Effectiveness of Antidepressants in Combination with Psychotherapy

BACKGROUND

Consensus-guidelines for prescribing antidepressants recommend that clinicians should be vigilant to match antidepressants to patient's medical history but provide no specific advice on which antidepressant is best for a given medical history.

AIMS OF THE STUDY

For patients with major depression who are in psychotherapy, this study provides an empirically derived guideline for prescribing antidepressant medications that fit patients' medical history.

METHODS

This retrospective, observational, cohort study analyzed a large insurance database of 3,678,082 patients. Data was obtained from healthcare providers in the U.S. between January 1, 2001, and December 31, 2018. These patients had 10,221,145 episodes of antidepressant treatments. This study reports the remission rates for the 14 most commonly prescribed single antidepressants (amitriptyline, bupropion, citalopram, desvenlafaxine, doxepin, duloxetine, escitalopram, fluoxetine, mirtazapine, nortriptyline, paroxetine, sertraline, trazodone, and venlafaxine) and a category named "Other" (other antidepressants/combination of antidepressants). The study used robust LASSO regressions to identify factors that affected remission rate and clinicians' selection of antidepressants. The selection bias in observational data was removed through stratification. We organized the data into 16,770 subgroups, of at least 100 cases, using the combination of the largest factors that affected remission and selection bias. This paper reports on 2,467 subgroups of patients who had received psychotherapy.

RESULTS

We found large, and statistically significant, differences in remission rates within subgroups of patients. Remission rates for sertraline ranged from 4.5% to 77.86%, for fluoxetine from 2.86% to 77.78%, for venlafaxine from 5.07% to 76.44%, for bupropion from 0.5% to 64.63%, for desvenlafaxine from 1.59% to 75%, for duloxetine from 3.77% to 75%, for paroxetine from 6.48% to 68.79%, for escitalopram from 1.85% to 65%, and for citalopram from 4.67% to 76.23%. Clearly these medications are ideal for patients in some subgroups but not others. If patients are matched to the subgroups, clinicians can prescribe the medication that works best in the subgroup. Some medications (amitriptyline, doxepin, nortriptyline, and trazodone) always had remission rates below 11% and therefore were not suitable as single antidepressant therapy for any of the subgroups.

DISCUSSIONS

This study provides an opportunity for clinicians to identify an optimal antidepressant for their patients, before they engage in repeated trials of antidepressants.

IMPLICATIONS FOR HEALTH CARE PROVISION AND USE

To facilitate the matching of patients to the most effective antidepressants, this study provides access to a free, non-commercial, decision aid at http://MeAgainMeds.com.

IMPLICATIONS FOR HEALTH POLICIES

 Policymakers should evaluate how study findings can be made available through fragmented electronic health records at point-of-care. Alternatively, policymakers can put in place an AI system that recommends antidepressants to patients online, at home, and encourages them to bring the recommendation to their clinicians at their next visit.

IMPLICATIONS FOR FURTHER RESEARCH

 Future research could investigate (i) the effectiveness of our recommendations in changing clinical practice, (ii) increasing remission of depression symptoms, and (iii) reducing cost of care. These studies need to be prospective but pragmatic. It is unlikely random clinical trials can address the large number of factors that affect remission.

The Gender Self-Report: A multidimensional gender characterization tool for gender-diverse and cisgender youth and adults

Gender identity is a core component of human experience, critical to account for in broad health, development, psychosocial research, and clinical practice. Yet, the psychometric characterization of gender has been impeded due to challenges in modeling the myriad gender self-descriptors, statistical power limitations related to multigroup analyses, and equity-related concerns regarding the accessibility of complex gender terminology. Therefore, this initiative employed an iterative multi-community-driven process to develop the Gender Self-Report (GSR), a multidimensional gender characterization tool, accessible to youth and adults, nonautistic and autistic people, and gender-diverse and cisgender individuals. In Study 1, the GSR was administered to 1,654 individuals, sampled through seven diversified recruitments to be representative across age (10-77 years), gender and sexuality diversity (∼33% each gender diverse, cisgender sexual minority, cisgender heterosexual), and autism status (> 33% autistic). A random half-split subsample was subjected to exploratory factor analytics, followed by confirmatory analytics in the full sample. Two stable factors emerged: Nonbinary Gender Diversity and Female-Male Continuum (FMC). FMC was transformed to Binary Gender Diversity based on designated sex at birth to reduce collinearity with designated sex at birth. Differential item functioning by age and autism status was employed to reduce item-response bias. Factors were internally reliable. Study 2 demonstrated the construct, convergent, and ecological validity of GSR factors. Of the 30 hypothesized validation comparisons, 26 were confirmed. The GSR provides a community-developed gender advocacy tool with 30 self-report items that avoid complex gender-related "insider" language and characterize diverse populations across continuous multidimensional binary and nonbinary gender traits. (PsycInfo Database Record (c) 2023 APA, all rights reserved).

Social and Medical Determinants of Diabetes: A Time-Constrained Multiple Mediator Analysis

Background A number of studies have shown an association between social determinants of health and the emergence of obesity and diabetes, but whether the relationship is causal is not clear. Objective To test whether social, environmental, and medical determinants directly or indirectly affect population-level diabetes prevalence after controlling for mediator-mediator interactions. Methods Data were obtained from the CDC and supplemented with nine other data sources for 3,109 US counties. The dependent variable was the prevalence of diabetes in 2017. Independent variables were a given county's 30 social, environmental, and medical characteristics in 2015 and 2016. A network multiple mediation analysis was conducted. First, we used Least Absolute Shrinkage and Selection Operator (LASSO) regression to relate the 2017 diabetes rate in each county to 30 predictors measured in 2016, identifying statistically significant and robust predictors as the mediators within the network model and as direct determinants of 2017 diabetes. Second, each of the direct causes of diabetes was taken as a new response variable and LASSO-regressed on the same 30 independent variables measured in 2015, identifying the indirect (mediated) causes of diabetes. Subsequently, these direct and indirect predictors were used to construct a network model. The completed network was then employed to estimate the direct and mediated impact of variables on diabetes. Results For 2017 diabetes rates, 63% of the variation was explained by five variables measured in 2016: the percentage of residents who were (1) obese, (2) African American, (3) physically inactive, (4) in poor health condition, and (5) had a history of diabetes. These five direct predictors, measured in 2016, mediated the effect of indirect variables measured in 2015, including the percentage of residents who were (1) Hispanic, (2) physically distressed, (3) smokers, (4) living with children in poverty, (5) experiencing limited access to healthy foods, and (6) had low income. Conclusion All of the direct predictors of diabetes prevalence, except the percentage of residents who were African American, were medical conditions potentially influenced by lifestyles. Counties characterized by higher levels of obesity, inactivity, and poor health conditions exhibited increased diabetes rates in the following year. The impact of social determinants of illness, such as low income, children in poverty, and limited access to healthy foods, had an indirect effect on the health of residents and, consequently, increased the prevalence of diabetes.

Impact of Political Leaning on COVID-19 Vaccine Hesitancy: A Network-Based Multiple Mediation Analysis

Prior studies have shown that political affiliation affected COVID-19 vaccine hesitancy. This study re-examined the data to see if these findings hold after controlling for alternative explanations. The dependent variable in the study was COVID-19 vaccination rates in 3,109 counties in the United States as of April 2022. The study examined 36 possible alternative explanations for vaccine hesitancy, including demographic, social, economic, environmental, and medical variables known to affect vaccine hesitancy. County-level political affiliation was measured as a percent of voters in the county who were affiliated with Democratic or Republican political parties. Data were analyzed using a temporally constrained multiple mediation network, which allowed for the identification of both direct and indirect predictors of vaccination rates. Despite controlling for alternative explanations of hesitancy, there was a statistically significant relationship between the percentage of Republican supporters and rates of vaccine hesitancy. The higher the Republican affiliation, the lower the vaccination rates. It is possible that the Republican Party has played an organizing role in encouraging vaccine hesitancy and patient harm.

Selecting Antidepressants Based on Medical History and Stress Mechanism

Purpose At present, clinicians typically prescribe antidepressants based on the widely accepted "serotonin hypothesis." This study explores an alternative mechanism, the stress mechanism, for selecting antidepressants based on patients' medical history. Methods This study investigated clinicians' prescribing patterns for the 15 most common antidepressants, including amitriptyline, bupropion, citalopram, desvenlafaxine, doxepin, duloxetine, escitalopram, fluoxetine, mirtazapine, nortriptyline, paroxetine, ropinirole, sertraline, trazodone, and Venlafaxine. The least absolute shrinkage and selection operator (LASSO) logistic regression was used to identify factors that affect the remission of depression symptoms after receiving an antidepressant. Results The study found that a wide range of factors influenced the propensity of clinicians to prescribe antidepressants, with the number of predictors ranging from 51 to 206 variables. The prevalence of prescribing an antidepressant ranged from 0.5% for doxepin to 24% for the combination of more than one antidepressant. The area under the receiver operating curves (AROC) ranged from 77.2% for venlafaxine to 90.5% for ropinirole, with an average AROC of 82% for predicting the propensity of medications. A variety of diagnoses and prior medications affected remission, in agreement that the central mechanism for the impact of medications on the brain is through stress reduction. For example, psychotherapy, whether done individually or in a group, whether done for a short or long time, and whether done with evaluation/assessment or not, had an impact on remission. Specifically, teenagers and octogenarians were less likely to benefit from bupropion, citalopram, escitalopram, fluoxetine, and sertraline compared to patients between 40 and 65 years old. The findings of this study suggest that considering a patient's medical history and individual characteristics is crucial for selecting the most effective antidepressant treatment. Conclusions Many studies have raised doubt about the serotonin hypothesis as the central mechanism for depression treatment. The identification of a wide range of predictors for prescribing antidepressants highlights the complexity of depression treatment and the need for individualized approaches that consider patients' comorbidities and previous treatments. The significant impact of comorbidities on the response to treatment makes it improbable that the mechanism of action of antidepressants is solely based on the serotonin hypothesis. It is hard to explain how comorbidities lead to the depletion of serotonin. These findings open up a variety of courses of action for the clinical treatment of depression, each addressing a different source of chronic stress in the brain. Overall, this study contributes to a better understanding of depression treatment and provides valuable insights for clinicians in selecting antidepressants based on patients' medical history.

