Assessing Mobility in Patients With Glioblastoma Using Digital Phenotyping-Piloting the Digital Assessment in Neuro-Oncology.

BACKGROUND AND OBJECTIVES: Digital phenotyping (DP) enables objective measurements of patient behavior and may be a useful tool in assessments of quality-of-life and functional status in neuro-oncology patients. We aimed to identify trends in mobility among patients with glioblastoma (GBM) using DP. METHODS: A total of 15 patients with GBM enrolled in a DP study were included. The Beiwe application was used to passively collect patient smartphone global positioning system data during the study period. We estimated step count, time spent at home, total distance traveled, and number of places visited in the preoperative, immediate postoperative, and late postoperative periods. Mobility trends for patients with GBM after surgery were calculated by using local regression and were compared with preoperative values and with values derived from a nonoperative spine disease group. RESULTS: One month postoperatively, median values for time spent at home and number of locations visited by patients with GBM decreased by 1.48 h and 2.79 locations, respectively. Two months postoperatively, these values further decreased by 0.38 h and 1.17 locations, respectively. Compared with the nonoperative spine group, values for time spent at home and the number of locations visited by patients with GBM 1 month postoperatively were less than control values by 0.71 h and 2.79 locations, respectively. Two months postoperatively, time spent at home for patients with GBM was higher by 1.21 h and locations visited were less than nonoperative spine group values by 1.17. Immediate postoperative values for distance traveled, maximum distance from home, and radius of gyration for patients with GBM increased by 0.346 km, 2.24 km, and 1.814 km, respectively, compared with preoperative values. CONCLUSIONS: :Trends in patients with GBM mobility throughout treatment were quantified through the use of DP in this study. DP has the potential to quantify patient behavior and recovery objectively and with minimal patient burden.

Precision Assessment of Real-World Associations Between Stress and Sleep Duration Using Actigraphy Data Collected Continuously for an Academic Year: Individual-Level Modeling Study.

BACKGROUND: Heightened stress and insufficient sleep are common in the transition to college, often co-occur, and have both been linked to negative health outcomes. A challenge concerns disentangling whether perceived stress precedes or succeeds changes in sleep. These day-to-day associations may vary across individuals, but short study periods and group-level analyses in prior research may have obscured person-specific phenotypes. OBJECTIVE: This study aims to obtain stable estimates of lead-lag associations between perceived stress and objective sleep duration in the individual, unbiased by the group, by developing an individual-level linear model that can leverage intensive longitudinal data while remaining parsimonious. METHODS: In total, 55 college students (n=6, 11% second-year students and n=49, 89% first-year students) volunteered to provide daily self-reports of perceived stress via a smartphone app and wore an actigraphy wristband for the estimation of daily sleep duration continuously throughout the academic year (median usable daily observations per participant: 178, IQR 65.5). The individual-level linear model, developed in a Bayesian framework, included the predictor and outcome of interest and a covariate for the day of the week to account for weekly patterns. We validated the model on the cohort of second-year students (n=6, used as a pilot sample) by applying it to variables expected to correlate positively within individuals: objective sleep duration and self-reported sleep quality. The model was then applied to the fully independent target sample of first-year students (n=49) for the examination of bidirectional associations between daily stress levels and sleep duration. RESULTS: Proof-of-concept analyses captured expected associations between objective sleep duration and subjective sleep quality in every pilot participant. Target analyses revealed negative associations between sleep duration and perceived stress in most of the participants (45/49, 92%), but their temporal association varied. Of the 49 participants, 19 (39%) showed a significant association (probability of direction>0.975): 8 (16%) showed elevated stress in the day associated with shorter sleep later that night, 5 (10%) showed shorter sleep associated with elevated stress the next day, and 6 (12%) showed both directions of association. Of note, when analyzed using a group-based multilevel model, individual estimates were systematically attenuated, and some even reversed sign. CONCLUSIONS: The dynamic interplay of stress and sleep in daily life is likely person specific. Paired with intensive longitudinal data, our individual-level linear model provides a precision framework for the estimation of stable real-world behavioral and psychological dynamics and may support the personalized prioritization of intervention targets for health and well-being.

