Bias and variance trade-offs when combining propensity score weighting and regression: with an application to HIV status and homeless men

On variance estimation of target population created by inverse probability weighting

Inverse probability weighting (IPW) is frequently used to reduce or minimize the observed confounding in observational studies. IPW creates a pseudo-sample by weighting each individual by the inverse of the conditional probability of receiving the treatment level that he/she has actually received. In the pseudo-sample there is no variation among the multiple individuals generated by weighting the same individual in the original sample. This would reduce the variability of the data and therefore bias the variance estimate in the target population. Conventional variance estimation methods for IPW estimators generally ignore this underestimation and tend to produce biased estimates of variance. We here propose a more reasonable method that incorporates this source of variability by using parametric bootstrapping based on intra-stratum variability estimates. This approach firstly uses propensity score stratification and intra-stratum standard deviation to approximate the variability among multiple individuals generated based on a single individual whose propensity score falls within the corresponding stratum. The parametric bootstrapping is then used to incorporate the target variability by re-generating outcomes after adding a random error term to the original data. The performance of the proposed method is compared with three existing methods including the naïve model-based variance estimator, the nonparametric bootstrap variance estimator, and the robust variance estimator in the simulation section. An example of patients with sarcopenia is used to illustrate the implementation of the proposed approach. According to the results, the proposed approach has desirable statistical properties and can be easily implemented using the provided R code.

Matching-adjusted indirect comparison of the pelabresib-ruxolitinib combination vs JAKi monotherapy in myelofibrosis

Janus kinase inhibitors (JAKis) ruxolitinib, fedratinib, and pacritinib are the current standard of care in symptomatic myelofibrosis (MF). However, progressive disease and toxicities frequently lead to JAKi discontinuation. Preclinical data indicate that combining JAK and bromodomain and extraterminal (BET) domain inhibition leads to overlapping effects in MF. Pelabresib (CPI-0610), an oral, small-molecule BET1,2 inhibitor (BETi), in combination with ruxolitinib showed improvements in spleen volume reduction (SVR35) and total symptom score reduction (TSS50) from baseline in the phase 2 MANIFEST study (NCT02158858) in patients with MF. Given the absence of a head-to-head clinical comparison between JAKi monotherapy and JAKi with BETi combination therapy, we performed an unanchored matching-adjusted indirect comparison analysis to adjust for differences between studies and allow for the comparison of SVR35, TSS50, and TSS measured at several timepoints in arm 3 of MANIFEST (pelabresib with ruxolitinib in JAKi treatment-naive patients with MF), with data from the following JAKi monotherapy studies in JAKi treatment-naive patients: COMFORT-I and COMFORT-II (ruxolitinib), SIMPLIFY-1 (ruxolitinib and momelotinib), and JAKARTA (fedratinib). Response rate ratios >1 were observed for pelabresib with ruxolitinib vs all comparators for SVR35 and TSS50 at week 24. Improvements in TSS were observed as early as week 12 and were durable. These results indicate that pelabresib with ruxolitinib may have a potentially higher efficacy than JAKi monotherapy in JAKi treatment-naive MF.

Metabolic-Related Outcomes After Switching From Tenofovir Disoproxil Fumarate to Tenofovir Alafenamide in Adults With Human Immunodeficiency Virus (HIV): A Multicenter Prospective Cohort Study

BACKGROUND

Tenofovir alafenamide (TAF) has replaced tenofovir disoproxil fumarate (TDF) in many clinical settings. However, concerns remain about potential metabolic complications of TAF. We aimed to evaluate changes in weight, laboratory markers, and metabolic-related clinical events after replacing TDF with TAF.

METHODS

Multicenter prospective cohort study in the Spanish CoRIS cohort. We included virologically suppressed adults with human immunodeficiency virus (HIV) receiving TDF for more than 12 months who either switched to TAF or maintained TDF, with no changes in the core agent. Participants were matched by propensity score. We fitted generalized equation models to assess changes in weight, blood lipids, and hepatic steatosis index, and to compare the incidence of diabetes, hypertension, and lipid-lowering drug use after 144 weeks.

