Article

Use of Instrumental Variables in the Presence of Heterogeneity and Self-Selection: An Application to Treatments of Breast Cancer Patients

November 2007
Health Economics 16(11):1133-57

November 2007
16(11):1133-57

DOI:10.1002/hec.1291

Source
PubMed

Authors:

Salvador Navarro

The University of Western Ontario

Instrumental variable (IV) methods are widely used in the health economics literature to adjust for hidden selection biases in observational studies when estimating treatment effects. Less attention has been paid in the applied literature to the proper use of IVs if treatment effects are heterogeneous across subjects and individuals select treatments based on expected idiosyncratic gains or losses from treatments. In this paper we compare conventional IV analysis with alternative approaches that use IVs to estimate treatment effects in models with response heterogeneity and self-selection. Instead of interpreting IV estimates as the effect of treatment at an unknown margin of patients, we identify the marginal patients and we apply the method of local IVs to estimate the average treatment effect and the effect on the treated on 5-year direct costs of breast-conserving surgery and radiation therapy compared with mastectomy in breast cancer patients. We use a sample from the Outcomes and Preferences in Older Women, Nationwide Survey which is designed to be representative of all female Medicare beneficiaries (aged 67 or older) with newly diagnosed breast cancer between 1992 and 1994. Our results reveal some of the advantages and limitations of conventional and alternative IV methods in estimating mean treatment effect parameters.

Assessing the properties of patient-specific treatment effect estimates from causal forest algorithms under essential heterogeneity

Article

Full-text available

Mar 2024
BMC MED RES METHODOL

Background Treatment variation from observational data has been used to estimate patient-specific treatment effects. Causal Forest Algorithms (CFAs) developed for this task have unknown properties when treatment effect heterogeneity from unmeasured patient factors influences treatment choice – essential heterogeneity. Methods We simulated eleven populations with identical treatment effect distributions based on patient factors. The populations varied in the extent that treatment effect heterogeneity influenced treatment choice. We used the generalized random forest application (CFA-GRF) to estimate patient-specific treatment effects for each population. Average differences between true and estimated effects for patient subsets were evaluated. Results CFA-GRF performed well across the population when treatment effect heterogeneity did not influence treatment choice. Under essential heterogeneity, however, CFA-GRF yielded treatment effect estimates that reflected true treatment effects only for treated patients and were on average greater than true treatment effects for untreated patients. Conclusions Patient-specific estimates produced by CFAs are sensitive to why patients in real-world practice make different treatment choices. Researchers using CFAs should develop conceptual frameworks of treatment choice prior to estimation to guide estimate interpretation ex post.

Remission and Relapse of Dyslipidemia After Vertical Sleeve Gastrectomy vs Roux-en-Y Gastric Bypass in a Racially and Ethnically Diverse Population

Article

Full-text available

Sep 2022

Importance: The comparative effectiveness of the most common operations in the long-term management of dyslipidemia is not clear. Objective: To compare 4-year outcomes associated with vertical sleeve gastrectomy (VSG) vs Roux-en-Y gastric bypass (RYGB) for remission and relapse of dyslipidemia. Design, setting, and participants: This retrospective comparative effectiveness study was conducted from January 1, 2009, to December 31, 2016, with follow-up until December 31, 2018. Participants included patients with dyslipidemia at the time of surgery who underwent VSG (4142 patients) or RYGB (2853 patients). Patients were part of a large integrated health care system in Southern California. Analysis was conducted from January 1, 2018, to December 31, 2021. Exposures: RYGB and VSG. Main outcomes and measures: Dyslipidemia remission and relapse were assessed in each year of follow-up for as long as 4 years after surgery. Results: A total of 8265 patients were included, with a mean (SD) age of 46 (11) years; 6591 (79.8%) were women, 3545 (42.9%) were Hispanic, 1468 (17.8%) were non-Hispanic Black, 2985 (36.1%) were non-Hispanic White, 267 (3.2%) were of other non-Hispanic race, and the mean (SD) body mass index (calculated as weight in kilograms divided by height in meters squared) was 44 (7) at the time of surgery. Dyslipidemia outcomes at 4 years were ascertained for 2168 patients (75.9%) undergoing RYGB and 3999 (73.9%) undergoing VSG. Remission was significantly higher for those who underwent RYGB (824 [38.0%]) compared with VSG (1120 [28.0%]) (difference in the probability of remission, 0.10; 95% CI, 0.01-0.19), with no differences in relapse (455 [21.0%] vs 960 [24.0%]). Without accounting for relapse, remission of dyslipidemia after 4 years was 58.9% (1279) for those who underwent RYGB and 51.9% (2079) for those who underwent VSG. Four-year differences between operations were most pronounced for patients 65 years or older (0.39; 95% CI, 0.27-0.51), those with cardiovascular disease (0.43; 95% CI, 0.24-0.62), or non-Hispanic Black patients (0.13; 95% CI, 0.01-0.25) and White patients (0.13; 95% CI, 0.03-0.22). Conclusions and relevance: In this large, racially and ethnically diverse cohort of patients who underwent bariatric and metabolic surgery in clinical practices, RYGB was associated with higher rates of dyslipidemia remission after 4 years compared with VSG. However, almost one-quarter of all patients experienced relapse, suggesting that patients should be monitored closely throughout their postoperative course to maximize the benefits of these operations for treatment of dyslipidemia.

Assessing the ability of an instrumental variable causal forest algorithm to personalize treatment evidence using observational data: the case of early surgery for shoulder fracture

Article

Full-text available

Jul 2022
BMC MED RES METHODOL

Background Comparative effectiveness research (CER) using observational databases has been suggested to obtain personalized evidence of treatment effectiveness. Inferential difficulties remain using traditional CER approaches especially related to designating patients to reference classes a priori. A novel Instrumental Variable Causal Forest Algorithm (IV-CFA) has the potential to provide personalized evidence using observational data without designating reference classes a priori, but the consistency of the evidence when varying key algorithm parameters remains unclear. We investigated the consistency of IV-CFA estimates through application to a database of Medicare beneficiaries with proximal humerus fractures (PHFs) that previously revealed heterogeneity in the effects of early surgery using instrumental variable estimators. Methods IV-CFA was used to estimate patient-specific early surgery effects on both beneficial and detrimental outcomes using different combinations of algorithm parameters and estimate variation was assessed for a population of 72,751 fee-for-service Medicare beneficiaries with PHFs in 2011. Classification and regression trees (CART) were applied to these estimates to create ex-post reference classes and the consistency of these classes were assessed. Two-stage least squares (2SLS) estimators were applied to representative ex-post reference classes to scrutinize the estimates relative to known 2SLS properties. Results IV-CFA uncovered substantial early surgery effect heterogeneity across PHF patients, but estimates for individual patients varied with algorithm parameters. CART applied to these estimates revealed ex-post reference classes consistent across algorithm parameters. 2SLS estimates showed that ex-post reference classes containing older, frailer patients with more comorbidities, and lower utilizers of healthcare were less likely to benefit and more likely to have detriments from higher rates of early surgery. Conclusions IV-CFA provides an illuminating method to uncover ex-post reference classes of patients based on treatment effects using observational data with a strong instrumental variable. Interpretation of treatment effect estimates within each ex-post reference class using traditional CER methods remains conditional on the extent of measured information in the data.

Local Instrumental Variable Methods to Address Confounding and Heterogeneity when Using Electronic Health Records: An Application to Emergency Surgery

Article

Full-text available

May 2022
MED DECIS MAKING

Background: Electronic health records (EHRs) offer opportunities for comparative effectiveness research to inform decision making. However, to provide useful evidence, these studies must address confounding and treatment effect heterogeneity according to unmeasured prognostic factors. Local instrumental variable (LIV) methods can help studies address these challenges, but have yet to be applied to EHR data. This article critically examines a LIV approach to evaluate the cost-effectiveness of emergency surgery (ES) for common acute conditions from EHRs. Methods: This article uses hospital episodes statistics (HES) data for emergency hospital admissions with acute appendicitis, diverticular disease, and abdominal wall hernia to 175 acute hospitals in England from 2010 to 2019. For each emergency admission, the instrumental variable for ES receipt was each hospital's ES rate in the year preceding the emergency admission. The LIV approach provided individual-level estimates of the incremental quality-adjusted life-years, costs and net monetary benefit of ES, which were aggregated to the overall population and subpopulations of interest, and contrasted with those from traditional IV and risk-adjustment approaches. Results: The study included 268,144 (appendicitis), 138,869 (diverticular disease), and 106,432 (hernia) patients. The instrument was found to be strong and to minimize covariate imbalance. For diverticular disease, the results differed by method; although the traditional approaches reported that, overall, ES was not cost-effective, the LIV approach reported that ES was cost-effective but with wide statistical uncertainty. For all 3 conditions, the LIV approach found heterogeneity in the cost-effectiveness estimates across population subgroups: in particular, ES was not cost-effective for patients with severe levels of frailty. Conclusions: EHRs can be combined with LIV methods to provide evidence on the cost-effectiveness of routinely provided interventions, while fully recognizing heterogeneity. Highlights: This article addresses the confounding and heterogeneity that arise when assessing the comparative effectiveness from electronic health records (EHR) data, by applying a local instrumental variable (LIV) approach to evaluate the cost-effectiveness of emergency surgery (ES) versus alternative strategies, for patients with common acute conditions (appendicitis, diverticular disease, and abdominal wall hernia).The instrumental variable, the hospital's tendency to operate, was found to be strongly associated with ES receipt and to minimize imbalances in baseline characteristics between the comparison groups.The LIV approach found that, for each condition, there was heterogeneity in the estimates of cost-effectiveness according to baseline characteristics.The study illustrates how an LIV approach can be applied to EHR data to provide cost-effectiveness estimates that recognize heterogeneity and can be used to inform decision making as well as to generate hypotheses for further research.

Personalizing Evidence for Shoulder Fracture Patients Using an Extended Instrumental Variable Causal Forest Algorithm

Preprint

Oct 2021

Objective: To assess the ability of an extended Instrumental Variable Causal Forest Algorithm (IV-CFA) to provide personalized evidence of early surgery effects on benefits and detriments for elderly shoulder fracture patients. Data Sources/Study Setting: Population of 72,751 fee-for-service Medicare beneficiaries with proximal humerus fractures (PHFs) in 2011 who survived a 60-day treatment window after an index PHF and were continuously Medicare fee-for-service eligible over the period 12 months prior to index to the minimum of 12 months after index or death. Study Design: IV-CFA estimated early surgery effects on both beneficial and detrimental outcomes for each patient in the study population. Classification and regression trees (CART) were applied to these estimates to create patient reference classes. Two-stage least squares (2SLS) estimators were applied to patients in each reference class to scrutinize the estimates relative to the known 2SLS properties. Principal Findings: This approach uncovered distinct reference classes of elderly PHF patients with respect to early surgery effects on benefit and detriment. Older, frailer patients with more comorbidities, and lower utilizers of healthcare were less likely to gain benefit and more likely to have detriment from early surgery. Reference classes were characterized by the appropriateness of early surgery rates with respect to benefit and detriment. Conclusions: Extended IV-CFA provides an illuminating method to uncover reference classes of patients based on treatment effects using observational data with a strong instrumental variable. This study isolated reference classes of new PHF patients in which changes in early surgery rates would improve patient outcomes. The inability to measure fracture complexity in Medicare claims means providers will need to discuss the appropriateness of these estimates to patients within a reference class in context of this missing information.

