A review of the application of propensity score methods yielded increasing use, advantages in specific settings, but not substantially different estimates compared with conventional multivariable methods

Prevalence and survival benefit of adjuvant chemotherapy in stage III colon cancer patients: Comparison of overall and age-stratified results by multivariable modeling and propensity score methodology in a population-based cohort

BACKGROUND

Few population-based studies have assessed the effectiveness of adjuvant chemotherapy (ACT) in stage III colon cancer patients according to age. We sought to quantify the prevalence of ACT use and the absolute and relative survival benefit of ACT overall and by age in a population-based cohort.

METHODS

Stage III patients with adenocarcinoma of the colon identified by the Georgia Comprehensive Cancer Registry for the years 2000-07 were eligible (final N=3057). We utilized Poisson regression to obtain adjusted mortality rates (MR) and Cox proportional hazards models to obtain adjusted hazard ratios (HRs) for 5-year overall survival. We evaluated control of confounding by comparing HRs obtained via multivariable modeling (MM), propensity score weighting (PSW), and propensity score matching (PSM).

RESULTS

Just over one-third of colon cancer patients did not receive ACT, and the proportion increased with age. Overall, receipt of ACT conferred an absolute (MR difference [No ACT rate-ACT rate] 25.4 deaths/1000 person-years [py], 95% confidence interval [CI]: 19.1-32.7 deaths/1000 py) and relative (MM HR=0.67, 95% CI: 0.59-0.76) survival benefit. The survival benefit was demonstrated across age groups. MM and propensity score methods yielded highly similar HRs.

CONCLUSION

Unless contraindicated, efforts to ensure receipt of ACT for stage III colon cancer patients up to 84 years of age are needed to improve the prognosis of patients with node-positive disease.

Outcomes of Induction Chemotherapy Plus Intensity-Modulated Radiotherapy (IMRT) Versus IMRT Plus Concurrent Chemotherapy for Locoregionally Advanced Nasopharyngeal Carcinoma: A Propensity Matched Study

PURPOSE

It deserves investigation whether induction chemotherapy (IC) followed by intensity-modulated radiotherapy (IMRT) is inferior to the current standard of IMRT plus concurrent chemotherapy (CC) in locoregionally advanced nasopharyngeal carcinoma.

METHODS

Patients who received IC (94 patients) or CC (302 patients) plus IMRT at our center between March 2003 and November 2012 were retrospectively analyzed. Propensity-score matching method was used to match patients in both arms at equal ratio. Failure-free survival (FFS), overall survival (OS), distant metastasis-free survival (DMFS), and locoregional relapse-free survival (LRFS) were assessed with Kaplan-Meier method, log-rank test, and Cox regression.

RESULTS

In the original cohort of 396 patients, IC plus IMRT resulted in similar FFS (P = .565), OS (P = .334), DMFS (P = .854), and LRFS (P = .999) to IMRT plus CC. In the propensity-matched cohort of 188 patients, no significant survival differences were observed between the two treatment approaches (3-year FFS 80.3% vs 81.0%, P = .590; OS 93.4% vs 92.1%, P = .808; DMFS 85.9% vs 87.7%, P = .275; and LRFS 93.1% vs 92.0%, P = .763). Adjusting for the known prognostic factors in multivariate analysis, IC plus IMRT did not cause higher risk of treatment failure, death, distant metastasis, or locoregional relapse.

CONCLUSIONS

IC plus IMRT appeared to achieve comparable survival to IMRT plus CC in locoregionally advanced nasopharyngeal carcinoma. Further investigations were warranted.

Corticosteroid Dosing in Pediatric Acute Severe Ulcerative Colitis: A Propensity Score Analysis

OBJECTIVES

We aimed to explore the optimal dosing of intravenous-corticosteroids (IVCS) using a robust statistical method on the largest pediatric cohort of acute severe colitis to date.

METHODS

Two hundred eighty-three children treated with IVCS for ulcerative colitis were included and studied for 1 year (46% boys, age 12.1 ± 3.9 years, disease duration 2 (interquartile range [IQR] 0-14) months, baseline Pediatric Ulcerative Colitis Activity Index 69 ± 13 points). Confounding by indication was addressed by matching high- and low-IVCS dose patients according to the propensity score method, using 3 cutoffs (1 mg · kg · methylprednisolone to 40 mg · day, 1.25 mg · kg to 50 mg · day and 2 mg · kg to 80 mg · day).

RESULTS

The median IVCS dose in the entire cohort was 1.0 mg · kg · day (IQR 0.8-1.4) and 44 mg · kg · day (32-60). Ninety-four of 283 children were matched in the low-dose cutoff (1 mg · kg · day), 218 of 283 were matched in the middle cutoff (1.25 mg · kg · day), and 86/283 in the high dose cutoff (2 mg · kg · day). No differences were found in 25 pretreatment baseline variables in the three cutoffs, implying successful matching. There were no statistical differences in the outcomes of the two lower cutoffs (including need for salvage therapy during admission and by 1 years, admission duration, and day-5 Pediatric Ulcerative Colitis Activity Index<35 points; all P > 0.05). In the high cutoff, the higher doses were somewhat better but this benefit reversed in a sensitivity analysis excluding one center. High doses were not associated with better outcome also in a propensity score-weighted regression model on the entire cohort.

CONCLUSIONS

Our data support present guidelines that doses of IVCS >1 to 1.5 mg · kg · day (maximum 40-60 mg · kg · day) are not justified in acute severe colitis.

A Comparison of Health Plan- and Provider-Delivered Chronic Care Management Models on Patient Clinical Outcomes

BACKGROUND

The real world implementation of chronic care management model varies greatly. One aspect of this variation is the delivery mode. Two contrasting strategies include provider-delivered care management (PDCM) and health plan-delivered care management (HPDCM).

OBJECTIVE

We aimed to compare the effectiveness of PDCM vs. HPDCM on improving clinical outcomes for patients with chronic diseases.

DESIGN

We used a quasi-experimental two-group pre-post design using the difference-in-differences method.

PATIENTS

Commercially insured patients, with any of the five chronic diseases-congestive heart failure, chronic obstructive pulmonary disease, coronary heart disease, diabetes, or asthma, who were outreached to and engaged in either PDCM or HPDCM were included in the study.

MAIN MEASURES

Outreached patients were those who received an attempted or actual contact for enrollment in care management; and engaged patients were those who had one or more care management sessions/encounters with a care manager. Effectiveness measures included blood pressure, low density lipoprotein (LDL), weight loss, and hemoglobin A1c (for diabetic patients only). Primary endpoints were evaluated in the first year of follow-up.

KEY RESULTS

A total of 4,000 patients were clustered in 165 practices (31 in PDCM and 134 in HPDCM). The PDCM approach demonstrated a statistically significant improvement in the proportion of outreached patients whose LDL was under control: the proportion of patients with LDL < 100 mg/dL increased by 3 % for the PDCM group (95 % CI: 1 % to 6 %) and 1 % for the HPDCM group (95 % CI: -2 % to 5 %). However, the 2 % difference in these improvements was not statistically significant (95 % CI: -2 % to 6 %). The HPDCM approach showed 3 % [95 % CI: 2 % to 6 %] improvement in overall diabetes care among outreached patients and significant reduction in obesity rates compared to PDCM (4 %, 95 % CI: 0.3 % to 8 %).

CONCLUSIONS

Both care management delivery modes may be viable options for improving care for patients with chronic diseases. In this commercially insured population, neither PDCM nor HPDCM resulted in substantial improvement in patients' clinical indicators in the first year. Different care management strategies within the provider-delivered programs need further investigation.