Cost-based feature selection for network model choice.

Selecting a small set of informative features from a large number of possibly noisy candidates is a challenging problem with many applications in machine learning and approximate Bayesian computation. In practice, the cost of computing informative features also needs to be considered. This is particularly important for networks because the computational costs of individual features can span several orders of magnitude. We addressed this issue for the network model selection problem using two approaches. First, we adapted nine feature selection methods to account for the cost of features. We show for two classes of network models that the cost can be reduced by two orders of magnitude without considerably affecting classification accuracy (proportion of correctly identified models). Second, we selected features using pilot simulations with smaller networks. This approach reduced the computational cost by a factor of 50 without affecting classification accuracy. To demonstrate the utility of our approach, we applied it to three different yeast protein interaction networks and identified the best-fitting duplication divergence model. Supplemental materials, including computer code to reproduce our results, are available online.

Insurance-Based Disparities in Stroke Center Access in California: A Network Science Approach.

BACKGROUND: Our objectives were to determine whether there is an association between ischemic stroke patient insurance and likelihood of transfer overall and to a stroke center and whether hospital cluster modified the association between insurance and likelihood of stroke center transfer. METHODS: This retrospective network analysis of California data included every nonfederal hospital ischemic stroke admission from 2010 to 2017. Transfers from an emergency department to another hospital were categorized based on whether the patient was discharged from a stroke center (primary or comprehensive). We used logistic regression models to examine the relationship between insurance (private, Medicare, Medicaid, uninsured) and odds of (1) any transfer among patients initially presenting to nonstroke center hospital emergency departments and (2) transfer to a stroke center among transferred patients. We used a network clustering method to identify clusters of hospitals closely connected through transfers. Within each cluster, we quantified the difference between insurance groups with the highest and lowest proportion of transfers discharged from a stroke center. RESULTS: Of 332 995 total ischemic stroke encounters, 51% were female, 70% were ≥65 years, and 3.5% were transferred from the initial emergency department. Of 52 316 presenting to a nonstroke center, 3466 (7.1%) were transferred. Relative to privately insured patients, there were lower odds of transfer and of transfer to a stroke center among all groups (Medicare odds ratio, 0.24 [95% CI, 0.22-0.26] and 0.59 [95% CI, 0.50-0.71], Medicaid odds ratio, 0.26 [95% CI, 0.23-0.29] and odds ratio, 0.49 [95% CI, 0.38-0.62], uninsured odds ratio, 0.75 [95% CI, 0.63-0.89], and 0.72 [95% CI, 0.6-0.8], respectively). Among the 14 identified hospital clusters, insurance-based disparities in transfer varied and the lowest performing cluster (also the largest; n=2364 transfers) fully explained the insurance-based disparity in odds of stroke center transfer. CONCLUSIONS: Uninsured patients had less stroke center access through transfer than patients with insurance. This difference was largely explained by patterns in 1 particular hospital cluster.

Smartphone-based Assessment of Preoperative Decision Conflict and Postoperative Physical Activity Among Patients Undergoing Cancer Surgery: A Prospective Cohort Study.

