LocalControl: An R Package for Comparative Safety and Effectiveness Research.

The LocalControl R package implements novel approaches to address biases and confounding when comparing treatments or exposures in observational studies of outcomes. While designed and appropriate for use in comparative safety and effectiveness research involving medicine and the life sciences, the package can be used in other situations involving outcomes with multiple confounders. LocalControl is an open-source tool for researchers whose aim is to generate high quality evidence using observational data. The package implements a family of methods for non-parametric bias correction when comparing treatments in observational studies, including survival analysis settings, where competing risks and/or censoring may be present. The approach extends to bias-corrected personalized predictions of treatment outcome differences, and analysis of heterogeneity of treatment effect-sizes across patient subgroups.

Systemic challenges in bipolar disorder management: A patient-centered approach.

OBJECTIVES: As part of a series of Patient-Centered Outcomes Research Institute-funded large-scale retrospective observational studies on bipolar disorder (BD) treatments and outcomes, we sought the input of patients with BD and their family members to develop research questions. We aimed to identify systemic root causes of patient-reported challenges with BD management in order to guide subsequent studies and initiatives. METHODS: Three focus groups were conducted where patients and their family members (total n = 34) formulated questions around the central theme, "What do you wish you had known in advance or over the course of treatment for BD?" In an affinity mapping exercise, participants clustered their questions and ranked the resulting categories by importance. The research team and members of our patient partner advisory council further rated the questions by expected impact on patients. Using a Theory of Constraints systems thinking approach, several causal models of BD management challenges and their potential solution were developed with patients using the focus group data. RESULTS: A total of 369 research questions were mapped to 33 categories revealing 10 broad themes. The top priorities for patient stakeholders involved pharmacotherapy and treatment alternatives. Analysis of causal relationships underlying 47 patient concerns revealed two core conflicts: for patients, whether or not to take pharmacotherapy, and for mental health services, the dilemma of care quality vs quantity. CONCLUSIONS: To alleviate the core conflicts identified, BD management requires a coordinated multidisciplinary approach including: improved access to mental health services, objective diagnostics, sufficient provider visit time, evidence-based individualized treatment, and psychosocial support.

Local Control Strategy: Simple Analyses of Air Pollution Data Can Reveal Heterogeneity in Longevity Outcomes.

Claims from observational studies that use traditional model specification searches often fail to replicate, partially because the available data tend to be biased. There is an urgent need for an alternative statistical analysis strategy, that is not only simple and easily understood but also is more likely to give reliable insights when the available data have not been designed and balanced. The alternative strategy known as local control first generates local, nonparametric effect-size estimates (fair treatment comparisons) and only then asks whether the observed variation in these local estimates can be predicted from potential confounding factors. Here, we illustrate application of local control to a historical air pollution data set describing a "natural experiment" initiated by the federal Clean Air Act Amendments of 1970. Our reanalysis reveals subgroup heterogeneity in the effects of air quality regulation on elderly longevity (one size does not fit all), and we show that this heterogeneity is largely explained by socioeconomic and environmental confounders other than air quality.

Hypothyroidism risk compared among nine common bipolar disorder therapies in a large US cohort.