Personalizing Evidence for Shoulder Fracture Patients Using an Extended Instrumental Variable Causal Forest Algorithm

Preprint

Sep 2021

Explaining the causal effect of adherence to medication on cholesterol through the marginal patient

Article

Full-text available

Jun 2020
HEALTH ECON

Domenico Depalo

This paper investigates the relation between adherence to prescribed medication and reduction of cholesterol in Italy, taking into account the possible sorting of patients into treatment and the heterogeneity of the effect. As predicted by a theoretical model, I find that patients who benefit most from medication are more likely to adhere to prescribed regime than those who benefit least. These results are used to study the effects of three hypothetical policies that aim at increasing the share of patients adherent to prescribed medication: one policy is directed towards patients, one towards physicians, and one towards both patients and physicians. For each policy I describe the observable characteristics of patients induced into treatment. Although the policy with the highest return is directed towards patients, the policies differ substantially with respect to the population affected. Therefore, a policy with lower return that targets better the desired population may be preferred to the policy with the highest return. The cost-effectiveness of these policies is presented.

Heterogeneity in Marginal Non-Monetary Returns to Higher Education

Article

Feb 2018

In this paper we estimate the effects of college education on cognitive abilities, health, and wages, exploiting exogenous variation in college availability. By means of semiparametric local instrumental variables techniques we estimate marginal treatment effects in an environment of essential heterogeneity. The results suggest positive average effects on cognitive abilities, wages, and physical health. Yet, there is heterogeneity in the effects, which points toward selection into gains. Although the majority of individuals benefits from more education, the average causal effect for individuals with the lowest unobserved desire to study is zero for all outcomes. Mental health effects, however, are absent for the entire population. (JEL: C31, H52, I10, I21)

Nonparametric Bayesian Instrumental Variable Analysis: Evaluating Heterogeneous Effects of Coronary Arterial Access Site Strategies

Article

Nov 2019

Percutaneous coronary interventions (PCIs) are nonsurgical procedures to open blocked blood vessels to the heart, frequently using a catheter to place a stent. The catheter can be inserted into the blood vessels using an artery in the groin or an artery in the wrist. Because clinical trials have indicated that access via the wrist may result in fewer post procedure complications, shortening the length of stay, and ultimately cost less than groin access, adoption of access via the wrist has been encouraged. However, patients treated in usual care are likely to differ from those participating in clinical trials, and there is reason to believe that the effectiveness of wrist access may differ between males and females. Moreover, the choice of artery access strategy is likely to be influenced by patient or physician unmeasured factors. To study the effectiveness of the two artery access site strategies on hospitalization charges, we use data from a state-mandated clinical registry including 7963 patients undergoing PCI. A hierarchical Bayesian likelihood-based instrumental variable analysis under a latent index modeling framework is introduced to jointly model outcomes and treatment status. Our approach accounts for unobserved heterogeneity via a latent factor structure, and permits nonparametric error distributions with Dirichlet process mixture models. Our results demonstrate that artery access in the wrist reduces hospitalization charges compared to access in the groin, with a higher mean reduction for male patients. Supplementary materials for this article are available online.

Nonparametric Bayesian Instrumental Variable Analysis: Evaluating Heterogeneous Effects of Arterial Access Sites for Opening Blocked Blood Vessels

Article

Apr 2018

Percutaneous coronary interventions (PCIs) are nonsurgical procedures to open blocked blood vessels to the heart, frequently using a catheter to place a stent. The catheter can be inserted into the blood vessels using an artery in the groin or an artery in the wrist. Because clinical trials have indicated that access via the wrist may result in fewer post procedure complications, shortening the length of stay, and ultimately cost less than groin access, adoption of access via the wrist has been encouraged. However, patients treated in usual care are likely to differ from those participating in clinical trials, and there is reason to believe that the effectiveness of wrist access may differ between males and females. Moreover, the choice of artery access strategy is likely to be influenced by patient or physician unmeasured factors. To study the effectiveness of the two artery access site strategies on hospitalization charges, we use data from a state-mandated clinical registry including 7,963 patients undergoing PCI. A hierarchical Bayesian likelihood-based instrumental variable analysis under a latent index modeling framework is introduced to jointly model outcomes and treatment status. Our approach accounts for unobserved heterogeneity via a latent factor structure, and permits nonparametric error distributions with Dirichlet process mixture models. Our results demonstrate that artery access in the wrist reduces hospitalization charges compared to access in the groin, with higher mean reduction for male patients.

2SLS Versus 2SRI: Appropriate Methods for Rare Outcomes and/or Rare Exposures

Article

Mar 2018

This study used Monte Carlo simulations to examine the ability of the two‐stage least squares (2SLS) estimator and two‐stage residual inclusion (2SRI) estimators with varying forms of residuals to estimate the local average and population average treatment effect parameters in models with binary outcome, endogenous binary treatment, and single binary instrument. The rarity of the outcome and the treatment was varied across simulation scenarios. Results showed that 2SLS generated consistent estimates of the local average treatment effects (LATE) and biased estimates of the average treatment effects (ATE) across all scenarios. 2SRI approaches, in general, produced biased estimates of both LATE and ATE under all scenarios. 2SRI using generalized residuals minimized the bias in ATE estimates. Use of 2SLS and 2SRI is illustrated in an empirical application estimating the effects of long‐term care insurance on a variety of binary health care utilization outcomes among the near‐elderly using the Health and Retirement Study.

A Machine-Learning Approach for Estimating Subgroup- and Individual-Level Treatment Effects: An Illustration Using the 65 Trial

Article

Full-text available

May 2022
MED DECIS MAKING

Highlights: This article examines a causal machine-learning approach, causal forests (CF), for exploring the heterogeneity of treatment effects, without prespecifying a specific functional form.The CF approach is considered in the reanalysis of the 65 Trial and was found to provide similar estimates of subgroup effects to using a fixed parametric model.The CF approach also provides estimates of individual-level treatment effects that suggest that for most patients in the 65 Trial, the intervention is expected to reduce 90-d mortality but with wide levels of statistical uncertainty.The study illustrates how individual-level treatment effect estimates can be analyzed to generate hypotheses for further research about those patients who are likely to benefit most from an intervention.

A sub-group evaluation of the multi-month dispensing strategy for differentiated HIV care: is personalization of care guidelines warranted in Haiti?

Article

Full-text available

Jan 2022
BMC HEALTH SERV RES

Background Differentiated care strategies are rapidly becoming the norm for HIV care delivery globally. Building upon an interest in tailoring antiretroviral therapy (ART) delivery for client-centered needs, the Ministry of Health and Population in Haiti formally endorsed multiple-month dispenses (MMD) in the 2016 national ART guidelines This study explores heterogeneity in retention in care with MMD for specific Haitian populations living with HIV and evaluates if a targeted algorithm for optimal ART prescription intervals is warranted in Haiti. Methods This study included ART-naïve individuals who started ART on or after January 1st, 2017 in Haiti. To identify subgroups in which to explore heterogeneity of retention, we implemented a double-lasso regression method to determine which individual characteristics would define the subgroups. Characteristics evaluated for potential subgroup definition included: sex, age category, WHO clinical stage, and body mass index category. We employed instrumental variable models to estimate the causal effect of increasing ART dispensing length on ART retention, by client subgroup. The outcome of interest was retention in care after one year in treatment. We then estimated the marginal effect of a 30-day increase to ART dispensing length to retention in care for each of these subgroups. Results There was evidence for heterogeneity in the effect of extending ART dispensing intervals on retention by WHO clinical stage. We observed significant improvements to retention in care at one year with a 30-day increase in ART dispense length for all subgroups defined by WHO clinical stages 1-4. The effects ranged from a 14.7% increase (95% CI: 12.4-17.0) to the likelihood of retention for people with HIV in WHO stage 1 to a 21.6% increase (95% CI: 18.7-24.5) to the likelihood of retention for those in WHO stage 3. Conclusions All the subgroups defined by WHO clinical stage experienced a benefit of extending ART intervals to retention in care at one year. Though the effect did differ slightly by WHO stage, the effects went in the same direction and were of similar magnitude. Therefore, a standardized recommendation for MMD among those living with HIV and new on ART is appropriate for Haiti treatment guidelines.

Instrumental variable approach for estimating a causal hazard ratio: application to the effect of postmastectomy radiotherapy on breast cancer patients

Article

Jan 2019

The use of postmastectomy radiotherapy (PMRT) on women with AJCC (American Joint Committee on Cancer) pT1-2pN1 breast cancer is controversial in practice. Huo et al. (2015) found that PMRT was associated with longer survival among a high-risk subgroup of AJCC pT1-2pN1 patients using a Cox model on data from the National Cancer Database. To address unmeasured confounding in this observational study, we consider the variation among facilities in the use of PMRT as an instrumental variable (IV). Recently, there has been widespread use of the two-stage residual inclusion (2SRI) method offered by Terza et al. (2008) for nonlinear models, and 2SRI has been the method of choice for analyzing proportional hazards model using IV in clinical studies. However, the causal parameter using 2SRI is only identified under a homogeneity assumption that goes beyond the standard assumptions of IV, and Wan et al. (2015) demonstrated that under standard IV assumptions, 2SRI could fail to consistently estimate the causal hazard ratio for compliers. In this paper, following Yu et al. (2015), we apply a model-based IV approach (Imbens and Rubin, 1997; Hirano et al., 2000) which allows consistent estimation of the causal hazard ratio for survival outcomes with a proportional hazards model specification under standard IV assumptions while flexibly incorporating the restrictions imposed by IV assumptions. Simulation studies show that when there is unmeasured confounding, both 2SRI and the standard Cox regression could provide biased estimates of the causal hazard ratio among compliers, while this model-based IV approach provides consistent estimates. We apply this IV method to the breast cancer study and our IV analysis did not find strong evidence to support the benefit of PMRT on survival among the targeted patients. In addition, we develop sensitivity analysis approaches to assess the sensitivity of causal conclusions to violations of the exclusion restrictions assumption for IV.

Moral hazard and selection for voluntary deductibles

Article

Full-text available

Jul 2020
HEALTH ECON

This paper investigates whether the voluntary deductible in the Dutch health insurance system reduces moral hazard or acts only as a cost reduction tool for low‐risk individuals. We use a sample of 14,089 observations, comprising 2,939 individuals over seven waves from the Longitudinal Internet Studies for the Social sciences panel for the analysis. We employ bivariate models that jointly model the choice of a deductible and health care utilization and supplement the identification with an instrumental variable strategy. The results show that the voluntary deductible reduces moral hazard, especially in the decision to visit a doctor (extensive margin) compared with the number of visits (intensive margin). In addition, a robustness test shows that selection on moral hazard is not present in this context.

Evaluating heterogeneity of treatment effects

Article

Jan 2020

Sandeep Vijan

Evaluation of treatment effects in randomized clinical trials typically focuses on the average difference in outcomes between arms of a trial. While this approach is the gold standard for establishing a causal relationship between treatment and outcome, reporting of average effects can mask important differences in benefits across various subpopulations, a phenomenon known as heterogeneity of treatment effects (HTE). The presence of HTE has been demonstrated in many settings and lack of consideration of HTE can lead to inappropriate treatment (or lack of treatment) for many patients. This paper describes approaches to analyzing and reporting trials with explicit consideration of heterogeneity, in order to improve our ability to treat individual patients more effectively.

