Characteristics of adult primary care patients as predictors of future health services charges

Maternal Schizophrenia and the Risk of a Childhood Chronic Condition

BACKGROUND AND HYPOTHESIS

Maternal schizophrenia heightens the risk for certain perinatal complications, yet it is not known to what degree future childhood chronic health conditions (Childhood-CC) might arise.

STUDY DESIGN

This population-based cohort study using health administrative data from Ontario, Canada (1995-2018) compared 5066 children of mothers with schizophrenia to 25 324 children of mothers without schizophrenia, propensity-matched on birth-year, maternal age, parity, immigrant status, income, region of residence, and maternal medical and psychiatric conditions other than schizophrenia. Cox proportional hazard models generated hazard ratios (HR) and 95% confidence intervals (CI) for incident Childhood-CCs, and all-cause mortality, up to age 19 years.

STUDY RESULTS

Six hundred and fifty-six children exposed to maternal schizophrenia developed a Childhood-CC (20.5/1000 person-years) vs. 2872 unexposed children (17.1/1000 person-years)-an HR of 1.18, 95% CI 1.08-1.28. Corresponding rates were 3.3 vs. 1.9/1000 person-years (1.77, 1.44-2.18) for mental health Childhood-CC, and 18.0 vs. 15.7/1000 person-years (1.13, 1.04-1.24) for non-mental health Childhood-CC. All-cause mortality rates were 1.2 vs. 0.8/1000 person-years (1.34, 0.96-1.89). Risk for children exposed to maternal schizophrenia was similar whether or not children were discharged to social service care. From age 1 year, risk was greater for children whose mothers were diagnosed with schizophrenia prior to pregnancy than for children whose mothers were diagnosed with schizophrenia postnatally.

CONCLUSIONS

A child exposed to maternal schizophrenia is at elevated risk of chronic health conditions including mental and physical subtypes. Future research should examine what explains the increased risk particularly for physical health conditions, and what preventive and treatment efforts are needed for these children.

Dimensionality, variance composition, and development of a brief form of the duke health profile, and its measurement invariance across five gender identity groups

OBJECTIVE

To conduct advanced psychometric analyses on the Duke Health Profile, a popular measure of health-related quality of life.

DESIGN

Online survey. Data (N = 1233, 34.3% transgender) were from community and college participants. Dimensionality was assessed for the first time using exploratory factor analysis (EFA) with part of the sample, followed by single- and multi-group confirmatory factor analyses (CFA) with the balance of the sample.

RESULTS

EFA resulted in a 14-item three factor structure: mental, physical and social health. CFA estimated four models (common factors, bifactor, hierarchical, unidimensional), none demonstrated adequate fit. From another EFA specifying one factor, the 6-item Duke Health Profile-Brief Form was developed based on updated guidelines for shortening composite measurement scales, which was assessed using CFA, finding good fit to the data. Measurement invariance by gender was assessed across the diverse gender spectrum, finding evidence for configural, metric, and partial scalar invariance.

CONCLUSIONS

There is insufficient evidence to use the general, mental, social and physical health scores of the DUKE Health Profile. However, there is evidence supporting the use of the unidimensional DUKE-BF, which is largely invariant between cisgender men and women, transgender men and women, and transgender men and non-binary participants.

Developing a new predictor of health expenditure: preliminary results from a primary healthcare setting

OBJECTIVES

Risk adjustment is a widely used tool for health expenditure prediction and control. Early approaches for estimating health expenditure were based on patient demographic variables alone, whereas more recent models incorporate patient information, such as chronic medical conditions, clinical diagnoses, and self-reported health status. Many studies have investigated the health expenditure predictive capacity of single demographic, morbidity, or health-related quality of life measures, but the best models prove to be those that include them all. The aim of this study was to develop an index that combines measures of perceived health and disease severity and to compare its efficacy in predicting health expenditure with that of the measures taken individually.

STUDY DESIGN

This is a linked cross-sectional study.