OBJECTIVE: To determine the prevalence of clinically significant decision conflict (CSDC) among patients undergoing cancer surgery and associations with postoperative physical activity, as measured through smartphone accelerometer data. BACKGROUND: Patients with cancer face challenging treatment decisions, which may lead to CSDC. CSDC negatively affects patient-provider relationships, psychosocial functioning, and health-related quality of life; however, physical manifestations of CSDC remain poorly characterized. METHODS: Adult smartphone-owners undergoing surgery for breast, skin-soft-tissue, head-and-neck, or abdominal cancer (July 2017-2019) were approached. Patients downloaded the Beiwe application that delivered the Decision Conflict Scale (DCS) preoperatively and collected smartphone accelerometer data continuously from enrollment through 6 months postop-eratively. Restricted-cubic-spline regression, adjusting for a priori potential confounders (age, type of surgery, support status, and postoperative complications) was used to determine trends in postoperative daily physical activity among patients with and without CSDC (DCS score >25/100). RESULTS: Among 99 patients who downloaded the application, 85 completed the DCS (86% participation rate). Twenty-three (27%) reported CSDC. These patients were younger (mean age 48.3 years [standard deviation 14.2]-vs-55.0 [13.3], P = 0.047) and more frequently lived alone (22%-vs-6%, P = 0.042). There were no differences in preoperative physical activity (115.4 minutes [95%CI 90.9, 139.9]-vs-110.8 [95%CI 95.7, 126.0], P = 0.753). Adjusted postoperative physical activity was lower among patients reporting CSDC at 30 days (difference 33.1 minutes [95%CI 5.93,60.2], P = 0.017), 60 days 35.5 [95%CI 8.50, 62.5], P = 0.010 and 90 days 31.8 [95%CI 5.44, 58.1], P = 0.018 postoperatively. CONCLUSIONS: CSDC was prevalent among patients who underwent cancer surgery and associated with lower postoperatively daily physical activity. These data highlight the importance of addressing modifiable decisional needs of patients through enhanced shared decision-making.

Decision Models and Technology Can Help Psychiatry Develop Biomarkers.

Why is psychiatry unable to define clinically useful biomarkers? We explore this question from the vantage of data and decision science and consider biomarkers as a form of phenotypic data that resolves a well-defined clinical decision. We introduce a framework that systematizes different forms of phenotypic data and further introduce the concept of decision model to describe the strategies a clinician uses to seek out, combine, and act on clinical data. Though many medical specialties rely on quantitative clinical data and operationalized decision models, we observe that, in psychiatry, clinical data are gathered and used in idiosyncratic decision models that exist solely in the clinician's mind and therefore are outside empirical evaluation. This, we argue, is a fundamental reason why psychiatry is unable to define clinically useful biomarkers: because psychiatry does not currently quantify clinical data, decision models cannot be operationalized and, in the absence of an operationalized decision model, it is impossible to define how a biomarker might be of use. Here, psychiatry might benefit from digital technologies that have recently emerged specifically to quantify clinically relevant facets of human behavior. We propose that digital tools might help psychiatry in two ways: first, by quantifying data already present in the standard clinical interaction and by allowing decision models to be operationalized and evaluated; second, by testing whether new forms of data might have value within an operationalized decision model. We reference successes from other medical specialties to illustrate how quantitative data and operationalized decision models improve patient care.

Assessment of Racial Disparities in Primary Care Physician Specialty Referrals.