Effectiveness of adjuvant supportive-expressive group therapy for breast cancer

Article

Full-text available

Feb 2020
BREAST CANCER RES TR

Purpose Randomized control trials exploring adjuvant supportive-expressive group therapy (SEGT) for breast cancer have yielded conflicting survival results. This retrospective cohort study was designed to explore the association of adjuvant SEGT performed at diagnosis with survival in real-world patients. Methods 3327 patients with breast cancer were divided between those who received oncologic treatment combined with SEGT-based intervention (referred to as BRBC [n = 354]) and those who only received oncologic treatment (referred to as OT [n = 2973]). Primary outcome was overall survival (OS) at 1-year, 3-year, 5-year. Propensity score-matched analysis (at a ratio of 1:3) and instrumental variable analysis (IVA) were performed. Results The median overall survival was 7.3 years (95% CI 7.0–7.7 years) in BRBC and 7.1 years (95% CI 6.9–7.4 years) in OT. BRBC was not significantly associated with improved 1-year (HR 0.74, 95% CI 0.49–1.10, P = 0.1748; NNT = 44.8, 95% CI − 118.5 to 22.6), 3-year (HR 0.98, 95% CI 0.75–1.27, P = 0.8640; NNT = 273.7, 95% CI − 21.0 to 21.3), or 5-year survival (HR 0.79, 95% CI 0.61–1.02, P = 0.0908; NNT = 36.0, 95% CI − 384.5 to 19.1) compared with OT. IVA indicated that BRBC had a survival benefit over OT in the 1-year, 3-year, and 5-year of 1.5% (95% CI 1.2–1.9%), 0.7% (95% CI 0.6–0.8%), and 2.6% (95% CI 2.0–3.4%), respectively. Conclusion Adjuvant SEGT cannot significantly prolong 5-year survival in breast cancer, though a longer observation period is warranted according to the marginal survival benefit identified at the end of the follow-up.

The heterogeneous causal effects of neonatal care: a model of endogenous demand for multiple treatment options based on geographical access to care

Article

Full-text available

Nov 2019
HEALTH ECON

Neonatal units in the UK are organised into three levels, from highest Neonatal Intensive Care Unit (NICU), to Local Neonatal Unit (LNU) to lowest Special Care Unit (SCU). We model the endogenous treatment selection of neonatal care unit of birth to estimate the average and marginal treatment effects of different neonatal designations on infant mortality, length of stay and hospital costs. We use prognostic factors, survival and hospital care use data on all preterm births in England for 2014–2015, supplemented by national reimbursement tariffs and instrumental variables of travel time from a geographic information system. The data were consistent with a model of demand for preterm birth care driven by physical access. In‐hospital mortality of infants born before 32 weeks was 8.5% overall, and 1.2 (95% CI: −0.7, 3.2) percentage points lower for live births in hospitals with NICU or SCU compared to those with an LNU according to instrumental variable estimates. We find imprecise differences in average total hospital costs by unit designation, with positive unobserved selection of those with higher unexplained absolute and incremental costs into NICU. Our results suggest a limited scope for improvement in infant mortality by increasing in‐utero transfers based on unit designation alone.

SPATIAL LIMITED DEPENDENT VARIABLE MODELS: A REVIEW FOCUSED ON SPECIFICATION, ESTIMATION, AND HEALTH ECONOMICS APPLICATIONS

Article

Full-text available

Jul 2019

Modeling individual choices is one of the main aim in microeconometrics. Discrete choice models have been widely used to describe economic agents' utility functions and most of them play a paramount role in applied health economics. On the other hand, spatial econometrics collects a series of econometric tools, which are particularly useful when we deal with spatially distributed data sets. Accounting for spatial dependence can avoid inconsistency problems of the commonly used statistical estimators. However, the complex structure of spatial dependence in most of the nonlinear models still precludes a large diffusion of these spatial techniques. The purpose of this paper is then twofold. The former is to review the main methodological problems and their different solutions in spatial nonlinear modeling. The latter is to review their applications to health issues, especially those appeared in the last few years, by highlighting the main reasons why spatial discrete neighboring effects should be considered and suggesting possible future lines of development in this emerging field. Particular attention has been paid to cross‐sectional spatial discrete choice modeling. However, discussions on the main methodological advancements in other spatial limited dependent variable models and spatial panel data models are also included.

The Distribution of the EQ-5D-5L Profiles and Values in Three Patient Groups

Article

Nov 2018

Background The distribution of EQ-5D-3L values (health state profiles, weighted by value sets) often shows two distinct groups, arising from both the distribution of profiles and the characteristics of value sets. To date, there is little evidence about the distribution of EQ-5D-5L values. Objectives To explore the distribution of EQ-5D-5L profiles; to compare the distributions of EQ-5D-5L values arising from the English value set (EVS) and a ‘mapped’ value set (MVS); and to develop further the methods used to investigate clustering within EQ-5D data. Methods We obtained data from Cambridgeshire Community Services NHS Trust containing EQ-5D-5L profiles before treatment for three patient groups: community rehabilitation (N=6919); musculoskeletal physiotherapy (N=19999); and specialist nursing services (N=3366). Values were calculated using the EVS and MVS. Clusters were examined using the k-means method and Calinski–Harabasz pseudo-F index stopping rule. Results We found no evidence for clustering of EQ-5D-5L values arising from the classification system and no strong or consistent evidence of clustering arising from the EVS. There was clearer evidence of clustering using the MVS, with two being the optimal number of clusters. The clusters that were found for the EVS were very different from the MVS clusters. Conclusions Unlike the EQ-5D-3L, clustering of EQ-5D-5L values does not seem to be driven by clustering of its profile. This suggests the EQ-5D-5L is superior in that it is less likely to generate artefactual clusters – however, clusters may still result from using value sets such as MVS that have the tendency to generate them.

Individual Results May Vary: Inequality-Probability Bounds for Some Health-Outcome Treatment Effects

Article

Jul 2018
J HEALTH ECON

Mullahy John

While many results from the treatment-effect and related literatures are familiar and have been applied productively in health economics evaluations, other potentially useful results from those literatures have had little influence on health economics practice. With the intent of demonstrating the value and use of some of these results in health economics applications, this paper focuses on one particular class of parameters that describe probabilities that one outcome is larger or smaller than other outcomes ("inequality probabilities"). While the properties of such parameters have been exposited in the technical literature, they have scarcely been considered in informing practical questions in health evaluations. This paper shows how such probabilities can be used informatively, and describes how they might be identified or bounded informatively given standard sampling assumptions and information only on marginal distributions of outcomes. The logic of these results and the empirical implementation thereof-sampling, estimation, and inference-are straightforward. Derivations are provided and several health-related applications are presented.

Understanding Treatment Effect Estimates When Treatment Effects Are Heterogeneous for More Than One Outcome

Article

Mar 2018

Background: Patient-centred care requires evidence of treatment effects across many outcomes. Outcomes can be beneficial (e.g. increased survival or cure rates) or detrimental (e.g. adverse events, pain associated with treatment, treatment costs, time required for treatment). Treatment effects may also be heterogeneous across outcomes and across patients. Randomized controlled trials are usually insufficient to supply evidence across outcomes. Observational data analysis is an alternative, with the caveat that the treatments observed are choices. Real-world treatment choice often involves complex assessment of expected effects across the array of outcomes. Failure to account for this complexity when interpreting treatment effect estimates could lead to clinical and policy mistakes. Objective: Our objective was to assess the properties of treatment effect estimates based on choice when treatments have heterogeneous effects on both beneficial and detrimental outcomes across patients. Methods: Simulation methods were used to highlight the sensitivity of treatment effect estimates to the distributions of treatment effects across patients across outcomes. Scenarios with alternative correlations between benefit and detriment treatment effects across patients were used. Regression and instrumental variable estimators were applied to the simulated data for both outcomes. Results: True treatment effect parameters are sensitive to the relationships of treatment effectiveness across outcomes in each study population. In each simulation scenario, treatment effect estimate interpretations for each outcome are aligned with results shown previously in single outcome models, but these estimates vary across simulated populations with the correlations of treatment effects across patients across outcomes. Conclusions: If estimator assumptions are valid, estimates across outcomes can be used to assess the optimality of treatment rates in a study population. However, because true treatment effect parameters are sensitive to correlations of treatment effects across outcomes, decision makers should be cautious about generalizing estimates to other populations.

Physicians' Multitasking and Incentives: Empirical Evidence from a Natural Experiment

Article

Full-text available

Dec 2008
J HEALTH ECON

We analyse how physicians respond to contractual changes and incentives within a multitasking environment. In 1999 the Quebec government (Canada) introduced an optional mixed compensation system, combining a xed per diem with a partial (relative to the traditional fee-for-service system) fee for services provided. We combine panel survey and administrative data on Quebec physicians to evaluate the impact of this change in incentives on their practice choices. We highlight the dierentiated impact of incentives on various dimensions of physician behaviour by considering a wide range of labour supply variables: time spent on seeing patients, time devoted to teaching, administrative tasks or research, as well as the volume of clinical services and average time per clinical service. Our results show that, on average, the reform induced physicians who changed from FFS to MC to reduce their volume of (billable) services by 6.15% and to reduce their hours of work spent on seeing patients by 2.57%. Their average time spent per service increased by 3.58%, suggesting a potential quality-quantity substitution. Also the reform induced these physicians to increase their time spent on teaching and administrative duties (tasks not remunerated under the fee-for-service system) by 7.9%.

Estimating risk attitude impact on allocative efficiency under different plantain-based cropping systems: A latent factor model with instrumental variables

Preprint

Full-text available

Apr 2024

Plantain production is subject to risk, emerging from diverse sources. Consequently, risk modifies farmers behavior, affecting their decision making and thereby their future results. These choices are not only related to investments, crop to produce, and technology use but also to resource allocation. The objective of this study is to analyze the impact of farmers’ risk attitude on allocative efficiency in plantain production. Primary data from 584 farmers, operating on 707 plots, selected through a four-level stratified random sampling was used. The study used the Latent factor models with instrumental variables to account for endogeneity and selection bias. We apply the causal inference methods to assess the impact of risk attitude on allocative efficiency. The result highlights a negative impact of risk attitude on the allocative efficiency. Compared to non-averse farmers, risk-averse farmers are less efficient. Also, as the level of risk aversion increases, the negative impact of risk aversion is more pronounced (-49.1% for extreme risk aversion). Furthermore, stronger impact of risk attitude regardless the level of risk aversion is found for the type 3 which stands for the intensive cropping system. The study provides empirical evidence of the extent of the impact of risk attitude on the allocative efficiency and confirms the heterogeneity of impact assumption. It also provides a useful insight into the key role risk attitude plays in productivity and efficiency in agriculture. These findings have critical policy implications and can help policymakers and practitioners make informed decisions and devise interventions to improve farmers allocative efficiency.

Advances in Addressing Patient Heterogeneity in Economic Evaluation: A Review of the Methods Literature

Article

Apr 2024

Cost-effectiveness analyses commonly use population or sample averages, which can mask key differences across subgroups and may lead to suboptimal resource allocation. Despite there being several new methods developed over the last decade, there is no recent summary of what methods are available to researchers. This review sought to identify advances in methods for addressing patient heterogeneity in economic evaluations and to provide an overview of these methods. A literature search was conducted using the Econlit, Embase and MEDLINE databases to identify studies published after 2011 (date of a previous review on this topic). Eligible studies needed to have an explicit methodological focus, related to how patient heterogeneity can be accounted for within a full economic evaluation. Sixteen studies were included in the review. Methodologies were varied and included regression techniques, model design and value of information analysis. Recent publications have applied methodologies more commonly used in other fields, such as machine learning and causal forests. Commonly noted challenges associated with considering patient heterogeneity included data availability (e.g., sample size), statistical issues (e.g., risk of false positives) and practical factors (e.g., computation time). A range of methods are available to address patient heterogeneity in economic evaluation, with relevant methods differing according to research question, scope of the economic evaluation and data availability. Researchers need to be aware of the challenges associated with addressing patient heterogeneity (e.g., data availability) to ensure findings are meaningful and robust. Future research is needed to assess whether and how methods are being applied in practice.