METHODS

In 2009 and 2010, the health-related quality of life questionnaire SF-36 (8 scales, two indices: Physical Component Summary [PCS] and Mental Component Summary [MCS]) was distributed to 886 patients of general practitioners in the Province of Siena, Italy. Severity of diseases was calculated for each patient using the Charlson Index (CH-I) and Cumulative Illness Rating Scale Severity Index (CIRS-SI). Siena Local Health Unit 2012 data on health expenditure were obtained for each patient. Multivariate linear regression was applied to test the performance of severity (CH-I, CIRS-SI) and perceived health (PCS and MCS) measures in predicting health expenditure. The indexes that predicted health expenditure best were then combined in a new tool, and its expenditure predictive capacity was tested.

RESULTS

The best health expenditure predictors proved to be PCS and SI (R² = 0.15 and R² = 0.17, respectively). When combined in a new index (PCS-SI), better predictive capacity of health expenditure was obtained than with the two single measures separately (R² = 0.19).

CONCLUSIONS

A multidimensional indicator proved to be a better predictor of healthcare expenditure than single health measures.

Predicting Health Care Utilization for Children With Respiratory Insufficiency Using Parent-Proxy Ratings of Children's Health-Related Quality of Life

INTRODUCTION

Children with chronic respiratory insufficiency and mechanical ventilation often experience acute illnesses requiring unscheduled hospitalizations. Health-related quality of life (HRQL) may predict future health care utilization.

METHODS

Participants were 30 days to 22 years old with chronic respiratory insufficiency (N = 120). Parent-proxies completed global HRQL and general health measures. Outcomes were total health care (emergency department, outpatient, inpatient) and inpatient days over 6 months. Adjusted negative binomial regression estimated the effects of global HRQL and general health on utilization.

RESULTS

Three quarters of children had any utilization; 32% had hospitalizations. Children with poor/fair global HRQL had 3.7 times more health care days than those with very good/excellent global HRQL. Children with poor/fair global HRQL had 6.3 times more inpatient days than those with very good/excellent global HRQL. Similar relationships existed between general health and utilization.

DISCUSSION

HRQL was associated with health care and inpatient days. Clinical teams can use HRQL as a marker for utilization risk, enabling potentially earlier intervention, better outcomes, and lower costs.

Factors associated with postpartum psychiatric admission in a population-based cohort of women with schizophrenia

OBJECTIVE

We aimed to identify factors associated with postpartum psychiatric admission in schizophrenia.

METHOD

In a population-based cohort study of 1433 mothers with schizophrenia in Ontario, Canada (2003-2011), we compared women with and without psychiatric admission in the 1st year postpartum on demographic, maternal medical/obstetrical, infant and psychiatric factors and identified factors independently associated with admission.

RESULTS

Admitted women (n = 275, 19%) were less likely to be adolescents, more likely to be low income and less likely to have received prenatal ultrasound before 20 weeks gestation compared to non-admitted women. They also had higher rates of predelivery psychiatric comorbidity and mental health service use. Factors independently associated with postpartum admission were age (<20 vs. ≥35 years: adjusted risk ratio, aRR, 0.48, 95% CI 0.24-0.96), income (lowest vs. highest income: aRR 1.67, 1.13-2.47) and the following mental health service use factors in pregnancy: admission (≥35 days/year vs. no days, aRR 4.54, 3.65-5.65), outpatient mental health care (no visits vs. ≥2 visits aRR 0.35, 0.27-0.47) and presence of a consistent mental health care provider during pregnancy (aRR 0.69, 0.54-0.89).

CONCLUSION

Certain subgroups of women with schizophrenia may benefit from targeted intervention to mitigate risk for postpartum admission.

Using Patient-Reported Outcome Measures to Capture the Patient's Voice in Research and Care of Juvenile Idiopathic Arthritis

Patient-reported outcome (PRO) measures provide a valuable window into how patients with juvenile idiopathic arthritis and their parents perceive their functioning, quality of life, and medication side effects in the context of their disease and treatment. Momentum behind adoption of PRO measures is increasing as these patient-relevant tools capture information pertinent to taking a patient-centered approach to health care and research. This article reviews the clinical and research utility of obtaining PROs across domains applicable to the experience of juvenile idiopathic arthritis and summarizes available self-report and parent-proxy PRO measures. Current challenges and limitations of PRO usage are discussed.

