Differences in medicaid expansion effects on buprenorphine treatment utilization by county rurality and income: A pharmacy data claims analysis from 2009-2018

Background

Although use of buprenorphine for treating opioid use disorder increased over the past decade, buprenorphine utilization remains limited in lower-income and rural areas. We examine how the Affordable Care Act Medicaid expansion influenced buprenorphine initiation rates by county income and evaluate how associations differ by county rural-urban status.

Methods

This study used nationwide 2009-2018 IQVIA retail pharmacy data and a comparative interrupted time series framework-a hybrid framework combining regression discontinuity and difference-in-difference approaches. We used piecewise linear estimation to quantify changes in buprenorphine initiation rates before and after Medicaid expansion.

Results

The sample included observations from 376,704 county-months. We identified 5,227,340 new buprenorphine treatment episodes, with an average of 9.2 new buprenorphine episodes per month per 100,000 county residents. Among urban counties, those with the lowest median incomes experienced significantly larger increases in buprenorphine initiation rates associated with Medicaid expansion than counties with higher median incomes (5-year rates difference est=3525.3, se=1695.3, p = 0.04). However, among rural counties, there was no significant association between buprenorphine initiation rates and county median income after Medicaid expansion (5-year rates difference est=979.0, se=915.8, p = 0.29).

Conclusions

Medicaid expansion was associated with a reduction in income-related buprenorphine disparities in urban counties, but not in rural counties. To achieve more equitable buprenorphine access, future policies should target low-income rural areas.

Buprenorphine Treatment Episodes During the First Year of COVID: a Retrospective Examination of Treatment Initiation and Retention

BACKGROUND

During the COVID pandemic, overall buprenorphine treatment appeared to remain relatively stable, despite some studies suggesting a decrease in patients starting buprenorphine. There is a paucity of empirical information regarding patterns of buprenorphine treatment during the pandemic.

OBJECTIVE

To better understand the patterns of buprenorphine episodes during the pandemic and how those patterns compared to pre-pandemic patterns.

DESIGN

Pharmacy claims representing approximately 92% of all prescriptions filled at retail pharmacies in all 50 US states and the District of Columbia.

PARTICIPANTS

Individuals filling buprenorphine prescriptions indicated for treatment of opioid use disorder.

MAIN MEASURES

The number of active, starting, and ending buprenorphine treatment episodes March 13 to December 1, 2020, and the expected number of such episodes in 2020 based on the growth in treatment episodes from March 13 to December 1, 2019.

KEY RESULTS

The observed number of active buprenorphine episodes in December 2020 was comparable to the expected number, but new treatment episodes starting between March 13 and December 1, 2020, were 17.2% fewer than expected based on the 2019 experience. Similarly, the number of episodes that ended between March 13 and December 1, 2020, was 16.0% fewer than expected. Decreases from expected episode starts and ends occurred throughout the period but were greatest in the 2 months after the declaration of the public health emergency.

CONCLUSIONS AND RELEVANCE

Beneath the apparent stability of buprenorphine patient numbers during the pandemic, the flow of individuals receiving buprenorphine treatment changed substantially. Our findings shed light on how policy changes meant to support buprenorphine prescribing influenced prescribing dynamics during that period, suggesting that while policy efforts may have been successful in maintaining existing patients in treatment, that success did not extend to individuals not yet in treatment.

Medicaid Payment and Fluoride Varnish Application During Pediatric Medical Visits

All Medicaid programs pay for fluoride varnish applications during medical visits for infants and toddlers, but receipt of care varies considerably across states. Using 2006-2014 Medicaid data from 22 states, this study examined the association between Medicaid payment and receipt of fluoride varnish during pediatric medical visits. Among 3,393,638 medical visits, fewer than one in 10 visits included fluoride varnish. Higher Medicaid payment was positively associated with receipt of fluoride varnish during pediatric medical visits. As policymakers consider strategies for increasing young children's access to preventive oral health services, as well as consider strategies for balancing budgets, attention should be paid to the effects of provider payment on access to pediatric oral health services.

Buprenorphine treatment episode duration, dosage, and concurrent prescribing of benzodiazepines and opioid analgesics: The effects of Medicaid prior authorization policies

BACKGROUND

Buprenorphine is an effective medication for the treatment of opioid use disorder (OUD), but the association between prior authorization policies and quality of care for individuals receiving buprenorphine treatment is not well-understood.

METHODS

Using 2006-2013 Medicaid Analytic eXtract (MAX) data from 34 states and the District of Columbia, we identified 294,031 episodes of buprenorphine treatment for OUD among individuals aged 14-64 years. We estimated generalized difference-in-differences models to examine the association between buprenorphine prior authorization policies and changes in buprenorphine treatment quality along four dimensions: (1) duration of at least 180 days, (2) dosage of at least 8 milligrams, and concurrent prescribing of (3) opioid analgesics and (4) benzodiazepines.

RESULTS

Buprenorphine prior authorization policies were associated with an 11-percentage point reduction (p < 0.01) in the likelihood of episodes with a duration of at least 180 days in the first four years after policy implementation. The policy was not associated with changes in effective dosage or concurrent prescribing of opioid analgesics or benzodiazepines.

CONCLUSIONS

Buprenorphine prior authorization policies were associated with a sizeable and significant reduction in episodes of at least 180 days duration, underscoring the importance of identifying and removing barriers to effective and appropriate OUD care.

Sociodemographic differences in quality of treatment to Medicaid enrollees receiving buprenorphine

Background: Buprenorphine is a key medication to treat opioid use disorder, but little is known about how treatment quality varies across sociodemographic groups. Objective: We examined measures of treatment quality and explored variation by sociodemographic factors. Methods: We used Medicaid MAX data from 50 states from 2006 to 2014 to identify buprenorphine treatment episodes (N = 317,494). We used multivariable logistic regression to examine the quality of buprenorphine treatment along four dimensions: (1) sufficient duration, (2) effective dosage, and concurrent prescribing of (3) opioid analgesics and (4) benzodiazepines. We explored how quality varied by race/ethnicity, age, sex, and urbanicity. Results: In adjusted models, compared to non-Hispanic White individuals, non-Hispanic Black and Hispanic individuals had lower odds of receiving effective dosage (aORs = 0.79 and 0.89, respectively) and sufficient duration (aORs = 0.64 and 0.71, respectively), and lower odds of concurrent prescribing of opioid analgesics (aORs = 0.86 and 0.85, respectively) and benzodiazepines (aORs = 0.51 and 0.59, respectively). Older individuals had higher odds of sufficient duration (aORs from 1.21-1.33), but also had higher odds of concurrent opioid analgesics prescribing (aORs from 1.29-1.56) and benzodiazepines (aORs from 1.44-1.99). Females had higher odds of sufficient duration (aOR = 1.12), but lower odds of effective dosage (aOR = 0.77) and higher odds of concurrent prescribing of opioid analgesics (aOR = 1.25) and benzodiazepines (aOR = 1.16). Compared to individuals living in metropolitan areas, individuals living in non-metropolitan areas had higher odds of sufficient duration (aORs = 1.11 and 1.24) and effective dosage (aORs = 1.06 and 1.33), and lower odds of concurrent prescribing (aORs from 0.81-0.98). Conclusions: Black and Hispanic individuals were less likely to receive effective buprenorphine dosage and sufficient duration. Quality results were mixed for older and female individuals; although these individuals were more likely to receive treatment of sufficient duration, they were also more likely to be concurrently prescribed potentially contraindicated medications, and females were less likely to receive effective dosage. Findings raise concerns about adequacy of care for minority and other at-risk populations.