Predicting 6-month mortality of patients from their medical history: Comparison of multimorbidity index to Deyo-Charlson index

While every disease could affect a patient's prognosis, published studies continue to use indices that include a selective list of diseases to predict prognosis, which may limit its accuracy. This paper compares 6-month mortality predicted by a multimorbidity index (MMI) that relies on all diagnoses to the Deyo version of the Charlson index (DCI), a popular index that utilizes a selective set of diagnoses. In this retrospective cohort study, we used data from the Veterans Administration Diabetes Risk national cohort that included 6,082,018 diabetes-free veterans receiving primary care from January 1, 2008 to December 31, 2016. For the MMI, 7805 diagnoses were assigned into 19 body systems, using the likelihood that the disease will increase risk of mortality. The DCI used 17 categories of diseases, classified by clinicians as severe diseases. In predicting 6-month mortality, the cross-validated area under the receiver operating curve for the MMI was 0.828 (95% confidence interval of 0.826-0.829) and for the DCI was 0.749 (95% confidence interval of 0.748-0.750). Using all available diagnoses (MMI) led to a large improvement in accuracy of predicting prognosis of patients than using a selected list of diagnosis (DCI).

Social Determinants of COVID-19 Vaccination Rates: A Time-Constrained Multiple Mediation Analysis

Objective To estimate the multiple direct/indirect effects of social, environmental, and economic factors on COVID-19 vaccination rates (series complete) in the 3109 continental counties in the United States (U.S.). Study design  The dependent variable was the COVID-19 vaccination rates in the U.S. (April 15, 2022). Independent variables were collected from reliable secondary data sources, including the Census and CDC. Independent variables measured at two different time frames were utilized to predict vaccination rates. The number of vaccination sites in a given county was calculated using the geographic information system (GIS) packages as of April 9, 2022. The Internet Archive (Way Back Machine) was used to look up data for historical dates. Methods  A chain of temporally-constrained least absolute shrinkage and selection operator (LASSO) regressions was used to identify direct and indirect effects on vaccination rates. The first regression identified direct predictors of vaccination rates. Next, the direct predictors were set as response variables in subsequent regressions and regressed on variables that occurred before them. These regressions identified additional indirect predictors of vaccination. Finally, both direct and indirect variables were included in a network model. Results  Fifteen variables directly predicted vaccination rates and explained 43% of the variation in vaccination rates in April 2022. In addition, 11 variables indirectly affected vaccination rates, and their influence on vaccination was mediated by direct factors. For example, children in poverty rate mediated the effect of (a) median household income, (b) children in single-parent homes, and (c) income inequality. For another example, median household income mediated the effect of (a) the percentage of residents under the age of 18, (b) the percentage of residents who are Asian, (c) home ownership, and (d) traffic volume in the prior year. Our findings describe not only the direct but also the indirect effect of variables. Conclusions  A diverse set of demographics, social determinants, public health status, and provider characteristics predicted vaccination rates. Vaccination rates change systematically and are affected by the demographic composition and social determinants of illness within the county. One of the merits of our study is that it shows how the direct predictors of vaccination rates could be mediators of the effects of other variables.

Guidelines for Triage of COVID-19 Patients Presenting With Multisystemic Symptoms

BACKGROUND AND OBJECTIVES

This article describes how multisystemic symptoms, both respiratory and nonrespiratory, can be used to differentiate coronavirus disease-2019 (COVID-19) from other diseases at the point of patient triage in the community. The article also shows how combinations of symptoms could be used to predict the probability of a patient having COVID-19.

METHODS

We first used a scoping literature review to identify symptoms of COVID-19 reported during the first year of the global pandemic. We then surveyed individuals with reported symptoms and recent reverse transcription polymerase chain reaction (RT-PCR) test results to assess the accuracy of diagnosing COVID-19 from reported symptoms. The scoping literature review, which included 81 scientific articles published by February 2021, identified 7 respiratory, 9 neurological, 4 gastrointestinal, 4 inflammatory, and 5 general symptoms associated with COVID-19 diagnosis. The likelihood ratio associated with each symptom was estimated from sensitivity and specificity of symptoms reported in the literature. A total of 483 individuals were then surveyed to validate the accuracy of predicting COVID-19 diagnosis based on patient symptoms using the likelihood ratios calculated from the literature review. Survey results were weighted to reflect age, gender, and race of the US population. The accuracy of predicting COVID-19 diagnosis from patient-reported symptoms was assessed using area under the receiver operating curve (AROC).

RESULTS

In the community, cough, sore throat, runny nose, dyspnea, and hypoxia, by themselves, were not good predictors of COVID-19 diagnosis. A combination of cough and fever was also a poor predictor of COVID-19 diagnosis (AROC = 0.56). The accuracy of diagnosing COVID-19 based on symptoms was highest when individuals presented with symptoms from different body systems (AROC of 0.74-0.81); the lowest accuracy was when individuals presented with only respiratory symptoms (AROC = 0.48).

CONCLUSIONS

There are no simple rules that clinicians can use to diagnose COVID-19 in the community when diagnostic tests are unavailable or untimely. However, triage of patients to appropriate care and treatment can be improved by reviewing the combinations of certain types of symptoms across body systems.

Order of Occurrence of COVID-19 Symptoms

BACKGROUND AND OBJECTIVES

COVID-19 symptoms change after onset-some show early, others later. This article examines whether the order of occurrence of symptoms can improve diagnosis of COVID-19 before test results are available.

METHODS

In total, 483 individuals who completed a COVID-19 test were recruited through Listservs. Participants then completed an online survey regarding their symptoms and test results. The order of symptoms was set according to (a) whether the participant had a "history of the symptom" due to a prior condition; and (b) whether the symptom "occurred first," or prior to, other symptoms of COVID-19. Two LASSO (Least Absolute Shrinkage and Selection Operator) regression models were developed. The first model, referred to as "time-invariant," used demographics and symptoms but not the order of symptom occurrence. The second model, referred to as "time-sensitive," used the same data set but included the order of symptom occurrence.

RESULTS

The average cross-validated area under the receiver operating characteristic (AROC) curve for the time-invariant model was 0.784. The time-sensitive model had an AROC curve of 0.799. The difference between the 2 accuracy levels was statistically significant (α < .05).

CONCLUSION

The order of symptom occurrence made a statistically significant, but small, improvement in the accuracy of the diagnosis of COVID-19.

The Role of Symptom Clusters in Triage of COVID-19 Patients

BACKGROUND AND OBJECTIVE

COVID-19 manifests with a broad range of symptoms. This study investigates whether clusters of respiratory, gastrointestinal, or neurological symptoms can be used to diagnose COVID-19.

METHODS

We surveyed symptoms of 483 subjects who had completed COVID-19 laboratory tests in the last 30 days. The survey collected data on demographic characteristics, self-reported symptoms for different types of infections within 14 days of onset of illness, and self-reported COVID-19 test results. Robust LASSO regression was used to create 3 nested models. In all 3 models, the response variable was the COVID-19 test result. In the first model, referred to as the "main effect model," the independent variables were demographic characteristics, history of chronic symptoms, and current symptoms. The second model, referred to as the "hierarchical clustering model," added clusters of variables to the list of independent variables. These clusters were established through hierarchical clustering. The third model, referred to as the "interaction-terms model," also added clusters of variables to the list of independent variables; this time clusters were established through pairwise and triple-way interaction terms. Models were constructed on a randomly selected 80% of the data and accuracy was cross-validated on the remaining 20% of the data. The process was bootstrapped 30 times. Accuracy of the 3 models was measured using the average of the cross-validated area under the receiver operating characteristic curves (AUROCs).

RESULTS

In 30 bootstrap samples, the main effect model had an AUROC of 0.78. The hierarchical clustering model had an AUROC of 0.80. The interaction-terms model had an AUROC of 0.81. Both the hierarchical cluster model and the interaction model were significantly different from the main effect model (α = .04). Patients with different races/ethnicities, genders, and ages presented with different symptom clusters.

CONCLUSIONS

Using clusters of symptoms, it is possible to more accurately diagnose COVID-19 among symptomatic patients.