Characteristics of Newly Enrolled Members of an Integrated Delivery System after the Affordable Care Act

Of 89,289 newly enrolled non-Medicare members, 25.3% completed the Brief Health Questionnaire between 1/1/2014, and 8/31/2014. Of these, 3593 respondents were insured through Medicaid, 9434 through the individual health exchange, and 9521 through primarily commercial plans. Of Medicaid, exchange, and commercial members, 19.5%, 7.1%, and 5.3%, respectively, self-reported fair or poor health; 12.9%, 2.0%, and 3.3% of each group self-reported 2 or more Emergency Department visits during the previous year; and 8.1%, 4.3%, and 4.4% self-reported an inpatient admission during the previous year.

Antipsychotic drug use in pregnancy: high dimensional, propensity matched, population based cohort study

OBJECTIVE

To evaluate maternal medical and perinatal outcomes associated with antipsychotic drug use in pregnancy.

DESIGN

High dimensional propensity score (HDPS) matched cohort study.

SETTING

Multiple linked population health administrative databases in the entire province of Ontario, Canada.

PARTICIPANTS

Among women who delivered a singleton infant between 2003 and 2012, and who were eligible for provincially funded drug coverage, those with ≥ 2 consecutive prescriptions for an antipsychotic medication during pregnancy, at least one of which was filled in the first or second trimester, were selected. Of these antipsychotic drug users, 1021 were matched 1:1 with 1021 non-users by means of a HDPS algorithm.

MAIN OUTCOME MEASURES

The main maternal medical outcomes were gestational diabetes, hypertensive disorders of pregnancy, and venous thromboembolism. The main perinatal outcomes were preterm birth (<37 weeks), and a birth weight <3rd or >97th centile. Conditional Poisson regression analysis was used to generate rate ratios and 95% confidence intervals, adjusting for additionally prescribed non-antipsychotic psychotropic medications.

RESULTS

Compared with non-users, women prescribed an antipsychotic medication in pregnancy did not seem to be at higher risk of gestational diabetes (rate ratio 1.10 (95% CI 0.77 to 1.57)), hypertensive disorders of pregnancy (1.12 (0.70 to 1.78)), or venous thromboembolism (0.95 (0.40 to 2.27)). The preterm birth rate, though high among antipsychotic users (14.5%) and matched non-users (14.3%), was not relatively different (rate ratio 0.99 (0.78 to 1.26)). Neither birth weight <3rd centile or >97th centile was associated with antipsychotic drug use in pregnancy (rate ratios 1.21 (0.81 to 1.82) and 1.26 (0.69 to 2.29) respectively).

CONCLUSIONS

Antipsychotic drug use in pregnancy had minimal evident impact on important maternal medical and short term perinatal outcomes. However, the rate of adverse outcomes is high enough to warrant careful assessment of maternal and fetal wellbeing among women prescribed an antipsychotic drug in pregnancy.

Population health needs assessment and healthcare services use in a 3 years follow-up on administrative and clinical data: results from the Brisighella Heart Study

INTRODUCTION

A large number of epidemiological trials clearly show the impact of the main cardiovascular disease risk factors in term of hospitalization and related cost, but relatively less frequently if this reflect the health needs of a given population.

AIM

To develop a model for the health needs-assessment that will be applied to verify if and how the prevalence of some classical risk factors for cardiovascular disease predicts mortality and hospitalisation episodes at 3 years, and if it could express the health need of that population. The long-life clinical record of 1,704 subjects, recruited during the 2004 Brisighella Heart Study survey, has been monitored. We defined the health profile of these subjects at 2004 (based on clinical history, smoking and dietary habits, physical activity, drug use, anthropometric data, blood pressure, and hematological data) and then sampled data relative to their hospitalisations, mortality, and general medical assistance.