State Policies and Buprenorphine Prescribing by Nurse Practitioners and Physician Assistants

Nurse practitioner (NP) and physician assistant (PA) prescribing can increase access to buprenorphine treatment for opioid use disorder. In this cross-sectional study, we used deidentified claims from approximately 90% of U.S. retail pharmacies (2017-2018) to examine the association of state policies with the odds of receiving buprenorphine treatment from an NP/PA versus a physician, overall and stratified by urban/rural status. From 2017 to 2018, the percentage of buprenorphine treatment episodes prescribed by NPs/PAs varied widely across states, from 0.4% in Alabama to 57.2% in Montana. Policies associated with greater odds of buprenorphine treatment from an NP/PA included full scope of practice (SOP) for NPs, full SOP for PAs, Medicaid pay parity for NPs (reimbursement at 100% of the fee-for-service physician rate), and Medicaid expansion. Although most findings with respect to policies were similar in urban and rural settings, the association of Medicaid expansion with NP/PA buprenorphine treatment was driven by rural counties.

Differences in buprenorphine treatment quality across physician provider specialties

BACKGROUND

The number and types of clinicians prescribing buprenorphine treatment for opioid use disorder (OUD) have increased over the past two decades, but there is little information on how potential indicators of quality of care to patients receiving buprenorphine vary by provider specialty.

METHODS

We used the Medicaid Analytic eXtract from 2009 to 2014 to identify buprenorphine treatment episodes. We assigned physician specialties to episodes based on whether an episode had at least one outpatient claim linked to specialists in addiction, behavioral health, opioid treatment program (OTP), pain, or primary care provider (PCP). We then used logistic regressions to estimate the association of linked physician specialty and achievement of the following process of care measures: at least 180-day duration, no co-occurring opioid analgesics, no co-occurring benzodiazepines, infectious disease screening, liver function test, drug and toxicology screenings, evaluation and management visits, and counseling.

RESULTS

Episodes linked to PCPs had significantly lower odds of achieving 180-day duration, an absence of opioid analgesics, an absence of benzodiazepines, drug and toxicology screenings, and counseling compared to addiction, behavioral health, and/or OTPs. Episodes linked to PCPs had significantly higher odds of undergoing infectious disease screenings, liver function tests, and evaluation and management visits compared to all specialty categories.

CONCLUSIONS

Episodes were more likely to achieve process of care measures related to the specialties of their physicians, but no specialty consistently demonstrated better performance compared to PCPs. Our findings highlight the need for models that can better integrate physical and behavioral health services for OUD treatment.

Subsequent Buprenorphine Treatment Following Emergency Physician Buprenorphine Prescription Fills: A National Assessment 2019 to 2020

STUDY OBJECTIVE

Buprenorphine treatment for opioid use disorder provided in the emergency department with subsequent buprenorphine treatment by community prescribers is associated with improved outcomes, but the frequency with which this occurs is unknown. We examined the rates of subsequent buprenorphine treatment for buprenorphine-naïve individuals filling buprenorphine prescriptions from emergency physicians and initiated buprenorphine treatment and how such rates varied before and during the coronavirus disease 2019 (COVID-19) pandemic.

METHODS

Using pharmacy claims capturing an estimated 92% of prescriptions filled at US retail pharmacies, we identified buprenorphine prescriptions filled between February 1, 2019, and November 30, 2020, written by emergency physicians. In this observational study, we calculated the rate at which patients subsequently filled buprenorphine prescriptions from other nonemergency clinicians, the frequency with which subsequent filled prescriptions were from different types of prescribers, and the changes in the rates of subsequent prescriptions following the declaration of the COVID-19 public health emergency.

RESULTS

We identified 22,846 prescriptions written by emergency physicians and filled by buprenorphine-naïve patients. They were most commonly paid for by Medicaid and were in metropolitan counties; 28.5% of patients subsequently filled buprenorphine prescriptions written by other clinicians. Adult primary care physicians and advanced practice providers (eg, physician assistants and nurse practitioners) were responsible for most of the subsequent prescriptions. The rates of subsequent prescriptions were 3.5% lower after the COVID-19 public health emergency declaration.

CONCLUSION

The majority of patients filling buprenorphine prescriptions written by emergency physicians do not subsequently fill prescriptions written by other clinicians, and the rates of subsequent prescriptions were lower after the declaration of the COVID-19 public health emergency. These findings highlight the need for a system of care that improves buprenorphine treatment continuity of care for patients with opioid use disorder from emergency settings to community treatment providers.

Individual and Community Factors Associated with Naloxone Co-prescribing Among Long-term Opioid Patients: a Retrospective Analysis

BACKGROUND

Naloxone co-prescribing to individuals at increased opioid overdose risk is a key component of opioid overdose prevention efforts.

OBJECTIVE

Examine naloxone co-prescribing in the general population and assess how co-prescribing varies by individual and community characteristics.

DESIGN

Retrospective cross-sectional study. We conducted a multivariable logistic regression of 2017-2018 de-identified pharmacy claims representing 90% of all prescriptions filled at retail pharmacies in 50 states and the District of Columbia.

PATIENTS

Individuals with opioid analgesic treatment episodes > 90 days MAIN MEASURES: Outcome was co-prescribed naloxone. Predictor variables included insurance type, primary prescriber specialty, receipt of concomitant benzodiazepines, high-dose opioid episode, county urbanicity, fatal overdose rates, poverty rates, and primary care health professional shortage areas.

KEY RESULTS

Naloxone co-prescribing occurred in 2.3% of long-term opioid therapy episodes. Medicaid (aOR 1.87, 95%CI 1.84 to 1.90) and Medicare (aOR 1.48, 95%CI 1.46 to 1.51) episodes had higher odds of naloxone co-prescribing than commercial insurance episodes, while cash pay (aOR 0.77, 95%CI 0.74 to 0.80) and other insurance episodes (aOR 0.81, 95%CI 0.79 to 0.83) had lower odds. Odds of naloxone co-prescribing were higher among high-dose opioid episodes (aOR 3.19, 95%CI 3.15 to 3.23), when concomitant benzodiazepines were prescribed (aOR 1.12, 95%CI 1.10 to 1.14), and in counties with higher fatal overdose rates.