OBJECTIVES: Thyroid abnormalities in patients with bipolar disorder (BD) have been linked to lithium treatment for decades, yet other drugs have been less well studied. Our objective was to compare hypothyroidism risk for lithium versus the anticonvulsants and second-generation antipsychotics commonly prescribed for BD. METHODS: Administrative claims data on 24,574 patients with BD were analyzed with competing risk survival analysis. Inclusion criteria were (i) one year of no prior hypothyroid diagnosis nor BD drug treatment, (ii) followed by at least one thyroid test during BD monotherapy on lithium carbonate, mood-stabilizing anticonvulsants (lamotrigine, valproate, oxcarbazepine, or carbamazepine) or antipsychotics (aripiprazole, olanzapine, risperidone, or quetiapine). The outcome was cumulative incidence of hypothyroidism per drug, in the presence of the competing risk of ending monotherapy, adjusted for age, sex, physician visits, and thyroid tests. RESULTS: Adjusting for covariates, the four-year cumulative risk of hypothyroidism for lithium (8.8%) was 1.39-fold that of the lowest risk therapy, oxcarbazepine (6.3%). Lithium was non-statistically significantly different from quetiapine. While lithium conferred a higher risk when compared to all other treatments combined as a group, hypothyroidism risk error bars overlapped for all drugs. Treatment (p = 3.86e-3), age (p = 6.91e-10), sex (p = 3.93e-7), and thyroid testing (p = 2.79e-87) affected risk. Patients taking lithium were tested for hypothyroidism 2.26-3.05 times more frequently than those on other treatments. CONCLUSIONS: Thyroid abnormalities occur frequently in patients with BD regardless of treatment. Therefore, patients should be regularly tested for clinical or subclinical thyroid abnormalities on all therapies and treated as indicated to prevent adverse effects of hormone imbalances on mood.

Item response analysis of the Positive and Negative Syndrome Scale.

BACKGROUND: Statistical models based on item response theory were used to examine (a) the performance of individual Positive and Negative Syndrome Scale (PANSS) items and their options, (b) the effectiveness of various subscales to discriminate among individual differences in symptom severity, and (c) the appropriateness of cutoff scores recently recommended by Andreasen and her colleagues (2005) to establish symptom remission. METHODS: Option characteristic curves were estimated using a nonparametric item response model to examine the probability of endorsing each of 7 options within each of 30 PANSS items as a function of standardized, overall symptom severity. Our data were baseline PANSS scores from 9205 patients with schizophrenia or schizoaffective disorder who were enrolled between 1995 and 2003 in either a large, naturalistic, observational study or else in 1 of 12 randomized, double-blind, clinical trials comparing olanzapine to other antipsychotic drugs. RESULTS: Our analyses show that the majority of items forming the Positive and Negative subscales of the PANSS perform very well. We also identified key areas for improvement or revision in items and options within the General Psychopathology subscale. The Positive and Negative subscale scores are not only more discriminating of individual differences in symptom severity than the General Psychopathology subscale score, but are also more efficient on average than the 30-item total score. Of the 8 items recently recommended to establish symptom remission, 1 performed markedly different from the 7 others and should either be deleted or rescored requiring that patients achieve a lower score of 2 (rather than 3) to signal remission. CONCLUSION: This first item response analysis of the PANSS supports its sound psychometric properties; most PANSS items were either very good or good at assessing overall severity of illness. These analyses did identify some items which might be further improved for measuring individual severity differences or for defining remission thresholds. Findings also suggest that the Positive and Negative subscales are more sensitive to change than the PANSS total score and, thus, may constitute a "mini PANSS" that may be more reliable, require shorter administration and training time, and possibly reduce sample sizes needed for future research.

Higher costs and therapeutic factors associated with adherence to NCQA HEDIS antidepressant medication management measures: analysis of administrative claims.