Importance: Disparities in quality of care according to patient race and socioeconomic status persist in the US. Differential referral patterns to specialist physicians might be associated with observed disparities. Objective: To examine whether differences exist between Black and White Medicare beneficiaries in the observed patterns of patient sharing between primary care physicians (PCPs) and physicians in the 6 specialties to which patients were most frequently referred. Design, Setting, and Participants: This cross-sectional observational study of Black and White Medicare beneficiaries used claims data from 2009 to 2010 on 100% of traditional Medicare beneficiaries who were seen by PCPs and selected high-volume specialists in 12 health care markets with at least 10% of the population being Black. Statistical analyses were conducted from December 20, 2017, to September 30, 2020. Exposures: Differences in patterns of patient sharing among Black and White patients. Main Outcomes and Measures: Primary care physician and specialist degree (the number of other PCPs or specialists to whom each physician is connected) and strength (the number of shared patients per connection, overall, for Black patients and White patients and after equalizing the numbers of Black and White patients per PCP), as well as distance between PCP and patient and specialist zip code centroids. Results: The 12 selected markets ranged in size from Manhattan, New York (187 054 Black or White beneficiaries seen by at least 2 physicians within an episode of care; 9794 total physicians), to Tallahassee, Florida (44 644 Black or White beneficiaries seen by at least 2 physicians within an episode of care; 847 total physicians). The percentage of Black beneficiaries ranged from 11.5% (Huntsville, Alabama) to 46.8% (Chicago, Illinois). The mean PCP-specialist degree (number of specialists with whom a PCP shares patients) was lower for Black patients than for White patients. For instance, the mean PCP-cardiologist degree across all markets for White patients was 17.5 compared with 8.8 for Black patients. After sampling White patients to equalize the numbers of patients seen, the degree differences narrowed but were still not equivalent in many markets (eg, for all specialties in Baton Rouge, Louisiana: 4.5 for Black patients vs 5.7 for White patients). Specialist networks among White patients were much larger than those constructed based just on Black patients (eg, for cardiology across all markets: 135 for Black patients vs 330 for White patients), even after equalizing the numbers of patients seen per PCP (123 for Black patients vs 211 for White patients). The overall test for differences in referral patterns was statistically significant for all 6 specialties examined in 7 of the 12 markets and in 5 specialties for another 3. Conclusions and Relevance: This study suggests that differences exist in specialist referral patterns by race among Medicare beneficiaries. This is an observational study, and thus some differences might have resulted from patient-initiated visits to specialists.

Smartphone Global Positioning System (GPS) Data Enhances Recovery Assessment After Breast Cancer Surgery.

PURPOSE: We sought to determine whether smartphone GPS data uncovered differences in recovery after breast-conserving surgery (BCS) and mastectomy, and how these data aligned with self-reported quality of life (QoL). METHODS: In a prospective pilot study, adult smartphone-owners undergoing breast surgery downloaded an application that continuously collected smartphone GPS data for 1 week preoperatively and 6 months postoperatively. QoL was assessed with the Short-Form-36 (SF36) via smartphone delivery preoperatively and 4 and 12 weeks postoperatively. Endpoints were trends in daily GPS-derived distance traveled and home time, as well as SF36 Physical (PCS) and Mental Component Scores (MCS) comparing BCS and mastectomy patients. RESULTS: Thirty-one patients were included. Sixteen BCS and fifteen mastectomy patients were followed for a mean of 201 (SD 161) and 174 (107) days, respectively. There were no baseline differences in demographics, PCS/MCS, home time, or distance traveled. Through 12 weeks postoperatively, mastectomy patients spent more time at home [e.g., week 4: 16.7 h 95% CI (14.3, 19.6) vs. 11.0 h (9.4, 12.9), p < 0.001] and traveled shorter distances [e.g., week 4: 52.5 km 95% CI (36.1, 76.0) vs. 107.7 km (75.8-152.9), p = 0.009] compared with BCS patients. There were no significant QoL differences throughout the study as measured by the MCS [e.g., week 4 difference: 7.83 95% CI (- 9.02, 24.7), p = 0.362] or PCS [e.g., week 4 difference: 8.14 (- 6.67, 22.9), p = 0.281]. GPS and QoL trends were uncorrelated (ρ < ± 0.26, p > 0.05). CONCLUSIONS: Differences in BCS and mastectomy recovery were successfully captured using smartphone GPS data. These data may describe currently unmeasured aspects of physical and mental recovery, which could supplement traditional and QoL outcomes to inform shared decision-making.

Digital Phenotyping to Quantify Psychosocial Well-Being Trajectories After Spinal Cord Injury.