Combined Symptom Screening and At-Home Tests for COVID-19

BACKGROUND AND OBJECTIVE

At-home rapid antigen tests provide a convenient and expedited resource to learn about severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) infection status. However, low sensitivity of at-home antigen tests presents a challenge. This study examines the accuracy of at-home tests, when combined with computer-facilitated symptom screening.

METHODS

The study used primary data sources with data collected during 2 phases at different periods (phase 1 and phase 2): one during the period in which the alpha variant of SARS-CoV-2 was predominant in the United States and another during the surge of the delta variant. Four hundred sixty-one study participants were included in the analyses from phase 1 and 374 subjects from phase 2. Phase 1 data were used to develop a computerized symptom screening tool, using ordinary logistic regression with interaction terms, which predicted coronavirus disease-2019 (COVID-19) reverse transcription polymerase chain reaction (RT-PCR) test results. Phase 2 data were used to validate the accuracy of predicting COVID-19 diagnosis with (1) computerized symptom screening; (2) at-home rapid antigen testing; (3) the combination of both screening methods; and (4) the combination of symptom screening and vaccination status. The McFadden pseudo-R2 was used as a measure of percentage of variation in RT-PCR test results explained by the various screening methods.

RESULTS

The McFadden pseudo-R2 for the first at-home test, the second at-home test, and computerized symptom screening was 0.274, 0.140, and 0.158, respectively. Scores between 0.2 and 0.4 indicated moderate levels of accuracy. The first at-home test had low sensitivity (0.587) and high specificity (0.989). Adding a second at-home test did not improve the sensitivity of the first test. Computerized symptom screening improved the accuracy of the first at-home test (added 0.131 points to sensitivity and 6.9% to pseudo-R2 of the first at-home test). Computerized symptom screening and vaccination status was the most accurate method to screen patients for COVID-19 or an active infection with SARS-CoV-2 in the community (pseudo-R2 = 0.476).

CONCLUSION

Computerized symptom screening could either improve, or in some situations, replace at-home antigen tests for those individuals experiencing COVID-19 symptoms.

Procedure for Organizing a Post-FDA-approval Evaluation of Antidepressants

Purpose: The study reports the construction of a cohort used to study the effectiveness of antidepressants.

Methods: The cohort includes experiences of 3,678,082 patients with depression in the United States on antidepressants between January 1, 2001, and December 31, 2018. A total of 10,221,145 antidepressant treatment episodes were analyzed. Patients who had no utilization of health services for at least two years, or who had died, were excluded from the analysis. Follow-up was passive, automatic, and collated from fragmented clinical services of diverse providers.

Results: The average follow-up was 2.93 years, resulting in 15,096,055 person-years of data. The mean age of the cohort was 46.54 years (standard deviation of 17.48) at first prescription of antidepressant, which was also the enrollment event (16.92% were over 65 years), and most were female (69.36%). In 10,221,145 episodes, within the first 100 days of start of the episode, 4,729,372 (46.3%) continued their treatment, 1,306,338 (12.8%) switched to another medication, 3,586,156 (35.1%) discontinued their medication, and 599,279 (5.9%) augmented their treatment.

Conclusions: We present a procedure for constructing a cohort using claims data. A surrogate measure for self-reported symptom remission based on the patterns of use of antidepressants has been proposed to address the absence of outcomes in claims. Future studies can use the procedures described here to organize studies of the comparative effectiveness of antidepressants.

Emerging approaches to multiple chronic condition assessment

Older adults experience a higher prevalence of multiple chronic conditions (MCCs). Establishing the presence and pattern of MCCs in individuals or populations is important for healthcare delivery, research, and policy. This report describes four emerging approaches and discusses their potential applications for enhancing assessment, treatment, and policy for the aging population. The National Institutes of Health convened a 2-day panel workshop of experts in 2018. Four emerging models were identified by the panel, including classification and regression tree (CART), qualifying comorbidity sets (QCS), the multimorbidity index (MMI), and the application of omics to network medicine. Future research into models of multiple chronic condition assessment may improve understanding of the epidemiology, diagnosis, and treatment of older persons.

Differential diagnosis of COVID-19 and influenza

This study uses two existing data sources to examine how patients' symptoms can be used to differentiate COVID-19 from other respiratory diseases. One dataset consisted of 839,288 laboratory-confirmed, symptomatic, COVID-19 positive cases reported to the Centers for Disease Control and Prevention (CDC) from March 1, 2019, to September 30, 2020. The second dataset provided the controls and included 1,814 laboratory-confirmed influenza positive, symptomatic cases, and 812 cases with symptomatic influenza-like-illnesses. The controls were reported to the Influenza Research Database of the National Institute of Allergy and Infectious Diseases (NIAID) between January 1, 2000, and December 30, 2018. Data were analyzed using case-control study design. The comparisons were done using 45 scenarios, with each scenario making different assumptions regarding prevalence of COVID-19 (2%, 4%, and 6%), influenza (0.01%, 3%, 6%, 9%, 12%) and influenza-like-illnesses (1%, 3.5% and 7%). For each scenario, a logistic regression model was used to predict COVID-19 from 2 demographic variables (age, gender) and 10 symptoms (cough, fever, chills, diarrhea, nausea and vomiting, shortness of breath, runny nose, sore throat, myalgia, and headache). The 5-fold cross-validated Area under the Receiver Operating Curves (AROC) was used to report the accuracy of these regression models. The value of various symptoms in differentiating COVID-19 from influenza depended on a variety of factors, including (1) prevalence of pathogens that cause COVID-19, influenza, and influenza-like-illness; (2) age of the patient, and (3) presence of other symptoms. The model that relied on 5-way combination of symptoms and demographic variables, age and gender, had a cross-validated AROC of 90%, suggesting that it could accurately differentiate influenza from COVID-19. This model, however, is too complex to be used in clinical practice without relying on computer-based decision aid. Study results encourage development of web-based, stand-alone, artificial Intelligence model that can interview patients and help clinicians make quarantine and triage decisions.

Modeling the Probability of COVID-19 Based on Symptom Screening and Prevalence of Influenza and Influenza-Like Illnesses

BACKGROUND

The importance of various patient-reported signs and symptoms to the diagnosis of coronavirus disease 2019 (COVID-19) changes during, and outside, of the flu season. None of the current published studies, which focus on diagnosis of COVID-19, have taken this seasonality into account.

OBJECTIVE

To develop predictive algorithm, which estimates the probability of having COVID-19 based on symptoms, and which incorporates the seasonality and prevalence of influenza and influenza-like illness data.

METHODS

Differential diagnosis of COVID-19 and influenza relies on demographic characteristics (age, race, and gender), and respiratory (eg, fever, cough, and runny nose), gastrointestinal (eg, diarrhea, nausea, and loss of appetite), and neurological (eg, anosmia and headache) signs and symptoms. The analysis was based on the symptoms reported by COVID-19 patients, 774 patients in China and 273 patients in the United States. The analysis also included 2885 influenza and 884 influenza-like illnesses in US patients. Accuracy of the predictions was calculated using the average area under the receiver operating characteristic (AROC) curves.

RESULTS

The likelihood ratio for symptoms, such as cough, depended on the flu season-sometimes indicating COVID-19 and other times indicating the reverse. In 30-fold cross-validated data, the symptoms accurately predicted COVID-19 (AROC of 0.79), showing that symptoms can be used to screen patients in the community and prior to testing.

CONCLUSION

Community-based health care providers should follow different signs and symptoms for diagnosing COVID-19 during, and outside of, influenza season.

Neighborhood Socioeconomic Environment and Risk of Type 2 Diabetes: Associations and Mediation Through Food Environment Pathways in Three Independent Study Samples

OBJECTIVE

We examined whether relative availability of fast-food restaurants and supermarkets mediates the association between worse neighborhood socioeconomic conditions and risk of developing type 2 diabetes (T2D).

RESEARCH DESIGN AND METHODS

As part of the Diabetes Location, Environmental Attributes, and Disparities Network, three academic institutions used harmonized environmental data sources and analytic methods in three distinct study samples: 1) the Veterans Administration Diabetes Risk (VADR) cohort, a national administrative cohort of 4.1 million diabetes-free veterans developed using electronic health records (EHRs); 2) Reasons for Geographic and Racial Differences in Stroke (REGARDS), a longitudinal, epidemiologic cohort with Stroke Belt region oversampling (N = 11,208); and 3) Geisinger/Johns Hopkins University (G/JHU), an EHR-based, nested case-control study of 15,888 patients with new-onset T2D and of matched control participants in Pennsylvania. A census tract-level measure of neighborhood socioeconomic environment (NSEE) was developed as a community type-specific z-score sum. Baseline food-environment mediators included percentages of 1) fast-food restaurants and 2) food retail establishments that are supermarkets. Natural direct and indirect mediating effects were modeled; results were stratified across four community types: higher-density urban, lower-density urban, suburban/small town, and rural.

RESULTS

Across studies, worse NSEE was associated with higher T2D risk. In VADR, relative availability of fast-food restaurants and supermarkets was positively and negatively associated with T2D, respectively, whereas associations in REGARDS and G/JHU geographies were mixed. Mediation results suggested that little to none of the NSEE-diabetes associations were mediated through food-environment pathways.