RESULTS

Our results shows that age over 65 years (OR 4.08; 95 % CI 2.74-6.08), hypertension (OR 3.44; 95 % CI 2.36-5.01) and hypercholesterolemia (OR 1.33; 95 % CI 0.92-1.94) increase the probability to get hospitalised. Furthermore, the burden of care was defined and computed for our sample. Vascular and respiratory diseases [Burden of health care (Bc) = 24.5 and 36.5, respectively] are the most costly DRGs which means that the biggest part of our resources directed to cardiovascular patients were provided for these diagnoses.

CONCLUSION

The application of the proposed model could help policy makers and researchers in directing resources and workforce in the treatment of cardiovascular diseases.

Measures of multimorbidity and morbidity burden for use in primary care and community settings: a systematic review and guide

PURPOSE

Many patients consulting in primary care have multiple conditions (multimorbidity). Aims of this review were to identify measures of multimorbidity and morbidity burden suitable for use in research in primary care and community populations, and to investigate their validity in relation to anticipated associations with patient characteristics, process measures, and health outcomes.

METHODS

Studies were identified using searches in MEDLINE and EMBASE from inception to December 2009 and bibliographies.

RESULTS

Included were 194 articles describing 17 different measures. Commonly used measures included disease counts (n = 98), Chronic Disease Score (CDS)/RxRisk (n = 17), Adjusted Clinical Groups (ACG) System (n = 25), the Charlson index (n = 38), the Cumulative Index Illness Rating Scale (CIRS; n = 10) and the Duke Severity of Illness Checklist (DUSOI; n = 6). Studies that compared measures suggest their predictive validity for the same outcome differs only slightly. Evidence is strongest for the ACG System, Charlson index, or disease counts in relation to care utilization; for the ACG System in relation to costs; for Charlson index in relation to mortality; and for disease counts or Charlson index in relation to quality of life. Simple counts of diseases or medications perform almost as well as complex measures in predicting most outcomes. Combining measures can improve validity.

CONCLUSIONS

The measures most commonly used in primary care and community settings are disease counts, Charlson index, ACG System, CIRS, CDS, and DUSOI. Different measures are most appropriate according to the outcome of interest. Choice of measure will also depend on the type of data available. More research is needed to directly compare performance of different measures.

Risk of acute complications of diabetes among people with schizophrenia in Ontario, Canada

OBJECTIVE

Diabetes mellitus (DM) is a complex, chronic disease requiring active self-management and coordinated care. This study aimed to evaluate the relationship between schizophrenia and risk of preventable, acute DM complications.

RESEARCH DESIGN AND METHODS

With the use of administrative data, a retrospective study assessed acute DM complications (emergency department [ED] visits or hospitalization for hypo- or hyperglycemia and hospital admissions for infections) among Ontario residents ages 18-50 with schizophrenia and newly diagnosed DM between 1995 and 2005, comparing people with and without pre-existing schizophrenia. Primary outcome was ED visit or hospitalization for hypo- or hyperglycemia. Secondary outcome was the first of either the primary outcome or hospitalization for infection.

RESULTS

People with schizophrenia had a 74% greater risk of requiring a hospital visit for hypo- or hyperglycemia (hazard ratio [HR] = 1.74, 95% confidence interval 1.42-2.12) compared with those without schizophrenia. The risk was similar when the outcome included infection (HR = 1.62, 95% CI 1.39-1.89). Outcomes remained significant after adjustment for baseline characteristics.

CONCLUSIONS

People with schizophrenia are at greater risk for developing an acute complication of DM. Understanding this relationship will direct future studies assessing barriers to care and implementation of individualized approaches to care for this population.

Using health-related quality of life to predict and manage pediatric healthcare

Increasing healthcare costs and the prevalence of managed care make population health management an imperative. Health-related quality of life (HRQOL) is a multidimensional construct that includes both physical and psychosocial (i.e., social, emotional and role) dimensions. Early studies suggest that HRQOL can predict costs of care for pediatric populations. A key issue is how to manage the care of those identified as high need. Here again, HRQOL measurement can be useful. HRQOL measurement in the clinical setting can streamline and structure the clinical interview, potentially leading to enhanced assessment. It can also make it easier for busy pediatricians to explore and address issues of psychosocial functioning. A particularly promising area for HRQOL is in identifying, proactively, suitable candidates for case management in large enrolled populations. Further research should extend the initial studies on HRQOL predicting utilization and cost, more thoroughly specify the proportion of identified costs that are manageable and care management's effect on care for different groups of children, document the causal links between physiologic variables and HRQOL on one hand and patient functioning on the other, and understand the conditions necessary for HRQOL assessment to affect clinical practice.