CONCLUSION

Co-prescription of naloxone represents a tangible clinical action that can be taken to help prevent opioid overdose deaths. However, despite recommendations to co-prescribe naloxone to patients at increased risk for opioid overdose, we found that co-prescribing rates remain low overall. States, insurers, and health systems should consider implementing strategies to facilitate increased co-prescribing of naloxone to at-risk individuals.

Trends in visits to substance use disorder treatment facilities in 2020

OBJECTIVE

To describe weekly changes in the number of substance use disorder treatment (SUDT) facility visits in 2020 compared to 2019 using cell phone location data.

METHODS

We calculated the percentage weekly change in visits to SUDT facilities from the week of January 5 through the week of October 11, 2020, relative to the week of January 6 through the week of October 13, 2019. We stratified facilities by county COVID-19 incidence per 10,000 residents in each week and by 2018 fatal drug overdose rate. Finally, we conducted a multivariable linear regression analysis examining percent change in visits per week as a function of county-level COVID-19 tercile, a series of calendar month indicators, and the interaction of county-level COVID-19 tercile and month. We repeated the regression analysis replacing COVID-19 tercile with overdose tercile.

RESULTS

Beginning the eleventh week of 2020, the number of visits to SUDT facilities declined substantially, reaching a nadir of 48% of 2019 visits in early July. In contrast to January, there were significantly fewer visits in 2020 compared to 2019 in all subsequent months (p < 0.01 in all months). Multivariable regression results found that facilities in the tercile of counties experiencing the most COVID-19 cases had a significantly greater reduction in the number of SUDT visits in 2020 for the months of June through August than facilities in counties with the fewest COVID-19 rates (p < 0.05). The study found no statistically significant difference in the change in the number of visits by facilities in counties with historically different overdose rates.

DISCUSSION

Our findings support the hypothesis that a reduction has occurred in the average weekly number of visits to SUDT facilities. The size of the effect differs based on the number of COVID-19 cases but not on historical overdose rate.

Advanced cognitive impairment among older nursing home residents

BACKGROUND

Though work has been done studying nursing home (NH) residents with either advanced Alzheimer's disease (AD) or Alzheimer's disease related dementia (ADRD), none have distinguished between them; even though their clinical features affecting survival are different. In this study, we compared mortality risk factors and survival between NH residents with advanced AD and those with advanced ADRD.

METHODS

This is a retrospective observational study, in which we examined a sample of 34,493 U.S. NH residents aged 65 and over in the Minimum Data Set (2011-2013). Incident assessment of advanced disease was defined as the first MDS assessment with severe cognitive impairment (Cognitive Functional Score equals to 4) and diagnoses of AD or ADRD. Demographics, functional limitations, and comorbidities were evaluated as mortality risk factors using Cox models. Survival was characterized with Kaplan-Maier functions.

RESULTS

Of those with advanced cognitive impairment, 35 % had AD and 65 % ADRD. At the incident assessment of advanced disease, those with AD had better health compared to those with ADRD. Mortality risk factors were similar between groups (shortness of breath, difficulties eating, substantial weight-loss, diabetes mellitus, heart failure, chronic obstructive pulmonary disease, and pneumonia; all p < 0.01). However, stroke and difficulty with transfer (for women) were significant mortality risk factors only for those with advanced AD. Urinary tract infection, and hypertension (for women) only were mortality risk factors for those with advanced ADRD. Median survival was significantly shorter for the advanced ADRD group (194 days) compared to the advanced AD group (300 days).

CONCLUSIONS

There were distinct mortality and survival patterns of NH residents with advanced AD and ADRD. This may help with care planning decisions regarding therapeutic and palliative care.

Concentration of Patient Care Among Buprenorphine-Prescribing Clinicians in the US

This study quantifies the total amount of buprenorphine care delivered by active prescribers and to what extent that varies across specialties.

Patient, prescriber, and Community factors associated with filled naloxone prescriptions among patients receiving buprenorphine 2017-18

BACKGROUND

Prescribing naloxone to patients at increased opioid overdose risk is a key component of opioid overdose prevention efforts, but little is known about naloxone fills among patients receiving buprenorphine for opioid use disorder, one such high risk group.

METHODS

This retrospective cross-sectional study used de-identified pharmacy claims representing 90% of all prescriptions filled at retail pharmacies in 50 states and the District of Columbia. We performed a multivariable logistic regression to examine filled naloxone prescriptions among patients receiving buprenorphine treatment and assessed how filled naloxone prescriptions vary by patient, prescriber, and community characteristics.

RESULTS

Filled naloxone prescriptions occurred among 4.5% of buprenorphine treatment episodes. Episodes paid through Medicaid (aOR 2.40, 95%CI 2.33-2.47) and Medicare (aOR 1.53, 95%CI 1.46-1.60) had higher odds of filled naloxone prescriptions than commercial insurance episodes. Compared to episodes where the primary prescriber was an adult primary care physician, odds of filling a naloxone prescription were higher among episodes prescribed by addiction specialists (aOR 1.30, 95% CI 1.24-1.37) and physician assistants/nurse practitioners (aOR 1.57, 95% CI 1.53-1.61).

CONCLUSIONS

Prescribing naloxone to patients receiving buprenorphine represents a tangible clinical action that can be taken to help prevent opioid overdose deaths. However, despite recommendations to co-prescribe naloxone to patients at increased risk for opioid overdose, rates of filling naloxone prescriptions remain low among patients dispensed buprenorphine. States, insurers, and health systems should consider implementing strategies to facilitate increased co-prescribing of naloxone to at-risk individuals.

Does Integration of Palliative Care and Infection Management Reduce Hospital Transfers among Nursing Home Residents?

Background: An estimated 50% of nursing home (NH) residents experience hospital transfers in their last year of life, often due to infections. Hospital transfers due to infection are often of little clinical benefit to residents with advanced illness, for whom aggressive treatments are often ineffective and inconsistent with goals of care. Integration of palliative care and infection management (i.e., merging the goals of palliative care and infection management at end of life) may reduce hospital transfers for residents with advanced illness. Objectives: Evaluate the association between integration and (1) all-cause hospital transfers and (2) hospital transfers due to infection. Design: Cross-sectional observational study. Setting/Subjects: 143,223 U.S. NH residents, including 42,761 residents in the advanced stages of dementia, congestive heart failure (CHF), and chronic obstructive pulmonary disease (COPD). Measurement: Cross-sectional, nationally representative NH survey data (2017-2018) were combined with resident data from the Minimum Data Set 3.0 and Medicare inpatient data (2016-2017). NH surveys measured integration of palliative care and infection management using an index of 0-100. Logistic regression models were used to estimate the relationships between integration intensity (i.e., the degree to which NHs follow best practices for integration) and all-cause hospital transfer and transfer due to infection. Results: Among residents with advanced dementia, integration intensity was inversely associated with all-cause hospital transfer and transfer due to infection (p < 0.001). Among residents with advanced COPD, integration intensity was inversely associated with all-cause hospital transfer (p < 0.05) but not transfers due to infection. Among residents with advanced CHF, integration intensity was not associated with either outcome. Conclusions: NH policies aimed to promote integration of palliative care and infection management may reduce burdensome hospital transfers for residents with advanced dementia. For residents with advanced CHF and COPD, alternative strategies may be needed to promote best practices for infection management at end of life.

