effects would reduce cross-sectional variation in life expectancy by 15 percent. Places with favorable life expectancy effects tend to have higher quality and quantity of health care, less extreme climates, lower crime rates, and higher socioeconomic status.

Mortality Effects and Choice Across Private Health Insurance Plans

Competition in health insurance markets may fail to improve health outcomes if consumers are not able to identify high quality plans. We develop and apply a novel instrumental variables framework to quantify the variation in causal mortality effects across plans and how much consumers attend to this variation. We first document large differences in the observed mortality rates of Medicare Advantage plans within local markets. We then show that when plans with high (low) mortality rates exit these markets, enrollees tend to switch to more typical plans and subsequently experience lower (higher) mortality. We derive and validate a novel "fallback condition" governing the subsequent choices of those affected by plan exits. When the fallback condition is satisfied, plan terminations can be used to estimate the relationship between observed plan mortality rates and causal mortality effects. Applying the framework, we find that mortality rates unbiasedly predict causal mortality effects. We then extend our framework to study other predictors of plan mortality effects and estimate consumer willingness to pay. Higher spending plans tend to reduce enrollee mortality, but existing quality ratings are uncorrelated with plan mortality effects. Consumers place little weight on mortality effects when choosing plans. Good insurance plans dramatically reduce mortality, and redirecting consumers to such plans could improve beneficiary health.

Social Determinants of Health and Geographic Variation in Medicare per Beneficiary Spending

IMPORTANCE

Despite substantial geographic variation in Medicare per beneficiary spending in the US, little is known about the extent to which social determinants of health (SDoH) are associated with this variation.

OBJECTIVE

To determine the associations between SDoH and county-level price-adjusted Medicare per beneficiary spending.

DESIGN, SETTING, AND PARTICIPANTS

This cross-sectional study used county-level data on 2017 Medicare fee-for-service (FFS) spending, patient demographic characteristics (eg, age and gender) and clinical risk score, supply of health care resources (eg, number of hospital beds), and SDoH measures (eg, median income and unemployment rate) from multiple sources. Multivariable regressions were used to estimate the association of the variation in spending across quintiles with SDoH.

MAIN OUTCOMES AND MEASURES

2017 county-level price-adjusted Medicare Parts A and B spending per beneficiary. SDoH measures included socioeconomic position, race/ethnicity, social relationships, and residential and community context.

RESULTS

Among 3038 counties with 33 495 776 Medicare FFS beneficiaries (18 352 336 [54.8%] women; mean [SD] age, 72 [1.5] years), mean Medicare price-adjusted per beneficiary spending for counties in the highest spending quintile was $3785 (95% CI, $3706-$3862) higher, or 49% higher, than spending for bottom-quintile counties (mean [SD] spending per beneficiary, $11 464 [735] vs $7679 [522]; P < .001). The total contribution (including through both direct and indirect pathways) of SDoH was 37.7% ($1428 of $3785) of this variation, compared with 59.8% ($2265 of $3785) by patient clinical risk, 14.5% ($549 of $3785) by supply of health care resources, and 19.8% ($751 of $3785) by patient demographic characteristics. When all factors were included within the same model, the direct contribution of SDoH was associated with 5.8% of the variation, compared with 4.6% by supply, 4.7% by patient demographic characteristics, and 62.0% by patient clinical risk.

CONCLUSIONS AND RELEVANCE

These findings suggest social determinants of health are associated with considerable proportions of geographic variation in Medicare spending. Policies addressing SDoH for disadvantaged patients in certain regions have the potential to contain health care spending and improve the value of health care; patient SDoH may need to be accounted for in publicly reported physician performance, and in value-based purchasing incentive programs for health care professionals.

The 2014 New York State Medicaid Expansion and Severe Maternal Morbidity During Delivery Hospitalizations

BACKGROUND

Medicaid expansions under the Affordable Care Act have increased insurance coverage and prenatal care utilization in low-income women. However, it is not clear whether they are associated with any measurable improvement in maternal health outcomes. In this study, we compared the changes in the incidence of severe maternal morbidity (SMM) during delivery hospitalizations between low- and high-income women associated with the 2014 Medicaid expansion in New York State.

METHODS

Data for this retrospective cohort study came from the 2006-2016 New York State Inpatient Database, a census of discharge records from community hospitals. The outcome was SMM during delivery hospitalizations, as defined by the Centers for Disease Control and Prevention. We used regression coefficients (β) from multivariable logistic models: (1) to compare independently in low-income women and in high-income women the changes in slopes in the incidence of SMM before (2006-2013) and after (2014-2016) the expansion, and (2) to compare low- and high-income women for the changes in slopes in the incidence of SMM before and after the expansion.