CONCLUSIONS

Worse neighborhood socioeconomic conditions were associated with higher T2D risk, yet associations are likely not mediated through food-environment pathways.

Effectiveness of common antidepressants: a post market release study

BACKGROUND

This study summarizes the experiences of patients, who have multiple comorbidities, with 15 mono-treated antidepressants.

METHODS

This is a retrospective, observational, matched case control study. The cohort was organized using claims data available through OptumLabs for depressed patients treated with antidepressants between January 1, 2001 and December 31, 2018. The cohort included patients from all states within United States of America. The analysis focused on 3,678,082 patients with major depression who had 10,221,145 antidepressant treatments. Using the robust, and large predictors of remission, and propensity to prescribe an antidepressant, the study created 16,770 subgroups of patients. The study reports the remission rate for the antidepressants within the subgroups. The overall impact of antidepressant on remission was calculated as the common odds ratio across the strata.

FINDINGS

The study accurately modelled clinicians' prescription patterns (cross-validated Area under the Receiver Operating Curve, AROC, of 82.0%, varied from 77% to 90%) and patients' remission (cross-validated AROC of 72.0%, varied from 69.5% to 78%). In different strata, contrary to published randomized studies, remission rates differed significantly and antidepressants were not equally effective. For example, in age and gender subgroups, the best antidepressant had an average remission rate of 50.78%, 1.5 times higher than the average antidepressant (30.30% remission rate) and 20 times higher than the worst antidepressant. The Breslow-Day chi-square test for homogeneity showed that across strata a homogenous common odds-ratio did not exist (alpha<0.0001). Therefore, the choice of the optimal antidepressant depended on the strata defined by the patient's medical history.

INTERPRETATION

Study findings may not be appropriate for specific patients. To help clinicians assess the transferability of study findings to specific patient, the web site http://hi.gmu.edu/ad assesses the patient's medical history, finds similar cases in our data, and recommends an antidepressant based on the experience of remission in our data. Patients can share this site's recommendations with their clinicians, who can then assess the appropriateness of the recommendations.

FUNDING

This project was funded by the Robert Wood Johnson foundation grant #76786. The development of related web site was supported by grant 247-02-20 from Virginia's Commonwealth Health Research Board.

Multi-omic molecular profiling guide's efficacious treatment selection in refractory metastatic breast cancer: a prospective phase II clinical trial

This prospective phase II clinical trial (Side Out 2) explored the clinical benefits of treatment selection informed by multi‐omic molecular profiling (MoMP) in refractory metastatic breast cancers (MBCs). Core needle biopsies were collected from 32 patients with MBC at trial enrollment. Patients had received an average of 3.94 previous lines of treatment in the metastatic setting before enrollment in this study. Samples underwent MoMP, including exome sequencing, RNA sequencing (RNA‐Seq), immunohistochemistry, and quantitative protein pathway activation mapping by Reverse Phase Protein Microarray (RPPA). Clinical benefit was assessed using the previously published growth modulation index (GMI) under the hypothesis that MoMP‐selected therapy would warrant further investigation for GMI ≥ 1.3 in ≥ 35% of the patients. Of the 32 patients enrolled, 29 received treatment based on their MoMP and 25 met the follow‐up criteria established by the trial protocol. Molecular information was delivered to the tumor board in a median time frame of 14 days (11–22 days), and targetable alterations for commercially available agents were found in 23/25 patients (92%). Of the 25 patients, 14 (56%) reached GMI ≥ 1.3. A high level of DNA topoisomerase I (TOPO1) led to the selection of irinotecan‐based treatments in 48% (12/25) of the patients. A pooled analysis suggested clinical benefit in patients with high TOPO1 expression receiving irinotecan‐based regimens (GMI ≥ 1.3 in 66.7% of cases). These results confirmed previous observations that MoMP increases the frequency of identifiable actionable alterations (92% of patients). The MoMP proposed allows the identification of biomarkers that are frequently expressed in MBCs and the evaluation of their role as predictors of response to commercially available agents. Lastly, this study confirmed the role of MoMP for informing treatment selection in refractory MBC patients: more than half of the enrolled patients reached a GMI ≥ 1.3 even after multiple lines of previous therapies for metastatic disease.

A surrogate measure for patient reported symptom remission in administrative data

BACKGROUND

In real-world pragmatic administrative databases, patient reported remission is often missing.

OBJECTIVE

We evaluate if, in administrative data, five features of antidepressant use patterns can replace patient-reported symptom remission.

METHOD

We re-examined data from Sequence Treatment Alternatives to Relieve Depression (STAR*D) study. Remission was measured using 50% reduction in Hamilton index. Pattern of antidepressant use was examined through five variables: (a) number of prior ineffective antidepressants, (b) duration of taking current antidepressant, (c) receiving therapeutic dose of the medication, and (d) switching to another medication, or (e) augmenting with another antidepressant. The likelihood ratio (LR) associated with each of these predictors was assessed in 90% of data (3329 cases) and evaluated in 10% of data (350 cases) set-aside for evaluation. The accuracy of predictions was calculated using Area under the Receiver Operating Curve (AROC).

RESULTS

Patients who took antidepressants for 14 weeks (LR = 2.007) were more likely to have symptom remission. Prior use of 3 antidepressants reduced the odds of remission (LR = 0.771). Patients who received antidepressants below therapeutic dose were 5 times less likely to experience remission (LR = 0.204). Antidepressant that were augment or switched, almost never led to remission (LR = 0.008, LR = 0.002 respectively). Patterns of antidepressant use accurately (AROC = 0.93) predicted symptom remission.

CONCLUSION

Within the first 100 days, antidepressants use patterns could serve as a surrogate measure for patient-reported remission of symptoms.

Computational Barthel Index: an automated tool for assessing and predicting activities of daily living among nursing home patients

Background

Assessment of functional ability, including activities of daily living (ADLs), is a manual process completed by skilled health professionals. In the presented research, an automated decision support tool, the Computational Barthel Index Tool (CBIT), was constructed that can automatically assess and predict probabilities of current and future ADLs based on patients’ medical history.

Methods

The data used to construct the tool include the demographic information, inpatient and outpatient diagnosis codes, and reported disabilities of 181,213 residents of the Department of Veterans Affairs’ (VA) Community Living Centers. Supervised machine learning methods were applied to construct the CBIT. Temporal information about times from the first and the most recent occurrence of diagnoses was encoded. Ten-fold cross-validation was used to tune hyperparameters, and independent test sets were used to evaluate models using AUC, accuracy, recall and precision. Random forest achieved the best model quality. Models were calibrated using isotonic regression.

Results

The unabridged version of CBIT uses 578 patient characteristics and achieved average AUC of 0.94 (0.93–0.95), accuracy of 0.90 (0.89–0.91), precision of 0.91 (0.89–0.92), and recall of 0.90 (0.84–0.95) when re-evaluating patients. CBIT is also capable of predicting ADLs up to one year ahead, with accuracy decreasing over time, giving average AUC of 0.77 (0.73–0.79), accuracy of 0.73 (0.69–0.80), precision of 0.74 (0.66–0.81), and recall of 0.69 (0.34–0.96). A simplified version of CBIT with 50 top patient characteristics reached performance that does not significantly differ from full CBIT.

Conclusion

Discharge planners, disability application reviewers and clinicians evaluating comparative effectiveness of treatments can use CBIT to assess and predict information on functional status of patients.

US veterans administration diabetes risk (VADR) national cohort: cohort profile

PURPOSE

The veterans administration diabetes risk (VADR) cohort facilitates studies on temporal and geographic patterns of pre-diabetes and diabetes, as well as targeted studies of their predictors. The cohort provides an infrastructure for examination of novel individual and community-level risk factors for diabetes and their consequences among veterans. This cohort also establishes a baseline against which to assess the impact of national or regional strategies to prevent diabetes in veterans.

PARTICIPANTS

The VADR cohort includes all 6 082 018 veterans in the USA enrolled in the veteran administration (VA) for primary care who were diabetes-free as of 1 January 2008 and who had at least two diabetes-free visits to a VA primary care service at least 30 days apart within any 5-year period since 1 January 2003, or veterans subsequently enrolled and were diabetes-free at cohort entry through 31 December 2016. Cohort subjects were followed from the date of cohort entry until censure defined as date of incident diabetes, loss to follow-up of 2 years, death or until 31 December 2018.

FINDINGS TO DATE

The incidence rate of type 2 diabetes in this cohort of over 6 million veterans followed for a median of 5.5 years (over 35 million person-years (PY)) was 26 per 1000 PY. During the study period, 8.5% of the cohort were lost to follow-up and 17.7% died. Many demographic, comorbidity and other clinical variables were more prevalent among patients with incident diabetes.

FUTURE PLANS

This cohort will be used to study community-level risk factors for diabetes, such as attributes of the food environment and neighbourhood socioeconomic status via geospatial linkage to residence address information.

Relative accuracy of social and medical determinants of suicide in electronic health records

OBJECTIVE

This paper compares the accuracy of predicting suicide from Social Determinants of Health (SDoH) or history of illness.

POPULATION STUDIED

5 313 965 Veterans who at least had two primary care visits between 2008 and 2016.