OBJECTIVE: To determine if the type of antidepressant drug is related to adherence to National Committee for Quality Assurance (NCQA) Antidepressant Medication Management (AMM) quality measures and to assess the 6-month health care costs among newly diagnosed depressed patients. METHODS: The MarketScan Commercial Claims and Encounter database for medical and pharmacy claims from January 2001 to September 2004 was used to assess adherence to the 3 AMM quality-of-care measures. AMM measures include (a) acute phase, the percentage of eligible members who remained on antidepressant medication continuously for 3 months after the initial diagnosis as determined by at least 84 days supply of antidepressant drugs during the first 114 days following receipt of the index antidepressant; (b) continuation phase, the percentage of eligible members who remained on antidepressant medication continuously for the 6 months after the initial diagnosis as determined by at least 180 days supply of antidepressants during the first 214 days following receipt of the index antidepressant; and (c) practitioner contacts, the percentage of members who received at least 3 follow-up office visits or telephone contacts with health care providers, including at least 1 contact with a practitioner licensed to prescribe (may not necessarily be the prescriber of the antidepressant). A fourth measure, overall adherence, was added, if all 3 AMM measures were met. Multivariate regression models determined demographic, clinical (such as receipt of mental health specialty care, the Charlson Comorbidity Index score, and co-occurring bipolar or schizophrenia), and therapy-related factors associated with outcomes of adherence and costs (paid amounts for insurance-reimbursable health care services for inpatient admissions, emergency department services, outpatient services, and outpatient prescription drugs). Health care expenditures (both total and mental-health-specific costs) were measured for each patient for 6 months following the date of service for the index antidepressant. RESULTS: A total of 60,386 adult patients (10.7%) of 562,898 patients with a depression diagnosis met NCQA inclusion criteria in the AMM Technical Specifications (e.g., aged 18 years or older, newly diagnosed with depression and initiating antidepressant therapy, 365 days of continuous enrollment; patients were excluded if there were missing data on dose or quantity of index drug in pharmacy claims or initiated therapy on 2 or more antidepressants as the index medication, exclusion criteria not in the AMM Technical Specifications). Only 19% of patients achieved overall adherence. Rates for the 3 AMM measures were 39% for practitioner contacts, 65% for acute phase, and 44% for continuation phase. Receipt of mental health specialty care was the only factor that was positively associated with greater adherence on all 4 measures (overall measure: odds ratio [OR]=3.895, 95% confidence interval [CI], 3.72-4.07; acute OR=1.38, 95% CI, 1.33-1.43; continuation OR=1.46, 95% CI, 1.41-1.51; contacts OR=5.83, 95% CI, 5.62-6.06). Most patients were initiated on selective serotonin reuptake inhibitors (SSRIs, 69.5%), followed by venlafaxine (21.4%), tricyclic antidepressants (TCAs, 21.4%), bupropion (11.0%), and other antidepressants (e.g., mirtazapine, nefazadone, trazadone; 7.2%). Before adjustment for confounding factors, patients initiated on venlafaxine, TCAs, or other antidepressants had higher rates of adherence on the overall performance measure versus initiators on SSRIs, but the absolute differences were relatively small: 21.4% for venlafaxine and TCAs and 23.1% for other antidepressants versus 18.5% for SSRIs (P <0.001). Patients initiated on venlafaxine, TCAs, or other antidepressants were also more likely to receive care from a mental health specialist, 16.8%, 15.0%, and 54.8%, respectively, compared with SSRIs (13.0%, all P <0.001). Regression analysis showed that only venlafaxine had a higher OR (1.13; 95% CI, 1.05-1.22) compared with SSRIs for adherence on the overall measure. Initiating dose level was in the target range for 70.0% of all patients (24.9% were below target dose and 5.2% above target dose), and adherent patients on all 3 AMM measures were less likely than nonadherent patients (70.4% vs. 68.4%, P <0.001) to be initiated in the target dose range. After multivariate adjustment, the initiating dose (target vs. high) was a significant factor in explaining adherence to the overall measure (OR=1.26; 95% CI, 1.16- 1.37). Adherent patients had 6-month median unadjusted total health care expenses that were nearly 2 times higher compared with nonadherent patients ($5,169 vs. $2,734) and mental health expenditures that were nearly 3 times higher ($1,922 vs. $677). After adjustment, adherent patients compared with nonadherent patients incurred an additional $644 in mental health expenditures and $806 in overall health care expenditures in the 6 months following initiation of antidepressant therapy. CONCLUSIONS: Only 19% of depressed patients initiated on antidepressants met all 3 criteria set forth in the NCQA Health Plan Employer Data and Information Set (HEDIS) AMM quality-of-care performance measures. Receipt of mental health specialty care was the single factor most strongly associated with quality treatment by these measures. Type and dosage level of initial antidepressant was associated with adherence to the NCQA HEDIS AMM measures, but the absolute difference in rates of adherence were relatively small among types of antidepressants. Costs were higher for guideline-adherent individuals in the 6 months following treatment initiation. These analyses were limited to administrative claims that lack indicators of depression disease severity.