RESULTS

In total, 1446 participants were matched in each group. Median age was 38 years, 85% were male, mean weight at baseline was 73 kg. Participants who switched to TAF had a mean weight increase of +0.5 kg at 144 weeks over those who maintained TDF, with no difference in the occurrence of overweight or obesity. Individuals who switched to TAF had a significant increase in total cholesterol (+7.9 mg/dL) and triglycerides (+11.2 mg/dL), with no differences in the total cholesterol-high-density lipoprotein (HDL) ratio. However, no increased incidence of diabetes, hypertension, or lipid-lowering drug use was observed after the follow-up period.

CONCLUSIONS

Switching from TDF to TAF is associated with modest weight gain and increases in total cholesterol and triglycerides, without an impact on the incidence of obesity or metabolic-related clinical events, in this Spanish cohort with a majority White male population.

Short- and long-term survival after ST-elevation myocardial infarction treated with pharmacoinvasive versus primary percutaneous coronary intervention strategy: a prospective cohort study

Objective

Compare survival in patients with ST-elevation myocardial infarction (STEMI) treated with a pharmacoinvasive (PI) or primary percutaneous coronary intervention (pPCI) strategy based on estimated time to PCI.

Design

Prospective observational cohort study. Consecutive STEMI patients were registered on admission to our PCI centre and classified in a PI or pPCI group, based on the reperfusion strategy chosen in the prehospital or local hospital location. Time and cause of death was provided by the Norwegian Cause of Death registry. Mortality at 30 days, Kaplan-Meier survival and incidence of cardiovascular (CV) death was estimated. Adjusted effect of PI versus pPCI strategy on survival was estimated using logistic and Cox regression and propensity score weighting.

Setting

Single-centre registry in Norway during 2005–2011, within a regional STEMI network allocating patients to a PI strategy if estimated time to PCI >120 min.

Primary outcomes

30-day mortality and survival during follow-up.

Secondary outcome

Incidence of CV death during follow-up.

Results

4061 STEMI patients <80 years were included, 527 (13%) treated with a PI strategy and 3534 (87%) with a pPCI strategy. Median symptom-to-needle time was 110 min (25–75th percentile 75–163) in the PI group vs symptom-to-balloon 230 min (149–435) in the pPCI group. 30-day mortality was 3.2% and 5.0% in the PI and pPCI groups (OR_adjusted0.58 (95% CI 0.30 to 1.13)) and 8-year survival was 85.9% (95% CI 80.9% to 89.6%) and 79.3% (95% CI 76.9% to 81.6%), respectively (HR_adjusted 0.72 (95% CI 0.53 to 0.99)). Unadjusted incidence of 8-year CV death was 7.0% (95% CI 4.4% to 10.4%) in the PI group vs 12.4% (95% CI 9.9% to 15.2%) in the pPCI group. Adjusted long-term CV death was also lower in the PI group.

Conclusion

STEMI patients treated with a PI strategy experienced better survival compared with a pPCI strategy, also when adjusting for baseline characteristics. This supports using a PI strategy for eligible STEMI patients when pPCI cannot be performed within 120 min.

Outcome of liver transplantation with grafts from brain-dead donors treated with dual hypothermic oxygenated machine perfusion, with particular reference to elderly donors