RESULTS

A total of 2,286,975 delivery hospitalizations were analyzed. The proportion of Medicaid beneficiaries in parturients increased a relative 12.1% (95% confidence interval [CI], 11.8-12.4), from 42.9% in the preexpansion period to 48.1% in the postexpansion period, whereas the proportion of the uninsured decreased a relative 4.8% (95% CI, 2.8-6.8). Multivariable logistic modeling revealed that implementation of the 2014 Medicaid expansion was associated with a decreased slope during the postexpansion period both in low-income women (β = -0.0161 or 1.6% decrease; 95% CI, -0.0190 to -0.0132) and in high-income women (β = -0.0111 or 1.1% decrease; 95% CI, -0.0130 to -0.0091). The decrease in slope during the postexpansion period was greater in low- than in high-income women (β = -0.0042 or 0.42% difference; 95% CI, -0.0076 to -0.0007).

CONCLUSIONS

Implementation of the Medicaid expansion in 2014 in New York State is associated with a small but statistically significant reduction in the incidence of SMM in low-income women compared with high-income women.

Geographic Variation In Medicare Per Capita Spending Narrowed From 2007 To 2017

We examined the trends in geographic variation in Medicare per capita spending and growth from 2007 to 2017 and found that the variation narrowed during this period. The difference in Medicare price- and risk-adjusted per capita spending between hospital referral regions (HRRs) in the top decile and those in the bottom decile decreased from $3,388 in 2007 to $2,916 in 2017-a reduction of $472, or 14 percent. The spending convergence occurred almost entirely between 2009 and 2014, during the early years of the Affordable Care Act (ACA). The highest-spending HRRs in 2007 had the lowest annual growth rates from 2007 to 2017, and the lowest-spending HRRs in 2007 had the highest annual growth rates. We also found that a greater supply of postacute care providers, especially hospice providers, significantly predicted lower spending growth across HRRs after the implementation of the ACA.

Patient Coded Severity and Payment Penalties Under the Hospital Readmissions Reduction Program: A Machine Learning Approach

OBJECTIVE

The objective of this study was to examine variation in hospital responses to the Centers for Medicare and Medicaid's expansion of allowable secondary diagnoses in January 2011 and its association with financial penalties under the Hospital Readmission Reduction Program (HRRP).

DATA SOURCES/STUDY SETTING

Medicare administrative claims for discharges between July 2008 and June 2011 (N=3102 hospitals).

RESEARCH DESIGN

We examined hospital variation in response to the expansion of secondary diagnoses by describing changes in comorbidity coding before and after the policy change. We used random forest machine learning regression to examine hospital characteristics associated with coded severity. We then used a 2-part model to assess whether variation in coded severity was associated with readmission penalties.

RESULTS

Changes in severity coding varied considerably across hospitals. Random forest models indicated that greater baseline levels of condition categories, case-mix index, and hospital size were associated with larger changes in condition categories. Hospital coding of an additional condition category was associated with a nonsignificant 3.8 percentage point increase in the probability for penalties under the HRRP (SE=2.2) and a nonsignificant 0.016 percentage point increase in penalty amount (SE=0.016).

CONCLUSION

Changes in patient coded severity did not affect readmission penalties.

Deriving Evidence from Secondary Data in Hand Surgery: Strengths, Limitations, and Future Directions

Health services research using secondary data is a powerful tool for guiding quality/performance measure development, payment reform, and health policy. Patient preferences, physical examination findings, use of postoperative care, and other factors specific to hand surgery research are critical pieces of information required to study quality of care and improve patient outcomes. These data often are missing from data sets, causing limitations and challenges when performing secondary data analyses in hand surgery. As the role of secondary data in surgical research expands, hand surgeons must apply novel strategies and become involved in collaborative initiatives to overcome the limitations of existing resources.

Evaluating Measures of Hospital Quality:Evidence from Ambulance Referral Patterns

Hospital quality measures are crucial to a key idea behind health care payment reforms: "paying for quality" instead of quantity. Nevertheless, such measures face major criticisms largely over the potential failure of risk adjustment to overcome endogeneity concerns when ranking hospitals. In this paper we test whether patients treated at hospitals that score higher on commonly-used quality measures have better health outcomes in terms of rehospitalization and mortality. To compare similar patients across hospitals in the same market, we exploit ambulance company preferences as an instrument for hospital choice. We find that a variety of measures used by insurers to measure provider quality are successful: choosing a high-quality hospital compared to a low-quality hospital results in 10-15% better outcomes.