Poor, persecuted, young, and alone: Toward explaining the elevated risk of alcohol problems among Black and Latino men who drink

BACKGROUND

Even given equivalent drinking patterns, Black and Latino men experience substantially more dependence symptoms and other consequences than White men, particularly at low/no heavy drinking. No known studies have identified factors driving these disparities. The current study examines this question.

METHODS

The 2005 and 2010 National Alcohol Surveys were pooled. Surveys are nationally representative, telephone interviews of the U.S. including Black and Latino oversamples; male drinkers were analyzed (N=4182). Preliminary analyses included negative binomial regressions of dependence symptom and consequence counts testing whether effects for race/ethnicity were diminished when entering potential explanatory factors individually. Additional analyses re-examined effects for race/ethnicity when using propensity score weighting to weight Blacks to Whites, and Latinos to Whites, first on heavy drinking alone, and then on heavy drinking and all explanatory factors supported by preliminary analyses.

RESULTS

Preliminary regressions suggested roles for lower individual SES, greater prejudice and unfair treatment, and younger age in the elevated risk of alcohol problems among Black and Latino (vs. White) men at low heavy drinking levels; additional support emerged for single (vs. married) status among Blacks and neighborhood disadvantage among Latinos. When Blacks and Latinos were weighted to Whites on the above variables, effects for race/ethnicity on dependence counts were reduced to nonsignificance, while racial/ethnic disparities in consequence counts were attenuated (by >43% overall).

CONCLUSIONS

Heavy drinking may be especially risky for those who are poor, exposed to prejudice and unfair treatment, young, and unmarried, and these factors may contribute to explaining racial/ethnic disparities in alcohol problems.

Antipsychotic Adherence and Rehospitalization in Schizophrenia Patients Receiving Oral Versus Long-Acting Injectable Antipsychotics Following Hospital Discharge

BACKGROUND

Antipsychotic medications are a central component of effective treatment for schizophrenia, but nonadherence is a significant problem for the majority of patients. Long-acting injectable (LAI) antipsychotic medications are a recommended treatment option for nonadherent patients, but evidence regarding their potential advantages has been mixed. Observational data on newer, second-generation LAI antipsychotic medications have been limited given their more recent regulatory approval and availability.

OBJECTIVE

To examine antipsychotic medication nonadherence, discontinuation, and rehospitalization outcomes in Medicaid patients receiving oral versus LAI antipsychotic medications in the 6 months after a schizophrenia-related hospitalization.

METHODS

The 2010-2013 Truven Health Analytics MarketScan Medicaid research claims database was used to identify adult patients with a recent history of nonadherence (prior 6 months) who received an oral or LAI antipsychotic medication within 30 days after an index schizophrenia-related hospitalization. Primary outcome measures were nonadherence (proportion of days covered less than  0.80), discontinuation (continuous medication gap ≥ 60 days), and schizophrenia-related rehospitalization, all in the 6 months after discharge. Descriptive analyses compared users of oral versus LAI antipsychotic medication on sociodemographic, clinical, and treatment characteristics. Logistic regressions were used to examine associations between use of oral versus LAI antipsychotics and each study outcome while controlling for observed differences in sample characteristics. All outcomes were compared at 3 levels of analysis: overall LAI class, LAI antipsychotic generation (first-generation [FGA] or second-generation [SGA] antipsychotics), and individual LAI agent (fluphenazine decanoate, haloperidol decanoate, risperidone LAI, and paliperidone palmitate).

RESULTS

Of the final sample, 91% (n = 3,428) received oral antipsychotics, and 9.0% (n = 340) received LAI antipsychotics after discharge. Slightly over half (n =183, 53.8%) of LAI users used an SGA LAI. A smaller percentage of patients receiving LAIs were nonadherent (51.8% vs. 67.7%, P  less than  0.001); had a 60-day continuous gap in medication (23.8% vs. 39.4%, P  less than  0.001); and were rehospitalized for schizophrenia (19.1% vs. 25.3%, P = 0.01) compared with patients receiving oral medications. The size of these differences was magnified when comparing SGA LAI users with users of oral antipsychotics for nonadherence. After controlling for all differences in measured covariates, LAI initiators had lower odds of being nonadherent (adjusted odds ratio [AOR] = 0.35, 95% CI = 0.27-0.46, P  less than  0.001) and of having continuous 60-day gaps (AOR = 0.45, 95% CI = 0.34-0.60, P  less than  0.001) when compared with patients receiving oral medications. Both FGA and SGA LAI users had lower odds of nonadherence compared with patients receiving oral antipsychotics. Similarly, FGA LAI users (AOR = 0.58, 95% CI = 0.40-0.85, P = 0.005) and SGA LAI initiators (AOR = 0.34, 95% CI =0.23-0.51, P  less than  0.001) had lower odds of a 60-day continuous gap compared with patients receiving oral antipsychotics. Compared with those receiving oral antipsychotics, LAI initiators also had lower odds of rehospitalization (AOR = 0.73, 95% CI = 0.54-0.99, P = 0.041); however, when examined separately, only patients receiving SGA LAIs (AOR = 0.59, 95% CI = 0.38-0.90, P = 0.015) and not FGA LAIs (AOR = 0.90, 95% CI = 0.60-1.34, P = 0.599) had a statistically significant reduction in odds of rehospitalization. Among individual LAIs, odds of rehospitalization only among initiators of paliperidone palmitate were statistically different from those among users of oral antipsychotics (AOR = 0.53, 95% CI = 0.30-0.94, P = 0.031). While odds of rehospitalization were 33% lower among patients receiving risperidone LAI compared with those receiving oral antipsychotics, the estimate did not reach statistical significance (AOR = 0.67, 95% CI = 0.37-1.22, P = 0.194). 

CONCLUSIONS

This claims-based analysis of posthospitalization adherence and rehospitalization outcomes in Medicaid patients with schizophrenia adds to the growing real-world evidence base of the benefits of LAI antipsychotic medications in routine clinical practice, particularly with regard to second-generation LAIs. As new SGA formulations become available for long-acting use, real-world studies with larger sample sizes will be needed to further delineate their potential advantages in terms of clinical outcomes and costs.

Pharmacosurveillance without borders: electronic health records in different countries can be used to address important methodological issues in estimating the risk of adverse events

OBJECTIVES

Evaluate methodological advantages and limitations of an international pharmacosurveillance system based on electronic health records (EHRs).

STUDY DESIGN AND SETTINGS

Type 2 diabetes was used as an exemplar. Cohorts of newly treated diabetics were followed in each country (Quebec, Canada; Massachusetts, United States; Manchester, UK) from 2009 to 2012 using local EHR systems. Cox proportional hazards models were used to assess the risk of cardiovascular events.

RESULTS

A total of 44,913 newly treated diabetics were identified; 82.6% (United States) to 93.1% (Canada) were started on biguanides; 13% of patients failed to fill initial prescriptions. An increased risk of cardiovascular events with sulfonylureas was observed when dispensing [hazard ratio (HR): 2.83] vs. EHR prescribing (HR: 2.47) data were used. The addition of clinical data produced a threefold to 10-fold increase in comorbidity for obesity and renal disease, but had no impact on the risk of different hypoglycemic therapies. The risk of cardiovascular events with sulfonylureas was higher in the United States [HR: 3.4; 95% confidence interval (CI): 2.1, 5.5] compared to England (HR: 1.3; 95% CI: 1.1, 1.6).

CONCLUSION

An international surveillance system based on EHRs may provide more timely information about drug safety and new opportunities to estimate potential sources of bias and health system effects on drug-related outcomes.

Get the most from your data: a propensity score model comparison on real-life data

Purpose

In the past, the propensity score has been in the middle of several discussions in terms of its abilities and limitations. With a comprehensive review and a practical example, this study examines the effect of propensity score analysis of real-life data and introduces a simple and effective clinical approach.