Treatment effect models with strategic interaction in treatment decisions

Article

Oct 2023
J ECONOMETRICS

Instrumental variable analysis for cost outcome: Application to the effect of primary care visit on medical cost among low‐income adults

Article

Aug 2023

Medical cost data often consist of zero values as well as extremely right‐skewed positive values. A two‐part model is a popular choice for analyzing medical cost data, where the first part models the probability of a positive cost using logistic regression and the second part models the positive cost using a lognormal or Gamma distribution. To address the unmeasured confounding in studies on cost outcome under two‐part models, two instrumental variable (IV) methods, two‐stage residual inclusion (2SRI) and two‐stage prediction substitution (2SPS) are widely applied. However, previous literature demonstrated that both the 2SRI and the 2SPS could fail to consistently estimate the causal effect among compliers under standard IV assumptions for binary and survival outcomes. Our simulation studies confirmed that it continued to be the case for a two‐part model, which is another nonlinear model. In this article, we develop a model‐based IV approach, Instrumental Variable with Two‐Part model (IV2P), to obtain a consistent estimate of the causal effect among compliers for cost outcome under standard IV assumptions. In addition, we develop sensitivity analysis approaches to allow the evaluation of the sensitivity of the causal conclusions to potential quantified violations of the exclusion restriction assumption and the randomization of IV assumption. We apply our method to a randomized cash incentive study to evaluate the effect of a primary care visit on medical cost among low‐income adults newly covered by a primary care program.

How does a local instrumental variable method perform across settings with instruments of differing strengths? A simulation study and an evaluation of emergency surgery

Article

Full-text available

Jun 2023
HEALTH ECON

Local instrumental variable (LIV) approaches use continuous/multi‐valued instrumental variables (IV) to generate consistent estimates of average treatment effects (ATEs) and Conditional Average Treatment Effects (CATEs). There is little evidence on how LIV approaches perform according to the strength of the IV or with different sample sizes. Our simulation study examined the performance of an LIV method, and a two‐stage least squares (2SLS) approach across different sample sizes and IV strengths. We considered four ‘heterogeneity’ scenarios: homogeneity, overt heterogeneity (over measured covariates), essential heterogeneity (unmeasured), and overt and essential heterogeneity combined. In all scenarios, LIV reported estimates with low bias even with the smallest sample size, provided that the instrument was strong. Compared to 2SLS, LIV provided estimates for ATE and CATE with lower levels of bias and Root Mean Squared Error. With smaller sample sizes, both approaches required stronger IVs to ensure low bias. We considered both methods in evaluating emergency surgery (ES) for three acute gastrointestinal conditions. Whereas 2SLS found no differences in the effectiveness of ES according to subgroup, LIV reported that frailer patients had worse outcomes following ES. In settings with continuous IVs of moderate strength, LIV approaches are better suited than 2SLS to estimate policy‐relevant treatment effect parameters.

Average Causal Effect Estimation Via Instrumental Variables: the No Simultaneous Heterogeneity Assumption

Article

Jan 2023

Background: Instrumental variables (IVs) can be used to provide evidence as to whether a treatment X has a causal effect on an outcome Y. Even if the instrument Z satisfies the three core IV assumptions of relevance, independence, and the exclusion restriction, further assumptions are required to identify the average causal effect (ACE) of X on Y. Sufficient assumptions for this include: homogeneity in the causal effect of X on Y; homogeneity in the association of Z with X; and no effect modification. Methods: We describe the no simultaneous heterogeneity assumption, which requires the heterogeneity in the X-Y causal effect to be mean independent of (i.e., uncorrelated with) both Z and heterogeneity in the Z-X association. This happens, for example, if there are no common modifiers of the X-Y effect and the Z-X association, and the X-Y effect is additive linear. We illustrate the assumption of no simultaneous heterogeneity using simulations and by re-examining selected published studies. Results: Under no simultaneous heterogeneity, the Wald estimand equals the ACE even if both homogeneity assumptions and no effect modification (which we demonstrate to be special cases of - and therefore stronger than - no simultaneous heterogeneity) are violated. Conclusions: The assumption of no simultaneous heterogeneity is sufficient for identifying the ACE using IVs. Since this assumption is weaker than existing assumptions for ACE identification, doing so may be more plausible than previously anticipated.

Information and Communication Technology, Hierarchy, and Job Design

Article

Jan 2022

Real‐world evidence in health technology assessment of high‐risk medical devices: Fit for purpose?

Article

Full-text available

Aug 2022
HEALTH ECON

Health technology assessment (HTA) of medical devices (MDs) increasingly rely on real‐world evidence (RWE). The aim of this study was to evaluate the type and the quality of the evidence used to assess the (cost‐)effectiveness of high risk MDs (Class III) by HTA agencies in Europe (four European HTA agencies and EUnetHTA), with particular focus on RWE. Data were extracted from HTA reports on the type of evidence demonstrating (cost‐)effectiveness, and the quality of observational studies of comparative effectiveness using the Good Research for Comparative Effectiveness principles. 25 HTA reports were included that incorporated 28 observational studies of comparative effectiveness. Half of the studies (46%) took important confounding and/or effect modifying variables into account in the design and/or analyses. The most common way of including confounders and/or effect modifiers was through multivariable regression analysis. Other methods, such as propensity score matching, were rarely employed. Furthermore, meaningful analyses to test key assumptions were largely omitted. Resulting recommendations from HTA agencies on MDs is therefore (partially) based on evidence which is riddled with uncertainty. Considering the increasing importance of RWE it is important that the quality of observational studies of comparative effectiveness are systematically assessed when used in decision‐making.

Genetic support of a causal relationship between iron status and atrial fibrillation: a Mendelian randomization study

Article

Full-text available

Dec 2022
GENES NUTR

Background Atrial fibrillation is the most common arrhythmia disease. Animal and observational studies have found a link between iron status and atrial fibrillation. However, the causal relationship between iron status and AF remains unclear. The purpose of this investigation was to use Mendelian randomization (MR) analysis, which has been widely applied to estimate the causal effect, to reveal whether systemic iron status was causally related to atrial fibrillation. Methods Single nucleotide polymorphisms (SNPs) strongly associated ( P < 5 × 10 ⁻⁸ ) with four biomarkers of systemic iron status were obtained from a genome-wide association study involving 48,972 subjects conducted by the Genetics of Iron Status consortium. Summary-level data for the genetic associations with atrial fibrillation were acquired from the AFGen (Atrial Fibrillation Genetics) consortium study (including 65,446 atrial fibrillation cases and 522,744 controls). We used a two-sample MR analysis to obtain a causal estimate and further verified credibility through sensitivity analysis. Results Genetically instrumented serum iron [OR 1.09; 95% confidence interval (CI) 1.02–1.16; p = 0.01], ferritin [OR 1.16; 95% CI 1.02–1.33; p = 0.02], and transferrin saturation [OR 1.05; 95% CI 1.01–1.11; p = 0.01] had positive effects on atrial fibrillation. Genetically instrumented transferrin levels [OR 0.90; 95% CI 0.86–0.97; p = 0.006] were inversely correlated with atrial fibrillation. Conclusion In conclusion, our results strongly elucidated a causal link between genetically determined higher iron status and increased risk of atrial fibrillation. This provided new ideas for the clinical prevention and treatment of atrial fibrillation.

Comparative Effectiveness of Gastric Bypass and Sleeve Gastrectomy on Predicted 10-Year Risk of Cardiovascular Disease 5 Years Following Surgery

Article

Mar 2022
SURG OBES RELAT DIS

Background Comparative evidence is needed when deciding on which bariatric operation to have for long-term cardiovascular risk reduction. Objective The Effectiveness of Gastric Bypass vs. Gastric Sleeve for Cardiovascular Disease (ENGAGE CVD) study compared the effectiveness of vertical sleeve gastrectomy (VSG) and Roux-en-Y gastric bypass (RYGB) operations for reduction of the American College of Cardiology (ACA) and the American Heart Association (AHA) predicted 10-year atherosclerotic cardiovascular disease (ASCVD) risk 5 years after surgery. Setting Data for this study came from a large integrated healthcare system in the Southern California region of the U.S. This is one of the most ethnically diverse (64% non-White) bariatric populations in the literature. Methods The ENGAGE CVD cohort consisted of 22,095 patients who underwent VSG or RYGB from 2009 - 2016. The VSG and RYGB were compared using a local instrumental variable (LIV) approach to address observed and unobserved confounding, as well as to conduct heterogeneity of treatment effects for patients of different age groups, baseline predicted 10-year CVD risk using the ASCVD risk score, and those who had T2DM at the time of surgery. Results Patients (2,771 RYGB and 6,256 VVSG) were primarily women (80.6%), Hispanic or non-Hispanic Black (63.7%), were 46+10 years old, with a BMI of 43.40+6.5 kg/m². The predicted 10-year ASCVD risk at surgery was 4.1% for VSG and 5.1% for RYGB, decreasing to 2.6% for VSG and 2.8% for RYGB 1-year postoperatively. By 5 years after surgery, patients remained with relatively low risk levels (3.0% for VSG and 3.3% for RYGB) and there were no significant differences in predicted 10-year ASCVD risk between VSG and RYGB at any time. Conclusions Predicted 10-year ASCVD risk was low in this population and remained low up to 5 years for those with diabetes, Black and Hispanic patients, and older adults. Literature reporting significant differences between VSG and RYGB in 10-year ASCVD risk may be a result of residual confounding.

Observational retrospective study calculating health service costs of patients receiving surgery for chronic rhinosinusitis in England, using linked patient-level primary and secondary care electronic data

Article

Full-text available

Feb 2022

Objectives Chronic rhinosinusitis (CRS) symptoms are experienced by an estimated 11% of UK adults, and symptoms have major impacts on quality of life. Data from UK and elsewhere suggest high economic burden of CRS, but detailed cost information and economic analyses regarding surgical pathway are lacking. This paper estimates healthcare costs for patients receiving surgery for CRS in England. Design Observational retrospective study examining cost of healthcare of patients receiving CRS surgery. Setting Linked electronic health records from the Clinical Practice Research Datalink, Hospital Episode Statistics and Office for National Statistics databases in England. Participants A phenotyping algorithm using medical ontology terms identified ‘definite’ CRS cases who received CRS surgery. Patients were registered with a general practice in England. Data covered the period 1997–2016. A cohort of 13 462 patients had received surgery for CRS, with 9056 (67%) having confirmed nasal polyps. Outcome measures Information was extracted on numbers and types of primary care prescriptions and consultations, and inpatient and outpatient hospital investigations and procedures. Resource use was costed using published sources. Results Total National Health Service costs in CRS surgery patients were £2173 over 1 year including surgery. Total costs per person-quarter were £1983 in the quarter containing surgery, mostly comprising surgical inpatient care costs (£1902), and around £60 per person-quarter in the 2 years before and after surgery, of which half were outpatient costs. Outpatient and primary care costs were low compared with the peak in inpatient costs at surgery. The highest outpatient expenditure was on CT scans, peaking in the quarter preceding surgery. Conclusions We present the first study of costs to the English healthcare system for patients receiving surgery for CRS. The total aggregate costs provide a further impetus for trials to evaluate the relative benefit of surgical intervention.