Major Neurologic Complications Associated With Postdural Puncture Headache in Obstetrics: A Retrospective Cohort Study

BACKGROUND

Increased risks of cerebral venous thrombosis or subdural hematoma, bacterial meningitis, persistent headache, and persistent low back pain are suggested in obstetric patients with postdural puncture headache (PDPH). Acute postpartum pain such as PDPH may also lead to postpartum depression. This study tested the hypothesis that PDPH in obstetric patients is associated with significantly increased postpartum risks of major neurologic and other maternal complications.

METHODS

This retrospective cohort study consisted of 1,003,803 women who received neuraxial anesthesia for childbirth in New York State hospitals between January 2005 and September 2014. The primary outcome was the composite of cerebral venous thrombosis and subdural hematoma. The 4 secondary outcomes were bacterial meningitis, depression, headache, and low back pain. PDPH and complications were identified during the delivery hospitalization and up to 1 year postdelivery. Adjusted odds ratios (aORs) and 95% confidence intervals (CIs) were estimated using the inverse probability of treatment weighting approach.

RESULTS

Of the women studied, 4808 (0.48%; 95% CI, 0.47-0.49) developed PDPH, including 264 cases (5.2%) identified during a readmission with a median time to readmission of 4 days. The incidence of cerebral venous thrombosis and subdural hematoma was significantly higher in women with PDPH than in women without PDPH (3.12 per 1000 neuraxial or 1:320 vs 0.16 per 1000 or 1:6250, respectively; P < .001). The incidence of the 4 secondary outcomes was also significantly higher in women with PDPH than in women without PDPH. The aORs associated with PDPH were 19.0 (95% CI, 11.2-32.1) for the composite of cerebral venous thrombosis and subdural hematoma, 39.7 (95% CI, 13.6-115.5) for bacterial meningitis, 1.9 (95% CI, 1.4-2.6) for depression, 7.7 (95% CI, 6.5-9.0) for headache, and 4.6 (95% CI, 3.3-6.3) for low back pain. Seventy percent of cerebral venous thrombosis and subdural hematoma were identified during a readmission with a median time to readmission of 5 days.

CONCLUSIONS

PDPH is associated with substantially increased postpartum risks of major neurologic and other maternal complications, underscoring the importance of early recognition and treatment of anesthesia-related complications in obstetrics.

Cumulative Doses of Ionizing Radiation From Computed Tomography: A Population-Based Study

OBJECTIVE

To assess cumulative radiation doses from computed tomography (CT), patient characteristics, and clinical indications for CT in a population-based sample.

PATIENTS AND METHODS

A cohort study using medical records linkage through the Rochester Epidemiology Project was conducted to ascertain all CT examinations in Olmsted County, Minnesota, performed between January 1, 2004, and December 31, 2013, among all adults who were alive for 3 or more years after the end of follow-up (to exclude exposures preceding death). Ten-year cumulative effective ionizing radiation doses were estimated on the basis of typical doses per CT modality. Among patients with high doses (≥100 mSv/10 years), CT scans were reviewed for clinical setting, indications, and results.

RESULTS

Of 54,447 adults (median age, 44.0 years at inclusion), 26,377 (48.4%) underwent at least one CT. Ten-year radiation doses from CT were 0.1 to 9.9 mSv in 15.8% of the population (8593 patients), 10 to 24.9 mSv in 16.9% (9502), 25 to 99.9 mSv in 13.8% (7492), and 100 mSv or greater in 1.9% (1041). Computed tomography of the abdomen and pelvis accounted for 67.2% of the estimated dose. In multivariable models, doses differed 1.21-fold to 2.16-fold between extreme categories of age, body mass index, education level, smoking status, and by race. Of 600 CTs in 200 patients with high doses, 70.5% were obtained for restaging of solid cancers and lymphoma, abdominal pain, infection, kidney stones, follow-up of nodules or masses, and chest pain/evaluation for pulmonary embolism.

CONCLUSION

Exposure to ionizing radiation from CT occurred disproportionally in specific subgroups of the population. A limited number of clinical indications contributed the majority of radiation among adults with high doses.

Local Area Variation in Morbidity Among Low-Income, Older Adults in the United States: A Cross-sectional Study

Background

Recent studies have reported that low-income adults living in more affluent areas of the United States have longer life expectancies. Less is known about the relationship between the affluence of a geographic area and morbidity of the low-income population.

