A. John Rush

Estimating Individualized Treatment Rules Using Outcome Weighted Learning

There is increasing interest in discovering individualized treatment rules (ITRs) for patients who have heterogeneous responses to treatment. In particular, one aims to find an optimal ITR that is a deterministic function of patient-specific characteristics maximizing expected clinical outcome. In this article, we first show that estimating such an optimal treatment rule is equivalent to a classification problem where each subject is weighted proportional to his or her clinical outcome. We then propose an outcome weighted learning approach based on the support vector machine framework. We show that the resulting estimator of the treatment rule is consistent. We further obtain a finite sample bound for the difference between the expected outcome using the estimated ITR and that of the optimal treatment rule. The performance of the proposed approach is demonstrated via simulation studies and an analysis of chronic depression data.

Combining Medications to Enhance Depression Outcomes (CO-MED): Acute and Long-Term Outcomes of a Single-Blind Randomized Study

OBJECTIVE Two antidepressant medication combinations were compared with selective serotonin reuptake inhibitor monotherapy to determine whether either combination produced a higher remission rate in first-step acute-phase (12 weeks) and long-term (7 months) treatment. METHOD The single-blind, prospective, randomized trial enrolled 665 outpatients at six primary and nine psychiatric care sites. Participants had at least moderately severe nonpsychotic chronic and/or recurrent major depressive disorder. Escitalopram (up to 20 mg/day) plus placebo, sustained-release bupropion (up to 400 mg/day) plus escitalopram (up to 20 mg/day), or extended-release venlafaxine (up to 300 mg/day) plus mirtazapine (up to 45 mg/day) was delivered (1:1:1 ratio) by using measurement-based care. The primary outcome was remission, defined as ratings of less than 8 and less than 6 on the last two consecutive applications of the 16-item Quick Inventory of Depressive Symptomatology--Self-Report. Secondary outcomes included side effect burden, adverse events, quality of life, functioning, and attrition. RESULTS Remission and response rates and most secondary outcomes were not different among treatment groups at 12 weeks. The remission rates were 38.8% for escitalopram-placebo, 38.9% for bupropion-escitalopram, and 37.7% for venlafaxine-mirtazapine, and the response rates were 51.6%-52.4%. The mean number of worsening adverse events was higher for venlafaxine-mirtazapine (5.7) than for escitalopram-placebo (4.7). At 7 months, remission rates (41.8%-46.6%), response rates (57.4%-59.4%), and most secondary outcomes were not significantly different. CONCLUSIONS Neither medication combination outperformed monotherapy. The combination of extended-release venlafaxine plus mirtazapine may have a greater risk of adverse events.

Prevalence of Major Depressive Episode in CKD

BACKGROUND Depression is prevalent in long-term dialysis patients and is associated with death and hospitalization. Whether depression is present through all chronic kidney disease (CKD) stages or appears after dialysis therapy initiation is not clear. We determined the prevalence of a major depressive episode and other psychiatric illnesses by using a structured gold-standard clinical interview and demographic and clinical variables associated with major depressive episode in patients with CKD. STUDY DESIGN Observational cross-sectional study using a Diagnostic and Statistical Manual of Mental Disorders (Fourth Edition)-based structured interview administered by trained persons to 272 consecutive participants. Multivariable logistic regression was used to determine demographic and clinical variables associated with major depressive episode. SETTING & PARTICIPANTS Patients with stages 2 to 5 CKD not treated by using dialysis were consecutively approached and enrolled from a Veterans Affairs CKD clinic. PREDICTORS Demographic and clinical variables. OUTCOME Major depressive episode diagnosed by using a structured Diagnostic and Statistical Manual of Mental Disorders (Fourth Edition)-based interview, the Mini International Neuropsychiatric Interview. RESULTS The cohort had a mean age of 64.5 +/- 12.0 years. Thirty-eight percent were African American, and 55% had diabetes mellitus. Percentages of patients with stages 2, 3, 4, and 5 CKD were 6%, 38%, 41%, and 14%, respectively. Mean hemoglobin level was 12.5 +/- 2.0 g/dL. The prevalence of a major depressive episode was 21% and did not vary significantly among different CKD stages. Variables associated with a major depressive episode were diabetes mellitus, comorbid psychiatric illness, and history of drug or alcohol abuse. LIMITATIONS Single-center study composed of primarily male veterans. CONCLUSIONS One in 5 patients with CKD had a major depressive episode. Patients with CKD should be screened routinely for depression given this high prevalence and the independent association of depression with poor outcomes in patients with end-stage renal disease receiving maintenance dialysis.