STUDY DESIGN

The dependent variable was suicide or intentional self-injury. The independent variables were 10 495 International Classification of Disease (ICD) Version 9 codes, age, and gender. The ICD codes included 40 V-codes used for measuring SDoH, such as family disruption, family history of substance abuse, lack of education, legal impediments, social isolation, unemployment, and homelessness. The sample was randomly divided into training (90 percent) and validation (10 percent) sets. Area under the receiver operating characteristic (AROC) was used to measure accuracy of predictions in the validation set.

PRINCIPAL FINDINGS

Separate analyses were done for inpatient and outpatient codes; the results were similar. In the hospitalized group, the mean age was 67.2 years, and 92.1 percent were male. The mean number of medical diagnostic codes during the study period was 37; and 12.9 percent had at least one SDoH V-code. At least one episode of suicide or intentional self-injury occurred in 1.89 percent of cases. SDoH V-codes, on average, elevated the risk of suicide or intentional self-injury by 24-fold (ranging from 4- to 86-fold). An index of 40 SDoH codes predicted suicide or intentional self-injury with an AROC of 0.64. An index of 10 445 medical diagnoses, without SDoH V-codes, had AROC of 0.77. The combined SDoH and medical diagnoses codes also had AROC of 0.77.

CONCLUSION

In predicting suicide or intentional self-harm, SDoH V-codes add negligible information beyond what is already available in medical diagnosis codes.

IMPLICATIONS FOR PRACTICE

Policies that affect SDoH (eg, housing policies, resilience training) may not have an impact on suicide rates, if they do not change the underlying medical causes of SDoH.

Worry less about the algorithm, more about the sequence of events

Background: Many algorithms exist for learning network structure and parameters from data. Some of these algorithms include Grow Shrink, Incremental Association Markov Blanket, IAMB, Fast IAMB, and Interleaved IAMB, Hill Climbing, Restricted Maximization and Maximum-Minimum Hill Climbing. These algorithms optimize the fit to the data, while ignoring the order of occurrences of the variables in the network structure. Objective: This paper examines if sequence information (i.e., one variable occurs before another) can make algorithms for learning directed acyclical graph networks more accurate. Methods: A 13- variable network was simulated, where information on sequence of occurrence of some of the variables was assumed to be known. In each simulation 10,000 observations were generated. These observations were used by 4 conditional dependency and 4 search and score algorithms to discover the network from the simulated data. Partial sequence was used to prohibit a directed arc from a later variable to an earlier one. The Area under the Receiver Operating Curve (AROC) was used to compare the accuracy of the sequence-constrained and unconstrained algorithms in predicting the last node in the network. In addition, we examined the performance of sequence constrained algorithms in a real data set. We analyzed 1.3 million disability assessments done on 296,051 residents in Veterans Affairs nursing homes; where the sequence of occurrence of variables was inferred from the average age of occurrence of disabilities. We constructed three networks using Grow-Shrink algorithm, one without and the other two use two permutation of the observed sequence. The fit of these three models to data was examined using Bayesian Information Criterion (BIC). Results: In simulated data, algorithms that used sequenced constraints (AROC = 0.94, confidence intervals, C.I. = 0.86 to 1) were significantly more accurate than the same algorithm without use of sequence constraints (AROC = 0.74, C.I. = 0.65 to 0.83). The agreement between discovered and observed networks improved from range of 0.54 to 0.97 to range of 0.88 to 1. In the real data set, the Bayesian network constructed with use of sequence had 6% lower BIC scores. Conclusions: Sequence information improved accuracy of all eight learning algorithms and should be routinely examined in learning network structure from data.

Prior antibiotic exposure and risk of type 2 diabetes among Veterans

BACKGROUND

Exposure to antibiotics may increase the risk of type 2 diabetes. Veterans are at increased risk for diabetes and for exposure to antibiotics.

OBJECTIVE

To determine the impact of antibiotic exposure for risk of diabetes.

DESIGN

Retrospective cohort study.

PARTICIPANTS

Veterans at the New York Harbor Healthcare System enrolled in primary care, 2004-2014, with ≥2 glycosylated hemoglobin test results <6.5%.

MAIN MEASURES

The primary exposure was any antimicrobial prescribed >6 months prior to the date of diabetes diagnosis, loss to follow-up, death, or the end of the study, measured as the number of courses of antimicrobial prescriptions filled and the mean daily dose (MDD). The primary outcome was incident diagnosis of diabetes through 2014, defined ≥2 ICD-9 codes for diabetes or ≥2 prescriptions of diabetes medications, other than metformin. Cox proportional hazards regression was used to model antimicrobial medications, demographic and anthropometric measures, and comorbid cardiovascular conditions to incident diabetes. Models incorporated time varying covariates of antimicrobial medication and MDD to analyze associations by antimicrobial class.

KEY RESULTS

Among 14,361 Veterans, 9922 (69.1%) were prescribed any antimicrobial medication during the study period. 1413 (9.8%) individuals developed type 2 diabetes. Increased risk of diabetes was associated with >1 prescription (HR 1.13 [1.01-1.26]) compared to none. Time varying analysis of the total number of cumulative courses prescribed showed increased diabetes risk for cephalosporin (HR 1.17 [1.04-1.31]), macrolide (HR 1.08 [1.03-1.13]) and penicillin (HR 1.05 [1.02-1.07]). MDD showed increased risk per 100-unit (mg) increase in antibiotic exposure from (HR 1.05 [1.02-1.08]) for sulfonamide to (HR 1.70 [1.51-1.92]) for cephalosporin.

CONCLUSION

Any and repeated exposure to certain antibiotics may increase diabetes risk among Veterans. Results from this study add to the growing evidence suggesting that antibiotic exposure increases risk for diabetes. Antibiotic stewardship may be enhanced by better understanding this risk, and may lower the incidence of diabetes in populations at risk.

HbA1c, lipid profiles and risk of incident type 2 Diabetes in United States Veterans

United States Veterans are at excess risk for type 2 diabetes, but population differentials in risk have not been characterized. We determined risk of type 2 diabetes in relation to prediabetes and dyslipidemic profiles in Veterans at the VA New York Harbor (VA NYHHS) during 2004-2014. Prediabetes was based on American Diabetes Association hemoglobin A1c (HbA1c) testing cut-points, one of several possible criteria used to define prediabetes. We evaluated transition to type 2 diabetes in 4,297 normoglycemic Veterans and 7,060 Veterans with prediabetes. Cox proportional hazards regression was used to relate HbA1c levels, lipid profiles, demographic, anthropometric and comorbid cardiovascular factors to incident diabetes (Hazard Ratio [HR] and 95% confidence intervals). Compared to normoglycemic Veterans (HbA1c: 5.0-5.6%; 31-38 mmol/mol), risks for diabetes were >2-fold in the moderate prediabetes risk group (HbA1c: 5.7-5.9%; 39-41 mmol/mol) (HR 2.37 [1.98-2.85]) and >5-fold in the high risk prediabetes group (HbA1c: 6.0-6.4%; 42-46 mmol/mol) (HR 5.59 [4.75-6.58]). Risks for diabetes were increased with elevated VLDL (≥40mg/dl; HR 1.31 [1.09-1.58]) and TG/HDL (≥1.5mg/dl; HR 1.34 [1.12-1.59]), and decreased with elevated HDL (≥35mg/dl; HR 0.80 [0.67-0.96]). Transition to diabetes in Veterans was related in age-stratified risk score analyses to HbA1c, VLDL, HDL and TG/HDL, BMI, hypertension and race, with 5-year risk differentials of 62% for the lowest (5-year risk, 13.5%) vs. the highest quartile (5-year risk, 21.9%) of the risk score. This investigation identified substantial differentials in risk of diabetes in Veterans, based on a readily-derived risk score suitable for risk stratification for type 2 diabetes prevention.

Electronic Health Record-Based Screening for Substance Abuse

Existing methods of screening for substance abuse (standardized questionnaires or clinician's simply asking) have proven difficult to initiate and maintain in primary care settings. This article reports on how predictive modeling can be used to screen for substance abuse using extant data in electronic health records (EHRs). We relied on data available through Veterans Affairs Informatics and Computing Infrastructure (VINCI) for the years 2006 through 2016. We focused on 4,681,809 veterans who had at least two primary care visits; 829,827 of whom had a hospitalization. Data included 699 million outpatient and 17 million inpatient records. The dependent variable was substance abuse as identified from 89 diagnostic codes using the Agency for Healthcare Quality and Research classification of diseases. In addition, we included the diagnostic codes used for identification of prescription abuse. The independent variables were 10,292 inpatient and 13,512 outpatient diagnoses, plus 71 dummy variables measuring age at different years between 20 and 90 years. A modified naive Bayes model was used to aggregate the risk across predictors. The accuracy of the predictions was examined using area under the receiver operating characteristic (AROC) curve in 20% of data, randomly set aside for the evaluation. Many physical/mental illnesses were associated with substance abuse. These associations supported findings reported in the literature regarding the impact of substance abuse on various diseases and vice versa. In randomly set-aside validation data, the model accurately predicted substance abuse for inpatient (AROC = 0.884), outpatient (AROC = 0.825), and combined inpatient and outpatient (AROC = 0.840) data. If one excludes information available after substance abuse is known, the cross-validated AROC remained high, 0.822 for inpatient and 0.817 for outpatient data. Data within EHRs can be used to detect existing or predict potential future substance abuse.