Comparison of costs and utilization between users of insulin lispro versus users of regular insulin in a managed care setting.

OBJECTIVE: To compare medical and pharmacy costs and utilization between patients with diabetes who received insulin lispro versus regular human insulin. METHODS: A retrospective analysis of medical and pharmacy claims was conducted among continuously enrolled users of insulin lispro or regular insulin during the identification period, March 1, 2000, through February 28, 2001, within a large managed care organization. This study improved upon the methodology used in previous studies by (a) stratifying (rather than 1:1 matching) individuals by their likelihood to use insulin lispro using the propensity score binning technique, and (b) refining the study inclusion criteria to include only patients with 3 or more fills of the insulin under study (lispro or regular) to exclude individuals who may have been on either product for a short time. Because the propensity score binning technique groups patients with similar baseline characteristics within strata (bins) and not among individual patients, almost the entire available sample is retained in the analysis, unlike propensity score matching, where large numbers of patients can be excluded depending on the matching scheme. Therefore, the propensity score binning technique, because it uses more complete information, is less likely to produce biased results. Patients were grouped into 5 bins (quintiles) based on their estimated likelihood to receive insulin lispro rather than regular insulin. The propensity score model used baseline characteristics of age, gender, comorbidities, use of oral antidiabetic medications, prescription copayment, and diabetes-related costs and utilization. Overall cost and utilization differences (lispro minus regular insulin) during the 12-month follow-up period were calculated using weights inversely proportional to variances of within-bin differences. RESULTS: Of 6,436 patients, 1,972 (30.6%) received insulin lispro and 4,464 (69.4%) received regular insulin. The propensity score estimation produced 5 bins, each containing between 1,287 and 1,288 patients, utilizing all patients in the analysis. Patients in the lower-numbered propensity score quintiles were older, more likely to use oral antidiabetic medications, and had more comorbidities than those in the higher-numbered quintiles. As quintile number increased, the percentage of insulin lispro users also increased. The weighted mean annual cost difference (lispro minus regular insulin) per patient was + USD 79 (P < 0.001) for diabetes-related pharmacy cost, + USD 212 (P < 0.001) for total pharmacy cost, USD 75 (P < 0.857) for diabetes-related medical cost, USD 2,286 (P <0.011) for nondiabetes medical cost, and USD 2,327 (P = 0.072) for total medical cost. CONCLUSIONS: Compared with regular insulin users, insulin lispro users incurred higher diabetes-related and total pharmacy costs but lower nondiabetes medical costs and similar total medical costs. Fewer hospitalizations among insulin lispro as compared with regular insulin users contributed to lower nondiabetes medical costs and similar total medical costs.

Appropriateness of prescribing practices for serotonergic antidepressants.

OBJECTIVE: Data from prescribing physicians were used to assess whether serotonergic antidepressants were used for appropriate indications and at appropriate initial dosages. METHODS: Data were derived from the confidential logs of psychiatrists and primary care physicians who provided prescription information from January 1, 1997, through June 30, 1999, as part of the National Disease and Therapeutic Index physician survey. The survey is not affiliated with a reimbursement system and therefore minimizes bias related to reimbursement. Data on the primary reason for use and the dosage at the time of first use were obtained for prescriptions of citalopram, fluoxetine, paroxetine, sertraline, venlafaxine, and extended-release venlafaxine. RESULTS: Depressive disorders accounted for the majority of the 3,206 prescriptions for the six antidepressants (74 percent to 86.2 percent). The next most common indications for use were anxiety (4.1 percent to 12.6 percent) and obsessive-compulsive disorder (1.3 percent to 3.3 percent). For patients with depressive disorders, psychiatrists prescribed slightly higher antidepressant dosages than primary care physicians. CONCLUSIONS: Serotonergic antidepressants are used primarily for the treatment of depression and depression-related disorders and are prescribed at the recommended starting dosages.

Cost and utilization comparisons among propensity score-matched insulin lispro and regular insulin users.