OBJECTIVE: The aim of the study was to explore feasibility of smartphone-based digital phenotyping methods to examine depression and its relation to psychosocial well-being indicators after spinal cord injury. DESIGN: Smartphone research platform obtained smartphone sensor and survey data among community-living adult wheelchair users with spinal cord injury. Weekly measurements for 4 mos included Patient Health Questionnaire 8, Spinal Cord Injury-Quality of Life Satisfaction with and ability to participate in social roles and activities, global positioning system-derived community mobility metrics, health conditions, and physical activity. RESULTS: Forty-three individuals were enrolled. Study retention was higher among individuals offered financial incentives (78%) compared with participants enrolled before incentives (50%). Participants who dropped out more commonly had nontraumatic or acute spinal cord injury, were older, and had less satisfaction and lower participation in social roles and activities. Among 15 individuals with complete data, half had 1 wk or more of mild depression. Those with depression had frequent health issues, less satisfaction, and lower participation in social roles and activities. Those without depression experienced increased social engagement over time. Average community mobility was similar across depression groups. Relationships were typically in-phase but also varied by individual. CONCLUSIONS: Smartphone-based digital phenotyping of psychosocial well-being after spinal cord injury is feasible but not without attrition challenges. Individual differences in depression patterns highlight clinical utility of scaling these methods.

Hospital Factors Associated With Interhospital Transfer Destination for Stroke in the Northeast United States.

Background We aimed to determine if there is an association between hospital quality and the likelihood of a given hospital being a preferred transfer destination for stroke patients. Methods and Results Data from Medicare claims identified acute ischemic stroke transferred between 394 northeast US hospitals from 2007 to 2011. Hospitals were categorized as transferring (n=136), retaining (n=241), or receiving (n=17) hospitals based on the proportion of acute ischemic stroke encounters transferred or received. We identified all 6409 potential dyads of sending and receiving hospitals, and categorized dyads as connected if ≥5 patients were transferred between the hospitals annually (n=82). We used logistic regression to identify hospital characteristics associated with establishing a connected dyad, exploring the effect of adjusting for different quality measures and outcomes. We also adjusted for driving distance between hospitals, receiving hospital stroke volume, and the number of hospitals in the receiving hospital referral region. The odds of establishing a transfer connection increased when rate of alteplase administration increased at the receiving hospital or decreased at the sending hospital, however this finding did not hold after applying a potential strategy to adjust for clustering. Receiving hospital performance on 90-day home time was not associated with likelihood of transfer connection. Conclusions Among northeast US hospitals, we found that differences in hospital quality, specifically higher levels of alteplase administration, may be associated with increased likelihood of being a transfer destination. Further research is needed to better understand acute ischemic stroke transfer patterns to optimize stroke transfer systems.

Association of Physician Peer Influence With Subsequent Physician Adoption and Use of Bevacizumab.

Importance: Understanding adoption of new cancer therapies may help identify opportunities to increase use for high-value indications. Objective: To determine whether use of bevacizumab in 2005 to 2006 by oncologists' peers was associated with greater bevacizumab use among oncologists in 2007 to 2010. Design, Setting, and Participants: This cohort study of physicians and their patients took place in 51 randomly selected hospital referral regions in the United States. Participants were 44 012 fee-for-service Medicare beneficiaries aged 65 years or older with cancers of the colorectum, lung, breast, kidney, brain, or ovary treated by 3261 oncologists in 2005 to 2010 and assigned to one of 252 communities. Data were analyzed in 2017 to 2018. Exposures: Among patients treated with chemotherapy during 2007 to 2010 by an oncologist who had not treated patients with bevacizumab in 2005 to 2006, models assessed the association of bevacizumab use with rates of bevacizumab use in their physician's community of connected physicians in 2005 to 2006. Models adjusted for patient and physician characteristics and physician, practice, and community random effects. Main Outcomes and Measures: Receipt of bevacizumab. Results: A total of 34 750 patients (14 126 [40.6%] aged ≥75 years; 21 321 [61.4%] female) with cancers of the colorectum, lung, breast, kidney, brain, and ovary were treated with chemotherapy in 2005 to 2006 in the 51 hospital referral regions. Among 9262 patients treated in 2007 to 2010 by 829 physicians whose patients did not use bevacizumab in 2005 to 2006, 3654 (39.5%) were aged 75 years or older and 6227 (67.2%) were female. The rate of bevacizumab use relative to other chemotherapy in 2007 to 2010 by tertile of use (bevacizumab for <4.4%, 4.4%-6.2%, and >6.2% of all patients receiving chemotherapy) among their physician's peers in 2005 to 2006 was 10.0%, 9.5%, and 13.6%, respectively. After adjustment, use of bevacizumab in 2007 to 2010 was greater among physicians in communities with the highest rates of bevacizumab use in 2005 to 2006 compared with those whose peers were in the lowest tertile of bevacizumab use in 2005 to 2006 (adjusted odds ratio, 1.64; 95% CI, 1.20-2.25). Conclusions and Relevance: This study found that an increase in oncologists' adoption and use of bevacizumab in the years after its approval was associated with their peer physicians being earlier adopters. As organizations seek to provide better care at lower costs, interventions that leverage physician ties may help to promote adoption of high-value use of new cancer treatments and deimplementation of low-value therapies.