Abstract 3290: The Side-Out Foundation Metastatic Breast Cancer Database, an open-access portal for multi-omic molecular data and more

Introduction: With the high volume of molecular data being generated daily from human malignancies, there is an increasing need to create novel internet-based portals where this information is readily and easily accessible to physicians, scientists, and the general public. Although numerous databases have been created to capture the molecular characteristics of primary tumors, limited resources are available when it comes to the metastatic lesions. We have developed a novel open-access database to capture demographic, clinical, and pathological information, outcome data, and multi-omic based molecular profiles from metastatic breast cancer (MBC) patients.

Materials and Methods: The portal was created using the open-source relational database management system MySQL and the custom-codes were written using the PHP server-side scripting language. User interface, management, and authentication were created in WordPress. The database is primarily used to record information collected through the Side-Out clinical trials, a series of prospective Phase II trials targeting refractory MBC (NCT01074814, NCT01919749, NCT03195192). De-identified patients' information includes patients' demographics, treatment history, pathological characteristics of the primary tumor, outcome data, and multi-omic molecular profiles. Molecular information collected from each lesion include genomic (NGS-based whole/targeted exome sequencing), transcriptomic (RNA microarray/RNA Seq), proteomic (protein expression by IHC), and phosphoproteomic (protein pathway activation mapping by Reverse Phase Protein Microarray) data. Over 700 data fields are collected for each patient. A higher level of security for the recorded information is achieved by using a secondary database along with custom-codes during the data entry process. Investigators are exploring how Block-chain database design can be used to make portion of the data public while encrypting patient identifiers and key variables.

Results: We present an overview of the portal, its usability, how access can be requested by interested third party, the user-friendly interface for downloading clinical, pathological and molecular data, and a few examples of how these data can be used to explore different aspects of metastatic breast cancers.

Conclusions: To our knowledge, this is the first web-based publicly accessible database portal where broad multi-omic profiles are captured from the MBC patients along with demographic, clinical and pathological data. This open-access portal represents a unique and highly valuable tool as it integrates different aspects of the disease and can be used for correlative analyses and hypothesis-generating studies. Finally, this web-based portal allows free-of-charge dissemination of data from existing or upcoming clinical and translational studies targeting breast cancer patients.

Citation Format: Elisa Baldelli, Shiva Ghaemi, Sanja Avramovic, Janusz Wojtusiak, Lance A. Liotta, Farrokh Alemi, Emanuel F. Petricoin, Bryant Dunetz, Mariaelena Pierobon. The Side-Out Foundation Metastatic Breast Cancer Database, an open-access portal for multi-omic molecular data and more [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2018; 2018 Apr 14-18; Chicago, IL. Philadelphia (PA): AACR; Cancer Res 2018;78(13 Suppl):Abstract nr 3290.

Abstract P453: Hypertension Mortality Risk May be Fake News for Nonagenarians

Background: Little is known regarding BP control and mortality risk in subjects greater than 90 years of age.

Objective: This paper assesses the association between systolic blood pressure and mortality risk for nonagenarians.

Methods: Data from the Veterans Administration Informatics and Computing Infrastructure were used to analyze the survival of 193,651 nonagenarians in 130 Veteran Administration medical centers. Following clinical guidelines, for each day, we selected the lowest Systolic reading of the day exceeding 90 mmHg and defined sustained pressure as the average of two consecutive readings at least one month apart. Kaplan Meier curves and Cox regression was used to analyze survival.

Results: Odds of mortality from hypertension declined with age. When patients were over 90 years old, the odds of mortality for hypertensives was below 1 to 1, suggesting a protective effect. Patients whose sustained systolic pressures exceeded 140 mmHg survived longer than patients whose highest sustained pressure was between 90 mmHg to 140 mmHg (p<.0001). Furthermore, strict control of hypertension of 90 year old patients (SBP 120) was associated with lower days of survival than hypertensives, raising questions about value of strict control of hypertension among 90 year olds.

Conclusions: For nonagenarians, mortality from hypertension may be a lower concern than mortality from other causes. Randomized clinical trials are needed to examine the impact of control of hypertension of patient over 90 year olds.

A Comprehensive Index for Predicting Risk of Anemia from Patients' Diagnoses

This article demonstrates how time-dependent, interacting, and repeating risk factors can be used to create more accurate predictive medicine. In particular, we show how emergence of anemia can be predicted from medical history within electronic health records. We used the Veterans Affairs Informatics and Computing Infrastructure database to examine a retrospective cohort of 9,738,838 veterans over an 11-year period. Using International Clinical Diagnoses Version 9 codes organized into 25 major diagnostic categories, we measured progression of disease by examining changes in risk over time, interactions in risk of combination of diseases, and elevated risk associated with repeated hospitalization for the same diagnostic category. The maximum risk associated with each diagnostic category was used to predict anemia. The accuracy of the model was assessed using a validation cohort. Age and several diagnostic categories significantly contributed to the prediction of anemia. The largest contributors were health status ([Formula: see text] = -1075, t = -92, p < 0.000), diseases of the endocrine ([Formula: see text] = -1046, t = -87, p < 0.000), hepatobiliary ([Formula: see text] = -1043, t = -72, p < 0.000), kidney ([Formula: see text] = -1125, t = -111, p < 0.000), and respiratory systems ([Formula: see text] = -1151, t = -89, p < 0.000). The AUC for the additive model was 0.751 (confidence interval 74.95%-75.26%). The magnitude of AUC suggests that the model may assist clinicians in determining which patients are likely to develop anemia. The procedures used for examining changes in risk factors over time may also be helpful in other predictive medicine projects.

Covariate Balancing through Naturally Occurring Strata

OBJECTIVE

To provide an alternative to propensity scoring (PS) for the common situation where there are interacting covariates.

SETTING

We used 1.3 million assessments of residents of the United States Veterans Affairs nursing homes, collected from January 1, 2000, through October 9, 2012.

DESIGN

In stratified covariate balancing (SCB), data are divided into naturally occurring strata, where each stratum is an observed combination of the covariates. Within each stratum, cases with, and controls without, the target event are counted; controls are weighted to be as frequent as cases. This weighting procedure guarantees that covariates, or combination of covariates, are balanced, meaning they occur at the same rate among cases and controls. Finally, impact of the target event is calculated in the weighted data. We compare the performance of SCB, logistic regression (LR), and propensity scoring (PS) in simulated and real data. We examined the calibration of SCB and PS in predicting 6-month mortality from inability to eat, controlling for age, gender, and nine other disabilities for 296,051 residents in Veterans Affairs nursing homes. We also performed a simulation study, where outcomes were randomly generated from treatment, 10 covariates, and increasing number of covariate interactions. The accuracy of SCB, PS, and LR in recovering the simulated treatment effect was reported.

FINDINGS

In simulated environment, as the number of interactions among the covariates increased, SCB and properly specified LR remained accurate but pairwise LR and pairwise PS, the most common applications of these tools, performed poorly. In real data, application of SCB was practical. SCB was better calibrated than linear PS, the most common method of PS.

CONCLUSIONS

In environments where covariates interact, SCB is practical and more accurate than common methods of applying LR and PS.

A Comprehensive Multimorbidity Index for Predicting Mortality in Intensive Care Unit Patients

BACKGROUND

Accurate prediction of mortality for patients admitted to the intensive care units (ICUs) is an important component of medical care. However, little is known about the role of multimorbidity in predicting end of life for high-risk and vulnerable patients.

OBJECTIVE

The aim of the study was to derive and validate a multimorbidity risk model in an attempt to predict all-cause mortality at 6 and 12 months posthospital discharge.

METHODS

This is a retrospective, observational, clinical cohort study. Data were collected on 442,692 ICU patients who received care through the Veterans Administration between January 2003 and December 2013. The primary outcome was all-cause mortality at 6 and 12 months posthospital discharge. We divided the data into derivation (80%) and validation (20%) sets. Using multivariable logistic regression models, we compared prognostic models based on age, principal diagnosis groups, physiological markers, immunosuppressants, comorbidity categories, and a newly developed multimorbidity index (MMI) based on 5695 comorbidities. The cross-validated area under the receiver operating characteristic curve (AUC) was used to report the accuracy of predicting all-cause mortality at 6 and 12 months of hospital discharge.

RESULTS

The average age of patients was 68.87 years (standard deviation = 12.1), 95.9% were males, 44.9% were widowed, divorced, or separated. The relative order of accuracy in predicting mortality was the MMI (AUC = 0.84, CI = 0.83-0.84), VA Inpatient Evaluation Center index (AUC = 0.80, CI = 0.79-0.81), principal diagnosis groups (AUC = 0.74, CI = 0.73-0.74), comorbidities (AUC = 0.69, CI = 0.68-0.70), physiological markers (AUC = 0.65, CI = 0.64-0.65), age (AUC = 0.60, CI = 0.60-0.61),and immunosuppressant use (AUC = 0.59, CI = 0.58-0.59).

CONCLUSIONS

The MMI improved the accuracy of predicting short- and long-term all-cause mortality for ICU patients. Further prospective studies are needed to validate the index in different clinical settings and test generalizability of results in patients outside the VA system of care.