OBJECTIVE: To compare cost and utilization among users of insulin lispro and regular (human) insulin. METHODS: This was a retrospective analysis using administrative claims data for continuously enrolled subjects using insulin lispro or regular insulin between January 1, 1998, and December 21, 1999. Subjects were matched 1 to 1 on the propensity to receive lispro versus regular insulin using a score estimated from baseline characteristics such as age, gender, comorbidities, and oral hypoglycemic use. Once matched, 12 months of follow-up pharmacy and medical cost and utilization data (e.g. prescriptions, office visits, hospitalizations) from July 1,1997, through December 31, 2000, were analyzed using univariate statistics. RESULTS: Of 11,443 subjects, 3,341 (29.2%) had received a prescription for insulin lispro, while 8,102 (70.8%) had received a prescription for regular insulin. At baseline, lispro subjects tended to be younger, more often had type 1 diabetes and a history of insulin use, had fewer comorbidities, visited endocrinologists more often than family practice physicians, and had lower total costs. After matching on propensity score to within +/-0.01, 1,832 subject pairs were retained. On average, lispro subjects had significantly more office visits (P=0.0022) and pharmacy prescriptions (P=0.0165) but fewer inpatient hospital visits (P=0.0028)compared to regular insulin subjects. Cost results were similar, with insulin lispro subjects having significantly higher average office visit costs (P=0.0237) and pharmacy costs (P<0.0001) but lower inpatient hospital costs (P=0.0227). Total costs were not significantly different between treatment groups (P=0.5266). CONCLUSION: Total direct health care costs were not different between insulin lispro and regular insulin users. An association was observed between higher direct drug product cost and more intensive ambulatory care for insulin lispro users and lower inpatient hospital cost in the short-term.

Antidepressant treatment for depression: total charges and therapy duration*

BACKGROUND: The economic costs of depression are significant, both the direct medical costs of care and the indirect costs of lost productivity. Empirical studies of antidepressant cost-effectiveness suggest that the use of selective serotonin reuptake inhibitors (SSRIs) may be no more costly than tricyclic antidepressants (TCAs), will improve tolerability, and is associated with longer therapy duration. However the success of depression care usually involves multiple factors, including source of care, type of care, and patient characteristics, in addition to drug choice. The cost-effective mix of antidepressant therapy components is unclear. AIMS OF THE STUDY: Our study evaluates cost and antidepressant-continuity outcomes for depressed patients receiving antidepressant therapy. Specifically, we determined the impact of provider choice for initial care, concurrent psychotherapy, and choice of SSRI versus TCA-based pharmacotherapies on the joint outcome of low treatment cost and continuous antidepressant therapy. METHODS: A database of private health insurance claims identifies 2678 patients who received both a diagnosis of depression and a prescription for an antidepressant during 1990-1994. Patients each fall into one of four groups according to whether their health care charges are high versus low (using the median value as the break point) and by whether their antidepressant usage pattern is continuous versus having discontinued pharmacotherapy early (filling fewer than six prescriptions). A bivariate probit model controlling for patient characteristics, co-morbidities, type of depression and concurrent treatment is the primary multivariate statistical vehicle for the cost-effective treatment situation. RESULTS: SSRIs substantially reduce the incidence of patients discontinuing pharmacotherapy while leaving charges largely unchanged. The relative effectiveness of SSRIs in depression treatment is independent of the patient's personal characteristics and dominates the consequences of other treatment dimensions such as seeing a mental health specialist and receiving concurrent psychotherapy. Initial provider specialty is irrelevant to the continuity of pharmacotherapy, and concurrent psychotherapy creates a tradeoff through reduced pharmacotherapy interruption with higher costs. DISCUSSION: Longer therapy duration is associated with SSRI-based pharmacotherapy (relative to TCA-based pharmacotherapy) and with concurrent psychotherapy. High cost is associated with concurrent psychotherapy and choice of a specialty provider for initial care. In our study cost-effective care includes SSRI-based pharmacotherapy initiated with a non-specialty provider. Previous treatment history and other unobserved factors that might affect antidepressant choice are not included in our model. IMPLICATIONS FOR HEALTH CARE PROVISION: The decision to use an SSRI-based pharmacotherapy need not consider carefully the patient's personal characteristics. Shifting depressed patients' pharmacotherapy away from TCAs to SSRIs has the effect of improving outcomes by lowering the incidence of discontinuation of pharmacotherapy while leaving largely unchanged the likelihood of having high overall health care charges. Targeted use of concurrent psychotherapy may be additionally cost-effective. IMPLICATIONS FOR HEALTH POLICIES: The interaction of various components of depression care can alter the cost-effectiveness of antidepressant therapy. Our results demonstrate a role for the non-specialty provider in initiating care and support increased use of SSRIs as first-line therapy for depression as a way of providing cost-effective care that is consistent with APA guidelines for continuous antidepressant treatment. IMPLICATIONS FOR FURTHER RESEARCH: Further research that improves our understanding of how decisions regarding provider choice, concurrent psychotherapy, and drug choice are made will improve our understanding of the effects treatment choices on the cost-effectiveness of depression care. We have suggested that targeted concurrent psychotherapy may prove to be cost-effective; research to determine groups most likely to benefit from the additional treatment would further enable clinicians and healthcare policy makers to form a consensus regarding a model for treating depression.