Materials and methods

After the authors reviewed current publications, they applied their insights to the data of a nonrandomized clinical trial in bariatric surgery. This study examined weight loss in 173 patients where 127 patients received Roux-en-Y gastric bypass surgery and 46 patients sleeve gastrectomy. Both groups underwent analysis in terms of their covariate distribution using Mann–Whitney U and χ² testing. Mean differences within excess weight loss in native data were examined with Student’s t-test. Three propensity score models were defined and matching was performed. Covariate distribution and mean differences in excess weight loss were checked with Mann–Whitney U and χ² testing.

Results

Native data implied a significant difference in excess weight loss. The propensity score models did not confirm this difference. All models proved that both surgical procedures were equal, due to their weight-loss induction. Covariate distribution improved after the matching procedure in terms of an equal distribution.

Conclusion

It seemed that a practical clinical approach with outcome-related covariates as a propensity score base is the ideal midpoint between an equal distribution in covariates and an acceptable loss of data. Nevertheless, propensity score models designed with clinical intent seemed to be absolutely suitable for overcoming heterogeneity in covariate distribution.

Not All Abdomens Are the Same: A Comparison of Damage Control Surgery for Intra-abdominal Sepsis versus Trauma

Damage control surgery (DCS) was developed to manage exsanguinating trauma patients, but is increasingly applied to the management of peritoneal sepsis and abdominal catastrophes. Few manuscripts compare the outcomes of these surgeries on disparate patient populations. A multi-institutional three group propensity score matched case cohort study comparing penetrating trauma (PT-DCS), blunt trauma (BT-DCS), and intraperitoneal sepsis (IPS-DCS) was performed comparing patients treated with DSC between 2008 and 2013. Propensity scoring was performed using demographic and presenting physiologic data. Four hundred and twelve patients were treated with DCS across two institutions. Propensity matching for age, gender, and initial Acute Physiology and Chronic Health Evaluation II score 80 identified 80 patients per group for comparison. Rate of primary fascial closure was lowest in the IPS-DCS group, and highest in the penetrating trauma DCS group. Intra-abdominal complication rates were highest in the IPS-DCS group. IPS-DCS had increased time to definitive closure compared with the other two groups (RR 1.8; 1.3–2.2; P < 0.03). Mortality at 90 days was highest in the IPS-DCS group and patients whose definitive closure was delayed >eight days were more than twice the risk of death at 90 days across all groups. (RR 2.15; 1.2–3.5; P < 0.002). Expected outcomes after the use of DCS for trauma and emergency general surgery are quite different. Despite this difference, prompt abdominal closure at the earliest possible opportunity afforded the best outcome in patients managed via DCS.

Proton Pump Inhibitors and Risk of Incident CKD and Progression to ESRD

The association between proton pump inhibitors (PPI) use and risk of acute interstitial nephritis has been described. However, whether exposure to PPI associates with incident CKD, CKD progression, or ESRD is not known. We used Department of Veterans Affairs national databases to build a primary cohort of new users of PPI (n=173,321) and new users of histamine H₂-receptor antagonists (H₂ blockers; n=20,270) and followed these patients over 5 years to ascertain renal outcomes. In adjusted Cox survival models, the PPI group, compared with the H₂ blockers group, had an increased risk of incident eGFR<60 ml/min per 1.73 m² and of incident CKD (hazard ratio [HR], 1.22; 95% confidence interval [95% CI], 1.18 to 1.26; and HR, 1.28; 95% CI, 1.23 to 1.34, respectively). Patients treated with PPI also had a significantly elevated risk of doubling of serum creatinine level (HR, 1.53; 95% CI, 1.42 to 1.65), of eGFR decline >30% (HR, 1.32; 95% CI, 1.28 to 1.37), and of ESRD (HR, 1.96; 95% CI, 1.21 to 3.18). Furthermore, we detected a graded association between duration of PPI exposure and risk of renal outcomes among those exposed to PPI for 31-90, 91-180, 181-360, and 361-720 days compared with those exposed for ≤30 days. Examination of risk of renal outcomes in 1:1 propensity score-matched cohorts of patients taking H₂ blockers versus patients taking PPI and patients taking PPI versus controls yielded consistent results. Our results suggest that PPI exposure associates with increased risk of incident CKD, CKD progression, and ESRD.

Dimension reduction and shrinkage methods for high dimensional disease risk scores in historical data

Background

Multivariable confounder adjustment in comparative studies of newly marketed drugs can be limited by small numbers of exposed patients and even fewer outcomes. Disease risk scores (DRSs) developed in historical comparator drug users before the new drug entered the market may improve adjustment. However, in a high dimensional data setting, empirical selection of hundreds of potential confounders and modeling of DRS even in the historical cohort can lead to over-fitting and reduced predictive performance in the study cohort. We propose the use of combinations of dimension reduction and shrinkage methods to overcome this problem, and compared the performances of these modeling strategies for implementing high dimensional (hd) DRSs from historical data in two empirical study examples of newly marketed drugs versus comparator drugs after the new drugs’ market entry—dabigatran versus warfarin for the outcome of major hemorrhagic events and cyclooxygenase-2 inhibitor (coxibs) versus nonselective non-steroidal anti-inflammatory drugs (nsNSAIDs) for gastrointestinal bleeds.

Results

Historical hdDRSs that included predefined and empirical outcome predictors with dimension reduction (principal component analysis; PCA) and shrinkage (lasso and ridge regression) approaches had higher c-statistics (0.66 for the PCA model, 0.64 for the PCA + ridge and 0.65 for the PCA + lasso models in the warfarin users) than an unreduced model (c-statistic, 0.54) in the dabigatran example. The odds ratio (OR) from PCA + lasso hdDRS-stratification [OR, 0.64; 95 % confidence interval (CI) 0.46–0.90] was closer to the benchmark estimate (0.93) from a randomized trial than the model without empirical predictors (OR, 0.58; 95 % CI 0.41–0.81). In the coxibs example, c-statistics of the hdDRSs in the nsNSAID initiators were 0.66 for the PCA model, 0.67 for the PCA + ridge model, and 0.67 for the PCA + lasso model; these were higher than for the unreduced model (c-statistic, 0.45), and comparable to the demographics + risk score model (c-statistic, 0.67).

Conclusions

hdDRSs using historical data with dimension reduction and shrinkage was feasible, and improved confounding adjustment in two studies of newly marketed medications.

Electronic supplementary material

The online version of this article (doi:10.1186/s12982-016-0047-x) contains supplementary material, which is available to authorized users.

On the use of propensity scores in case of rare exposure

Background

Observational post-marketing assessment studies often involve evaluating the effect of a rare treatment on a time-to-event outcome, through the estimation of a marginal hazard ratio. Propensity score (PS) methods are the most used methods to estimate marginal effect of an exposure in observational studies. However there is paucity of data concerning their performance in a context of low prevalence of exposure.

Methods

We conducted an extensive series of Monte Carlo simulations to examine the performance of the two preferred PS methods, known as PS-matching and PS-weighting to estimate marginal hazard ratios, through various scenarios.

Results

We found that both PS-weighting and PS-matching could be biased when estimating the marginal effect of rare exposure. The less biased results were obtained with estimators of average treatment effect in the treated population (ATT), in comparison with estimators of average treatment effect in the overall population (ATE). Among ATT estimators, PS-weighting using ATT weights outperformed PS-matching. These results are illustrated using a real observational study.

Conclusions

When clinical objectives are focused on the treated population, applied researchers are encouraged to estimate ATT with PS-weighting for studying the relative effect of a rare treatment on time-to-event outcomes.

Electronic supplementary material

The online version of this article (doi:10.1186/s12874-016-0135-1) contains supplementary material, which is available to authorized users.