Prompted by the utilization of extended criteria donors, dual hypothermic oxygenated machine perfusion (D-HOPE) was introduced in liver transplantation to improve preservation. When donors after neurological determination of death (DBD) are used, D-HOPE effect on graft outcomes is unclear. To assess D-HOPE value in this setting and to identify ideal scenarios for its use, data on primary adult liver transplant recipients from January 2014 to April 2021 were analyzed using inverse probability of treatment weighting, comparing outcomes of D-HOPE-treated grafts (n = 121) with those preserved by static cold storage (n = 723). End-ischemic D-HOPE was systematically applied since November 2017 based on donor and recipient characteristics and transplant logistics. D-HOPE use was associated with a significant reduction of early allograft failure (OR: 0.24; 0.83; p = .024), grade ≥3 complications (OR: 0.57; p = .046), comprehensive complication index (-7.20 points; p = .003), and improved patient and graft survival. These results were confirmed in the subset of elderly donors (>75-year-old). Although D-HOPE did not reduce the incidence of biliary complications, its use was associated with a reduced severity of ischemic cholangiopathy. In conclusion, D-HOPE improves postoperative outcomes and reduces early allograft loss in extended criteria DBD grafts.

Informing power and sample size calculations when using inverse probability of treatment weighting using the propensity score

Propensity score weighting is increasingly being used in observational studies to estimate the effects of treatments. The use of such weights induces a within‐person homogeneity in outcomes that must be accounted for when estimating the variance of the estimated treatment effect. Knowledge of the variance inflation factor (VIF), which describes the extent to which the effective sample size has been reduced by weighting, allows for conducting sample size and power calculations for observational studies that use propensity score weighting. However, estimation of the VIF requires knowledge of the weights, which are only known once the study has been conducted. We describe methods to estimate the VIF based on two characteristics of the observational study: the anticipated prevalence of treatment and the anticipated c‐statistic of the propensity score model. We considered five different sets of weights: those for estimating the average treatment effect (ATE), the average treated effect in the treated (ATT), and three recently described sets of weights: overlap weights, matching weights, and entropy weights. The VIF was substantially smaller for the latter three sets of weights than for the first two sets of weights. Once the VIF has been estimated during the design phase of the study, sample size and power calculations can be done using calculations appropriate for a randomized controlled trial with similar prevalence of treatment and similar outcome variable, and then multiplying the requisite sample size by the estimated VIF. Implementation of these methods allows for improving the design and reporting of observational studies that use propensity score weighting.

Adherence to Antiretroviral Treatment Among People Who Started Treatment on the Same-Day of HIV Diagnosis in Ethiopia: A Multicenter Observational Study

Introduction

Methods

Results

Conclusion

Virologic outcomes of people living with human immunodeficiency virus who started antiretroviral treatment on the same-day of diagnosis in Ethiopia: A multicenter observational study

INTRODUCTION

There have been tremendous achievements in scaling-up antiretroviral therapy (ART) for treatment of human immunodeficiency virus (HIV), following universal "test and treat" policy implementation in low- and middle-income countries. However, its effects on virologic outcomes is not yet well investigated. We compared low viral load status in people living with HIV between those who were initiated on ART on the same-day and after 7 days of being diagnosed with HIV infection.

METHODS

We conducted a retrospective cohort study of persons age ≥15 years-old who were newly diagnosed and started on ART between October 2016 and July 2018 at 11 public health facilities in northwest Ethiopia. Exposure was initiation of ART on the same-day of HIV diagnosis. The outcome was low viral load at 12-months following ART initiation. We used double-robust estimator using inverse-probability-weighted regression adjustment to compare the groups.

RESULTS

A total of 398 people who started ART on the same-day of HIV diagnosis and 479 people who started 7 days after the initial diagnosis were included in this study. By 12-months following ART initiation, 73.4% (292) in the same-day group vs 83.7% (401) in the >7 days group achieved low viral load (absolute difference = 10.3% (95% CI: 4.9%, 15.8%)). After adjusting for baseline and follow-up covariates, there was statistically significant difference in low viral load status (adjusted difference = 8.3% (95% CI: 3.5%, 13.0%)) between the same-day group and the >7 days group.

CONCLUSIONS

Achievement of low viral load by 12-months post-initiation of ART was not optimal among participants who started ART on the same-day of HIV diagnosis. Efforts should be made to reinforce treatment adherence while initiating same-day ART.

Postpartum psychiatric readmissions: A nationwide study in women with and without epilepsy

OBJECTIVE

To assess whether epilepsy is associated with increased odds of 30-day readmission due to psychiatric illness during the postpartum period.