Comparative Effectiveness of Gastric Bypass and Vertical Sleeve Gastrectomy for Hypertension Remission and Relapse: The ENGAGE CVD Study

Article

Aug 2021

Weight loss is an effective strategy for the management of hypertension, and bariatric surgery is the most effective weight loss and maintenance strategy for obesity. The importance of bariatric surgery in the long-term management of hypertension and which operation is most effective is less clear. We compared the effectiveness of vertical sleeve gastrectomy (VSG) and Roux-en-Y gastric bypass (RYGB) for remission and relapse of hypertension after surgery in the ENGAGE CVD cohort study (Effectiveness of Gastric Bypass Versus Gastric Sleeve for Cardiovascular Disease). Operations were done by 23 surgeons across 9 surgical practices. Hypertension remission and relapse were assessed in each year of follow-up beginning 30 days and up to 5 years postsurgery. We used a local instrumental variable approach to account for selection bias in the choice of VSG or RYGB. The study population included 4964 patients with hypertension at the time of surgery (n=3186 VSG and n=1778 RYGB). At 1 year, 27% of patients with RYGB and 28% of patients with VSG achieved remission. After 5 years, without accounting for relapse, 42% of RYGB and 43% of VSG patients had experienced hypertension remission. After accounting for relapse, only 17% of RYGB and 18% of VSG patients remained in remission 5 years after surgery. There were no statistically significant differences between VSG and RYGB for hypertension remission, relapse, or mean systolic and diastolic blood pressure at any time during follow-up.

Tracking and specialization of high schools: Heterogeneous effects of school choice

Article

Jul 2021

We analyze the impact of choosing an elite school on high school graduation in an early tracking system in Flanders (Belgium). While elite schools offer only an academic track, most other schools offer multiple tracks. On average, students experience a 3.3 percentage point increase in the likelihood of obtaining a degree. We find that the effects are heterogeneous. On average, students who self‐select into elite schools do not experience an effect. However, students who do not choose an elite school would experience positive effects. Our results can be explained by different tracking decisions in both types of schools.

Estimation and inference for policy relevant treatment effects

Article

Jun 2021
J ECONOMETRICS

The policy relevant treatment effect (PRTE) measures the average effect of switching from a status-quo policy to a counterfactual policy under consideration. Estimation of the PRTE involves estimation of multiple preliminary parameters, including propensity scores, conditional expectation functions of the outcome and covariates given the propensity score, and marginal treatment effects. These preliminary estimators can affect the asymptotic distribution of the PRTE estimator in complicated and intractable manners. In this light, we propose an orthogonal score for double debiased estimation of the PRTE, whereby the asymptotic distribution of the PRTE estimator is obtained without any influence of preliminary parameter estimators as far as they satisfy mild requirements of convergence rates. To our knowledge, this paper is the first to develop limit distribution theories for inference about the PRTE.

Decomposition analysis as a framework for understanding heterogeneity of treatment effects in n on‐randomized health care studies

Article

Mar 2021

William H Crown

This paper uses the decomposition framework from the economics literature to examine the statistical structure of treatment effects estimated with observational data compared to those estimated from randomized studies. It begins with the estimation of treatment effects using a dummy variable in regression models and then presents the decomposition method from economics which estimates separate regression models for the comparison groups and recovers the treatment effect using bootstrapping methods. This method shows that the overall treatment effect is a weighted average of structural relationships of patient features with outcomes within each treatment arm and differences in the distributions of these features across the arms. In large randomized trials, it is assumed that the distribution of features across arms is very similar. Importantly, randomization not only balances observed features but also unobserved. Applying high dimensional balancing methods such as propensity score matching to the observational data causes the distributional terms of the decomposition model to be eliminated but unobserved features may still not be balanced in the observational data. Finally, a correction for non‐random selection into the treatment groups is introduced via a switching regime model. Theoretically, the treatment effect estimates obtained from this model should be the same as those from a randomized trial. However, there are significant challenges in identifying instrumental variables that are necessary for estimating such models. At a minimum, decomposition models are useful tools for understanding the relationship between treatment effects estimated from observational versus randomized data.

Forest commons, vertical integration and smallholder’s saving and investment responses: Evidence from a quasi-experiment

Article

Aug 2020
WORLD DEV

Dambala Gelo

As the result of prohibitively high transaction costs, smallholder farmers are only partly integrated into agricultural and forest commodity markets, a situation that may leave them in a lower level of development equilibrium (i.e., a poverty trap). For the most part, many users of forest commons extract forest products, typically non-timber products, for subsistence use or safety net purposes. To overcome this problem, in recent years, collective vertical integration (VI) of forest product marketing cooperative structures have been promoted and, in some cases, adopted by users of forest commons. Although this type of program has been observed to raise smallholder incomes, there is little evidence available on saving/investment responses to such income gains. This paper investigates precautionary saving and investment responses to collective forest product marketing programs among users of forest commons in Ethiopian villages. To identify the causal effects of the program, I applied propensity score matching, difference-in-difference (DID) and change-in-change (CIC) estimators to household survey data collected from randomly selected households in the Gimbo district (south-western Ethiopia). I find strong evidence that participation in the program reduces savings in the form of livestock holdings and that effect is limited to non-poor households. When interpreted in terms of the Permanent Income Hypothesis (PIH), the results imply that participants felt the current income gains to be non-transient, which led to reduced precautionary savings and to a gain in consumption/welfare. Moreover, I found that the program has spurred investment in child education and participation in off-farm self-employment. These results point to the importance of the safety net/insurance channel of the program. Overall, the findings underscore the program’s potential to raise the standard of living via ancillary mechanisms beyond directly raising income outcome.

Collective action and heterogeneous welfare effects: Evidence from Ethiopian villages

Article

Nov 2019

In this study, welfare and distributional impacts associated with a forest users cooperative (FUC) programs in Ethiopian villages were examined. We employed covariate balancing propensity scores (CBPS), instrumental variable (IV) and selection models to estimate both the average treatment effect and quantile treatment effects. Our results revealed that the program was found to raise the welfare of the average program participating households and that result is robust to alternative specifications. Furthermore, the analysis confirmed that no-targeted households would similarly benefit from the program, underscoring the importance of expanding the current programs. Results of quantile treatment effect evaluations confirmed that return to the program participation is heterogeneous across income distribution. Specifically, the program was found to raise welfare only for households in the middle and uppers spans of income distribution, without bearing effect along the bottom income quantiles. This suggests that FUC programs is not pro-poor, and, therefore, is not equity enhancing.

Instrumental Variable Analysis: Methods and Protocols

Chapter

Feb 2019

Robust Causal Inference with Continuous Instruments Using the Local Instrumental Variable Curve

Article

Nov 2018

Instrumental variables are commonly used to estimate effects of a treatment afflicted by unmeasured confounding, and in practice instruments are often continuous (e.g. measures of distance, or treatment preference). However, available methods for continuous instruments have important limitations: they either require restrictive parametric assumptions for identification, or else rely on modelling both the outcome and the treatment process well (and require modelling effect modification by all adjustment covariates). In this work we develop the first semiparametric doubly robust estimators of the local instrumental variable effect curve, i.e. the effect among those who would take treatment for instrument values above some threshold and not below. In addition to being robust to misspecification of either the instrument or treatment or outcome processes, our approach also incorporates information about the instrument mechanism and allows for flexible data‐adaptive estimation of effect modification. We discuss asymptotic properties under weak conditions and use the methods to study infant mortality effects of neonatal intensive care units with high versus low technical capacity, using travel time as an instrument.

Treatment Eﬀect Models with Strategic Interaction in Treatment Decisions

Article

Jan 2018

A conversation including 39 questions with Anirban Basu

Article

Full-text available

Dec 2018
Health Serv Outcome Res Meth

At the 2018 International Conference on Health Policy Statistics (ICHPS) held in Charleston, South Carolina, Anirban Basu was awarded the Mid-Career Excellence Award from the American Statistical Association Section on Health Policy Statistics (HPSS). Anirban was exceptionally and uniquely qualified for this award. Highlights include his providing outstanding service to the HPSS, advancing statistical methodology, advancing methodology in other domains of health policy, and performing extensive and highly impactful applied work in medicine and health care. In this interview, we trace Anirban’s upbringing, schooling, early career, and mid-career phases to gain insights into his success. We also sought his opinions on salient topics or issues.

Instrumental Variable Analysis

Chapter

Jul 2017

Statistical methods for cost-effectiveness analysis that use cluster-randomised trials

Thesis

Jan 2012

Manuel Gomes

This thesis considers alternative statistical methods for cost-effectiveness analysis (CEA) that use cluster randomised trials (CRTs). The thesis has four objectives: firstly to develop criteria for identifying appropriate methods for CEA that use CRTs; secondly to critically appraise the methods used in applied CEAs that use CRTs; thirdly to assess the performance of alternative methods for CEA that use CRTs in settings where baseline covariates are balanced; fourthly to compare statistical methods that adjust for systematic covariate imbalance in CEA that use CRTs. The thesis developed a checklist to assess the methodological quality of published CEAs that use CRTs. This checklist was informed by a conceptual review of statistical methods, and applied in a systematic literature review of published CEAs that use CRTs. The review found that most studies adopted statistical methods that ignored clustering or correlation between costs and health outcomes. A simulation study was conducted to assess the performance of alternative methods for CEA that use CRTs across different circumstances where baseline covariates are balanced. This study considered: seemingly unrelated regression (SUR) and generalised estimating equations (GEEs), both with a robust standard error; multilevel models (MLMs) and a non-parametric 'two-stage' bootstrap (TS8). Performance was reported as, for example, bias and confidence interval (Cl) coverage of the incremental net benefit. The MLMs and the TSB performed well across all settings; SUR and GEEs reported poor Cl coverage in CRTs with few clusters. The thesis compared methods for CEA that use CRTs when there are systematic differences in baseline covariates between the treatment groups. In a case study and further simulations, the thesis considered SUR, MLMs, and TSB combined with SUR to adjust for covariate imbalance. The case-study showed that cost-effectiveness results can differ according to adjustment method. The simulations reported that MLMs performed well across all settings, and unlike the other methods, provided Cl coverage close to nominal levels, even with few clusters and unequal cluster sizes. The thesis concludes that MLMs are the most appropriate method across the circumstances considered. This thesis presents methods for improving the quality ofCEA that use CRTs, to help future studies provide a sound basis for policy making.

Who benefits most from university education in Switzerland?

Article

Full-text available

Apr 2014

Lionel Perini

Recent literature on private returns to education considers diversity in the population, heterogeneity in wage gains and self-selection into schooling. This research addresses these issues by analyzing to what extent returns associated with completing a university degree in Switzerland depend on the propensity to attend and complete this degree. Using data from the Swiss Household Panel and propensity score matching models, I find that low propensity men — after controlling for labor market variables — benefit most from a university degree while returns for women are rather homogenous along the propensity score distribution. This finding suggests that completing university increases more the earning capability of men with disadvantaged family backgrounds than that of men with more favorable background, refuting the hypothesis of comparative advantage. An auxiliary analysis focusing on the relationship between returns to education and inherent ability within a quantile regression framework leads to similar conclusions.

The Effect of Education on Health Behavior after Screening for Colorectal Cancer

Chapter

May 2017

Misinterpretation of a negative test results in health screening may initiate less preventive effort and more future lifestyle-related disease. We predict that misinterpretation occurs more frequently among individuals with a low level of education compared with individuals with a high level of education. The empirical analyses are based on unique data from a randomized controlled screening experiment in Norway, NORCCAP (NORwegian Colorectal Cancer Prevention). The dataset consists of approximately 50,000 individuals, of whom 21,000 were invited to participate in a once only screening with sigmoidoscopy. For all individuals, we also have information on outpatient consultations and inpatient stays and education. The result of health behaviour is mainly measured by lifestyle-related diseases, such as COPD, hypertension and diabetes type 2, identified by ICD-10 codes. The results according to intention-to-treat indicate that screening does not increase the occurrence of lifestyle related diseases among individuals with a high level of education, while there is an increase for individuals with low levels of education. These results are supported by the further analyses among individuals with a negative screening test.