Risk adjustment using administrative data-based and survey-derived methods for explaining physician utilization

OBJECTIVES

The objective of this study was to evaluate an administrative data-based risk adjustment method for predicting physician utilization and the contribution of survey-derived indicators of health status. The results of this study will support the use of administrative data for planning, reimbursement, and assessing equity of physician utilization.

METHODS

The Ontario portion of the 2000-2001 Canadian Community Health Survey was linked with administrative physician claims data from 2002-2003 and 2003-2004. Explanatory models of family physician (FP) and specialist physician (SP) utilization were run using demographic information and The Johns Hopkins University Adjusted Clinical Groups (ACG) Case-mix System. Survey-based measures of health status were then added to the models. The coefficient of determination, R, indicated the models' explanatory power.

RESULTS

The study sample consisted of 25,558 individuals aged 20 to 79 years representing approximately 7.8 million people. Over the 2 years of study period, 82.5% of the study population had a FP visit with a median of 6 visits and 53.2% had a SP visit with a median of 1 visit. The R values based on administrative data alone were 33% and 21% for the frequency of FP and SP visits and 16% and 35% for having one or more visit to an FPs and SPs, respectively. The addition of the survey-based measures to the administrative data-based models produced less than a 2% increase in explanatory power for any outcome.

CONCLUSION

Administrative data-based measures of morbidity burden are valid and useful indicators of future physician utilization. The survey-derived measures used in this study did not contribute significantly to models on the basis of administrative data-based measures. These findings support the future use of administrative data-based data and Adjusted Clinical Groups for planning, reimbursement, and research.

An in-depth assessment of a diagnosis-based risk adjustment model based on national health insurance claims: the application of the Johns Hopkins Adjusted Clinical Group case-mix system in Taiwan

BACKGROUND

Diagnosis-based risk adjustment is becoming an important issue globally as a result of its implications for payment, high-risk predictive modelling and provider performance assessment. The Taiwanese National Health Insurance (NHI) programme provides universal coverage and maintains a single national computerized claims database, which enables the application of diagnosis-based risk adjustment. However, research regarding risk adjustment is limited. This study aims to examine the performance of the Adjusted Clinical Group (ACG) case-mix system using claims-based diagnosis information from the Taiwanese NHI programme.

METHODS

A random sample of NHI enrollees was selected. Those continuously enrolled in 2002 were included for concurrent analyses (n = 173,234), while those in both 2002 and 2003 were included for prospective analyses (n = 164,562). Health status measures derived from 2002 diagnoses were used to explain the 2002 and 2003 health expenditure. A multivariate linear regression model was adopted after comparing the performance of seven different statistical models. Split-validation was performed in order to avoid overfitting. The performance measures were adjusted R2 and mean absolute prediction error of five types of expenditure at individual level, and predictive ratio of total expenditure at group level.

RESULTS

The more comprehensive models performed better when used for explaining resource utilization. Adjusted R2 of total expenditure in concurrent/prospective analyses were 4.2%/4.4% in the demographic model, 15%/10% in the ACGs or ADGs (Aggregated Diagnosis Group) model, and 40%/22% in the models containing EDCs (Expanded Diagnosis Cluster). When predicting expenditure for groups based on expenditure quintiles, all models underpredicted the highest expenditure group and overpredicted the four other groups. For groups based on morbidity burden, the ACGs model had the best performance overall.

CONCLUSIONS

Given the widespread availability of claims data and the superior explanatory power of claims-based risk adjustment models over demographics-only models, Taiwan's government should consider using claims-based models for policy-relevant applications. The performance of the ACG case-mix system in Taiwan was comparable to that found in other countries. This suggested that the ACG system could be applied to Taiwan's NHI even though it was originally developed in the USA. Many of the findings in this paper are likely to be relevant to other diagnosis-based risk adjustment methodologies.