Association Between Major Depressive Episodes in Patients With Chronic Kidney Disease and Initiation of Dialysis, Hospitalization, or Death

CONTEXT Patients with chronic kidney disease (CKD) experience increased rates of hospitalization and death. Depressive disorders are associated with morbidity and mortality. Whether depression contributes to poor outcomes in patients with CKD not receiving dialysis is unknown. OBJECTIVE To determine whether the presence of a current major depressive episode (MDE) is associated with poorer outcomes in patients with CKD. DESIGN, SETTING, AND PATIENTS Prospective cohort study of 267 consecutively recruited outpatients with CKD (stages 2-5 and who were not receiving dialysis) at a VA medical center between May 2005 and November 2006 and followed up for 1 year. An MDE was diagnosed by blinded personnel using the Diagnostic and Statistical Manual of Mental Disorders (Fourth Edition) criteria. MAIN OUTCOME MEASURES The primary outcome was event-free survival defined as the composite of death, dialysis initiation, or hospitalization. Secondary outcomes included each of these events assessed separately. RESULTS Among 267 patients, 56 had a current MDE (21%) and 211 did not (79%). There were 127 composite events, 116 hospitalizations, 38 dialysis initiations, and 18 deaths. Events occurred more often in patients with an MDE compared with those without an MDE (61% vs 44%, respectively, P = .03). Four patients with missing dates of hospitalization were excluded from survival analyses. The mean (SD) time to the composite event was 206.5 (19.8) days (95% CI, 167.7-245.3 days) for those with an MDE compared with 273.3 (8.5) days (95% CI, 256.6-290.0 days) for those without an MDE (P = .003). The adjusted hazard ratio (HR) for the composite event for patients with an MDE was 1.86 (95% CI, 1.23-2.84). An MDE at baseline independently predicted progression to dialysis (HR, 3.51; 95% CI, 1.77-6.97) and hospitalization (HR, 1.90; 95% CI, 1.23-2.95). CONCLUSION The presence of an MDE was associated with an increased risk of poor outcomes in CKD patients who were not receiving dialysis, independent of comorbidities and kidney disease severity.

Association Between Bipolar Spectrum Features and Treatment Outcomes in Outpatients With Major Depressive Disorder

CONTEXT It has been suggested that patients with major depressive disorder (MDD) who display pretreatment features suggestive of bipolar disorder or bipolar spectrum features might have poorer treatment outcomes. OBJECTIVE To assess the association between bipolar spectrum features and antidepressant treatment outcome in MDD. DESIGN Open treatment followed by sequential randomized controlled trials. SETTING Primary and specialty psychiatric outpatient centers in the United States. PARTICIPANTS Male and female outpatients aged 18 to 75 years with a DSM-IV diagnosis of nonpsychotic MDD who participated in the Sequenced Treatment Alternatives to Relieve Depression (STAR*D) study. INTERVENTIONS Open treatment with citalopram followed by up to 3 sequential next-step treatments. MAIN OUTCOME MEASURES Number of treatment levels required to reach protocol-defined remission, as well as failure to return for the postbaseline visit, loss to follow-up, and psychiatric adverse events. For this secondary analysis, putative bipolar spectrum features, including items on the mania and psychosis subscales of the Psychiatric Diagnosis Screening Questionnaire, were examined for association with treatment outcomes. RESULTS Of the 4041 subjects who entered the study, 1198 (30.0%) endorsed at least 1 item on the psychosis scale and 1524 (38.1%) described at least 1 recent maniclike/hypomaniclike symptom. Irritability and psychoticlike symptoms at entry were significantly associated with poorer outcomes across up to 4 treatment levels, as were shorter episodes and some neurovegetative symptoms of depression. However, other indicators of bipolar diathesis including recent maniclike symptoms and family history of bipolar disorder as well as summary measures of bipolar spectrum features were not associated with treatment resistance. CONCLUSION Self-reported psychoticlike symptoms were common in a community sample of outpatients with MDD and strongly associated with poorer outcomes. Overall, the data do not support the hypothesis that unrecognized bipolar spectrum illness contributes substantially to antidepressant treatment resistance.