Decomposition of the US black/white inequality in premature mortality, 2010-2015: an observational study.

OBJECTIVE: Decompose the US black/white inequality in premature mortality into shared and group-specific risks to better inform health policy. SETTING: All 50 US states and the District of Columbia, 2010 to 2015. PARTICIPANTS: A total of 2.85 million non-Hispanic white and 762 639 non-Hispanic black US-resident decedents. PRIMARY AND SECONDARY OUTCOME MEASURES: The race-specific county-level relative risks for US blacks and whites, separately, and the risk ratio between groups. RESULTS: There is substantial geographic variation in premature mortality for both groups and the risk ratio between groups. After adjusting for median household income, county-level relative risks ranged from 0.46 to 2.04 (median: 1.03) for whites and from 0.31 to 3.28 (median: 1.15) for blacks. County-level risk ratios (black/white) ranged from 0.33 to 4.56 (median: 1.09). Half of the geographic variation in white premature mortality was shared with blacks, while only 15% of the geographic variation in black premature mortality was shared with whites. Non-Hispanic blacks experience substantial geographic variation in premature mortality that is not shared with whites. Moreover, black-specific geographic variation was not accounted for by median household income. CONCLUSION: Understanding geographic variation in mortality is crucial to informing health policy; however, estimating mortality is difficult at small spatial scales or for small subpopulations. Bayesian joint spatial models ameliorate many of these issues and can provide a nuanced decomposition of risk. Using premature mortality as an example application, we show that Bayesian joint spatial models are a powerful tool as researchers grapple with disentangling neighbourhood contextual effects and sociodemographic compositional effects of an area when evaluating health outcomes. Further research is necessary in fully understanding when and how these models can be applied in an epidemiological setting.

Passive data collection and use in healthcare: A systematic review of ethical issues.

INTRODUCTION: Passive data refers to data generated without the active participation of the subject. This includes data from global positioning systems and accelerometers or metadata on phone call and text activity. Although the potential healthcare applications are far-reaching, passive data raises numerous ethical challenges. MATERIALS AND METHODS: We performed a systematic review to identify all ethical concerns, normative standpoints, and underlying arguments related to the use of passive data in healthcare. RESULTS: Among the various challenges discussed in the ethical literature, informational privacy, informed consent, and data security were the primary focus of the current debate. Other topics of discussion were the evaluation and regulation of products, equity in access, vulnerable patient groups, ownership, and secondary use. CONCLUSION: No clear ethical framework has been established that stimulates passive data-driven innovation while protecting patient integrity. The consensus in the ethical literature, as well as the parallels with similar concerns and solutions in other fields, can lay a foundation for the construction of an ethical framework. The future debate should focus on conflicts between two or more ethical, technical, or clinical values to ensure a safe and effective implementation of passive data in healthcare.

Influence of Peer Physicians on Intensity of End-of-Life Care for Cancer Decedents.