Local control for identifying subgroups of interest in observational research: persistence of treatment for major depressive disorder.

Caregivers are regularly faced with decisions between competing treatments. Large observational health care databases provide a golden opportunity for research on heterogeneity in patient response to guide caregiver decisions, due to their sample size, diverse populations, and real-world setting. Local control is a promising tool for using observational data to detect patient subgroups with differential response on one treatment relative to another. While standard data mining approaches find subgroups with optimal responses for a particular population, detecting subgroups that reveal treatment differences while also adjusting for confounding in observational data is challenging. Local control utilizes unsupervised clustering to form non-parametric patient-level counterfactual treatment differences and displays them as an observed distribution of effect-size estimates. Classification and regression trees (CART) then find the factors that drive the greatest outcome differentiation between treatments. In this manuscript, we demonstrate the use of this two-step strategy using local control plus CART to identify depression patients most (least) likely to benefit from treatment with duloxetine relative to extended-release venlafaxine. Prior medication costs and age were found to be factors most associated with differential outcome, with prior medication costs remaining as an important factor after sensitivity analyses using a second dataset.

Cost-effectiveness inferences from bootstrap quadrant confidence levels: three degrees of dominance.

When with at least 95% confidence a new treatment is shown to be not only less costly (LC), but also more effective (ME), than a current treatment, that new treatment can be said to "strictly dominate" the current treatment statistically. But what can be said when head-to-head treatment comparisons turn out to be less clear-cut than this? Here, we propose two additional sets of specific LC and/or ME confidence thresholds to define the concepts of "some dominance" and "much dominance." Confidence levels associated with entire quadrants of the incremental cost-effectiveness (ICE) plane are easily computed using the same bootstrapping techniques used to estimate an "acceptability curve." Our two proposed additional "degrees" of dominance, although less stringent than strict dominance, are nevertheless more stringent than commonly accepted approaches using ICE ratio or net benefit calculations. To illustrate analysis concepts, we use data from a randomized, double-blind, placebo- and active comparator-controlled clinical registration trial for treatment of major depressive disorder (MDD). As is typical, our case study is rather small and short term, providing outcome information for a total of only 264 patients during their initial 8 weeks of acute-phase MDD treatment. Thus, we focus attention on sensitivity analyses, showing that the bootstrap distribution of cost-effectiveness uncertainty is robust across two alternative ways of measuring overall effectiveness and three alternative ways of imputing missing values. Evaluation of the balance between cost and benefit is particularly difficult when a new pharmacological treatment is first introduced, yet information of this sort is highly desired by decision makers. We show that, even with only a relatively modest amount of clinical trial information, sensitivity analyses can still confirm that cost-effectiveness comparisons are being made in a consistent fashion. In contrast, extensive follow-up comparisons using data from actual clinical practice will almost always ultimately be needed to better inform health policy makers.