The VA Medical Foster Home Program, Ambulatory Care Sensitive Conditions, and Avoidable Hospitalizations

This quality control study analyzes whether the Veterans Administration Medical Foster Home (VA MFH) program has been successful in improving access and effectiveness of ambulatory care. Individuals hospitalized for one or more of 22 adult ambulatory care sensitive conditions were identified. Pre and post comparisons of a specified population of participants in the program were conducted to determine rates of avoidable hospitalizations for 6 months prior to and following MFH enrollment. The overall rate of avoidable hospitalizations declined from 18.5 to 14.9 per 100 enrollees following enrollment. The number of bed days used declined by 39%, as did the cost associated with avoidable hospitalizations. Enrollment in the VA MFH program resulted in an overall reduction in the rate of avoidable hospitalizations, resource utilization, and costs. Studies are needed comparing these results with other matched cohorts of nursing home eligible veterans.

The Multimorbidity Index: A Tool for Assessing the Prognosis of Patients from Their History of Illness

BACKGROUND

The Multimorbidity (MM) Index predicts the prognosis of patients from their diagnostic history. In contrast to existing approaches with broad diagnostic categories, it treats each diagnosis as a separate independent variable using individual International Classification of Disease, Revision 9 (ICD-9) codes.

OBJECTIVE

This paper describes the MM Index, reviews the published data on its accuracy, and provides procedures for implementing the Index within electronic health record (EHR) systems. Methods: The MM Index was tested on various patient populations by using data from the United States Department of Veterans Affairs data warehouse and claims data within the Healthcare Cost and Utilization Project of the Agency for Health Care Research and Quality.

RESULTS

In cross-validated studies, the MM Index outperformed prognostic indices based on physiological markers, such as CD4 cell counts in HIV/AIDS, HbAlc levels in diabetes, ejection fractions in heart failure, or the 13 physiological markers commonly used for patients in intensive care units. When predicting the prognosis of nursing home patients by using the cross-validated area under a receiver operating characteristic (ROC) curve, the MM Index was 15 percent outperformed the Quan variant of the Charlson Index, 27 percent more accurate than the Deyo variant of the Charlson Index, and 22 percent more accurate than the von Walraven variant of the Elixhauser Index. For patients in intensive care units, the MM Index was 13 percent outperformed the cross-validated area under ROC associated with Elixhauser's categories. The MM Index also demonstrated greater accuracy than a number of commercially available measures of illness severity; including a fivefold greater accuracy than the All Patient Refined Diagnosis-Related Groups and a threefold greater accuracy than All Payer Severity-Adjusted Diagnosis-Related Groups.

CONCLUSION

The MM Index is statistically more accurate than many existing measures of prognosis. The magnitude of improvement is large and may lead to a clinically meaningful difference in patient care. Given the large improvements in accuracy, the use of the MM Index for policy and comparative effectiveness analysis is recommended.

Improving Prognostic Web Calculators: Violation of Preferential Risk Independence

BACKGROUND

Web-based applications are available for prognostication of individual patients. These prognostic models were developed for groups of patients. No one is the average patient, and using these calculators to inform individual patients could provide misleading results.

OBJECTIVE

This article gives an example of paradoxical results that may emerge when indices used for prognosis of the average person are used for care of an individual patient.

METHODS

We calculated the expected mortality risks of stomach cancer and its associated comorbidities. Mortality risks were calculated using data from 140,699 Veterans Administration nursing home residents.

RESULTS

On average, a patient with hypertension has a higher risk of mortality than one without hypertension. Surprisingly, among patients with lung cancer, hypertension is protective and reduces risk of mortality. This paradoxical result is explained by how group-level, average prognosis could mislead individual patients. In particular, average prognosis of lung cancer patients reflects the impact of various comorbidities that co-occur in lung cancer patients. The presence of hypertension, a relatively mild comorbidity of lung cancer, indicates that more serious comorbidities have not occurred. It is not that hypertension is protective; it is the absence of more serious comorbidities that is protective. The article shows how the presence of these anomalies can be checked through the mathematical concept of preferential risk independence.

CONCLUSION

Instead of reporting average risk scores, web-based calculators may improve accuracy of predictions by reporting the unconfounded risks.

Using observed sequence to orient causal networks

In learning causal networks, typically cross-sectional data are used and the sequence among the network nodes is learned through conditional independence. Sequence is inherently a longitudinal concept. We propose to learn sequence of events in longitudinal data and use it to orient arc directions in a network learned from cross-sectional data. The network is learned from cross-sectional data using various established algorithms, with one modification. Arc directions that do not agree with the longitudinal sequence were prohibited. We established longitudinal sequence through two methods: Probabilistic Contrast, and Goodman and Kruskal error reduction methods. In simulated data, the error reduction method was used to learn the sequence in the data. The procedure reduced the number of arc direction errors and larger improvements were observed with increasing number of events in the network. In real data, different algorithms were used to learn the network from cross-sectional data, while prohibiting arc directions not supported by longitudinal information. The agreement among learned networks increased significantly. It is possible to combine sequence information learned from longitudinal data with algorithms organized for learning network models from cross-sectional data. Such models may have additional causal interpretation as they more explicitly take into account observed sequence of events.

Family self-tailoring: Applying a systems approach to improving family healthy living behaviors

The adoption and maintenance of healthy living behaviors by individuals and families is a major challenge. We describe a new model of health behavior change, SystemCHANGE (SC), which focuses on the redesign of family daily routines using system improvement methods. In the SC intervention, families are taught a set of skills to engage in a series of small, family self-designed experiments to test ideas to change their daily routines. The family system-oriented changes brought about by these experiments build healthy living behaviors into family daily routines so that these new behaviors happen as a matter of course, despite wavering motivation, willpower, or personal effort on the part of individuals. Case stories of the use of SC to improve family healthy living behaviors are provided. Results of several pilot tests of SC indicate its potential effectiveness to change health living behaviors across numerous populations.

Shared Homes as an Alternative to Nursing Home Care: Impact of VA's Medical Foster Home Program on Hospitalization

PURPOSE OF THE STUDY

This study compares hospitalization rates for common conditions in the Veteran Affairs (VA) Medical Foster Home (MFH) program to VA nursing homes, known as Community Living Centers (CLCs).

DESIGN AND METHODS

We used a nested, matched, case control design. We examined 817 MFH residents and matched each to 3 CLC residents selected from a pool of 325,031. CLC and MFH cases were matched on (a) baseline time period, (b) follow-up time period, (c) age, (d) gender, (e) race, (f) risk of mortality calculated from comorbidities, and (g) history of hospitalization for the selected condition during the baseline period. Odds ratio (OR) and related confidence interval (CI) were calculated to contrast MFH cases and matched CLC controls.

RESULTS

Compared with matched CLC cases, MFH residents were less likely to be hospitalized for adverse care events, (OR = 0.13, 95% CI = 0.03-0.53), anxiety disorders (OR = 0.52, 95% CI = 0.33-0.80), mood disorders (OR = 0.57, 95% CI = 0.42-0.79), skin infections (OR = 0.22, 95% CI = 0.10-0.51), pressure ulcers (OR = 0.22, 95% CI = 0.09-0.50) and bacterial infections other than tuberculosis or septicemia (OR = 0.54, 95% CI = 0.31-0.92). MFH cases and matched CLC controls did not differ in rates of urinary tract infections, pneumonia, septicemia, suicide/self-injury, falls, other injury besides falls, history of injury, delirium/dementia/cognitive impairments, or adverse drug events. Hospitalization rates were not higher for any conditions studied in the MFH cohort compared with the CLC cohort.

IMPLICATIONS

MFH participants had the same or lower rates of hospitalizations for conditions examined compared with CLC controls suggesting that noninstitutional care by a nonfamilial caregiver does not increase hospitalization rates for common medical conditions.

Sequence of Functional Loss and Recovery in Nursing Homes

PURPOSE OF THE STUDY

This study provides benchmarks for likelihood, number of days until, and sequence of functional decline and recovery.

DESIGN AND METHODS

We analyzed activities of daily living (ADLs) of 296,051 residents in Veteran Affairs nursing homes between January 1, 2000 and October 9, 2012. ADLs were extracted from standard minimum data set assessments. Because of significant overlap between short- and long-stay residents, we did not distinguish between these populations. Twenty-five combinations of ADL deficits described the experience of 84.3% of all residents. A network model described transitions among these 25 combinations. The network was used to calculate the shortest, longest, and maximum likelihood paths using backward induction methodology. Longitudinal data were used to derive a Bayesian network that preserved the sequence of occurrence of 9 ADL deficits.

RESULTS

The majority of residents (57%) followed 4 pathways in loss of function. The most likely sequence, in order of occurrence, was bathing, grooming, walking, dressing, toileting, bowel continence, urinary continence, transferring, and feeding. The other three paths occurred with reversals in the order of dressing/toileting and bowel/urinary continence. ADL impairments persisted without any change for an average of 164 days (SD = 62). Residents recovered partially or completely from a single impairment in 57% of cases over an average of 119 days (SD = 41). Recovery rates declined as residents developed more than 4 impairments.