Impact of a Medicaid policy on preventive oral health services for children with intellectual disabilities, developmental disabilities, or both

BACKGROUND

Integrating preventive oral health services (POHS) into medical offices may ease access to care for children with intellectual and developmental disabilities (IDD). The authors examined the impact of state policies allowing delivery of POHS in medical offices on receipt of POHS among Medicaid enrollees with IDD.

METHODS

The authors used 2006 through 2014 Medicaid data for children with IDD aged 6 months through 5 years from 38 states. IDD were defined using 14 condition codes from Centers for Medicare & Medicaid Services Chronic Conditions Data Warehouse. The length of the state's medical POHS policy (no policy, < 1 year, 1 year, 2 years, 3 years, or ≥ 4 years) was interacted with an indicator that the child was younger than 3 years. The authors used logistic regression models to estimate the likelihood that a child received POHS in a medical office or in a medical or dental office in a given year.

RESULTS

Among 447,918 children with IDD, 1.6% received POHS in medical offices. Children younger than 3 years in states with longer-enacted policies had higher rates of receiving POHS. For example, the predicted probability of receiving POHS was 40.6% (95% confidence interval, 36.3% to 44.9%) for children younger than 3 years in states with a medical POHS policy for more than 4 years compared with 30.6% (95% confidence interval, 27.8% to 33.5%) for children in states without a policy.

CONCLUSIONS

State Medicaid policies allowing delivery of POHS in medical offices increased receipt of POHS among Medicaid-enrolled children with IDD who were younger than 3 years.

PRACTICAL IMPLICATIONS

Few children with IDD receive POHS in any setting. Efforts are needed to reduce barriers to POHS for publicly insured children with IDD.

Changes in US Nursing Home Infection Prevention and Control Programs From 2014 to 2018

OBJECTIVE

Burgeoning rates of antibiotic resistance have resulted in a shift in national focus to improve infection prevention and control programs in US nursing homes (NHs). We sought to evaluate the changes in nursing home infection prevention and control programs over time.

DESIGN

Retrospective comparative analysis of national nursing home survey data from 2014 and 2018.

SETTING AND PARTICIPANTS

We used survey data from 2 nationally representative samples of US nursing homes (945 NHs in 2014 and 888 in 2018).

METHODS

Three indices measuring antibiotic stewardship, outbreak control, and urinary tract infection prevention (ranging from 0 to 100) were developed to measure the change in infection prevention and control programs. Multivariable linear regression models were used to identify facility and infection preventionist characteristics associated with each index. Decomposition models were used to identify contributions of factors on the differences in each index over time.

RESULTS

From 2014 to 2018, we saw strengthening of antibiotic stewardship practices by 33 percentage points, outbreak control practices by 13 percentage points, and urinary tract infection prevention practices by 6 percentage points. Although we found several predictors of these improvements, much of the improvement was due to the difference in time.

CONCLUSIONS AND IMPLICATIONS

Policy mandates and greater national attention are likely important factors in improving nursing home infection prevention and control practices. Further work is needed to evaluate the effect of these programs on resident outcomes.

ADVANCED ILLNESS AMONG ELDERLY NURSING HOME RESIDENTS WITH ALZHEIMER’S DISEASE AND RELATED DEMENTIA

Alzheimer’s disease and related dementia (AD/ADRD) are leading causes of mortality in the United States. Identifying advanced illness (AI) in NH residents is key for developing therapeutic and palliative care plans for end of life. We refined and extended existing measures of AI in NH residents with AD/ADRD and described patterns of survival for each measure. Using the Minimum Data Set (MDS; 2011 to 2013) linked to vital status (through 2016), we defined categories of AD/ADRD residents at AI onset: (1) those with ADRD, (2) and those with both, AD and ADRD. We estimated survival functions and multivariable duration models to describe patterns of survival from AI onset until death, stratified by AD/ADRD classifications, sex and functional status at AI onset, conditional on socio-demographics and co-morbidities. We limited our sample to adults ages >64 for whom we observed the incident AI assessment in the MDS. Median survival was 229 days for all classifications of AI, but higher for those with only ADRD (300 days). Survival declined substantially for residents with eating difficulties; to 122 days for residents with AD and ADRD. A stark survival decline (40 days) occurred among residents with shortness of breath. Across all AI classifications, survival was negatively associated with male sex, age, diabetes, substantial weight-loss and events such as heart failure. Depression, hypertension, and UTI were associated with small or insignificant increases in mortality risk. AI can be defined using MDS data, allowing for examination of policies designed to improve end of life care.

Integration of Palliative Care and Infection Management at the End of Life in U.S. Nursing Homes

CONTEXT

Infections in nursing home (NH) residents are often terminal illnesses. Integration of palliative care (PC) and infection management (IM) is a new concept that can help reduce burdensome treatments and improve quality of care for NH residents at the end of life.

OBJECTIVES

To develop measures of integration, describe the integration in U.S. NHs, and examine predictors of integration.

METHODS

A nationally representative sample of NHs was surveyed. An instrument to measure integration was tested using factor analyses. Descriptive analyses of each integration factor were conducted, construct validity was examined using correlations between the integration factors and validated measures of PC and IM, and multivariable linear regression models were developed to identify NH characteristics associated with integration.

RESULTS

A total of 892 NH surveys were returned (49% response rate), 859 with complete data. Three integration factors were identified: patient involvement in care planning (Involvement), formalized advance care planning (Advance Care Planning), and routine practices of integration (Routine Practices). The highest level of integration in NHs was reported for Involvement (mean (μ) = 73.2, standard error [SE] = 1.57), with lower rates for Advance Care Planning and Routine Practices (respectively, μ = 34.1, SE = 1.05; μ = 31.4, SE = 1.48). Each integration measure was weakly, positively associated with the PC and IM measures (r ≤ 0.25, P ≤ 0.01). There were few associations between NH characteristics and integration.