Second-Line Intraperitoneal Chemotherapy for Recurrent Epithelial Ovarian, Tubal and Peritoneal Cancer: A Propensity Score-Matching Study

BACKGROUND

The superiority of frontline intraperitoneal (IP) over intravenous (IV) chemotherapy is well established in the treatment of epithelial ovarian cancer. However, the role of IP chemotherapy in the second-line setting has rarely been investigated.

METHODS

Consecutive patients diagnosed with recurrent epithelial, tubal and peritoneal cancers between January 2000 and December 2012 were recruited using a propensity score-matching technique to adjust relevant risk factors.

RESULTS

In total, 310 patients were included in the final analysis (94 for platinum-refractory/resistant disease and 216 for platinum-sensitive disease). IP chemotherapy demonstrated significantly longer median progression-free survival than IV chemotherapy (4.9 vs. 2.4 months, p < 0.001, for platinum-refractory/resistant disease, and 9.8 vs. 6.9 months, p < 0.001, for platinum-sensitive disease).

CONCLUSIONS

Second-line IP chemotherapy confers longer progression-free survival than IV chemotherapy. Large-scale clinical trials should be conducted to validate the true efficacy.

Current state of clinical end-points assessment in transplant: Key points

Solid organ transplantation is the treatment of choice for patients with end-stage organ disease. However, organ transplantation can stress the cardiovascular system and decrease immune surveillance, leading to early mortality and graft loss due to multiple underlying comorbidities. Clinical end-points in transplant include death and graft failure. Thus, generating accurate predictive models through regression models is crucial to test for definitive clinical post-transplantation end-points. Survival predictive models should assemble efficient surrogate markers or prognostic factors to generate a minimal set of variables derived from a proper modeling strategy through regression models. However, a few critical points should be considered when reporting survival analyses and regression models to achieve proper discrimination and calibration of the predictive models. Additionally, population-based risk scores may underestimate risk prediction in transplant. The application of predictive models in these patients should therefore incorporate both classical and non-classical risk factors, as well as community-based health indicators and transplant-specific factors to quantify the outcomes in terms of survival properly. This review focuses on assessment of clinical end-points in transplant through regression models by combining predictive and surrogate variables, and considering key points in these analyses to accurately predict definitive end-points, which could aid clinicians in decision making.

Propensity score matching for selection of local areas as controls for evaluation of effects of alcohol policies in case series and quasi case-control designs

OBJECTIVES

Area-level public health interventions can be difficult to evaluate using natural experiments. We describe the use of propensity score matching (PSM) to select control local authority areas (LAU) to evaluate the public health impact of alcohol policies for (1) prospective evaluation of alcohol policies using area-level data, and (2) a novel two-stage quasi case-control design.

STUDY DESIGN

Ecological.

METHODS

Alcohol-related indicator data (Local Alcohol Profiles for England, PHE Health Profiles and ONS data) were linked at LAU level. Six LAUs (Blackpool, Bradford, Bristol, Ipswich, Islington, and Newcastle-upon-Tyne) as sample intervention or case areas were matched to two control LAUs each using PSM. For the quasi case-control study a second stage was added aimed at obtaining maximum contrast in outcomes based on propensity scores. Matching was evaluated based on average standardized absolute mean differences (ASAM) and variable-specific P-values after matching.

RESULTS

The six LAUs were matched to suitable control areas (with ASAM < 0.20, P-values >0.05 indicating good matching) for a prospective evaluation study that sought areas that were similar at baseline in order to assess whether a change in intervention exposure led to a change in the outcome (alcohol related harm). PSM also generated appropriate matches for a quasi case-control study--whereby the contrast in health outcomes between cases and control areas needed to be optimized in order to assess retrospectively whether differences in intervention exposure were associated with the outcome.

CONCLUSIONS

The use of PSM for area-level alcohol policy evaluation, but also for other public health interventions, will improve the value of these evaluations by objective and quantitative selection of the most appropriate control areas.

Anastomotic leakage after curative rectal cancer resection has no impact on long-term survival: a propensity score analysis

PURPOSE

Anastomotic leakage (AL) is a severe and frequent complication of rectal cancer resection, with an incidence rate of approximately 9 %. Although the impact of AL on morbidity and short-term mortality has been established, the literature is contradictory regarding its influence on long-term, cancer-specific survival. The present investigation assessed the long-term survival of 584 patients with stage I-III rectal cancer.

METHODS

The 10-year overall survival and cancer-specific survival were analyzed in 584 patients from a single tertiary center. All patients had undergone curative rectal cancer resection between 1991 and 2010. Patients with and without AL were compared using both a multivariate Cox hazards model and propensity score analysis.

RESULTS

A total of 64 patients developed AL (11.0 %, 95 % confidence interval (CI) = 8.7 to 13.8 %). The median follow-up was 5.2 years for all patients; and 7.4 years for patients still alive at the end of the investigated period. AL did persistently not impair cancer-specific survival based on unadjusted Cox regression (hazard ratio of death (HR) = 1.27, 95 % CI = 0.65 to 2.48, P = 0.489); risk-adjusted Cox regression (HR = 1.10, 95 % CI = 0.54 to 2.20, P = 0.799); and propensity score matching (HR = 1.18, 95 % CI = 0.57 to 2.43, P = 0.660).

CONCLUSIONS

Based on the present propensity score analysis, the oncologic outcomes in patients undergoing curative rectal cancer resections were not impaired by the development of anastomotic leakage.

Developmental differences in intra-individual variability in children with ADHD and ASD

BACKGROUND

Intra-individual variability reflects temporal variation within an individual's performance on a cognitive task. Children with developmental disorders, such as ADHD and ASD show increased levels of intra-individual variability. In typical development, intra-individual variability decreases sharply between the ages 6 and 20. The tight link between intra-individual variability and age has led to the suggestion that it may be marker of neural development. As there is accumulating evidence that ADHD and ASD are characterised by atypical neurodevelopmental trajectories, we set out to explore developmental changes in intra-individual variability in subjects with ADHD and ASD.

METHOD

We used propensity score matching to match a cross-sectional sample of children with ADHD, ASD and control subjects (N = 405, aged 6-19 years old) for age, IQ and gender. We used ex-Gaussian distribution parameters to characterise intra-individual variability on fast responses (sigma) and slow responses (tau).

RESULT

Results showed that there was a similar decrease in mean response times with age across groups, and an interaction between age and group for measures of variability, where there was a much lower rate of change in the variability parameters (sigma and tau) for subjects with ASD compared with the other two groups. Subjects with ADHD had higher intra-individual variability, reflected by both sigma and tau, but the rate of decrease in variability with age was similar to that of the controls.

CONCLUSION

These results suggest that subjects with ADHD, ASD and controls differ in the rate at which intra-individual variability decreases during development, and support the idea that intra-individual variability may be a marker of neural development, mimicking the neurodevelopmental changes in these disorders.

The effect of renin-angiotensin system blockade on the incidence of end-stage renal disease in IgA nephropathy

BACKGROUND

The impact of renin-angiotensin system blockade (RASB) on the incidence of end-stage renal disease (ESRD) remains unclear in IgA nephropathy (IgAN).

METHODS

This study assessed associations between RASB treatment and the incidence of ESRD in IgAN using propensity score approaches. We retrospectively analyzed 1273 patients with IgAN biopsied between 1979 and 2010. Propensity scores were calculated using logistic regression. Associations between RASB and ESRD were examined using a Cox regression model adjusted by inverse probability of treatment weighted, regression, stratification and matching.