METHODS

The 2014 Nationwide Readmissions Database and the International Classification of Disease, Ninth Revision, Clinical Modification codes were used to identify postpartum women up to 50 years old in the United States, including the subgroup with epilepsy. The primary outcome was 30-day readmission and was categorized as (1) readmission due to psychiatric illness, (2) readmission due to all other causes, or (3) no readmission. Secondary outcome was diagnosis at readmission. The association of the primary outcome and presence of epilepsy was examined using multinomial logistic regression.

RESULTS

Of 1 558 875 women with admissions for delivery identified, 6745 (.45%) had epilepsy. Thirteen of every 10 000 women had 30-day psychiatric readmissions in the epilepsy group compared to one of every 10 000 in the no-epilepsy group (p < .0001). Of every 10 000 women with epilepsy, 256 had 30-day readmissions due to other causes compared to 115 of every 10 000 women in the no-epilepsy group (p < .0001). The odds ratio for readmission due to psychiatric illness was 10.13 (95% confidence interval = 5.48-18.72) in those with epilepsy compared to those without. Top psychiatric causes for 30-day readmissions among women with epilepsy were mood disorders, schizophrenia and other psychotic disorders, and substance-related disorders.

SIGNIFICANCE

This large-scale study demonstrated that postpartum women with epilepsy have higher odds of readmission due to a psychiatric illness compared to women without epilepsy. Postpartum treatment strategies and interventions to prevent psychiatric readmissions are necessary in this vulnerable population.

Effectiveness of same-day antiretroviral therapy initiation in retention outcomes among people living with human immunodeficiency virus in Ethiopia: empirical evidence

BACKGROUND

In August 2016, Ethiopia endorsed a universal "test and treat" strategy for people living with human immunodeficiency virus (PLHIV) based on World Health Organization recommendation. However, there is limited evidence on the routine application of the same-day "test and treat" recommendation in low-income settings. This study assessed the effect of same-day treatment initiation on individual-level retention at 6- and 12-months follow-up.

METHODS

A multicenter facility-based retrospective cohort study was conducted to compare retention-in-care between PLHIV who started antiretroviral therapy (ART) on the same-day and those started ART > 7 days following HIV diagnoses. Participants were at least 15 years-old and were newly diagnosed and started on ART between October 2016 and July 2018 in 11 health facilities in the Amhara region of Ethiopia. Multivariable logistic regression controlling for potential confounders and Kaplan-Meier survival analysis were used to assess differences in outcomes between the groups.

RESULTS

In total, 433 PLHIV started ART on the same-day of diagnosis and 555 PLHIV who started ART > 7 days after HIV diagnosis were included in the study. At 6-months, 82.0% (355) in the same-day group vs 89.4% (496) in the > 7 days group were retained-in-care (absolute risk difference (RD) = 7.4%; 95% confidence interval (CI): 2.9-11.8%). At 12-months, 75.8% (328) in the same-day group vs 82.0% (455) in the > 7 days group were retained-in-care (absolute RD = 6.2%; 95% CI: 1.1, 11.4%). The major drop in retention was in the first 30 days following ART initiation among same-day group. After adjusting for baseline and non-baseline covariates, the same-day group was less likely to be retained-in-care at 6- and 12-months (adjusted risk ratio (RR) = 0.89; 95% CI: 0.87, 0.90 and adjusted RR = 0.86; 95% CI: 0.83, 0.89, respectively).

CONCLUSIONS

Reduced retention-in-care can threaten the benefit of the same-day "test and treat" policy. The policy needs to be implemented cautiously with greater emphasis on assessment and preparation of PLHIV for ART to ensure treatment readiness before starting them on same-day ART and close monitoring of patients during early follow-up periods.