CONCLUSION

Integration is a distinct concept that is associated with, but different from, PC and IM. Results serve as a baseline assessment of integration in NHs. Continued refinement of the integration instrument is recommended, as is studying if higher integration leads to better resident outcomes.

Measuring Quality in Home Healthcare

BACKGROUND/OBJECTIVES

Improving quality performance in home health is an increasingly high priority. The objective of this study was to examine trends in industry performance over time using three quality measures: a composite quality metric (Q index), an infection prevention measure (vaccination verification), and an outcome measure (hospital avoidance).

DESIGN/SETTING/PARTICIPANTS/MEASURES

We linked Home Health Compare and Provider of Services data from 2012 to 2016, which included 39 211 observations during the 5-year study period and 7670 agencies in 2016. The Q index was developed to allow comparability over time, equally weighting the contributions of each element. After examining summary statistics, we developed three regression models stratified by ownership (for-profit/nonprofit agency) and included two constructs of nurse staffing, in addition to controlling for known confounders.

RESULTS

Most agencies (80.4%) were for-profit agencies. The Q index and vaccination verification improved substantially over time, but there was no change in hospital avoidance. Ownership status was associated with all three measures (P < .001). Registered nurse staffing (relative to licensed practical nurses and home health aides) was associated with higher Q index and vaccination verification (P < .001).

CONCLUSION

The Q index allows for assessment of trends over time in home healthcare. Ownership and nurse staffing are important factors in the quality of care. The overall home care market is driven by for-profit agencies, but their characteristics and outcomes differ from nonprofit agencies. J Am Geriatr Soc 67:1859-1865, 2019.

Delivery of Preventive Oral Health Services by Rurality: A Cross-Sectional Analysis

PURPOSE

Integrating oral health care into primary care has been promoted as a strategy to increase delivery of preventive oral health services (POHS) to young children, particularly in rural areas where few dentists practice. Using a multistate sample of Medicaid claims, we examined a child's odds of receiving POHS in a medical office by county rurality.

METHODS

We used 2012-2014 Medicaid Analytic extract claims data for 6,275,456 children younger than 6 years in 39 states that allowed Medicaid payment for POHS in medical offices. We used county-level characteristics from the Area Health Resources Files, including a 3-level measure of county rurality. We used logistic regression to estimate a child's odds of receiving POHS in a medical office by county rurality, while controlling for other patient and county characteristics.

FINDINGS

POHS in medical offices were received by 7.8% of children. Rates of POHS in medical offices were higher in metropolitan (metro) counties (8.4%) than nonmetro adjacent to metro (5.8%) and nonmetro not adjacent to metro (4.3%). In adjusted analysis, children living in nonmetro not adjacent to metro (OR = 0.79, 95% CI: 0.64-0.99) and adjacent to metro counties (OR = 0.70, 95% CI: 0.59-0.82) were significantly less likely to receive POHS in medical offices than children living in metro counties.

CONCLUSIONS

In this study of POHS in medical offices among young Medicaid-enrolled children, we found POHS rates were lowest in nonmetro counties. Given barriers to dental care in rural areas, states should take additional steps beyond allowing Medicaid reimbursement to increase delivery of POHS in medical offices.

Impact of Medicaid Policy on the Oral Health of Publicly Insured Children

Objective Fluoride varnish (FV) applications among non-dentist primary care providers has increased due to state Medicaid policies. In this study we examine the impact of FV policies on the oral health of publicly insured children aged 2-6 years old. Methods Using three waves of the National Survey of Children's Health (2003, 2007, 2011/12), we used a logistic regression model with state and year fixed effects, adjusting for relevant child characteristics, to examine the association between years since a state implemented a FV policy and the odds of a publicly insured child having very good or excellent teeth. We compared children with public insurance in states with FV policies to children with public insurance in states without FV policies, controlling for the same difference among children with private insurance who were unlikely to be affected by Medicaid FV policies. Results Among 68,890 children aged 2-6 years, 38% had public insurance. Compared to privately insured children, publicly insured children had significantly lower odds of having very good or excellent teeth [odds ratio (OR) 0.70, 95% CI 0.62-0.81]. Publicly insured children in states with FV policies implemented for four or more years had significantly greater odds of having very good or excellent teeth (OR 1.28, 95% CI 1.03-1.60) compared to publicly insured children in states without FV policies. Conclusions for Practice State policies supporting non-dental primary care providers application of FV were associated with improvements in oral health for young children with public insurance.

Nursing home adoption of the National Healthcare Safety Network Long-term Care Facility Component

BACKGROUND

Health care-associated infections pose a significant problem in nursing homes (NHs). The Long-term Care Facility Component of the National Healthcare Safety Network (NHSN) was launched in 2012, and since then, enrollment of NHs into NHSN has been deemed a national priority. Our goal was to understand the characteristics of NHs reporting to the NHSN compared to other NHs across the country.

METHODS

To meet this goal, we quantified the characteristics of NHs by NHSN enrollment status and reporting consistency using the Certification and Survey Provider Enhanced Reporting (CASPER) data linked to NHSN enrollment and reporting data.

RESULTS

Of the 16,081 NHs in our sample, 262 (or 1.6% of NHs) had enrolled in NHSN by the end of 2015; these early adopting facilities were more likely to be for-profit and had a higher percentage of Medicare residents. By the end of 2016, enrollment expanded by more than 5-fold to 1,956 facilities (or 12.2% of NHs). In our analysis, the characteristics of those later adopting NHs were more similar to NHs nationally than the early adopters. Specifically, bed size and hospital-based facilities were related to both early and late adoption of NHSN.

CONCLUSIONS

The types of NHs that have enrolled in NHSN have changed substantially since the program began. The increased enrollment was likely due to the Centers for Medicare & Medicaid (CMS)-funded "C. difficile Infection (CDI) Reporting and Reduction Project" that incentivized Quality Innovation Network-Quality Improvement Organizations (QIN-QIOs) to support NH enrollment and participation in NHSN. Further understanding of a facility's ability to enroll in and maintain reporting to NHSN, and how this relates to infection prevention staffing and infrastructure in NHs and infection rates among NH residents, is needed.

Association between type of health insurance and children's oral health, NHANES 2011-2014

OBJECTIVES

To examine the association between type of health insurance (public, uninsured, private, or other) and oral health outcomes for children in the United States using nationally representative surveillance data.

METHODS

Using the National Health and Nutrition Examination Survey (2011/12-2013/14), logistic regression models were used to estimate the odds of any dental caries and any untreated caries by type of health insurance (public, uninsured, private, and other) for children aged 2-19 years, with adjustment for relevant individual and socioeconomic characteristics.

RESULTS

Among 6,057 children, the odds of having any dental caries or untreated caries was not significantly different for publicly insured and uninsured children compared to privately insured children, when adjusting for family income and education. Children in families with income to poverty ratios <200 percent had greater odds of caries and untreated caries relative to children in families with income to poverty ratios ≥400 percent. Children with less educated parents also experienced greater odds of caries and untreated caries.