BACKGROUND: The intensity of end-of-life care varies substantially both within and between areas. Differing practice patterns of individual physicians are likely influenced by their peers. OBJECTIVE: To assess whether intensity of end-of-life care previously provided by a physician's peers influences patterns of care at the end-of-life for that physician's patients. RESEARCH DESIGN: Observational study. SUBJECTS: A total of 185,947 fee-for-service Medicare enrollees with cancer who died during 2006-2010 who were treated by 26,383 physicians. MEASURES: Spending in the last month of life, >1 emergency room visit, >1 hospitalization, intensive care unit admission in the last month of life, chemotherapy within 2 weeks of death, no/late hospice, terminal hospitalization. RESULTS: Mean (SD) spending in the last month of life was $16,237 ($17,124). For each additional $1000 of spending for a peer physician's patients in the prior year, spending for the ego physician's patients was $83 higher (P<0.001). Among physicians with peers both in and out of their practice, more of the peer effect was explained by physicians outside of the practice ($72 increase for each $1000 increase by peer physicians' patients, P<0.001) than peer physicians in the practice ($27 for each $1000 increase by within-practice peer physicians' patients, P=0.01). Results were similar across the other measures of end-of-life care intensity. CONCLUSIONS: Physician's peers exert influence on the intensity of care delivered to that physician's patients at the end-of-life. Physician education efforts led by influential providers and provider organizations may have potential to improve the delivery of high-value end-of-life care.

A multilevel approach to modeling health inequalities at the intersection of multiple social identities.

RATIONALE: Examining interactions between numerous interlocking social identities and the systems of oppression and privilege that shape them is central to health inequalities research. Multilevel models are an alternative and novel approach to examining health inequalities at the intersection of multiple social identities. This approach draws attention to the heterogeneity within and between intersectional social strata by partitioning the total variance across two levels. METHOD: Utilizing a familiar empirical example from social epidemiology-body mass index among U.S. adults (N = 32,788)-we compare the application of multilevel models to the conventional fixed effects approach to studying high-dimension interactions. Researchers are often confronted with the need to explore numerous interactions of identities and social processes. We explore the interactions of five dimensions of social identity and position-gender, race/ethnicity, income, education, and age-for a total of 384 unique intersectional social strata. RESULTS: We find that the multilevel approach provides advantages over conventional models, including scalability for higher dimensions, adjustment for sample size of social strata, model parsimony, and ease of interpretation. CONCLUSION: Considerable variation is attributable to the within-strata level, indicating the low discriminatory accuracy of these intersectional identities and the high within-strata heterogeneity of risk that remains unexplained. Multilevel modeling is an innovative and valuable tool for evaluating the intersectionality of health inequalities.

Patient-Sharing Networks of Physicians and Health Care Utilization and Spending Among Medicare Beneficiaries.

Importance: Physicians are embedded in informal networks in which they share patients, information, and behaviors. Objective: We examined the association between physician network properties and health care spending, utilization, and quality of care among Medicare beneficiaries. Design, Setting, and Participants: In this cross-sectional study, we applied methods from social network analysis to Medicare administrative data from 2006 to 2010 for an average of 3 761 223 Medicare beneficiaries per year seen by 40 241 physicians practicing in 51 hospital referral regions (HRRs) to identify networks of physicians linked by shared patients. We improved on prior methods by restricting links to physicians who shared patients for distinct episodes of care, thereby excluding potentially spurious linkages between physicians treating common patients but for unrelated reasons. We also identified naturally occurring communities of more tightly linked physicians in each region. We examined the relationship between network properties measured in the prior year and outcomes in the subsequent year using regression models. Main Outcomes and Measures: Spending on total medical services, hospital, physician, and other services, use of services, and quality of care. Results: The mean patient age across the 5 years of study was 72.3 years and 58.5% of the participants were women. The mean age across communities of included physicians was 49 years and approximately 78% were men. Mean total annual spending per patient was $10 051. Total spending was higher for patients of physicians with more connections to other physicians ($1009 for a 1-standard deviation increase, P < .001) and more shared care outside of their community ($172, P < .001). Spending on inpatient care was slightly lower for patients of physicians whose communities had higher proportions of primary care physicians (-$38, P < .001). Patients cared for by physicians linked to more physicians also had more hospital admissions and days (0.02 and 0.18, respectively; both P < .001 for a 1-standard deviation increase in the number of connected physicians), more emergency visits (0.02, P < .001), more visits to specialists (0.37, P < .001), and more primary care visits (0.11, P < .001). Patients whose physicians' networks had more primary care physicians had more primary care visits (0.44, P < .001) and fewer specialist and emergency visits (-0.33 [P < .001] and -0.008 [P = .008], respectively). The various measures of quality were inconsistently related to the network measures. Conclusions and Relevance: Characteristics of physicians' networks and the position of physicians in the network were associated with overall spending and utilization of services for Medicare beneficiaries.