An evaluation of whether propensity score adjustment can remove the self-selection bias inherent to web panel surveys addressing sensitive health behaviours

Background

In health research, population estimates are generally obtained from probability-based surveys. In market research surveys are frequently conducted from volunteer web panels. Propensity score adjustment (PSA) is often used at analysis to try to remove bias in the web survey, but empirical evidence of its effectiveness is mixed. We assess the ability of PSA to remove bias in the context of sensitive sexual health research and the potential of web panel surveys to replace or supplement probability surveys.

Methods

Four web panel surveys asked a subset of questions from the third British National Survey of Sexual Attitudes and Lifestyles (Natsal-3). Five propensity scores were generated for each web survey. The scores were developed from progressively larger sets of variables, beginning with demographic variables only and ending with demographic, sexual identity, lifestyle, attitudinal and sexual behaviour variables together. The surveys were weighted to match Natsal-3 based on propensity score quintiles. The performance of each survey and weighting was assessed by calculating the average ‘absolute’ odds ratio (inverse of the odds ratio if less than 1) across 22 pre-specified sexual behaviour outcomes of interest comparing the weighted web survey with Natsal-3. The average standard error across odds ratios was examined to assess the impact of weighting upon variance.

Results

Propensity weighting reduced bias relative to Natsal-3 as more variables were added for males, but had little effect for females, and variance increased for some surveys. Surveys with more biased estimates before propensity weighting showed greater reduction in bias from adjustment. Inconsistencies in performance were evident across surveys and outcomes. For most surveys and outcomes any reduction in bias was only partial and for some outcomes the bias increased.

Conclusions

Even after propensity weighting using a rich range of information, including some sexual behaviour variables, some bias remained and variance increased for some web surveys. Whilst our findings support the use of PSA for web panel surveys, the reduction in bias is likely to be partial and unpredictable, consistent with the findings from market research. Our results do not support the use of volunteer web panels to generate unbiased population health estimates.

Influenza increases invasive meningococcal disease risk in temperate countries

OBJECTIVES

Invasive meningococcal disease (IMD) is a severe bacterial infection that displays wintertime seasonality in temperate countries. Mechanisms driving seasonality are poorly understood and may include environmental conditions and/or respiratory virus infections. We evaluated the contribution of influenza and environmental conditions to IMD risk, using standardized methodology, across multiple geographical regions.

METHODS

We evaluated 3276 IMD cases occurring between January 1999 and December 2011 in 11 jurisdictions in Australia, Canada, France and the United States. Effects of environmental exposures and normalized weekly influenza activity on IMD risk were evaluated using a case-crossover design. Meta-analytic methods were used to evaluate homogeneity of effects and to identify sources of between-region heterogeneity.

RESULTS

After adjustment for environmental factors, elevated influenza activity at a 2-week lag was associated with increased IMD risk (adjusted odds ratio (OR) per standard deviation increase 1.29; 95% confidence interval, 1.04-1.59). This increase was homogeneous across the jurisdictions studied. By contrast, although associations between environmental exposures and IMD were identified in individual jurisdictions, none was generalizable.

CONCLUSIONS

Using a self-matched design that adjusts for both coseasonality and case characteristics, we found that surges in influenza activity result in an acute increase in population-level IMD risk. This effect is seen across diverse geographic regions in North America, France and Australia. The impact of influenza infection on downstream meningococcal risk should be considered a potential benefit of influenza immunization programmes.

Long-term outcomes of colorectal cancer surgery for elderly patients: a propensity score-matched analysis

PURPOSE

To investigate the effect of old age on the perioperative, short-term, and long-term surgical outcomes of elderly patients undergoing curative surgery for colorectal cancer (CRC).

METHODS

The subjects of this retrospective study were 526 patients who underwent curative resections for stage I-III CRC between March 2005 and March 2016. We divided the patients into a young group (< 75 years old, n = 361) and an elderly group (≥ 75 years old, n = 165) and compared the clinicopathological factors and prognoses of the two groups. We performed a propensity score-matched (PSM) analysis with inverse probability of treatment weighting (IPTW) to avoid confounding bias.