CONCLUSIONS

Oral health outcomes, after adjusting for covariates, were similar for children with public and private health insurance. However, children in low-income families and with less educated parents had greater odds of untreated caries and dental caries, suggesting that initiatives focused on publicly insured populations may miss other vulnerable children of low socioeconomic status.

A population-based examination of trends and disparities in medication treatment for opioid use disorders among Medicaid enrollees

BACKGROUND

Medication treatment (MT) with methadone and buprenorphine are effective treatments for opioid use disorders, but little information is available regarding the extent to which buprenorphine's approval resulted in more individuals receiving MT nor to what extent receipt of such treatment was equitable across communities.

METHODS

To examine changes in MT utilization and the association between MT utilization and county-level indicators of poverty, race/ethnicity, and urbanicity, we used Medicaid claims of non-dually eligible Medicaid enrollees aged 18-64 from 14 states for 2002-2009. We generated county-level aggregate counts of MT (methadone, buprenorphine, and any MT) by year (N = 7760 county-years). We estimated count data models to identify associations between MT and county characteristics, including levels of poverty and racial/ethnic concentration.

RESULTS

The number of Medicaid enrollees receiving MT increased 62% from 2002 to 2009. The number of enrollees receiving methadone increased 20%, with the remaining increase resulting from buprenorphine. Urban county residents were significantly more likely to receive MT in both 2002 and 2009 than rural county residents. However, buprenorphine substantially increased MT in rural counties from 2002 to 2009. Receipt of MT increased at a much higher rate for residents of counties with lower poverty rates and lower concentrations of black and Hispanic individuals than for residents of counties without those characteristics.

CONCLUSIONS

The increase in Medicaid enrollees receiving MT in the years following buprenorphine's approval is encouraging. However, it is concerning that MT trends varied so dramatically by characteristics of the county population and that increases in utilization were substantially lower in counties with populations that historically have been disadvantaged with respect to health care access and quality. Concerted efforts are needed to ensure that MT benefits are equitably distributed across society and reach disadvantaged individuals who may be at higher risk of experiencing opioid use disorders.

Infection Trends in US Nursing Homes, 2006-2013

OBJECTIVES

The objectives of this study were to estimate trends in the prevalence of infections in nursing home (NH) residents using 2006-2013 Minimum Data Set (MDS) data, estimate the number of all infections in 2013, and evaluate differences in trends between MDS versions 2.0 and 3.0.

DESIGN

Retrospective study.

SETTING

NHs in the United States.

PARTICIPANTS

All NH residents with a quarterly or annual MDS assessment in 2006-2013 (n = 30,366,807 assessments).

MEASUREMENTS

MDS 2.0 and 3.0 quarterly and annual assessment data (2006-2013) from over 15,000 NHs were used to estimate the 7-day prevalence of multidrug-resistant organism (MDRO) infection, pneumonia, septicemia, viral hepatitis, and wound infection and 30-day prevalence of urinary tract infection (UTI). Admission assessments were excluded. Annual infection counts were estimated using 2013 data. Changes in the prevalence of reported infections over time and differences in trends between MDS 2.0 and 3.0 were examined using tests of linear trends.

RESULTS

In 2013, there were an estimated 1.13 to 2.68 million infections in NH residents. UTI and pneumonia were the most commonly reported infections in every quarter, ranging from 5.6% to 8.1% and 1.4% to 2.5%, respectively. Prevalence of all infections increased in 2006-2010 (P values < .01). In 2011-2013, prevalence of UTI, MDRO, and wound infections decreased and viral hepatitis increased (P values < .0001). Between MDS 2.0 and 3.0, the prevalence of UTI, MDRO, and wound infections decreased and the prevalence of viral hepatitis increased (P values < .0001).

CONCLUSION

Infections are a major and persistent problem in NHs. Although MDS data are useful for identifying trends in infection prevalence, revisions in definitions need to be accounted for when evaluating trends over time. Additional research is needed to identify factors that contribute to changes in infection prevalence.

Opioid analgesic and benzodiazepine prescribing among Medicaid-enrollees with opioid use disorders: The influence of provider communities

Opioid analgesic and benzodiazepine use in individuals with opioid use disorders can increase the risk for medical consequences and relapse. Little is known about rates of use of these medications or prescribing patterns among communities of prescribers. The goal of this study was to examine rates of prescribing to Medicaid-enrollees in the calendar year after an opioid use disorder diagnosis, and to examine individual, county, and provider community factors associated with such prescribing. 2008 Medicaid claims data were used from 12 states to identify enrollees diagnosed with opioid use disorders, and 2009 claims data were used to identify rates of prescribing of each drug. Social network analysis was used to identify provider communities, and multivariate regression analyses was used to to identify patient, county, and provider community level factors associated with prescribing these drugs. The authors also examined variation in rates of prescribing across provider communities. Among Medicaid-enrollees identified with an opioid use disorder, 45% filled a prescription for an opioid analgesic, 37% filled a prescription for a benzodiazepine, and 21% filled a prescription for both in the year following their diagnosis. Females, older individuals, individuals with pain syndromes, and individuals residing in counties with higher rates of poverty were more likely to fill prescriptions. Prescribing rates varied substantially across provider communities, with rates in the highest quartile of prescribing communities over 2.5 times the rates in the lowest prescribing communities. Prescribing opioid analgesics and benzodiazepines to individuals diagnosed with opioid use disorders may increase risk of relapse and overdose. Interventions should be considered that target provider communities with the highest rates of prescribing and individuals at the highest risk.

Growth In Buprenorphine Waivers For Physicians Increased Potential Access To Opioid Agonist Treatment, 2002-11

Opioid use disorders are a significant public health problem, affecting two million people in the United States. Treatment with buprenorphine, methadone, or both is predominantly offered in methadone clinics, yet many people do not receive the treatment they need. In 2002 the Food and Drug Administration approved buprenorphine for prescription by physicians who completed a course and received a waiver from the Drug Enforcement Administration, exempting them from requirements in the Controlled Substances Act. To determine the waiver program's impact on the availability of opioid agonist treatment, we analyzed data for the period 2002-11 to identify counties with opioid treatment shortages. We found that the percentage of counties with a shortage of waivered physicians fell sharply, from 98.9 percent in 2002 to 46.8 percent in 2011. As a result, the percentage of the US population residing in what we classified as opioid treatment shortage counties declined from 48.6 percent in 2002 to 10.4 percent in 2011. These findings suggest that the increase in waivered physicians has dramatically increased potential access to opioid agonist treatment. Policy makers should focus their efforts on further increasing the number and geographical distribution of physicians, particularly in more rural counties, where prescription opioid misuse is rapidly growing.