RESULTS

During follow-up (median 5.1 years), 130 patients developed ESRD. With Cox regression adjusted by inverse probability of treatment weighted, RASB use was significantly associated with a lower risk of ESRD (hazard ratio 0.58; 95 % confidence interval 0.42-0.80). Significant associations were observed for other propensity score-based approaches. In stratified analysis, a beneficial association between RASB and ESRD was observed in patients ≥35 years, with hypertension, reduced estimated glomerular filtration rate (<60 mL/min/1.73 m²), mesangial proliferation and segmental glomerulosclerosis (P for interaction <0.05), and tended to be greater in patients with proteinuria (≥1.0 g/24 h), extracapillary proliferation and receiving methylprednisolone pulse therapy (P for interaction <0.10).

CONCLUSION

Treatment with RASB was associated with a lower incidence of ESRD in the real-world practice of IgAN.

Performance of the disease risk score in a cohort study with policy-induced selection bias

AIM

To examine the performance of the disease risk score (DRS) in a cohort study with evidence of policy-induced selection bias.

METHODS

We examined two cohorts of new users of bisphosphonates. Estimates for 1-year hip fracture rates between agents using DRS, exposure propensity scores and traditional multivariable analysis were compared.

RESULTS

The results for the cohort with no evidence of policy-induced selection bias showed little variation across analyses (-4.1-2.0%). Analysis of the cohort with evidence of policy-induced selection bias showed greater variation (-13.5-8.1%), with the greatest difference seen with DRS analyses.

CONCLUSION

Our findings suggest that caution may be warranted when using DRS methods in cohort studies with policy-induced selection bias, further research is needed.

Statistical power in parallel group point exposure studies with time-to-event outcomes: an empirical comparison of the performance of randomized controlled trials and the inverse probability of treatment weighting (IPTW) approach

Background

Estimating statistical power is an important component of the design of both randomized controlled trials (RCTs) and observational studies. Methods for estimating statistical power in RCTs have been well described and can be implemented simply. In observational studies, statistical methods must be used to remove the effects of confounding that can occur due to non-random treatment assignment. Inverse probability of treatment weighting (IPTW) using the propensity score is an attractive method for estimating the effects of treatment using observational data. However, sample size and power calculations have not been adequately described for these methods.

Methods

We used an extensive series of Monte Carlo simulations to compare the statistical power of an IPTW analysis of an observational study with time-to-event outcomes with that of an analysis of a similarly-structured RCT. We examined the impact of four factors on the statistical power function: number of observed events, prevalence of treatment, the marginal hazard ratio, and the strength of the treatment-selection process.

Results

We found that, on average, an IPTW analysis had lower statistical power compared to an analysis of a similarly-structured RCT. The difference in statistical power increased as the magnitude of the treatment-selection model increased.

Conclusions

The statistical power of an IPTW analysis tended to be lower than the statistical power of a similarly-structured RCT.

Association between guideline recommended drugs and death in older adults with multiple chronic conditions: population based cohort study

OBJECTIVE

To estimate the association between guideline recommended drugs and death in older adults with multiple chronic conditions.

DESIGN

Population based cohort study.

SETTING

Medicare Current Beneficiary Survey cohort, a nationally representative sample of Americans aged 65 years or more.

PARTICIPANTS

8578 older adults with two or more study chronic conditions (atrial fibrillation, coronary artery disease, chronic kidney disease, depression, diabetes, heart failure, hyperlipidemia, hypertension, and thromboembolic disease), followed through 2011.

EXPOSURES

Drugs included β blockers, calcium channel blockers, clopidogrel, metformin, renin-angiotensin system (RAS) blockers; selective serotonin reuptake inhibitors (SSRIs) and serotonin norepinephrine reuptake inhibitors (SNRIs); statins; thiazides; and warfarin.

MAIN OUTCOME MEASURE

Adjusted hazard ratios for death among participants with a condition and taking a guideline recommended drug relative to participants with the condition not taking the drug and among participants with the most common combinations of four conditions.

RESULTS

Over 50% of participants with each condition received the recommended drugs regardless of coexisting conditions; 1287/8578 (15%) participants died during the three years of follow-up. Among cardiovascular drugs, β blockers, calcium channel blockers, RAS blockers, and statins were associated with reduced mortality for indicated conditions. For example, the adjusted hazard ratio for β blockers was 0.59 (95% confidence interval 0.48 to 0.72) for people with atrial fibrillation and 0.68 (0.57 to 0.81) for those with heart failure. The adjusted hazard ratios for cardiovascular drugs were similar to those with common combinations of four coexisting conditions, with trends toward variable effects for β blockers. None of clopidogrel, metformin, or SSRIs/SNRIs was associated with reduced mortality. Warfarin was associated with a reduced risk of death among those with atrial fibrillation (adjusted hazard ratio 0.69, 95% confidence interval 0.56 to 0.85) and thromboembolic disease (0.44, 0.30 to 0.62). Attenuation in the association with reduced risk of death was found with warfarin in participants with some combinations of coexisting conditions.

CONCLUSIONS

Average effects on survival, particularly for cardiovascular study drugs, were comparable to those reported in randomized controlled trials but varied for some drugs according to coexisting conditions. Determining treatment effects in combinations of conditions may guide prescribing in people with multiple chronic conditions.

Impact of a Suicide-Specific Intervention within Inpatient Psychiatric Care: The Collaborative Assessment and Management of Suicidality

A growing body of literature indicates that suicidal patients differ from other psychiatric patients with respect to specific psychological vulnerabilities and that suicide-specific interventions may offer benefits beyond conventional care. This naturalistic controlled-comparison trial (n = 52) examined outcomes of intensive psychiatric hospital treatment (mean length of stay 58.8 days), comparing suicidal patients who received individual therapy from clinicians utilizing the Collaborative Assessment and Management of Suicidality (CAMS) to patients whose individual therapists did not utilize CAMS. Propensity score matching was used to control for potential confounds, including age, sex, treatment unit, and severity of depression and suicidality. Results showed that both groups improved significantly over the course of hospitalization; however, the group receiving CAMS showed significantly greater improvement on measures specific to suicidal ideation and suicidal cognition. Results are discussed in terms of the potential advantages of treating suicide risk with a suicide-specific intervention to make inpatient psychiatric treatment more effective in reducing risk for future suicidal crises.

The Impact of Inpatient Palliative Care Consultations on 30-Day Hospital Readmissions

BACKGROUND

Inpatient palliative care consultations have been shown to reduce acute care utilization by reducing length of stay, but less is known about their impact on subsequent costs including hospital readmissions.

OBJECTIVE

The study's objective was to examine the impact of inpatient palliative care consultations on 30-day hospital readmissions to a large urban academic medical center.

METHODS

The hospital's electronic medical record system was used to identify all live discharges between August 2013 and November 2014. After adjusting for a propensity score, readmission rates were compared between palliative care and usual care groups.

RESULTS

Of the 34,541 hospitalizations included in the study, 1430 (4.1%) involved a palliative care consult. After adjusting for the propensity score, patients seen by palliative care had a lower 30-day readmission rate-adjusted odds ratio (AOR) 0.66, 0.55-0.78; p<0.001. Adjusted rates were 10.3% (95% confidence interval [CI] 8.9%-12.0%) for palliative care and 15.0% (95% CI 14.4%-15.4%) for usual care. Among all palliative care patients, consultations that involved goals of care discussions were associated with a lower readmission rate (AOR 0.36, 0.27-0.48; p<0.001), but consultations involving symptom management were not (AOR 1.05, 0.82-1.35; p=0.684).

CONCLUSIONS

Palliative care palliative care consultations facilitate goals discussions, which in turn are associated with reduced rates of 30-day readmissions.