Objective

To evaluate the association between the prevalence of chronic conditions among low-income, older adults and the economic affluence of a local area.

Design

Cross-sectional association study.

Setting

Medicare in 2015.

Participants

6 363 097 Medicare beneficiaries aged 66 to 100 years with a history of low-income support under Medicare Part D.

Measurements

Adjusted prevalence of 48 chronic conditions was computed for 736 commuting zones (CZs). Factor analysis was used to assess spatial covariation of condition prevalence and to construct a composite condition prevalence index for each CZ. The association between morbidity and area affluence was measured by comparing the average of condition prevalence index across deciles of median CZ house values.

Results

The mean age of study participants was 77.7 years (SD, 8.2); 67% were women, and 61% were white. The crude prevalence of 48 chronic conditions ranged from 72.5 per 100 for hypertension to 0.6 per 100 for posttraumatic stress disorder. The prevalence of these 48 chronic conditions was highly spatially correlated. Composite condition prevalence was on average substantially lower in more affluent CZs.

Limitation

Low-income status measured on the basis of receipt of Medicare Part D low-income subsidies and not capturing persons not enrolled in Medicare Part D.

Conclusion

Low-income, older adults living in more affluent areas of the country are healthier, and areas with poor health in the low-income, older adult population tend to have a high prevalence of most chronic conditions.

Primary Funding Source

National Institute on Aging.

Engaging Primary Care Providers to Reduce Unwanted Clinical Variation and Support ACO Cost and Quality Goals: A Unique Provider-Payer Collaboration

This project was undertaken to reduce unneeded variation among practicing primary care clinicians participating in an accountable care organization (ACO) and to raise quality and reduce costs. This real-world, quasi-controlled experiment compared ACO target improvements between 3 participating geographic regions and members within the ProHealth ACO against nonparticipating regions and members. The authors used a novel care standardization initiative to engage participating providers. This was a 2-year longitudinal study with 6 rounds of serially measured provider care decisions and customized individual and group improvement feedback. Participating providers cared for online patient simulations as they would actual patients, and their care decisions were scored against evidence-based guidelines. This approach generated significant increases in evidence-based quality scores (+27%) and reductions in unneeded testing (-55%) in the patient simulations. Improvements in the online simulated patients correlated with improvements in patient-level ACO quality measures, which showed gains above and beyond the quasi-control group. Reductions calculated for spending on unneeded tests and specialist referrals exceeded $4.8 million. This study found that supporting practicing physicians in ACOs with evidence-based feedback significantly improved care and cost-efficiency.

Adoption and learning across hospitals: The case of a revenue-generating practice

Performance-raising practices tend to diffuse slowly in the health care sector. To understand how incentives drive adoption, I study a practice that generates revenue for hospitals: submitting detailed documentation about patients. After a 2008 reform, hospitals could raise their Medicare revenue over 2% by always specifying a patient's type of heart failure. Hospitals only captured around half of this revenue, indicating that large frictions impeded takeup. Exploiting the fact that many doctors practice at multiple hospitals, I find that four-fifths of the dispersion in adoption reflects differences in the ability of hospitals to extract documentation from physicians. A hospital's adoption of coding is robustly correlated with its heart attack survival rate and its use of inexpensive survival-raising care. Hospital-physician integration and electronic medical records are also associated with adoption. These findings highlight the potential for institution-level frictions, including agency conflicts, to explain variations in health care performance across providers.

Predictive modeling of U.S. health care spending in late life

That one-quarter of Medicare spending in the United States occurs in the last year of life is commonly interpreted as waste. But this interpretation presumes knowledge of who will die and when. Here we analyze how spending is distributed by predicted mortality, based on a machine-learning model of annual mortality risk built using Medicare claims. Death is highly unpredictable. Less than 5% of spending is accounted for by individuals with predicted mortality above 50%. The simple fact that we spend more on the sick-both on those who recover and those who die-accounts for 30 to 50% of the concentration of spending on the dead. Our results suggest that spending on the ex post dead does not necessarily mean that we spend on the ex ante "hopeless."

Missing data and chance variation in public reporting of cancer stage at diagnosis: Cross-sectional analysis of population-based data in England

BACKGROUND

The percentage of cancer patients diagnosed at an early stage is reported publicly for geographically-defined populations corresponding to healthcare commissioning organisations in England, and linked to pay-for-performance targets. Given that stage is incompletely recorded, we investigated the extent to which this indicator reflects underlying organisational differences rather than differences in stage completeness and chance variation.