Assessing the impact of colonoscopy complications on use of colonoscopy among primary care physicians and other connected physicians: an observational study of older Americans.

OBJECTIVES: Psychological biases can distort treatment decision-making. The availability heuristic is one such bias, wherein events that are recent, vivid or easily imagined are readily 'available' to memory and are therefore judged more likely to occur than expected based on epidemiological data. We assessed if the occurrence of a serious colonoscopy complication for a primary care physician's patient influenced colonoscopy rates for the physician's other patients. DESIGN: Longitudinal study with time-varying exposure variables. SETTING/PARTICIPANTS: Individuals living in 51 hospital referral regions across the USA identified based on enrolment in fee-for-service Medicare during 2005-2010. We assigned patients to a primary care physician based on office visits during the prior 2 years. EXPOSURES: For each physician in each month, we calculated the proportion of patients assigned to them who had a colonoscopy. We identified two serious complications of which the primary care provider would very likely be aware: gastrointestinal bleed or perforation leading to hospitalisation or death within 14 days of colonoscopy. MAIN OUTCOME MEASURES: We employed Poisson regression models including physician fixed effects to assess the change in number of colonoscopies in the four quarters following an adverse colonoscopy event. RESULTS: We identified 5 360 191 patients assigned to 30 704 physicians. 4864 physicians (16%) had at least one patient with an adverse event. The estimated change in the quarterly number of colonoscopies among physicians' patients was significantly lower in quarter 2 following an adverse colonoscopy event (change=-2.1% (95% CI -3.4 to -0.8%)), before returning to the rate expected in the absence of an adverse event. CONCLUSIONS: Having a patient experience a serious adverse colonoscopy event was associated with a small and temporary decline in colonoscopy rates among a physician's other patients. This finding provides empirical evidence for the influence of notable adverse events on care, possibly due to the availability heuristic.

Multiple contexts and adolescent body mass index: Schools, neighborhoods, and social networks.

Adolescent health and behaviors are influenced by multiple contexts, including schools, neighborhoods, and social networks, yet these contexts are rarely considered simultaneously. In this study we combine social network community detection analysis and cross-classified multilevel modeling in order to compare the contributions of each of these three contexts to the total variation in adolescent body mass index (BMI). Wave 1 of the National Longitudinal Study of Adolescent to Adult Health is used, and for robustness we conduct the analysis in both the core sample (122 schools; N = 14,144) and a sub-set of the sample (16 schools; N = 3335), known as the saturated sample due to its completeness of neighborhood data. After adjusting for relevant covariates, we find that the school-level and neighborhood-level contributions to the variance are modest compared with the network community-level (σ(2)school = 0.069, σ(2)neighborhood = 0.144, σ(2)network = 0.463). These results are robust to two alternative algorithms for specifying network communities, and to analysis in the saturated sample. While this study does not determine whether network effects are attributable to social influence or selection, it does highlight the salience of adolescent social networks and indicates that they may be a promising context to address in the design of health promotion programs.