Panel Data Methods and Applications to Health Economics

Chapter

Jan 2009

Andrew M Jones

Much of the empirical analysis done by health economists seeks to estimate the impact of specific health policies, and the greatest challenge for successful applied work is to find appropriate sources of variation to identify the treatment effects of interest. Estimation can be prone to selection bias when the assignment to treatments is associated with the potential outcomes of the treatment. Overcoming this bias requires variation in the assignment of treatments that is independent of the outcomes. One source of independent variation comes from randomized controlled experiments. But, in practice, most economic studies have to draw on non-experimental data. Many studies seek to use variation across time and events that takes the form of a quasi-experimental design, or “natural experiment,” that mimics the features of a genuine experiment. This chapter reviews the data and methods that are used in applied health economics with a particular emphasis on the use of panel data. The focus is on nonlinear models and methods that can accommodate unobserved heterogeneity. These include conditional estimators, maximum simulated likelihood, Bayesian MCMC, finite mixtures and copulas.

Individualization of Treatment and Comparative Effectiveness Research

Chapter

Jan 2016

Anirban Basu

Comparative effectiveness research (CER) comprises of the generation and synthesis of evidence that compares the benefits and harms of alternative methods to prevent, diagnose, treat, and monitor a clinical condition or to improve the delivery of care. Its purpose is to assist consumers, clinicians, purchasers, and policymakers to make informed decisions that will improve healthcare at both the individual and population levels. The agenda for CER is very ambitious considering its limited resources and the starkly different informational requirements of the various decision-makers. How results emanating from a single or a few CER studies can inform all levels of decision-making remains the biggest challenge in the designs of CER studies. This chapter discusses the role of CER in generating individualized information on the value of medical products and how such information has the potential for improving decision-making at all levels. In practice, this notion of generating individualized information and using it to deliver care is denoted as personalized medicine (PM), which allows for the possibility of variation in medical quality based on demographics, comorbidities, preferences, genomics, and even environmental contexts within which care is delivered. However, traditionally, CER and PM are thought to be disparate research strategies. Recognizing that this distinction may be artificial and created by silos in research practices, this chapter discusses some of the key behavioral and economic issues that encourage the adoption of PM in practice and how the current infrastructure for CER studies can be leveraged to evaluate PM and also foster innovation in PM. Indeed, the fields of CER and PM appear to be morphing into the single paradigm of patient-centered outcomes research (PCOR) (also denoted as precision medicine). The chapter ends with discussing some of the tools available in order to prioritize PM research in a prospective manner.

Does more intensive treatment of acute myocardial infarction in the elderly reduce mortality?

Article

Full-text available

Jan 1994

Instrumental Variables: A Study of Implicit Behavioral Assumptions Used in Making Program Evaluations

Article

Full-text available

Jun 1997

James Heckman

This paper considers the use of instrumental variables to estimate the mean effect of treatment on the treated, the mean effect of treatment on randomly selected persons and the local average treatment effect. It examines what economic questions these parameters address. When responses to treatment vary, the standard argument justifying the use of instrumental variables fails unless person-specific responses to treatment do not influence decisions to participate in the program being evaluated. This requires that individual gains from the program that cannot be predicted from variables in outcome equations do not influence the decision of the persons being studied to participate in the program. In the likely case in which individuals possess and act on private information about gains from the program that cannot be fully predicted by variables in the outcome equation, instrumental variables methods do not estimate economically interesting evaluation parameters. Instrumental variable methods are extremely sensitive to assumptions about how people process information. These arguments are developed for both continuous and discrete treatment variables and several explicit economic models are presented.

Identification of Causal Effects Using Instrumental Variables: Rejoinder

Article

Full-text available

Jul 1993

We outline a framework for causal inference in setting where assignment to a binary treatment is ignorable, but compliance with the assignment is not perfect so that the receipt of treatment is nonignorable. To address the problems associated with comparing subjects by the ignorable assignment--an "intention-to-treat analysis"--we make use of instrumental variables, which have long been used by economists in the context of regression models with constant treatment effects. We show that the instrumental variables (IV) estimand can be embedded within the Rubin Causal Model (RCM) and that under some simple and easily interpretable assumptions, the IV estimand is the average causal effect for a subgroup of units, the compliers. Without these assumptions, the IV estimand is simply the ratio of intention-to-treat causal estimands with no interpretation as an average causal effect. The advantages of embedding the IV approach in the RCM are that it clarifies the nature of critical assumptions needed for a causal interpretation, and moreover allows us to consider sensitivity of the results to deviations from key assumptions in a straightforward manner. We apply our analysis to estimate the effect of veteran status in the Vietnam era on mortality, using the lottery number assigned priority for the draft as an instrument, and we use our results to investigate the sensitivity of the conclusions to critical assumptions. Statistics Version of Record

Heterogeneity, Aggregation, and Market Wage Functions: An Empirical Model of Self-Selection in the Labor Market

Article

Full-text available

Feb 1985

Micro Data, Heterogeneity, and the Evaluation of Public Policy: Nobel Lecture

Article

Full-text available

Feb 2001

James Heckman

This paper summarizes the contributions of microeconometrics to economic knowledge. Four main themes are developed. (1) Microeconometricians developed new tools to respond to econometric problems raised by the analysis of the new sources of micro data produced after the Second World War. (2) Microeconometrics improved on aggregate time-series methods by building models that linked economic models for individuals to data on individual behavior. (3) An important empirical regularity detected by the field is the diversity and heterogeneity of behavior. This heterogeneity has profound consequences for economic theory and for econometric practice. (4) Microeconometrics has contributed substantially to the scientific evaluation of public policy.

Understanding Instrumental Variables in Models with Essential Heterogeneity

Article

Full-text available

Feb 2006
REV ECON STAT

This paper examines the properties of instrumental variables (IV) applied to models with essential heterogeneity, that is, models where responses to interventions are heterogeneous and agents adopt treatments (participate in programs) with at least partial knowledge of their idiosyncratic response. We analyze two-outcome and multiple-outcome models, including ordered and unordered choice models. We allow for transition-specific and general instruments. We generalize previous analyses by developing weights for treatment effects for general instruments. We develop a simple test for the presence of essential heterogeneity. We note the asymmetry of the model of essential heterogeneity: outcomes of choices are heterogeneous in a general way; choices are not. When both choices and outcomes are permitted to be symmetrically heterogeneous, the method of IV breaks down for estimating treatment parameters. Copyright by the President and Fellows of Harvard College and the Massachusetts Institute of Technology.

The Estimation of Wage Gains and Welfare Gains in Self-Selection Models

Article

Full-text available

Feb 1987
REV ECON STAT

The authors modify the basic self-sele ction model for the effects ofeducation, training, unions, and other activities on wages, by including "heterogeneity of rewards" to the activity-i.e., diffe rences across individuals in the rate of return to the activity. The authors sho w that such heterogeneity creates a new form of selectionbias. They provide tes ts for its presence and draw out its implications for the wage and welfare gains to the activity. An empirical application provides strong support for such hete rogeneityin one particular training program. Copyright 1987 by MIT Press.

Does More Intensive Treatment of Acute Myocardial Infarction Reduce Mortality?

Article

Full-text available

Oct 1994

To determine the effect of more intensive treatments on mortality in elderly patients with acute myocardial infarction (AMI). Analysis of incremental treatment effects using differential distances as instrumental variables to account for unobserved case-mix variation (selection bias) in observational Medicare claims data (1987 through 1991). Survival to 4 years after AMI. Patients who receive different treatments differ in observable and unobservable health characteristics, biasing estimates of treatment effects based on standard methods of adjusting for observable differences. Patients' differential distances to alternative types of hospitals are strong independent predictors of how intensively an AMI patient will be treated and appear uncorrelated with health status. Thus, differential distances approximately randomize patients to different likelihoods of receiving intensive treatments. Comparisons of patient groups that differ only in differential distances show that the impact on mortality at 1 to 4 years after AMI of the incremental ("marginal") use of invasive procedures in Medicare patients was at most 5 percentage points; this gain was achieved during the first day of hospitalization and therefore appears attributable to treatments other than the procedures. Admission to a hospital treating a high volume of AMI patients was associated with an effect on mortality at 4 years of less than 1 percentage point, again arising on day 1. Patients living in rural areas experienced acute mortality that was an additional 0.6 percentage-point higher, after controlling for less access to intensive treatments. For elderly patients with AMI, the aspects of treatment most affecting long-term survival relate to care within the first 24 hours of admission. The survival benefits from greater use of catheterization and revascularization procedures appear minimal in marginal patients.

Econometrics in Outcomes Research: The Use of Instrumental Variables

Article

Full-text available

Feb 1998

We describe an econometric technique, instrumental variables, that can be useful in estimating the effectiveness of clinical treatments in situations when a controlled trial has not or cannot be done. This technique relies upon the existence of one or more variables that induce substantial variation in the treatment variable but have no direct effect on the outcome variable of interest. We illustrate the use of the technique with an application to aggressive treatment of acute myocardial infarction in the elderly.

Local Instrumental Variables and Latent Variable Models for Identifying and Bounding Treatment Effec

Article

Full-text available

May 1999

This paper examines the relationship between various treatment parameters within a latent variable model when the effects of treatment depend on the recipient's observed and unobserved characteristics. We show how this relationship can be used to identify the treatment parameters when they are identified and to bound the parameters when they are not identified.

Twenty-Year Follow-up of a Randomized Study Comparing Breast-Conserving Surgery with Radical Mastectomy for Early Breast Cancer

Article

Full-text available

Oct 2002
NEW ENGL J MED

We conducted 20 years of follow-up of women enrolled in a randomized trial to compare the efficacy of radical (Halsted) mastectomy with that of breast-conserving surgery. From 1973 to 1980, 701 women with breast cancers measuring no more than 2 cm in diameter were randomly assigned to undergo radical mastectomy (349 patients) or breast-conserving surgery (quadrantectomy) followed by radiotherapy to the ipsilateral mammary tissue (352 patients). After 1976, patients in both groups who had positive axillary nodes also received adjuvant chemotherapy with cyclophosphamide, methotrexate, and fluorouracil. Thirty women in the group that underwent breast-conserving therapy had a recurrence of tumor in the same breast, whereas eight women in the radical-mastectomy group had local recurrences (P<0.001). The crude cumulative incidence of these events was 8.8 percent and 2.3 percent, respectively, after 20 years. In contrast, there was no significant difference between the two groups in the rates of contralateral-breast carcinomas, distant metastases, or second primary cancers. After a median follow-up of 20 years, the rate of death from all causes was 41.7 percent in the group that underwent breast-conserving surgery and 41.2 percent in the radical-mastectomy group (P=1.0). The respective rates of death from breast cancer were 26.1 percent and 24.3 percent (P=0.8). The long-term survival rate among women who undergo breast-conserving surgery is the same as that among women who undergo radical mastectomy. Breast-conserving surgery is therefore the treatment of choice for women with relatively small breast cancers.

Evaluating the Welfare State

Article

Full-text available

Jun 1998

A variety of criteria are relevant for evaluating alternative policies in democratic societies composed of persons with diverse values and perspectives. In this paper, we consider alternative criteria for evaluating the welfare state, and the data required to operationalize them. We examine sets of identifying assumptions that bound, or exactly produce, these alternative criteria given the availability of various types of data. We consider the economic questions addressed by two widely-used econometric evaluation estimators and relate them to the requirements of a comprehensive cost-benefit analysis. We present evidence on how the inference from the most commonly used econometric evaluation estimator is modified when the direct costs of a program are fully assessed, including the welfare costs of the taxes required to support the program. Finally, we present evidence of the empirical inconsistency of alternative criteria derived from evaluations based on on self-selection and attrition decisions, and on self-reported evaluations from questionnaires when applied to a prototypical job training program.