IMPLICATIONS

Recovery of deficits among those studied followed a relatively predictable path, and although more than half recovered from a single functional deficit, recovery exceeded 100 days suggesting time to recover often occurs over many months.

The development and application of an oncology Therapy-Related Symptom Checklist for Adults (TRSC) and Children (TRSC-C) and e-health applications

Background

Studies found that treatment symptoms of concern to oncology/hematology patients were greatly under-identified in medical records. On average, 11.0 symptoms were reported of concern to patients compared to 1.5 symptoms identified in their medical records. A solution to this problem is use of an electronic symptom checklist that can be easily accessed by patients prior to clinical consultations. Purpose: Describe the oncology Therapy-Related Symptom Checklists for Adults (TRSC) and Children (TRSC-C), which are validated bases for e-Health symptom documentation and management. The TRSC has 25 items/symptoms; the TRSC-C has 30 items/symptoms. These items capture up to 80% of the variance of patient symptoms. Measurement properties and applications with outpatients are presented. E-Health applications are indicated.

Methods

The TRSC was developed for adults (N = 282) then modified for children (N = 385). Statistical analyses have been done using correlational, epidemiologic, and qualitative methods. Extensive validation of measurement properties has been reported.

Results

Research has found high levels of patient/clinician satisfaction, no increase in clinic costs, and strong correlations of TRSC/TRSC-C with medical outcomes. A recently published sequential cohort trial with adult outpatients at a Mayo Clinic community cancer center found TRSC use produced a 7.2% higher patient quality of life, 116% more symptoms identified/managed, and higher functional status.

Discussion, implications, and follow-up

An electronic system has been built to collect TRSC symptoms, reassure patients, and enhance patient-clinician communications. This report discusses system design and efforts made to provide an electronic system comfortable to patients. Methods used by clinicians to promote comfort and patient engagement were examined and incorporated into system design. These methods included (a) conversational data collection as opposed to survey style or standardized questionnaires, (b) short response phrases indicating understanding of the reported symptom, (c) use of open-ended questions to reduce long lists of symptoms, (d) directed questions that ask for confirmation of expected symptoms, (e) review of symptoms at designated stages, and (d) alerting patients when the computer has informed clinicians about patient-reported symptoms.

Conclusions

An e-Health symptom checklist (TRSC/TRSC-C) can facilitate identification, monitoring, and management of symptoms; enhance patient-clinician communications; and contribute to improved patient outcomes.

Predicting Functional Decline and Recovery for Residents in Veterans Affairs Nursing Homes

PURPOSE OF THE STUDY

This article describes methods and accuracy of predicting change in activities of daily living (ADLs) for nursing home patients following hospitalization.

DESIGN AND METHODS

Electronic Health Record data for 5,595 residents of Veterans Affairs' (VAs') Community Living Centers (CLCs) aged 70 years and older were analyzed within the VA Informatics and Computing Infrastructure. Data included diagnoses from 7,106 inpatient records, 21,318 functional status evaluations, and 69,140 inpatient diagnoses. The Barthel Index extracted from CLC's Minimum Data Set was used to assess ADLs loss and recovery. Patients' diagnoses on hospital admission, ADL status prior to hospitalization, age, and gender were used alone or in combination to predict ADL loss/gain following hospitalization. Area under the Receiver-Operator Curve (AUC) was used to report accuracy of predictions in short (14 days) and long-term (15-365 days) follow-up post-hospitalization.

RESULTS

Admissions fell into 7 distinct patterns of recovery and loss: early recovery 19%, delayed recovery 9%, delayed recovery after temporary decline 9%, early decline 29%, delayed decline 10%, delayed decline after temporary recovery 6%, and no change 18%. Models accurately predicted ADL's 14-day post-hospitalization (AUC for bathing 0.917, bladder 0.842, bowels 0.875, dressing 0.871, eating 0.867, grooming 0.902, toileting 0.882, transfer 0.852, and walking deficits was 0.882). Accuracy declined but remained relatively high when predicting 14-365 days post-hospitalization (AUC ranging from 0.798 to 0.875).

IMPLICATIONS

Predictive modeling may allow development of more personalized predictions of functional loss and recovery after hospitalization among nursing home patients.

Predictors of six-month mortality among nursing home residents: diagnoses may be more predictive than functional disability

OBJECTIVE

Loss of daily living functions can be a marker for end of life and possible hospice eligibility. Unfortunately, data on patient's functional abilities is not available in all settings. In this study we compare predictive accuracy of two indices designed to predict 6-month mortality among nursing home residents. One is based on traditional measures of functional deterioration and the other on patients' diagnoses and demography.

METHODS

We created the Hospice ELigibility Prediction (HELP) Index by examining mortality of 140,699 Veterans Administration (VA) nursing home residents. For these nursing home residents, the available data on history of hospital admissions were divided into training (112,897 cases) and validation (27,832 cases) sets. The training data were used to estimate the parameters of the HELP Index based on (1) diagnoses, (2) age on admission, and (3) number of diagnoses at admission. The validation data were used to assess the accuracy of predictions of the HELP Index. The cross-validated accuracy of the HELP Index was compared with the Barthel Index (BI) of functional ability obtained from 296,052 VA nursing home residents. A receiver operating characteristic curve was used to examine sensitivity and specificity of the predicted odds of mortality.

RESULTS

The area under the curve (AUC) for the HELP Index was 0.838. This was significantly (α <0.01) higher than the AUC for the BI of 0.692.

CONCLUSIONS

For nursing home residents, comorbid diagnoses predict 6-month mortality more accurately than functional status. The HELP Index can be used to estimate 6-month mortality from hospital data and can guide prognostic discussions prior to and following nursing home admission.

Comparative effectiveness of hypoglycemic medications among veterans

BACKGROUND

The efficacy of diabetic medications among patients with multiple comorbidities is not tested in randomized clinical studies. It is important to monitor the performance of these medications after marketing approvals.

OBJECTIVE

To investigate the risk of all-cause mortality associated with prescription of hypoglycemic agents.

METHODS

We retrospectively examined data from 17,773 type 2 diabetic patients seen from March 2, 1998, to December 13, 2010, in 3 Veterans Administration medical centers. Severity was measured using patients' inpatient and outpatient comorbidities during the last year of visits. Severity-adjusted logistic regression was used to measure the odds ratio for mortality within the study period.

RESULTS

Patients' severity of illness correctly classified mortality for 89.8% of the patients (P less than 0.0001). Being younger, married, and white decreased severity adjusted risk of mortality. Exposure to the following medications increased severity adjusted risk of mortality: glyburide (odds ratio [OR] = 1.804, 95% CI from 1.518 to 2.145), glipizide (OR = 1.566, 95% CI from 1.333 to 1.839), rosiglitazone (OR = 1.805, 95% CI from 1.378 to 2.365), chlorpropamide (OR = 3.026, 95% CI from 1.096 to 8.351), insulin (OR = 2.382, 95% CI from 2.112 to 2.686). None of the other medications (metformin, acarbose, glimepiride, pioglitazone, repaglinide, troglitazone, or dipeptidyl peptidase-4) were associated with excess mortality beyond what could be expected from the patients' severity of illness or demographic characteristics. The reported excess mortality could not be explained away by use of other concurrent, nondiabetic classes of medications.

CONCLUSION

Our findings suggest chlorpropamide, glipizide, glyburide, insulin, and rosiglitazone increased severity-adjusted mortality in veterans with type 2 diabetes. A decision aid that could optimize selection of hypoglycemic medications based on patients' comorbidities might increase patients' survival.

Continuously Rethinking the Definition of Influenza for Surveillance Systems

Objective. In the Electronic Surveillance System for the Early Notification of Community-based Epidemics (ESSENCE), influenza was originally defined by a list of 29 and later by a list of 12 diagnosis codes. This article describes a dependent Bayesian procedure designed to improve the ESSENCE system and exploit multiple sources of information without being biased by redundancy. Methods. We obtained 13,096 cases within the Armed Forces Health Longitudinal Technological Application electronic medical records that included an influenza laboratory test. A Dependent Bayesian Expert System (D-BESt) was used to predict influenza from diagnoses, symptoms, reason for visit, temperature, month of visit, category of enrollment, and demographics. For each case, D-BESt sequentially selects the most discriminating piece of information, calculates its likelihood ratio conditioned on previously selected information, and updates the case’s probability of influenza. Results. When the analysis was limited to definitions based on diagnoses and was applied to a sample of patients for whom laboratory tests had been ordered, the areas under the receiver operating characteristic curve (AUCs) for the previous (29-diagnosis) and current (12-diagnosis) ESSENCE lists and the D-BESt algorithm were, respectively, 0.47, 0.36, and 0.77. Including other sources of information further improved the AUC for D-BESt to 0.79. At the best cutoff point for D-BESt, where the receiver operating characteristic curve for D-BESt is farthest from the diagonal line, the D-BESt algorithm correctly classified 84% of cases (specificity = 88%, sensitivity = 62%). In comparison, the current ESSENCE approach of using a list of 12 diagnoses correctly classified only 31% of this sample of cases (specificity = 29%, sensitivity = 42%). Conclusions. False alarms in ESSENCE surveillance systems can be reduced if a probabilistic dynamic learning system is used.