METHODS

We used population-based data on patients diagnosed with one of ten cancer sites in 2013 (bladder, breast, colorectal, endometrial, lung, ovarian, prostate, renal, NHL, and melanoma). We assessed the degree of bias in CCG (Clinical Commissioning Group) indicators introduced by missing-is-late and complete-case specifications compared with an imputed 'gold standard'. We estimated the Spearman-Brown (organisation-level) reliability of the complete-case specification. We assessed probable misclassification rates against current pay-for-performance targets.

RESULTS

Under the missing-is-late approach, bias in estimated CCG percentage of tumours diagnosed at an early stage ranged from -2 to -30 percentage points, while bias under the complete-case approach ranged from -2 to +7 percentage points. Using an annual reporting period, indicators based on the least biased complete-case approach would have poor reliability, misclassifying 27/209 (13%) CCGs against a pay-for-performance target in current use; only half (53%) of CCGs apparently exceeding the target would be correctly classified in terms of their underlying performance.

CONCLUSIONS

Current public reporting schemes for cancer stage at diagnosis in England should use a complete-case specification (i.e. the number of staged cases forming the denominator) and be based on three-year reporting periods. Early stage indicators for the studied geographies should not be used in pay-for-performance schemes.

Hospital costs associated with psychiatric comorbidities: a retrospective study

BACKGROUND

Psychiatric comorbidities are relevant for the diagnostic and therapeutic regimes in somatic hospital care. The main aim of this study was to analyse the association between psychiatric comorbidities and hospital costs per inpatient episode. A further aim was to discuss and address the methodological challenges in the estimation of these outcomes based on retrospective data.

METHODS

The study included 338,162 inpatient episodes consecutively discharged between 2011 and 2014 at a German university hospital. We used detailed resource use data to calculate day-specific hospital costs. We adjusted analyses for sex, age, somatic comorbidities and main diagnoses. We addressed potential time-related bias in retrospective diagnosis data with sensitivity analyses.

RESULTS

Psychiatric comorbidities were associated with an increase in hospital costs per episode of 40% and an increase of reimbursement per episode of 28%, representing marginal effects of 1344 € and 1004 €, respectively. After controlling for length of stay, sensitivity analyses provided a lower bound increase in daily costs and reimbursement of 207 € and 151 €, respectively.

CONCLUSION

If differences in hospital costs between patient groups are not adequately accounted for in DRG-systems, perverse incentives are created that can reduce the efficiency of care. Therefore, we suggest intensifying the inclusion of psychiatric comorbidities in the German DRG system. Future research should investigate the appropriate inclusion of psychiatric comorbidities in other health care systems' payment schemes.

Consistency of pressure injury documentation across interfacility transfers

BACKGROUND

Hospital-acquired pressure injuries (HAPIs) are publicly reported in the USA and used to adjust Medicare payment to acute inpatient facilities. Current methods used to identify HAPIs in administrative claims rely on hospital-reported present-on-admission (POA) data instead of prior patient health information.

OBJECTIVE

To study the reliability of claims data for HAPIs and pressure injury (PI) stage by evaluating diagnostic coding agreement across interfacility transfers.

METHODS

Using the 2012 100% Medicare Provider and Analysis Review file, we identified all fee-for-service acute inpatient discharge records with a PI diagnosis among Medicare patients 65 years and older. We then identified additional facility claims (eg, acute inpatient, long-stay inpatient or skilled nursing facility) belonging to the same patient who had either (1) admission within 1day of hospital discharge or (2) discharge within 1day of hospital admission. Multivariable logistic regression and stratified kappa statistics were used to measure coding agreement between transferring and receiving facilities in the presence or absence of a PI diagnosis at the time of patient transfer and PI stage category (early vs advanced).

RESULTS

In our comparison of claims data between transferring and receiving facilities, we observed poor agreement in the presence or absence of a PI diagnosis at the time of transfer (36.3%, kappa=0.03) and poor agreement in PI stage category (74.3%, kappa=0.17). Among transfers with a POA PI reported by the receiving hospital, only 34.0% had a PI documented at the prior transferring facility.

CONCLUSIONS

The observed discordance in PI documentation and staging between transferring and receiving facilities may indicate inaccuracy of HAPI identification in claims data. Future research should evaluate the accuracy of hospital-reported POA data and its impact on PI quality measurement.