International Study to Predict Optimized Treatment for Depression (iSPOT-D), a randomized clinical trial: rationale and protocol

AbstractBackgroundClinically useful treatment moderators of Major Depressive Disorder (MDD) have not yet been identified, though some baseline predictors of treatment outcome have been proposed. The aim of iSPOT-D is to identify pretreatment measures that predict or moderate MDD treatment response or remission to escitalopram, sertraline or venlafaxine; and develop a model that incorporates multiple predictors and moderators.Methods/DesignThe International Study to Predict Optimized Treatment - in Depression (iSPOT-D) is a multi-centre, international, randomized, prospective, open-label trial. It is enrolling 2016 MDD outpatients (ages 18-65) from primary or specialty care practices (672 per treatment arm; 672 age-, sex- and education-matched healthy controls). Study-eligible patients are antidepressant medication (ADM) naïve or willing to undergo a one-week wash-out of any non-protocol ADM, and cannot have had an inadequate response to protocol ADM. Baseline assessments include symptoms; distress; daily function; cognitive performance; electroencephalogram and event-related potentials; heart rate and genetic measures. A subset of these baseline assessments are repeated after eight weeks of treatment. Outcomes include the 17-item Hamilton Rating Scale for Depression (primary) and self-reported depressive symptoms, social functioning, quality of life, emotional regulation, and side-effect burden (secondary). Participants may then enter a naturalistic telephone follow-up at weeks 12, 16, 24 and 52. The first half of the sample will be used to identify potential predictors and moderators, and the second half to replicate and confirm.DiscussionFirst enrolment was in December 2008, and is ongoing. iSPOT-D evaluates clinical and biological predictors of treatment response in the largest known sample of MDD collected worldwide.Trial registrationInternational Study to Predict Optimised Treatment - in Depression (iSPOT-D) ClinicalTrials.gov Identifier: NCT00693849 URL:http://clinicaltrials.gov/ct2/show/NCT00693849?term=International+Study+to+Predict+Optimized+Treatment+for+Depression&rank=1

Association of Polymorphisms in Genes Regulating the Corticotropin-Releasing Factor System With Antidepressant Treatment Response

CONTEXT The corticotropin-releasing factor (CRF, or corticotropin-releasing hormone) and arginine vasopressin systems have been implicated in the pathophysiology of anxiety and depressive disorders and response to antidepressant treatment. OBJECTIVE To study the association of genetic variants in 10 genes that regulate the CRF and arginine vasopressin systems with treatment response to citalopram in the Sequenced Treatment Alternatives to Relieve Depression (STAR*D) sample (N = 1768). DESIGN Pharmacogenetic association study derived from the STAR*D study, a multicenter, prospective, open, 12-week effectiveness trial. SETTING Outpatient primary care and psychiatric clinics. Patients Individuals with nonpsychotic major depressive disorder for whom DNA was available who were subsequently treated with citalopram hydrobromide for 4 to 12 weeks. Intervention Flexible doses of citalopram. Main Outcome Measure Association of genetic polymorphisms in genes encoding the CRF system with response and remission to citalopram treatment at exit visit. RESULTS One single-nucleotide polymorphism (SNP) (rs10473984) within the CRHBP locus showed a significant association with both remission (P = 6.0 x 10(-6); corrected, P = .0026) and reduction in depressive symptoms (P = 7.0 x 10(-7); corrected, P = .00031) in response to citalopram. The T allele of this SNP was associated with poorer treatment outcome in 2 of the 3 ethnic subsamples (African American and Hispanic), despite large differences in minor allele frequency. This association was more pronounced in patients with features of anxious depression (P = .008). The nonresponse allele was shown to be associated with overall higher plasma corticotropin levels and more pronounced dexamethasone suppression of corticotropin. CONCLUSIONS These data indicate that a genetic variant within the CRHBP locus affects response to citalopram in African American and Hispanic patients, suggesting a role for this gene and for the CRF system in antidepressant treatment response.