The effects of prior authorization policies on medicaid-enrolled children's use of antipsychotic medications: evidence from two mid-Atlantic states

OBJECTIVE

The purpose of this study was to examine the impact of prior authorization policies on the receipt of antipsychotic medication for Medicaid-enrolled children.

METHODS

Using de-identified administrative Medicaid data from two large, neighboring, mid-Atlantic states from November 2007 through June 2011, we identified subjects <18 years of age using antipsychotics, from the broader group of children and adolescents receiving behavioral health services or any psychotropic medication. Prior authorization for antipsychotics was required for children in State A <6 years of age from September 2008, and for children <13 years of age from August 2009. No such prior authorizations existed in State B during that period. Filled prescriptions were identified in the data using national drug codes. Using a triple-difference strategy (using differences among the states, time periods, and differences in antidepressant prescribing rates among states over the same time periods), we examined the effect of the prior authorization policy on the rate at which antipsychotic prescriptions were filled for Medicaid-enrolled children and adolescents.

RESULTS

The impact of prior authorization policies on antipsychotic medication use varied by age: Among 6-12 year old children, the impact of the prior authorization policy on antipsychotic medication prescribing was a modest but statistically significant decrease of 0.47% after adjusting for other factors; there was no effect of the prior authorization among children 0-5 years.

CONCLUSIONS

Prior authorization policies had a modest but statistically significant effect on antipsychotic use in 6-12 year old children, but had no impact in younger children. Future research is needed to understand the utilization and clinical effects of prior authorization and other policies and interventions designed to influence antipsychotic use in children.

Concurrent mental health therapy among medicaid-enrolled youths starting antipsychotic medications

OBJECTIVE

The use of antipsychotic medications among children and adolescents has increased substantially in recent years, predominantly in disorders for which effective psychosocial interventions exist. The authors of this study examined the extent to which youths being prescribed antipsychotic medications were receiving concurrent mental health therapy.

METHODS

Using administrative data, the authors identified 6,110 Medicaid-enrolled youths starting antipsychotic medications from November 1, 2006, through October 31, 2009, and identified youths who had received any concurrent mental health therapy. Multivariate regression models were used to examine the relationship between concurrent therapy and sociodemographic and clinical variables.

RESULTS

Sixty-eight percent (N=4,155) of youths starting antipsychotic medications received concurrent therapy. Multivariate regression findings are that concurrent therapy was more common with younger children, recently hospitalized youths, children from urban communities, youths from racial-ethnic minority groups, children with an antipsychotic-indicated diagnosis, and youths eligible for Medicaid because of family income.

CONCLUSIONS

The finding that 68% of youths starting antipsychotic medications received concurrent therapy suggests that for a majority of children, these medications complemented rather than substituted for nonpharmacologic interventions. However, with almost one-third of youths not receiving concurrent therapy, a better understanding of the factors contributing to the lack of concurrent therapy for youths starting antipsychotic medications is needed. Children and families should be aware of and have access to effective psychosocial treatments for disorders such as attention-deficit hyperactivity disorder and depression that are common among children receiving antipsychotic medications.

Obesity and Angiolymphatic Invasion in Primary Breast Cancer

Background and Objectives: Obesity is associated both with higher rates of breast cancer and with unfavorable breast cancer outcomes. Small studies have suggested that obesity may be associated with unfavorable tumor biologic characteristics, such as angiolymphatic invasion. The purpose of this study was to characterize the independent association of obesity with the presence of angiolymphatic invasion, stage at presentation, estrogen receptor (ER) status, the triple-negative phenotype, and tumor grade after controlling for clinical factors.Methods: Detailed clinical and pathologic data were abstracted from the medical records of all 1361 patients with 1398 Stage I, II, or III primary breast cancers who had breast surgery at the University of Michigan Comprehensive Cancer Center between January 1, 2000 and December 31, 2006. A second abstractor, who did not have access to the clinical and pathologic tumor characteristics, collected patient height and weight. Bivariate and multivariate analyses were conducted to investigate the relationships between body mass index (BMI) and tumor biologic features controlling for menopausal status, comorbid diabetes and hypertension, use of hormone replacement therapy before the diagnosis of breast cancer, race, and ethnicity. Results: In bivariate analyses, obesity was not associated with hormone receptor status, HER2 status, triple-negative phenotype, or histologic grade. There was a borderline significant association between obesity category and the presence of angiolymphatic invasion (p = 0.06) in the sample of 1398 tumors, severely obese women being the most likely to have angiolymphatic invasion. Among pre- and perimenopausal women, the likelihood of angiolymphatic invasion in the tumors varied by obesity category (p = 0.02) with angiolymphatic invasion more likely in obese (20.0%) compared with the tumors of healthy weight (14.4%) women. In multivariate analyses, obesity status was independently associated with the presence of angiolymphatic invasion (joint test of significance, p = 0.04) after controlling for menopausal status, use of hormone replacement therapy at the time of diagnosis, diabetes and hypertension, and tumor features, including grade, stage, estrogen receptor status, and HER2 status. Obesity was associated with lower odds of triple-negative disease (odds ratio, OR, among severely obese women 0.37, 95% C.I. 0.15 – 0.92). There were significant associations between diabetes and tumor biology in multivariate analyses restricted to tumors in pre- and perimenopausal women. Diabetes was associated with ER-negative tumor status (OR 5.66, 95% C.I. 1.05 to 30.55) and with the triple-negative phenotype (OR 13.73, 95% C.I. 2.23 to 84.50).Discussion: In this large sample of invasive breast cancers, obesity was independently associated with the presence of angiolymphatic invasion. Furthermore, diabetes was found to be associated with ER-negative and triple-negative breast cancer among premenopausal women. Higher rates of angiolymphatic invasion among obese women may account in part for the poorer outcomes among obese women with breast cancer, and higher rates of triple-negative breast cancer may account for poorer outcomes among women with diabetes.

Citation Information: Cancer Res 2009;69(24 Suppl):Abstract nr 5151.

Substance abuse detoxification and residential treatment among Medicaid-enrolled adults: rates and duration of subsequent treatment

BACKGROUND

Substance use disorders are chronic disorders with substantial public health significance, yet the treatment provided is often episodic despite ongoing need. Among the more severely ill individuals requiring detoxification or residential treatment, little empirical information is available about rates and predictors of subsequent engagement in necessary subsequent treatment.

METHODS

Using administrative data from the largest Medicaid managed behavioral health organization in a large mid-Atlantic state, we used multivariate regression to examine rates and predictors of subsequent treatment engagement and retention following new episodes of detoxification or residential substance abuse treatment among 5670 Medicaid-enrolled adults during 2004-2006.