Local Instrumental Variables

Article

Full-text available

May 2000

This paper unites the treatment effect literature and the latent variable literature. The economic questions answered by the commonly used treatment effect parameters are considered. We demonstrate how the marginal treatment effect parameter can be used in a latent variable framework to generate the average treatment effect, the effect of treatment on the treated and the local average treatment effect, thereby establishing a new relationship among these parameters. The method of local instrumental variables directly estimates the marginal treatment effect parameters, and thus can be used to estimate all of the conventional treatment effect parameters when the index condition holds and the parameters are identified. When they are not, the method of local instrumental variables can be used to produce bounds on the parameters with the width of the bounds depending on the width of the support for the index generating the choice of the observed potential outcome.

Estimating Distributions of Treatment Effects with an Application to the Returns to Schooling and Measurement of the Effects of Uncertainty on College Choice

Article

Full-text available

May 2003

This paper uses factor models to identify and estimate distributions of counterfactuals. We extend LISREL frameworks to a dynamic treatment effect setting, extending matching to account for unobserved conditioning variables. Using these models, we can identify all pairwise and joint treatment effects. We apply these methods to a model of schooling and determine the intrinsic uncertainty facing agents at the time they make their decisions about enrollment in school. Reducing uncertainty in returns raises college enrollment. We go beyond the “Veil of Ignorance” in evaluating educational policies and determine who benefits and loses from commonly proposed educational reforms.

Identification and Estimation of Local Average Treatment Effects

Article

Jan 1994
ECONOMETRICA

Simultaneous equations models with continuous and discrete endogenous variables and structural shifts

Article

Jan 1976

James J Heckman

Econometric evaluation of social programs

Article

Jan 2007
Handbook Econometrics

Alternative Methods for Evaluating the Impact of Interventions.

Article

Jan 1985

The Effects of Cognitive and Non-Cognitive Abilities on Labor Market Outcomes and Social Behavior

Article

Jul 2006
J LABOR ECON

This article establishes that a low-dimensional vector of cognitive and noncognitive skills explains a variety of labor market and behavioral outcomes. Our analysis addresses the problems of measurement error, imperfect proxies, and reverse causality that plague conventional studies. Noncognitive skills strongly influence schooling decisions and also affect wages, given schooling decisions. Schooling, employment, work experience, and choice of occupation are affected by latent noncognitive and cognitive skills. We show that the same low-dimensional vector of abilities that explains schooling choices, wages, employment, work experience, and choice of occupation explains a wide variety of risky behaviors.

The effects of cognitive and noncognitive abilities on labor market outcomes and socialbehavior

Article

Jan 2006
J LABOR ECON

Dealing with skewed data on costs and expenditures

Article

Jan 2006

W.G. Manning

Identification of Causal Effects Using Instrumental Variables

Article

Jun 1996
J AM STAT ASSOC

We outline a framework for causal inference in settings where assignment to a binary treatment is ignorable, but compliance with the assignment is not perfect so that the receipt of treatment is nonignorable. To address the problems associated with comparing subjects by the ignorable assignment - an "intention-to-treat analysis" - we make use of instrumental variables, which have long been used by economists in the context of regression models with constant treatment effects. We show that the instrumental variables (IV) estimand can be embedded within the Rubin Causal Model (RCM) and that under some simple and easily interpretable assumptions, the IV estimand is the average causal effect for a subgroup of units, the compliers. Without these assumptions, the IV estimand is simply the ratio of intention-to-treat causal estimands with no interpretation as an average causal effect. The advantages of embedding the IV approach in the RCM are that it clarifies the nature of critical assumptions needed for a causal interpretation, and moreover allows us to consider sensitivity of the results to deviations from key assumptions in a straightforward manner. We apply our analysis to estimate the effect of veteran status in the Vietnam era on mortality, using the lottery number that assigned priority for the draft as an instrument, and we use our results to investigate the sensitivity of the conclusions to critical assumptions.

Factors influencing a woman's choice to undergo breast conserving surgery versus modified radical mastectomy - Closing

Article

Dec 2000
AM J SURG

RJ Nold

Identification of Causal Effects Using Instrumental Variables: Comment

Article

Jun 1996

James J. Heckman

Does More Intensive Treatment of Acute Myocardial Infarction in the Elderly Reduce Mortality?

Article

Sep 1994

Mark McClellan

Objective. —To determine the effect of more intensive treatments on mortality in elderly patients with acute myocardial infarction (AMI).Design. —Analysis of incremental treatment effects using differential distances as instrumental variables to account for unobserved case-mix variation (selection bias) in observational Medicare claims data (1987 through 1991).Main Outcome Measures. —Survival to 4 years after AMI.Results. —Patients who receive different treatments differ in observable and unobservable health characteristics, biasing estimates of treatment effects based on standard methods of adjusting for observable differences. Patients' differential distances to alternative types of hospitals are strong independent predictors of how intensively an AMI patient will be treated and appear uncorrelated with health status. Thus, differential distances approximately randomize patients to different likelihoods of receiving intensive treatments. Comparisons of patient groups that differ only in differential distances show that the impact on mortality at 1 to 4 years after AMI of the incremental ("marginal") use of invasive procedures in Medicare patients was at most 5 percentage points; this gain was achieved during the first day of hospitalization and therefore appears attributable to treatments other than the procedures. Admission to a hospital treating a high volume of AMI patients was associated with an effect on mortality at 4 years of less than 1 percentage point, again arising on day 1. Patients living in rural areas experienced acute mortality that was an additional 0.6 percentage-point higher, after controlling for less access to intensive treatments.Conclusions. —For elderly patients with AMI, the aspects of treatment most affecting long-term survival relate to care within the first 24 hours of admission. The survival benefits from greater use of catheterization and revascularization procedures appear minimal in marginal patients.(JAMA. 1994;272:859-866)

The Design Experiments

Article

Jan 1960

R. A. S. Fisher

Measurements, Regression, and Calibration.

Article

Mar 1996

Philip J. Brown

A New Approach to Estimating Switching Regressions

Article

Jun 1972

Richard E. Quandt

In recent years much attention has been focussed on the problem of discontinuous shifts in regression regimes at unknown points in the data series. This article approaches this problem by assuming that nature chooses between regimes with probabilities λ and 1 — λ. This allows formulation of the appropriate likelihood function maximized with respect to the parameters in the regression equations and λ. The method is compared to another recent procedure in some sampling experiments and in a realistic economic problem and is found satisfactory.

The Design of Experiments

Book

Jan 1966

R. A. Fisher

The Econometrics of Disequilibrium

Article

Jan 1988

Richard E. Quandt

Sur les applications de la th'eorie des probabilit'es aux experiences agricoles: Essai des principes

Article

Jan 1923

J. Neyman

Statistical Problems in Agricultural Experimentation

Article

Jan 1935

Jerzy Neyman

Sample Selection Bias Specification Error

Article

Jan 1979
ECONOMETRICA

James J. Heckman

Perspectives on mean-based evaluation of health care

Article

Some Thoughts On The Distribution Of Earnings

Article

Jun 1951

AD Roy

Identification of Causal Effects Using Instrumental Variables (Disc: P456-472)

Article

Jan 1996

Empirical Strategies in Labor Economics

Article

Jul 1998
Handbook Labor Econ

This chapter provides an overview of the methodological and practical issues that arise when estimating causal relationships that are of interest to labor economists. The subject matter includes identification, data collection, and measurement problems. Four identification strategies are discussed, and five empirical examples — the effects of schooling, unions, immigration, military service, and class size — illustrate the methodological points. In discussing each example, we adopt an experimentalist perspective that emphasizes the distinction between variables that have causal effects, control variables, and outcome variables. The chapter also discusses secondary datasets, primary data collection strategies, and administrative data. The section on measurement issues focuses on recent empirical examples, presents a summary of empirical findings on the reliability of key labor market data, and briefly reviews the role of survey sampling weights and the allocation of missing values in empirical research.

Bayesian Inference for Causal Effects: The Role of Randomization

Article

Jan 1978
ANN STAT

Donald Rubin

Causal effects are comparisons among values that would have been observed under all possible assignments of treatments to experimental units. In an experiment, one assignment of treatments is chosen and only the values under that assignment can be observed. Bayesian inference for causal effects follows from finding the predictive distribution of the values under the other assignments of treatments. This perspective makes clear the role of mechanisms that sample experimental units, assign treatments and record data. Unless these mechanisms are ignorable (known probabilistic functions of recorded values), the Bayesian must model them in the data analysis and, consequently, confront inferences for causal effects that are sensitive to the specification of the prior distribution of the data. Moreover, not all ignorable mechanisms can yield data from which inferences for causal effects are insensitive to prior specifications. Classical randomized designs stand out as especially appealing assignment mechanisms designed to make inference for causal effects straightforward by limiting the sensitivity of a valid Bayesian analysis.

Nonparametric Estimation of Average Treatment Effects Under Exogeneity: A Review

Article

Feb 2004
REV ECON STAT

Guido W. Imbens

Recently there has been a surge in econometric work focusing on estimating average treatment effects under various sets of assumptions. One strand of this literature has developed methods for estimating average treatment effects for a binary treatment under assumptions variously described as exogeneity, unconfoundedness, or selection on observables. The implication of these assumptions is that systematic (for example, average or distributional) differences in outcomes between treated and control units with the same values for the covariates are attributable to the treatment. Recent analysis has considered estimation and inference for average treatment effects under weaker assumptions than typical of the earlier literature by avoiding distributional and functional-form assumptions. Various methods of semiparametric estimation have been proposed, including estimating the unknown regression functions, matching, methods using the propensity score such as weighting and blocking, and combinations of these approaches. In this paper I review the state of this literature and discuss some of its unanswered questions, focusing in particular on the practical implementation of these methods, the plausibility of this exogeneity assumption in economic applications, the relative performance of the various semiparametric estimators when the key assumptions (unconfoundedness and overlap) are satisfied, alternative estimands such as quantile treatment effects, and alternate methods such as Bayesian inference. Copyright (c) 2004 President and Fellows of Harvard College and the Massachusetts Institute of Technology.

Who Is the Marginal Patient? Understanding Instrumental Variables Estimates of Treatment Effects

Article

Jan 1999

To clarify the issues of generalizability arising from the use of instrumental variable (IV) methods to estimate treatment effects in nonexperimental medical outcome studies. We generate Monte Carlo data designed to resemble typical data sets where detailed health status information is unavailable and the treatment assignment process is unobserved. The model used to generate our data makes the realistic assumption that unobservable health status characteristics of patients influence the treatment assignment process and the effectiveness of treatment. We use Monte Carlo data to illustrate the circumstances where IV estimates generalize to an unobservable patient subpopulation and those where IV estimates generalize to the entire patient population represented by the sample used in the analysis. We also simulate the effect of two policy changes that affect practice patterns. Further, we show that IV estimates are useful for predicting the effect of these changes on treatment effectiveness when the subpopulation to which the IV estimate refers is the same or very similar to the population whose treatment status is affected by the policy change. Health services researchers cannot take for granted that IV estimates generalize to the same population represented by the sample used for analysis. Instead, researchers must rely on their knowledge of clinical practice and theory regarding the treatment assignment process in interpreting their results and in predicting the effect of changes in practice patterns.