Validation of Depression Screening Scales in Patients With CKD

BACKGROUND Depressive symptoms, assessed by using self-report scales, are present at a striking rate of 45% in patients with chronic kidney disease (CKD) at dialysis therapy initiation. These scales may emphasize somatic symptoms of anorexia, sleep disturbance, and fatigue, which may coexist with chronic disease symptoms and lead to overestimation of depression diagnosis. No study has validated these scales in patients with CKD before dialysis therapy initiation. STUDY DESIGN We conducted a diagnostic test study in participants with CKD to investigate the screening characteristics of 2 depression self-report scales against a gold-standard structured psychiatric interview. SETTING & PARTICIPANTS 272 consecutively recruited outpatients with stages 2 to 5 CKD not treated by dialysis were studied. INDEX TESTS The Beck Depression Inventory (BDI) and the 16-item Quick Inventory of Depressive Symptomatology-Self Report (QIDS-SR(16)) depression screening scales were administered to all participants. REFERENCE TEST A structured Diagnostic and Statistical Manual of Mental Disorders (Fourth Edition)-based interview, the Mini International Neuropsychiatric Interview, was administered by trained persons blinded to self-report scale scores. RESULTS 57 of 272 (21%) patients had major depression according to the reference test. The best cutoff scores by means of receiver/responder operating characteristic curves to identify a major depressive episode were 11 for the BDI and 10 for the QIDS-SR(16). Sensitivities were 89% (95% confidence interval [CI], 78 to 96; BDI) and 91% (95% CI, 80 to 97; QIDS-SR(16)), whereas specificities were 88% (95% CI, 83 to 92; BDI) and 88% (95% CI, 83 to 92; QIDS-SR(16)). The positive and negative likelihood ratios for these cutoff scores were 7.6 and 0.1 (BDI) and 7.5 and 0.1 (QIDS-SR(16)). LIMITATIONS Single-center study and a sample not representative of US demographics. CONCLUSIONS We found that a BDI score of 11 or higher was a sensitive and specific cutoff value for identifying a major depressive episode in patients with CKD not on dialysis therapy. Both the BDI and QIDS-SR(16) are effective screening tools.

Measurement-Based Care in Psychiatric Practice: A Policy Framework for Implementation

This article describes the need for measurement-based care (MBC) in psychiatric practice and defines a policy framework for implementation. Although measurement in psychiatric treatment is not new, it is not standard clinical practice. Thus a gap exists between research and practice outcomes. The current standards of psychiatric clinical care are reviewed and illustrated by a case example, along with MBC improvements. Measurement-based care is defined for clinical practice along with limitations and recommendations. This article provides a policy top 10 list for implementing MBC into standard practice, including establishing clear expectations and guidelines, fostering practice-based implementation capacities, altering financial incentives, helping practicing doctors adapt to MBC, developing and expanding the MBC science base, and engaging consumers and their families. Measurement-based care as the standard of care could transform psychiatric practice, move psychiatry into the mainstream of medicine, and improve the quality of care for patients with psychiatric illness.

Pharmacogenetics Studies in STAR*D: Strengths, Limitations, and Results

Several lines of evidence support an important genetic contribution to the wide individual variation in therapeutic response to antidepressant medications. The Sequenced Treatment Alternatives to Relieve Depression (STAR*D) study provided the largest cohort assembled to date of DNA from patients with nonpsychotic major depressive disorder, uniformly treated with citalopram and followed prospectively for up to 12 weeks. This pivotal study changed the face of pharmacogenetics research by increasing the sample size by an order of magnitude as well as by providing detailed prospective information about antidepressant response and tolerability. Several groups have identified markers in genes and tested the replication of previous findings of genes associated with outcome and side effects of antidepressant treatment. Variants in HTR2A, GRIK4, and KCNK2 were associated with citalopram treatment outcome. Replication was achieved in markers in the FKBP5 gene. Other findings in PDE11A and BDNF were not successfully replicated, and reports of potential confounders in previous associations with serotonin transporter variation (SLC6A4) were identified. Polymorphisms in pharmacokinetic genes involved in metabolism and transmembrane transport were also not associated with antidepressant response. Adverse events were also tested. Treatment-emergent suicidal ideation was associated with GRIK2, GRIA3, PAPLN, IL28RA, and CREB1. Sexual dysfunction was linked with variation in GRIN3A, GRIA1 GRIA3, and GRIK2. Reported and future findings of pharmacogenetics studies in STAR*D could help elucidate pathways involved in major depression and those pertinent to antidepressant outcome and side effects. Replication of these findings in independent samples could lead to the development of new treatments and to optimization of available treatments.