Optimal full matching for survival outcomes: a method that merits more widespread use

Matching on the propensity score is a commonly used analytic method for estimating the effects of treatments on outcomes. Commonly used propensity score matching methods include nearest neighbor matching and nearest neighbor caliper matching. Rosenbaum (1991) proposed an optimal full matching approach, in which matched strata are formed consisting of either one treated subject and at least one control subject or one control subject and at least one treated subject. Full matching has been used rarely in the applied literature. Furthermore, its performance for use with survival outcomes has not been rigorously evaluated. We propose a method to use full matching to estimate the effect of treatment on the hazard of the occurrence of the outcome. An extensive set of Monte Carlo simulations were conducted to examine the performance of optimal full matching with survival analysis. Its performance was compared with that of nearest neighbor matching, nearest neighbor caliper matching, and inverse probability of treatment weighting using the propensity score. Full matching has superior performance compared with that of the two other matching algorithms and had comparable performance with that of inverse probability of treatment weighting using the propensity score. We illustrate the application of full matching with survival outcomes to estimate the effect of statin prescribing at hospital discharge on the hazard of post‐discharge mortality in a large cohort of patients who were discharged from hospital with a diagnosis of acute myocardial infarction. Optimal full matching merits more widespread adoption in medical and epidemiological research. © 2015 The Authors. Statistics in Medicine Published by John Wiley & Sons Ltd.

Treatment biases in traumatic neurosurgical care: a retrospective study of the Nationwide Inpatient Sample from 1998 to 2009

OBJECT

This study was designed to assess the relationship between insurance status and likelihood of receiving a neurosurgical procedure following admission for either extraaxial intracranial hemorrhage or spinal vertebral fracture.

METHODS

A retrospective analysis of the Nationwide Inpatient Sample (NIS; 1998–2009) was performed. Cases of traumatic extraaxial intracranial hematoma and spinal vertebral fracture were identified using International Classification of Diseases, Ninth Revision (ICD-9) diagnosis codes. Within this cohort, those patients receiving a craniotomy or spinal fusion and/or decompression in the context of an admission for traumatic brain or spine injury, respectively, were identified using the appropriate ICD-9 procedure codes.

RESULTS

A total of 190,412 patients with extraaxial intracranial hematoma were identified between 1998 and 2009. Within this cohort, 37,434 patients (19.7%) received a craniotomy. A total of 477,110 patients with spinal vertebral fracture were identified. Of these, 37,302 (7.8%) received a spinal decompression and/or fusion. On multivariate analysis controlling for patient demographics, severity of injuries, comorbidities, hospital volume, and hospital characteristics, uninsured patients had a reduced likelihood of receiving a craniotomy (odds ratio [OR] 0.76, 95% confidence interval [CI] 0.71–0.82) and spinal fusion (OR 0.67, 95% CI 0.64–0.71) relative to insured patients. This statistically significant trend persisted when uninsured and insured patients were matched on the basis of mortality propensity score. Uninsured patients demonstrated an elevated risk-adjusted mortality rate relative to insured patients in cases of extraaxial intracranial hematoma. Among patients with spinal injury, mortality rates were similar between patients with and without insurance.

CONCLUSIONS

In this study, uninsured patients were consistently less likely to receive a craniotomy or spinal fusion for traumatic intracranial extraaxial hemorrhage and spinal vertebral fracture, respectively. This difference persisted after accounting for overall injury severity and patient access to high- or low-volume treatment centers, and potentially reflects a resource allocation bias against uninsured patients within the hospital setting. This information adds to the growing literature detailing the benefits of health reform initiatives seeking to expand access for the uninsured.

Clinical outcomes of early and later physical therapist services for older adults with back pain

BACKGROUND CONTEXT

The timing of physical therapy (PT) services and its association with later function and pain are not clear, especially in older adults.

PURPOSE

The purpose of this study was to compare clinical outcomes of patients receiving early or later PT services with those not receiving PT among older adults presenting to primary care for a new visit for back pain.

STUDY DESIGN/SETTING

Prospective cohort study using the Back Pain Outcomes Using Longitudinal Data registry.

PATIENT SAMPLE

A total of 3,705 adults 65 years and older with a new visit for back pain were included.

OUTCOME MEASURES

The outcome measures were Roland-Morris Disability Questionnaire (RMDQ), Pain Numerical Rating Scales, and EuroQol-5D.

METHODS

We studied two phases of PT utilization: early (0-28 days) and later (3-6 months). At baseline, we selected the participants with complete 12 months of patient-reported outcomes and electronic medical record data. Early PT was defined as initiating PT less than or equal to 28 days from the index visit for back pain. The no early PT group consisted of patients with no PT, no injections, no surgery, and no chiropractic within 28 days. We restricted the later phase analysis to patients with pain greater than 2 of 10 and an RMDQ score greater than 4 to create a subsample of patients with continuing clinically important back pain. We defined later PT as initiating PT between 3 and 6 months after the index visit. The no later PT group consisted of patients without any PT during this time. We used propensity score matching followed by multiple linear regression to estimate the mean difference in outcome. Sensitivity analysis examined clinically important change and dose of PT use among the early PT group.

RESULTS

The early PT group had better functional status with an adjusted mean RMDQ of 1.1 points less than the no early PT group (95% confidence interval: -2.2, -0.1) and less back pain of -0.5 (-0.9, -0.1) at 12 months. There was no difference between early PT groups at 3 and 6 months. The odds of a 30% improvement in function or pain were not different between these matched groups at 12 months, but the early PT group had increased odds of a 50% improvement in function at 12 months (odds ratio: 1.58, 95% confidence interval: 1.04, 2.40). There was no difference between later groups at 12 months. Greater dose of PT use within the early PT group was associated with better functional status (p= .01).

CONCLUSIONS

We found that among older adults presenting to their primary care providers for a new episode of back pain, early referral to PT resulted in no or minimal differences in pain, function, or health-related quality at 3, 6, or 12 months compared with a matched group that did not receive early PT. Secondary analysis show that patients initiating early PT may be somewhat more likely to experience 50% improvement in function at 12 months.

Effect of taxanes-based induction chemotherapy in locoregionally advanced nasopharyngeal carcinoma: A large scale propensity-matched study

OBJECTIVES

The effect of taxanes-based induction chemotherapy (IC) in locoregionally advanced nasopharyngeal carcinoma (LA-NPC) was quite contradictory in two phase II randomized controlled trials with small sample size. We aimed to investigate it in this large scale propensity-matched study.

MATERIALS AND METHODS

Totally, 779 LA-NPC patients who underwent intensity-modulated radiotherapy (IMRT) plus concurrent chemotherapy with or without taxanes-based IC were included. Patients in both treatment arms were matched using propensity score matching method at the ratio of 1:1. Failure-free survival (FFS), overall survival (OS), distant metastasis-free survival (DMFS) and locoregional relapse-free survival (LRFS) were assessed with Kaplan-Meier method, log-rank test and Cox regression analysis.

RESULTS

After matching, 534 patients were identified for analysis. In univariate analysis, both treatment arms resulted in parallel survival (4-years FFS 78.0% vs 74.1%, P = 0.304; OS 87.5% vs 87.3%, P = 0.595; DMFS 88.2% vs 84.4%, P = 0.154; and LRFS 91.2% vs 90.1%, P = 0.960). In multivariate analysis, taxanes-based IC did not improve any survival (P ⩾ 0.139). And this association remained unchanged in subgroup analysis by age, sex and histology, and among patients with stage III and T4N0M0. But among patients with T4N1-2M0 and stage IVb, taxanes-based IC significantly prolonged the 4-year DMFS by 11.2% (86.1% vs 74.9%, P = 0.034), and marginally improved FFS (P = 0.133) and OS (P = 0.215) in both univariate and multivariate analysis.

CONCLUSIONS

In this large scale propensity-matched study, LA-NPC patients could not benefit from taxanes-based IC on the whole. But the risk of distant metastasis significantly decreased by above 10% for patients with T4N1-2M0 and stage IVb.