RESULTS

Slightly less than half (49%) of the sample received follow-up care within 30 days of discharge. Rates of follow-up were significantly higher in individuals with a serious mental illness, and significantly lower in African-American individuals, males, individuals with disabilities, and those who received detoxification without residential treatment. The mean duration of follow-up treatment was 84 days, and was longer among individuals with a serious mental illness and Caucasians. Even after controlling for individuals' sociodemographic and clinical characteristics, there was substantial variation in follow-up rates among discharging providers.

CONCLUSION

The relatively low rates of follow-up care and relatively brief duration of treatment for many of those who received such follow-up care are concerning in a population receiving substance abuse detoxification or residential treatment. The markedly lower rates among those receiving detoxification alone without subsequent residential treatment and among those without a comorbid serious mental illness suggest that efforts specifically targeting those individuals may be of particular benefit.

Physician referral rates: style without much substance?

BACKGROUND

Primary care physicians (PCPs) exhibit widely varying referral rates, resulting in dramatic differences in the exposure of their patients to specialists. The relationships between this physician behavior and costs and patient outcomes are unknown.

OBJECTIVES

To examine the relationships between PCP referral rates and costs, risk of avoidable hospitalization, health status, and satisfaction.

DESIGN

Cross-sectional analyses of claims and patient survey data.

SETTING AND SUBJECTS

Independent practice association (IPA)-style managed care organization in the Rochester, NY, metropolitan area. The 1995 claims data included 457 PCPs in the IPA and 217,606 adult patients assigned to their panels. Approximately 50 consecutive patients of each of a random sample of 100 PCPs completed a patient survey in 1997-1998.

MEASURES

From the claims data, total expenditures per panel member, the risk of avoidable hospitalization, and physician referral rate were measured. Measures derived from the survey included SF-12 scores, satisfaction, and physician referral rate.

RESULTS

The relationship between physician referral rate and per-panel-member costs was not statistically significant after case-mix adjustment of the referral rate. There was no relationship between the case-mix-adjusted referral rate and risk of avoidable hospitalization. In the survey data, there was no adjusted relationship between the physicians' referral rate and their patients' self-rated physical or mental health. There was a modest direct relationship between patient satisfaction and survey-derived referral rate.

CONCLUSIONS

Despite stable, wide variations in PCP referral rates, there are few discemible relationships between this physician behavior and costs and patient outcomes. Efforts to constrain PCP referrals to specialists may be misguided.

Do physicians who diagnose more mental health disorders generate lower health care costs?

BACKGROUND

Underrecognition and undertreatment of mental health disorders in primary care have been associated with poor health outcomes and increased health care costs, but little is known about the impact of the diagnoses of mental health disorders on health care expenditures or outcomes. Our goal was to examine the relationships between the proportion of mental health diagnoses by primary care physicians and both health care expenditures and the risk of avoidable hospitalizations.

METHODS

We used cross-sectional analyses of claims data from an independent practice association-style (IPA) managed care organization in Rochester, New York, in 1995. The sample was made up of the 457 primary care physicians in the IPA and the 243,000 adult patients assigned to their panels. We looked at total expenditures per panel member per year generated by each primary care physician and avoidable hospitalizations among their patients.

RESULTS

After adjustment for case mix, physicians who recorded a greater proportion of mental health diagnoses generated significantly lower per panel member expenditures. For physicians in the highest quartile of recording mental health diagnoses, expenditures were 9% lower than those of physicians in the lowest quartile (95% confidence interval, 5% - 13%). There was a trend (P = .051) for patients of physicians in the highest quartile of recording mental health diagnoses to be at lower risk for an avoidable hospitalization than those of physicians in the lowest quartile.

CONCLUSIONS

Primary care physicians with higher proportions of recorded mental health diagnoses generate significantly lower panel member costs, and their patients may be less likely to be admitted for avoidable hospitalization conditions.

Why do physicians vary so widely in their referral rates?

OBJECTIVE

To determine which physician practice and psychological factors contribute to observed variation in primary care physicians' referral rates.

DESIGN

Cross-sectional questionnaire-based survey and analysis of claims database.

SETTING

A large managed care organization in the Rochester, NY, metropolitan area.

PARTICIPANTS

Internists and family physicians.

MEASUREMENTS AND MAIN RESULTS

Patient referral status (referred or not) was derived from the 1995 claims database of the managed care organization. The claims data were also used to generate a predicted risk of referral based on patient age, gender, and case mix. A physician survey completed by a sample of 182 of the physicians (66% of those eligible) included items on their practice and validated psychological scales on anxiety from uncertainty, risk aversiveness, fear of malpractice, satisfaction with practice, autonomous and controlled motivation for referrals and test ordering, and psycho-social beliefs. The relation between the risk of referral and the physician practice and psychological factors was examined using logistic regression. After adjustment for predicted risk of referral (case mix), patients were more likely to be referred if their physician was female, had more years in practice, was an internist, and used a narrower range of diagnoses (a higher Herfindahl index, also derived from the claims data). Of the psychological factors, only greater psychosocial orientation and malpractice fear was associated with greater likelihood of referral. When the physician practice factors were excluded from the analysis, risk aversion was positively associated with referral likelihood.

CONCLUSIONS

Most of the explainable variation in referral likelihood was accounted for by patient and physician practice factors like case mix, physician gender, years in practice, specialty, and the Herfindahl index. Relatively little variation was explained by any of the examined physician psychological factors.

Variations in primary care physician referral rates

OBJECTIVE

To examine primary care physician referral rate variations, including their extent and their stability over time and across diagnostic categories.

DATA SOURCES

1995/1996 claims data for adult patients from a large Independent Practitioner Association (IPA) model managed care organization (MCO) in the Rochester, NY metropolitan area. The IPA includes over 95 percent of area primary care physicians (PCPs), and the MCO includes over 50 percent area residents.

STUDY DESIGN

Referral rates (patients referred to and seen by specialists one or more times/patients seen by PCP/year) were developed for the PCPs (457 general practitioners, family physicians, and internists) in the MCO, including observed referral rates, expected referral rates based on case-mix adjustment across the whole sample, physician-specific case mix-adjusted referral rates (empirical Bayes estimates), and diagnostic category-specific case mix-adjusted referral rates.

PRINCIPAL FINDINGS

Wide variations in observed referral rates (0.01-0.69 patients referred/patients seen/year) were attenuated relatively little by case-mix adjustment and persisted in case mix-adjusted empirical Bayes estimates (0.02-0.65). The year-to-year case mix-adjusted referral rate correlation was .90. Correlations of case mix adjusted-referral rates across diagnostic categories were moderate (r=.46-.67).

CONCLUSIONS

PCP referral rates exhibit wide variations that are independent of case mix, remain stable over time, and are generalizable across diagnostic categories. Understanding this physician practice variation and its relationship to costs and outcomes is critical to evaluating the effect of current efforts to reduce PCP referral rates.