RESULTS

The elderly group had more right-sided tumors and more comorbidities than the young group. After PSM, there were 148 patients in each group. Although the elderly group had significantly shorter overall survival than the young group, the two groups did not differ significantly in cancer-specific survival (CSS; P = 0.136) or recurrence rate (RR; P = 0.317). Multivariate analysis with IPTW also revealed no significant difference in CSS (P = 0.171) or RR (P = 0.284) between the young and elderly groups. Our findings were limited by the study's retrospective single-institute conditions, and the inclusion of only patients who underwent radical resections.

CONCLUSION

Primary tumor resection is appropriate for elderly patients with CRC.

Tracks as frames of reference for academic self-concept

This study compared the development of academic self-concepts between different educational programs. A longitudinal cohort study in Flanders (3205 students in 46 schools) was used to compare students' academic self-concepts during the first three years of secondary education. General academic self-concept, self-concept in mathematics and self-concept in Dutch were measured. The investigated educational programs, called tracks, differ in the extent they are academically or vocationally focused and differ in average student academic ability. To control for selection effects, students who are comparable across the four tracks were matched using propensity score matching, Mahalanobis distance matching and coarsened exact matching. By means of multiple indicator quadratic latent growth curves, pairs of tracks that are hierarchically consecutive were compared regarding the development in academic self-concepts. For the two highest tracks, it was beneficial to be allocated to the highest track, whereas the pairwise comparisons between the three lower tracks indicated a detrimental effect of being in a higher track. The findings from this study do not support the big-fish-little-pond hypothesis or the basking in reflected glory hypothesis. Differences between tracks for the development of self-concepts only became apparent after two years.

Optimizing Variance-Bias Trade-off in the TWANG Package for Estimation of Propensity Scores

While propensity score weighting has been shown to reduce bias in treatment effect estimation when selection bias is present, it has also been shown that such weighting can perform poorly if the estimated propensity score weights are highly variable. Various approaches have been proposed which can reduce the variability of the weights and the risk of poor performance, particularly those based on machine learning methods. In this study, we closely examine approaches to fine-tune one machine learning technique (generalized boosted models [GBM]) to select propensity scores that seek to optimize the variance-bias trade-off that is inherent in most propensity score analyses. Specifically, we propose and evaluate three approaches for selecting the optimal number of trees for the GBM in the twang package in R. Normally, the twang package in R iteratively selects the optimal number of trees as that which maximizes balance between the treatment groups being considered. Because the selected number of trees may lead to highly variable propensity score weights, we examine alternative ways to tune the number of trees used in the estimation of propensity score weights such that we sacrifice some balance on the pre-treatment covariates in exchange for less variable weights. We use simulation studies to illustrate these methods and to describe the potential advantages and disadvantages of each method. We apply these methods to two case studies: one examining the effect of dog ownership on the owner's general health using data from a large, population-based survey in California, and a second investigating the relationship between abstinence and a long-term economic outcome among a sample of high-risk youth.

Rejoinder: Remaining Challenges in Investigating Grade-Retention Effectiveness

This rejoinder, in response to the commentaries of Steiner, Park, and Kim (this issue) and Reshetnyak, Cham, and Hughes (this issue), discusses remaining challenges in grade retention research. First, a same-age comparison assumes that the instruments used in different grades measure ability equally well. We discuss the importance of evaluating the properties of the scaling process to address whether this assumption has been met. Second, we discuss issues in the selection of covariates to be included in the weights. Third, we discuss the unconfoundedness assumption and the problem of remaining imbalance. Finally, we provide an empirical illustration showing that studying grade-retention effectiveness comes with multiple methodological decisions that are rooted in a bias-variance trade-off.

Long-term survival in laparoscopic vs open resection for colorectal liver metastases: inverse probability of treatment weighting using propensity scores

BACKGROUND

This study compares long-term outcomes between intention-to-treat laparoscopic and open approaches to colorectal liver metastases (CLM), using inverse probability of treatment weighting (IPTW) based on propensity scores to control for selection bias.