Matching on the disease risk score in comparative effectiveness research of new treatments

PURPOSE

We use simulations and an empirical example to evaluate the performance of disease risk score (DRS) matching compared with propensity score (PS) matching when controlling large numbers of covariates in settings involving newly introduced treatments.

METHODS

We simulated a dichotomous treatment, a dichotomous outcome, and 100 baseline covariates that included both continuous and dichotomous random variables. For the empirical example, we evaluated the comparative effectiveness of dabigatran versus warfarin in preventing combined ischemic stroke and all-cause mortality. We matched treatment groups on a historically estimated DRS and again on the PS. We controlled for a high-dimensional set of covariates using 20% and 1% samples of Medicare claims data from October 2010 through December 2012.

RESULTS

In simulations, matching on the DRS versus the PS generally yielded matches for more treated individuals and improved precision of the effect estimate. For the empirical example, PS and DRS matching in the 20% sample resulted in similar hazard ratios (0.88 and 0.87) and standard errors (0.04 for both methods). In the 1% sample, PS matching resulted in matches for only 92.0% of the treated population and a hazard ratio and standard error of 0.89 and 0.19, respectively, while DRS matching resulted in matches for 98.5% and a hazard ratio and standard error of 0.85 and 0.16, respectively.

CONCLUSIONS

When PS distributions are separated, DRS matching can improve the precision of effect estimates and allow researchers to evaluate the treatment effect in a larger proportion of the treated population. However, accurately modeling the DRS can be challenging compared with the PS.

Proton Pump Inhibitor Usage and the Risk of Myocardial Infarction in the General Population

Background and Aims

Proton pump inhibitors (PPIs) have been associated with adverse clinical outcomes amongst clopidogrel users after an acute coronary syndrome. Recent pre-clinical results suggest that this risk might extend to subjects without any prior history of cardiovascular disease. We explore this potential risk in the general population via data-mining approaches.

Methods

Using a novel approach for mining clinical data for pharmacovigilance, we queried over 16 million clinical documents on 2.9 million individuals to examine whether PPI usage was associated with cardiovascular risk in the general population.

Results

In multiple data sources, we found gastroesophageal reflux disease (GERD) patients exposed to PPIs to have a 1.16 fold increased association (95% CI 1.09–1.24) with myocardial infarction (MI). Survival analysis in a prospective cohort found a two-fold (HR = 2.00; 95% CI 1.07–3.78; P = 0.031) increase in association with cardiovascular mortality. We found that this association exists regardless of clopidogrel use. We also found that H₂ blockers, an alternate treatment for GERD, were not associated with increased cardiovascular risk; had they been in place, such pharmacovigilance algorithms could have flagged this risk as early as the year 2000.

Conclusions

Consistent with our pre-clinical findings that PPIs may adversely impact vascular function, our data-mining study supports the association of PPI exposure with risk for MI in the general population. These data provide an example of how a combination of experimental studies and data-mining approaches can be applied to prioritize drug safety signals for further investigation.

A practical guide for using registry data to inform decisions about the cost effectiveness of new cancer drugs: lessons learned from the PHAROS registry

Decision makers increasingly request evidence on the real-world cost effectiveness of a new treatment. There is, however, a lack of practical guidance on how to conduct an economic evaluation based on registry data and how this evidence can be used in actual decision making. This paper explains the required steps on how to perform a sound economic evaluation using examples from an economic evaluation conducted with real-world data from the Dutch Population based HAematological Registry for Observational Studies. There are three main issues related to using registry data: confounding by indication, missing data, and insufficient numbers of (comparable) patients. If encountered, it is crucial to accurately deal with these issues to maximize the internal validity and generalizability of the outcomes and their value to decision makers. Multivariate regression modeling, propensity score matching, and data synthesis are well-established methods to deal with confounding. Multiple imputation methods should be used in cases where data are missing at random. Furthermore, it is important to base the incremental cost-effectiveness ratio of a new treatment compared with its alternative on comparable groups of (matched) patients, even if matching results in a small analytical population. Unmatched real-world data provide insights into the costs and effects of a treatment in a real-world setting. Decision makers should realize that real-world evidence provides extremely valuable and relevant policy information, but needs to be assessed differently compared with evidence derived from a randomized clinical trial.

Economics of the Iceberg: Informal Care Provided to French Elderly with Dementia

OBJECTIVES

Dementia has a substantial effect on patients and their relatives, who have to cope with medical, social, and economic changes. In France, most elderly people with dementia live in the community and receive informal care, which has not been well characterized.

METHODS

Using a sample of 4680 people aged 75 years and older collected in 2008 through a national comprehensive survey on health and disability, we compared the economic value of the care received by 513 elderly people with dementia to that received by a propensity score- matched set of older people without dementia.

RESULTS

More than 85% of elderly people with dementia receive informal care; the estimation of its economic value ranges from €4.9 billion (proxy good method) to €6.7 billion (opportunity cost method) per year.

CONCLUSIONS

The informal care provided to people with dementia has substantial annual costs; further work should be done to examine the social and economic roles foregone as a result of this care.

Severe COPD Exacerbation Risk and Long-Acting Bronchodilator Treatments: Comparison of Three Observational Data Analysis Methods

OBJECTIVE

Results from three observational methods for assessing effectiveness of long-acting bronchodilator therapies for reducing severe exacerbations of chronic obstructive pulmonary disease (COPD) were compared: intent-to-treat (ITT), as protocol (AP), and an as-treated analysis that utilized a marginal structural model (MSM) incorporating time-varying covariates related to treatment adherence and moderate exacerbations.

STUDY DESIGN AND SETTING

Severe exacerbation risk was assessed over a 2-year period using claims data for patients aged ≥40 years who initiated long-acting muscarinic antagonist (LAMA), inhaled corticosteroid/long-acting beta-agonist (ICS/LABA), or triple therapy (LAMA + ICS/LABA).

RESULTS

A total of 5475 COPD patients met inclusion criteria. Six months post-initiation, 53.5 % of patients discontinued using any therapy. The ITT analysis found an increased severe exacerbation risk for triple therapy treatment (hazard ratio [HR] 1.24; 95 % confidence interval [CI] 1.00-1.53). No increased risk was found in the AP (HR 1.00; 95 % CI 0.73-1.36), or MSM analyses (HR 1.11; 95 % CI 0.68-1.81). The MSM highlighted important associations among post-index events.

CONCLUSION

Neglecting to adjust for treatment discontinuation may produce biased risk estimates. The MSM approach is a promising tool to compare chronic disease management by illuminating relationships between treatment decisions, adherence, patient choices, and outcomes.

Hormone therapy and young-onset breast cancer

Estrogen plus progestin hormone therapy (HT) is associated with an increased risk of postmenopausal breast cancer, but few studies have examined the impact of HT use on the risk of breast cancer in younger women. We assessed the association between estrogen plus progestin HT or unopposed estrogen HT and young-onset breast cancer using data from the Two Sister Study (2008-2010), a sister-matched study of 1,419 cases diagnosed with breast cancer before the age of 50 years and 1,665 controls. We assessed exposures up to a family-specific index age to ensure comparable opportunities for exposures and used propensity scores to control for birth cohort effects on HT use. Ever HT use was uncommon (7% and 11% in cases and controls, respectively). Use of estrogen plus progestin was not associated with an increased risk of young-onset breast cancer (odds ratio = 0.80, 95% confidence interval: 0.41, 1.59). Unopposed estrogen use was inversely associated with the risk of young-onset breast cancer (odds ratio = 0.58, 95% confidence interval: 0.34, 0.99). Duration of use, age at first use, and recency of use did not modify these associations.