The impact of trauma exposure and post-traumatic stress disorder on healthcare utilization among primary care patients

BACKGROUND

Trauma exposure and post-traumatic stress disorder (PTSD) increase healthcare utilization in veterans, but their impact on utilization in other populations is uncertain.

OBJECTIVES

To examine the association of trauma exposure and PTSD with healthcare utilization, in civilian primary care patients.

RESEARCH DESIGN

Cross-sectional study.

SUBJECTS

English speaking patients at an academic, urban primary care clinic.

MEASURES

Trauma exposure and current PTSD diagnoses were obtained from the Composite International Diagnostic Interview. Outcomes were nonmental health outpatient and emergency department visits, hospitalizations, and mental health outpatient visits in the prior year from an electronic medical record. Analyses included bivariate unadjusted and multivariable Poisson regressions adjusted for age, gender, income, substance dependence, depression, and comorbidities.

RESULTS

Among 592 subjects, 80% had > or =1 trauma exposure and 22% had current PTSD. In adjusted regressions, subjects with trauma exposure had more mental health visits [incidence rate ratio (IRR), 3.9; 95% confidence interval (CI), 1.1-14.1] but no other increased utilization. After adjusting for PTSD, this effect of trauma exposure was attenuated (IRR, 3.2; 95% CI, 0.9-11.7). Subjects with PTSD had more hospitalizations (IRR, 2.2; 95% CI, 1.4-3.7), more hospital nights (IRR, 2.6; 95% CI, 1.4-5.0), and more mental health visits (IRR, 2.2; 95% CI, 1.1-4.1) but no increase in outpatient and emergency department visits.

CONCLUSIONS

PTSD is associated with more hospitalizations, longer hospitalizations, and greater mental healthcare utilization in urban primary care patients. Although trauma exposure is independently associated with greater mental healthcare utilization, PTSD mediates a portion of this association.

Concurrent Vaccinations and U.S. Military Hospitalizations

PURPOSE

To investigate whether concurrent vaccinations (> or =2 vaccinations on consecutive days) are associated with hospitalization risk among U.S. military personnel.

METHODS

We analyzed Defense Medical Surveillance System data from January 1998 through December 2003 for 117,876 active component U.S. military personnel. We performed a time-to-event analysis of a historical cohort using a Cox proportional hazards model comparing hospitalizations during a 120-day postvaccination exposure interval to hospitalizations within a 120-day pre-exposure interval. We excluded personnel who were deployed during these intervals and those having hospitalizations 60 days prior to the concurrent vaccination exposure. Hazards ratios (HRs) with 95% confidence intervals were calculated, adjusting for demographic, occupational, health, and calendar variables.

RESULTS

We analyzed 19,743 persons having concurrent vaccinations. Receiving two or more vaccinations concurrently was not statistically associated with the adjusted risk of hospitalization (HR = 0.90 [0.75, 1.09]). Furthermore, no statistically significant associations were detected for 3 concurrent vaccinations (HR = 0.86 [0.58, 1.28]), 4 concurrent vaccinations (HR = 1.08, [0.66, 1.74]), or five or more concurrent vaccinations (HR = 0.86 [0.49, 1.51]).

CONCLUSIONS

No evidence was found that the concurrent receipt of multiple vaccinations is related to hospitalization risk among this sample of U.S. military personnel.

Clinical validity of the SWAL-QOL and SWAL-CARE outcome tools with respect to bolus flow measures

The aim of this study was to quantify the association between a dysphagia-specific quality of life (SWAL-QOL) and quality of care (SWAL-CARE) questionnaire and four measures of bolus flow. Three hundred eighty-six people with oropharyngeal dysphagia completed a videofluoroscopic examination of their swallowing structure and physiology. They also completed the SWAL-QOL and SWAL-CARE surveys. Measures of bolus flow patterns for each swallow were analyzed from videofluoroscopic recordings and correlated with the SWAL-QOL and SWAL-CARE scale scores. The SWAL-QOL and SWAL-CARE scales were modestly related to the four measures of the bolus flow. The SWAL-QOL and SWAL-CARE were most related to measures of oral transit duration and total swallow duration. The SWAL-QOL and SWAL-CARE scales were least related to pharyngeal transit duration. Results were stronger for semisolid trials than for liquid trials. Results were generally weak for the Penetration Aspiration Scale. For all of the significant relationships, the greater the bolus flow severity, the worse the quality of life. The observed modest correlations suggest that patient-centered quality-of-life measures and clinician-driven bolus flow measures provide distinct yet complementary information about oropharyngeal dysphagia. Both sets of measures should be used in dysphagia effectiveness and outcomes research.