METHOD

Patients undergoing liver resection for CLM by 5 surgeons at 3 institutions from 2000 to early 2014 were analysed. IPTW based on propensity scores were generated and used to assess the marginal treatment effect of the laparoscopic approach via a weighted Cox proportional hazards model.

RESULTS

A total of 298 operations were performed in 256 patients. 7 patients with planned two-stage resections were excluded leaving 284 operations in 249 patients for analysis. After IPTW, the population was well balanced. With a median follow up of 36 months, 5-year overall survival (OS) and recurrence-free survival (RFS) for the cohort were 59% and 38%. 146 laparoscopic procedures were performed in 140 patients, with weighted 5-year OS and RFS of 54% and 36% respectively. In the open group, 138 procedures were performed in 122 patients, with a weighted 5-year OS and RFS of 63% and 38% respectively. There was no significant difference between the two groups in terms of OS or RFS.

CONCLUSION

In the Brisbane experience, after accounting for bias in treatment assignment, long term survival after LLR for CLM is equivalent to outcomes in open surgery.

In-office magnetic resonance imaging (MRI) equipment ownership and MRI volume among medicare patients in orthopedic practices

BACKGROUND

Concerns have been raised about physician ownership of onsite advanced imaging equipment as allowed under Stark laws by the in-office ancillary service exception (IOASE).

METHODS

A web-based survey of orthopedic practices in the United States was used to assign a first date of onsite MRI capacity acquisition (if any) to specific orthopedic practices. Medicare claims data for 2006-2010 was obtained for providers in orthopedic practices acquiring onsite MRI capacity and in matched orthopedic practices without an onsite MRI over the same period of time. Multivariate regression was used to estimate the change in provider Medicare MRI volume one year before and one year after the onsite MRI acquisition year for providers in MRI practices compared to providers in propensity-score matched non-MRI practices.

RESULTS

In all of the MRI volume change models estimated, the association between onsite MRI acquisition and the change in provider Medicare MRI volume (one-year post-onsite-MRI-acquisition less one year pre-acquisition) was consistently small and not statistically significant. This lack of association was robust to changes in model specification in terms of types of MRI exams considered, specific covariates included in the multivariate model, or the process used to confirm individual provider affiliation with study practices in study years.

CONCLUSIONS

Our analysis of Medicare claims data provides no empirical support for the proposition that acquisition of onsite MRI capacity within an orthopedic surgery practice induces an increase in the rate of MRI use for Medicare patients among practice providers, relative to physicians in practices without MRI capacity over the same time period.

[The first KiGGS follow-up (KiGGS Wave 1): study conduct, sample design, and response]

The "German Health Interview and Examination Survey for Children and Adolescents" (KiGGS) is part of the health monitoring system of the Robert Koch Institute (RKI). Following the KiGGS baseline study (2003 - 06), which comprised interviews and physical examinations of 0- to 17-year-old participants, KiGGS Wave 1 (2009 - 2012) was carried out as a telephone-based survey. In addition to providing longitudinal data, a second essential aim of KiGGS is to regularly provide population-based cross-sectional data on the health situation of children and adolescents aged 0-17 years living in Germany. Therefore, the study population of KiGGS Wave 1 consists of re-invited participants from the baseline study (KiGGS cohort), supplemented by newly invited children aged 0-6 years. The newly invited participants were randomly chosen from local population registries in the 167 baseline sample points. This method was chosen to supplement the sample with younger age groups. This article focuses on the age groups from 0 to 17 years, which are relevant for prevalence estimations among children and adolescents. In total 12,368 children and adolescents took part; among them 4,455 newly invited and 7,913 re-invited participants (response 38.8 and 72.9%, respectively). A comparison of the net sample with the resident German population (0-17 years) regarding particular population characteristics and an analysis of the relationship between the re-participation rate and certain characteristics collected in the baseline study (7-17 years) suggest a mostly unbiased sample. To account for certain aspects of the population and nonresponse, cross-sectional and trend analyses were partially corrected by weighting factors.