Impact of adjuvant chemotherapy with radiation for node-positive vulvar cancer: A National Cancer Data Base (NCDB) analysis

BACKGROUND

For node-positive vulvar cancer, adjuvant radiotherapy has an established benefit, whereas the impact of chemotherapy is unknown. A National Cancer Data Base (NCDB) analysis was conducted to determine patterns of care and evaluate the survival impact of adjuvant chemotherapy.

METHODS

The NCDB was queried for vulvar cancer patients diagnosed from 1998-2011 who underwent extirpative surgery with confirmed inguinal nodal involvement treated with adjuvant radiotherapy. Patients with inadequate follow-up or non-squamous histologies were excluded. Chi-square test, logistic regression analysis, log-rank test and multivariable Cox proportional regression modeling with adjustment using propensity score with inverse probability of treatment weights (IPTW) were conducted to establish factors associated with utilization and survival.

RESULTS

A total of 1797 patients were identified: 26.3% received adjuvant chemotherapy and 76.6% had 1-3 involved lymph nodes. Adoption of adjuvant chemotherapy significantly increased over time, from 10.8% in 1998 to 41.0% in 2006 (p<0.001). Lower utilization was seen in older patients, Northeast or Southern facilities, and patients with more extensive nodal dissection, whereas greater number of involved nodes, stage IVA disease and positive surgical margins led to a higher probability of receiving chemotherapy. Unadjusted median survival without and with adjuvant chemotherapy was 29.7months and 44.0months (p=0.001). On IPTW-adjusted Cox proportional regression modeling, delivery of adjuvant chemotherapy resulted in a 38% reduction in the risk of death (HR 0.62, 95% CI 0.48-0.79, p<0.001).

CONCLUSION

In a large population-based analysis, adjuvant chemotherapy resulted in a significant reduction in mortality risk for node-positive vulvar cancer patients who received adjuvant radiotherapy.

Performing both propensity score and instrumental variable analyses in observational studies often leads to discrepant results: a systematic review

OBJECTIVES

Propensity score (PS) and instrumental variable (IV) are analytical techniques used to adjust for confounding in observational research. More and more, they seem to be used simultaneously in studies evaluating health interventions. The present review aimed to analyze the agreement between PS and IV results in medical research published to date.

STUDY DESIGN AND SETTING

Review of all published observational studies that evaluated a clinical intervention using simultaneously PS and IV analyses, as identified in MEDLINE and Web of Science.

RESULTS

Thirty-seven studies, most of them published during the previous 5 years, reported 55 comparisons between results from PS and IV analyses. There was a slight/fair agreement between the methods [Cohen's kappa coefficient = 0.21 (95% confidence interval: 0.00, 0.41)]. In 23 cases (42%), results were nonsignificant for one method and significant for the other, and IV analysis results were nonsignificant in most situations (87%).

CONCLUSION

Discrepancies are frequent between PS and IV analyses and can be interpreted in various ways. This suggests that researchers should carefully consider their analytical choices, and readers should be cautious when interpreting results, until further studies clarify the respective roles of the two methods in observational comparative effectiveness research.

Attachment insecurity and obsessive-compulsive personality disorder among inpatients with serious mental illness

BACKGROUND

Obsessive-compulsive personality disorder (OCPD) is characterized by traits such as extreme rigidity, perfectionism, and controlling behavior, all of which have a negative impact on interpersonal functioning. Attachment theory provides a useful framework to elucidate the interpersonal dysfunction characteristic of OCPD; yet, there is a dearth of attachment research on OCPD in the context of severe mental illness.

METHODS

Attachment security and personality disorders were assessed in adult inpatients with severe mental illness. Propensity Score Matching (PSM) was used to match OCPD and control subjects on age, gender, number of psychiatric disorders, and number of criteria endorsed for borderline personality disorder.

RESULTS

Consistent with hypotheses, the OCPD group (n=61) showed greater attachment avoidance than controls (n=61), and the avoidance was manifested in a predominance of the most insecure attachment style, fearful attachment. Correlations between attachment anxiety/avoidance with specific OCPD diagnostic criteria revealed that attachment avoidance was correlated with four of eight OCPD criteria across the full sample. Within the subset of OCPD patients, attachment avoidance was significantly correlated with OCPD criterion 3 (is excessively devoted to work and productivity to the exclusion of leisure activities and friendships).

LIMITATIONS

The use of self-report measure of attachment and the high burden of illness in the SMI population may not generalize to interview based assessment or outpatients, respectively.

CONCLUSIONS

Findings attest to the severity of impairment in interpersonal functioning and attachment avoidance, in particular, is characteristic of OCPD patients. These results suggest that viable treatment targets include interpersonal functioning along with more classical features of OCPD such as perfectionism and obsessiveness in task performance.

Longitudinal studies 4: Matching strategies to evaluate risk

Matching is a strategy that can be used to control for confounding at the design stage of observational studies that examine exposure-outcome relationships. In case-control studies, matching can be used to generate subsamples of case and control units that are similar with respect to one or more confounders. In cohort studies, matching can balance confounder(s) so that they are the same in exposed and unexposed groups. Matching methods have been extended to include multivariable approaches, the most common being propensity score matching in observation studies of interventions. This chapter describes the major principles of matching applied to case-control, cohort, and propensity score studies. Matched study designs provide several advantages for controlling confounding in observational studies; however, they remain vulnerable to residual confounding and can even introduce bias when implemented incorrectly.

Arteriovenous fistula placement in the elderly: when is the optimal time?

Arteriovenous fistula (AVF) is the preferred vascular access for hemodialysis (HD). However, many AVFs fail before starting dialysis. To assess the optimal time for AVF placement in the elderly, we linked data from the US Renal Data System with Medicare claims data to identify 17,511 patients ≥67 years old on incident HD who started dialysis between January 1, 2005, and December 31, 2008, with an AVF placed as the first predialysis access. AVF success was defined as dialysis initiation using the AVF, with time between AVF placement and dialysis start as our primary variable of interest. The mean age was 76.1±6.0 years, and 58.3% of subjects were men. Overall, 54.9% of subjects initiated dialysis using an AVF, and 45.1% of subjects used a catheter or graft. The success rate increased as time from AVF creation to HD initiation increased from 1-3 months (odds ratio [OR], 0.49; 95% confidence interval [95% CI], 0.44 to 0.53) to 3-6 months (OR, 0.93; 95% CI, 0.85 to 1.02) to 6-9 months (OR, 0.99; 95% CI, 0.88 to 1.11) but stabilized after that time. Furthermore, the number of interventional access procedures increased over time starting at 1-3 months, with a mean of 0.64 procedures/patient for AVFs created 6-9 months predialysis compared with 0.72 for AVFs created >12 months predialysis (P<0.001). Although limited by the observational nature of this study, our results suggest that placing an AVF >6-9 months predialysis in the elderly may not associate with a better AVF success rate.

Improving propensity score estimators' robustness to model misspecification using super learner

The consistency of propensity score (PS) estimators relies on correct specification of the PS model. The PS is frequently estimated using main-effects logistic regression. However, the underlying model assumptions may not hold. Machine learning methods provide an alternative nonparametric approach to PS estimation. In this simulation study, we evaluated the benefit of using Super Learner (SL) for PS estimation. We created 1,000 simulated data sets (n = 500) under 4 different scenarios characterized by various degrees of deviance from the usual main-term logistic regression model for the true PS. We estimated the average treatment effect using PS matching and inverse probability of treatment weighting. The estimators' performance was evaluated in terms of PS prediction accuracy, covariate balance achieved, bias, standard error, coverage, and mean squared error. All methods exhibited adequate overall balancing properties, but in the case of model misspecification, SL performed better for highly unbalanced variables. The SL-based estimators were associated with the smallest bias in cases of severe model misspecification. Our results suggest that use of SL to estimate the PS can improve covariate balance and reduce bias in a meaningful manner in cases of serious model misspecification for treatment assignment.