Alternative models for academic family practices

Background

The Future of Family Medicine Report calls for a fundamental redesign of the American family physician workplace. At the same time, academic family practices are under economic pressure. Most family medicine departments do not have self-supporting practices, but seek support from specialty colleagues or hospital practice plans. Alternative models for academic family practices that are economically viable and consistent with the principles of family medicine are needed. This article presents several "experiments" to address these challenges.

Methods

The basis of comparison is a traditional academic family medicine center. Apart of the faculty practice plan, our center consistently operated at a deficit despite high productivity. A number of different practice types and alternative models of service delivery were therefore developed and tested. They ranged from a multi-specialty office arrangement, to a community clinic operated as part of a federally-qualified health center, to a team of providers based in and providing care for residents of an elderly public housing project. Financial comparisons using consistent accounting across models are provided.

Results

Academic family practices can, at least in some settings, operate without subsidy while providing continuity of care to a broad segment of the community. The prerequisites are that the clinicians must see patients efficiently, and be able to bill appropriately for their payer mix.

Conclusion

Experimenting within academic practice structure and organization is worthwhile, and can result in economically viable alternatives to traditional models.

Predicting persistently high primary care use

PURPOSE

We wanted to identify risk factors for persistently high use of primary care.

METHODS

We analyzed outpatient office visits to practitioners in family medicine, general internal medicine, general pediatrics, and obstetrics for 1997-1999 among patients in a small Midwestern city covered by a fee-for-service insurance plan with no co-payments for physician visits and no requirement for referral to specialty care. Logistic regression was used to predict which patients with 10 or more primary care visits in 1997 would repeat high use in 1998 based on demographic and diagnostic categories (adjusted clinical groups [ACGs]). A confirmatory data set (high primary care use in 1998 persistent into 1999) was used to evaluate the model.

RESULTS

Two percent of the 54,074 patients had 10 or more primary care visits in 1997, and of these, almost 19% had 10 or more visits in the next year. Among adults, 4 ambulatory diagnosis groups (ADGs) were simultaneously positive predictors of repeated high primary care visits: unstable chronic medical conditions, see and reassure conditions, minor time-limited psychosocial conditions, and minor signs and symptoms. Meanwhile, pregnancy was negatively associated. The area under the receiver operating characteristic (ROC) curve was 0.794 for adults in the developmental data set and 0.752 in the confirmatory data set, indicating a moderately accurate assessment. A satisfactory model was not developed for pediatric patients.

CONCLUSIONS

Many persistently high primary care users appear to be overserviced but underserved, with underlying problems not addressed by a medical approach. Some may benefit from psychosocial support, whereas others may be good candidates for disease management interventions.

Risk Classification of Adult Primary Care Patients by Self-Reported Quality of Life

BACKGROUND

: Although patient-reported health-related quality of life (HRQOL) is known to predict health services utilization, most risk assessment systems use provider-reported diagnoses as predictors rather than HRQOL.

OBJECTIVE

: We sought to classify adult primary care patients prospectively by utilization risk based on age, gender, and HRQOL at a single clinic visit.

RESEARCH DESIGN

: Patients completed the Duke Health Profile. Providers completed the Duke Severity of Illness Checklist. Diagnoses were grouped with the Ambulatory Care Groups system. Predictive coefficients for 1-year primary care charges calculated from the age, gender, and HRQOL of 728 reference patients were used to classify 474 test patients into 4 risk classes. Comparisons were made with models that used diagnoses or severity of illness as predictors.