The irrelevance of inference: A decision-making approach to the stochastic evaluation of health care technologies

Article

Jul 1999
J HEALTH ECON

Karl Claxton

The literature which considers the statistical properties of cost-effectiveness analysis has focused on estimating the sampling distribution of either an incremental cost-effectiveness ratio or incremental net benefit for classical inference. However, it is argued here that rules of inference are arbitrary and entirely irrelevant to the decisions which clinical and economic evaluations claim to inform. Decisions should be based only on the mean net benefits irrespective of whether differences are statistically significant or fall outside a Bayesian range of equivalence. Failure to make decisions in this way by accepting the arbitrary rules of inference will impose costs which can be measured in terms of resources or health benefits forgone. The distribution of net benefit is only relevant to deciding whether more information is required. A framework for decision making and establishing the value of additional information is presented which is consistent with the decision rules in CEA. This framework can distinguish the simultaneous but conceptually separate steps of deciding which alternatives should be chosen, given existing information, from the question of whether more information should be acquired. It also ensures that the type of information acquired is driven by the objectives of the health care system, is consistent with the budget constraint on service provision and that research is designed efficiently.

Nold RJ, Beamer L, Helmer SD, McBoyle MFFactors influencing a women's choice to undergo breast-conserving surgery versus mastectomy. Am J Surg 180: 413-418

Article

Jan 2001
AM J SURG

The use of breast-conserving surgery (BCS) rather than modified radical mastectomy (MRM) for the treatment of breast carcinoma is an option for the majority of women (75%) with early stage breast cancer, but only 20% to 50% choose to undergo this procedure nationwide. The objective of this study was to identify factors influencing a woman's choice between BCS and MRM, and specifically, the surgeon's influence on this choice. A total of 134 women eligible for BCS were sent a survey. Data obtained included demographics, influential factors in treatment choice, and satisfaction with preoperative discussion and postoperative results. Ninety-six women completed the questionnaire. Mean patient age was 62 years. Most women surveyed felt their treatment options were satisfactorily explained to them. BCS, MRM with reconstruction (MRM-R), and MRM without reconstruction (MRM-NR) were performed in 45%, 15%, and 40% of patients, respectively. Overall, the most influential factor was the fear of cancer. Women choosing BCS indicated that the surgeon, cosmetic result, and psychological aspects were more influential in their decision than in women undergoing MRM-NR (P <0.02). Fear of cancer was the most important factor affecting the choice to undergo MRM-NR. In comparing MRM-R with MRM-NR, there was a similar fear of cancer; however, MRM-R had much greater concern with cosmesis (P = 0.0002). The surgeon's input is important in a woman's choice to undergo BCS or MRM-R. However, it appears that if a woman wants to have MRM-NR, even when she is a candidate for BCS, the surgeon's input is overshadowed by the patient's fear of cancer.

Effectiveness of Chemotherapy for Advanced Lung Cancer in the Elderly: Instrumental Variable and Propensity Analysis

Article

Mar 2001

To compare the effectiveness of chemotherapy given to elderly patients in routine practice for stage IV non-small-cell lung cancer (NSCLC) with the efficacy observed in randomized trials. We used instrumental variable analysis (IVA) and propensity scores (PS) to simulate the conditions of a randomized trial in a retrospective cohort of patients over age 65 from the Survival, Epidemiology, and End Results (SEER) tumor registry. Geographic variation in chemotherapy use served as the instrument for the IVA analysis, and propensity scores were calculated with a logistic model based on patient disease and sociodemographic characteristics. Among 6,232 elderly patients, the instrumental variable estimate indicated an increase in median survival of 33 days and an improvement in 1-year survival of 9% attributable to chemotherapy. In a Cox regression model, chemotherapy administration was associated with a hazard ratio of 0.81 (95% confidence interval, 0.76 to 0.85). When survival was analyzed separately within propensity score quintiles, the hazard ratios were all similar, ranging from 0.78 to 0.85. These results are comparable with those of a large meta-analysis, which found a hazard ratio of 0.87 in the subgroup of patients over age 65. Chemotherapy for stage IV NSCLC seems to have effectiveness for elderly patients and those with comorbid conditions that is similar to the efficacy seen in randomized trials containing mostly younger, highly selected patients. All suitable patients should be given the opportunity to consider palliative chemotherapy for metastatic NSCLC.

An Exploratory Instrumental Variable Analysis of the Outcomes of Localized Breast Cancer Treatments in a Medicare Population

Article

Mar 2003

This study is motivated by the potential problem of using observational data to draw inferences about treatment outcomes when experimental data are not available. We compare two statistical approaches, ordinary least‐squares (OLS) and instrumental variables (IV) regression analysis, to estimate the outcomes (three‐year post‐treatment survival) of three treatments for early stage breast cancer in elderly women: mastectomy (MST), breast conserving surgery with radiation therapy (BCSRT), and breast conserving surgery only (BCSO). The primary data source was Medicare claims for a national random sample of 2907 women (age 67 or older) with localized breast cancer who were treated between 1992 and 1994. Contrary to randomized clinical trial (RCT) results, analysis with the observational data found highly significant differences in survival among the three treatment alternatives: 79.2% survival for BCSO, 85.3% for MST, and 93.0% for BCSRT. Using OLS to control for the effects of observable characteristics narrowed the estimated survival rate differences, which remained statistically significant. In contrast, the IV analysis estimated survival rate differences that were not significantly different from 0. However, the IV‐point estimates of the treatment effects were quantitatively larger than the OLS estimates, unstable, and not significantly different from the OLS results. In addition, both sets of estimates were in the same quantitative range as the RCT results. We conclude that unadjusted observational data on health outcomes of alternative treatments for localized breast cancer should not be used for cost‐effectiveness studies. Our comparisons suggest that whether one places greater confidence in the OLS or the IV results depends on at least three factors: (1) the extent of observable health information that can be used as controls in OLS estimation, (2) the outcomes of statistical tests of the validity of the instrumental variable method, and (3) the similarity of the OLS and IV estimates. In this particular analysis, the OLS estimates appear to be preferable because of the instability of the IV estimates. Copyright © 2002 John Wiley & Sons, Ltd.

Economic Evaluation of Breast Cancer Treatment: Considering the Value of Patient Choice

Article

Apr 2003

To use 5 years of primary data to compare the incremental cost-effectiveness of breast conservation and radiation versus mastectomy with the restriction of choice to a single therapy versus providing a choice of either therapy. We evaluated a random retrospective cohort of 2,517 Medicare beneficiaries treated for newly diagnosed stage I or II breast cancer from 1992 through 1994. The outcome measures were quality-adjusted life-years (QALYs) and 5-year medical costs. Risk and propensity score adjustments were used in the analysis. A breast conservation and radiation regimen has significantly higher costs than mastectomy in the first year after surgery; the adjusted 5-year costs are $14,054 (95% confidence interval, $9,791 to $18,312) greater than those of mastectomy. The adjusted incremental cost-effectiveness ratio comparing breast conservation and radiation to mastectomy was $219,594 per QALY for the comparison of the two strategies. If the possibility of patient choice from maintaining the availability of multiple treatments versus restricting choice to mastectomy alone provides a quality-of-life gain of 0.031 QALYs, then the cost-effectiveness ratio of this choice option is $80,440 per QALY. The current system of providing a choice between mastectomy and breast conservation surgery is economically attractive when the economic analysis includes the benefit of patient choice of treatment.

Was Breast Conserving Surgery Underutilized for Early Stage Breast Cancer? Instrumental Variables Evidence for Stage II Patients from Iowa

Article

Jan 2004

To estimate the average survival effects of breast conserving surgery plus irradiation relative to mastectomy for marginal stage II breast cancer patients in Iowa from 1989-1994. DATA SOURCES/DATA SETTING: Secondary linked Iowa SEER Cancer Registry--Iowa Hospital Association discharge abstract data for women in Iowa with stage II breast cancer from 1989-1994. Observational instrumental variables (IV) analysis. Women with stage II breast cancer from the Iowa SEER Cancer Registry 1989-1994 who received all of their inpatient care in Iowa were linked with their respective hospital discharge abstracts. Breast conserving surgery plus irradiation decreased survival relative to mastectomy for marginal stage II breast cancer patients in Iowa during the early 1990s. In this study marginal patients were those whose surgery choices were affected by differences in area treatment rates and access to radiation facilities. If marginal patients are representative of patients whose treatment choices would be affected by changes in treatment rates, an increase in the breast conserving surgery plus irradiation rate for stage II early stage breast cancer patients would have decreased survival in Iowa during the early 1990s. Further research with newer data and broader samples is needed to make more current and specific assessments.

Comparing alternative models: Log vs Cox proportional hazard?

Article

Aug 2004

Health economists often use log models (based on OLS or generalized linear models) to deal with skewed outcomes such as those found in health expenditures and inpatient length of stay. Some recent studies have employed Cox proportional hazard regression as a less parametric alternative to OLS and GLM models, even when there was no need to correct for censoring. This study examines how well the alternative estimators behave econometrically in terms of bias when the data are skewed to the right. Specifically we provide evidence on the performance of the Cox model under a variety of data generating mechanisms and compare it to the estimators studied recently in Manning and Mullahy (2001). No single alternative is best under all of the conditions examined here. However, the gamma regression model with a log link seems to be more robust to alternative data generating mechanisms than either OLS on ln(y) or Cox proportional hazards regression. We find that the proportional hazard assumption is an essential requirement to obtain consistent estimate of the E(y|x) using the Cox model.

Generalized Modeling Approaches to Risk Adjustment of Skewed Outcomes Data

Article

Jun 2005
J HEALTH ECON

There are two broad classes of models used to address the econometric problems caused by skewness in data commonly encountered in health care applications: (1) transformation to deal with skewness (e.g., ordinary least square (OLS) on ln(y)); and (2) alternative weighting approaches based on exponential conditional models (ECM) and generalized linear model (GLM) approaches. In this paper, we encompass these two classes of models using the three parameter generalized Gamma (GGM) distribution, which includes several of the standard alternatives as special cases-OLS with a normal error, OLS for the log-normal, the standard Gamma and exponential with a log link, and the Weibull. Using simulation methods, we find the tests of identifying distributions to be robust. The GGM also provides a potentially more robust alternative estimator to the standard alternatives. An example using inpatient expenditures is also analyzed.

Using Matching, Instrumental Variables and Control Functions to Estimate Economic Choice Models

Article

May 2003
REV ECON STAT

This paper investigates four topics. (1) It examines the different roles played by the propensity score (probability of selection) in matching, instrumental variable and control functions methods. (2) It contrasts the roles of exclusion restrictions in matching and selection models. (3) It characterizes the sensitivity of matching to the choice of conditioning variables and demonstrates the greater robustness of control function methods to misspecification of the conditioning variables. (4) It demonstrates the problem of choosing the conditioning variables in matching and the failure of conventional model selection criteria when candidate conditioning variables are not exogenous.

2001 Lawrence R. Klein Lecture Estimating Distributions of Treatment Effects with an Application to the Returns to Schooling and Measurement of the Effects of Uncertainty on College Choice

Article

May 2003
INT ECON REV

This article uses factor models to identify and estimate the "distributions" of counterfactuals. We extend "LISREL" frameworks to a dynamic treatment effect setting, extending matching to account for unobserved conditioning variables. Using these models, we can identify all pairwise and joint treatment effects. We apply these methods to a model of schooling and determine the intrinsic uncertainty facing agents at the time they make their decisions about enrollment in school. We go beyond the "Veil of Ignorance" in evaluating educational policies and determine who benefits and who loses from commonly proposed educational reforms. Copyright 2003 By The Economics Department Of The University Of Pennsylvania And Osaka University Institute Of Social And Economic Research Association.

Use of Instrumental Variables in the Presence of Heterogeneity and Self-Selection: An Application to Treatments of Breast Cancer Patients

Abstract

No full-text available

Recommended publications

MammoSite® Radiation Therapy System

Angiosarcoma after tylectomy and radiation therapy for carcinoma of the breast

Surgery or radiation therapy: A comparative cost analysis for early carcinoma of the prostate and br...

Lumpectomy and radiation therapy for breast cancer