RESULTS

: The positive likelihood ratio for predicting highest risk was 2.2 for the HRQOL model, compared with 1.8 for the diagnoses model, 1.6 for the severity model, and 1.5 for age and gender alone. One-year actual primary care visits and charges increased step-wise from lowest to highest risk class. Highest risk patients were older and more likely to be women, black, or Medicaid recipients. Although the highest-risk patients represented only 18.6% of the test group, they accounted for 26.7% of the primary care clinic visits, 31.6% of the clinic charges, 34.6% of the hospital days, 35.1% of hospital charges, and 30.8% of total charges at all healthcare sites.

CONCLUSION

: The HRQOL risk classification system can identify primary care patients at risk for high future health services utilization.

Health-related quality of life as a predictor of pediatric healthcare costs: a two-year prospective cohort analysis

Background

The objective of this study was to test the primary hypothesis that parent proxy-report of pediatric health-related quality of life (HRQL) would prospectively predict pediatric healthcare costs over a two-year period. The exploratory hypothesis tested anticipated that a relatively small group of children would account for a disproportionately large percent of healthcare costs.

Methods

317 children (157 girls) ages 2 to 18 years, members of a managed care health plan with prospective payment participated in a two-year prospective longitudinal study. At Time 1, parents reported child HRQL using the Pediatric Quality of Life Inventory™ (PedsQL™ 4.0) Generic Core Scales, and chronic health condition status. Costs, based on health plan utilization claims and encounters, were derived for 6, 12, and 24 months.

Results

In multiple linear regression equations, Time 1 parent proxy-reported HRQL prospectively accounted for significant variance in healthcare costs at 6, 12, and 24 months. Adjusted regression models that included both HRQL scores and chronic health condition status accounted for 10.1%, 14.4%, and 21.2% of the variance in healthcare costs at 6, 12, and 24 months. Parent proxy-reported HRQL and chronic health condition status together defined a 'high risk' group, constituting 8.7% of the sample and accounting for 37.4%, 59.2%, and 62% of healthcare costs at 6, 12, and 24 months. The high risk group's per member per month healthcare costs were, on average, 12 times that of other enrollees' at 24 months.

Conclusions

While these findings should be further tested in a larger sample, our data suggest that parent proxy-reported HRQL can be used to prospectively predict healthcare costs. When combined with chronic health condition status, parent proxy-reported HRQL can identify an at risk group of children as candidates for proactive care coordination.

The best of both worlds? Potential of hybrid prospective/concurrent risk adjustment

BACKGROUND

There remains considerable uncertainty about whether prospective or concurrent risk adjustment (RA) is preferable. Although concurrent models have better predictive power than prospective models, the large payments associated with concurrent RA create incentives for fraudulent coding. A hybrid strategy--in which prospective payments were used for patients with low expected costs and concurrent payments were available upon the diagnosis of a small number of common, expensive conditions--might improve predictive performance while requiring less auditing than fully concurrent RA. In addition, within-condition RA (using clinical data) for the selected conditions could further improve predictive power.

OBJECTIVES

To assess how such a hybrid strategy might perform, focusing on a small number of chronic, expensive conditions that are verifiable (hence auditable).

SUBJECTS AND MEASURES

All patients from seven health plans who had two complete years of utilization data were considered. RA models were estimated among patients younger than 65 (n = 319,209) using the Hierarchical Coexisting Conditions (HCC) model with and without stratification of the sample based on the presence of one or more of 100 verifiable, expensive, predictive conditions (VEP100). R2 and predictive ratios were calculated for each model studied.

RESULTS

Patients with a VEP100 condition (9.3% of the population) accounted for 84.3% of the variation in cost. R2 was 0.08 using a prospective HCC model on the entire population, but increased to 0.26 for a hybrid using prospective HCCs on the 90.7% of the sample without a VEP100 condition and a simple concurrent model consisting of dummy variables for each of the VEP100 conditions.

CONCLUSION

Combined with targeted auditing, a hybrid approach to RA could improve our ability to match payments to costs. However, because this would require additional, costly data collection, more research is needed to determine whether this benefit justifies the data collection and auditing burden.