Michele Jonsson Funk

Lifetime risk of stress urinary incontinence or pelvic organ prolapse surgery.

OBJECTIVE: To estimate the lifetime risk of stress urinary incontinence (SUI) surgery, pelvic organ prolapse (POP) surgery, or both using current, population-based surgical rates from 2007 to 2011. METHODS: We used a 2007–2011 U.S. claims and encounters database. We included women aged 18–89 years and estimated age-specific incidence rates and cumulative incidence (lifetime risk) of SUI surgery, POP surgery, and either incontinence or prolapse surgery with 95% confidence intervals (CIs). We estimated lifetime risk until the age of 80 years to be consistent with prior studies. RESULTS: From 2007 to 2011, we evaluated 10,177,480 adult women who were followed for 24,979,447 person-years. Among these women, we identified 65,397 incident, or first, SUI and 57,755 incident prolapse surgeries. Overall, we found that the lifetime risk of any primary surgery for SUI or POP was 20.0% (95% CI 19.9–20.2) by the age of 80 years. Separately, the cumulative risk for SUI surgery was 13.6% (95% CI 13.5–13.7) and that for POP surgery was 12.6% (95% CI 12.4–2.7). For age-specific annual risk, SUI demonstrated a bimodal peak at age 46 years and then again at age 70–71 years with annual risks of 3.8 and 3.9 per 1,000 women, respectively. For POP, the risk increased progressively until ages 71 and 73 years when the annual risk was 4.3 per 1,000 women. CONCLUSION: Based on a U.S. claims and encounters database, the estimated lifetime risk of surgery for either SUI or POP in women is 20.0% by the age of 80 years. LEVEL OF EVIDENCE: III

Discordance between patient-predicted and model-predicted life expectancy among ambulatory patients with heart failure

CONTEXT Patients with chronic heart failure have impaired long-term survival, but their own expectations regarding prognosis have not been well studied. OBJECTIVES To quantify expectations for survival in patients with heart failure, to compare patient expectations to model predictions, and to identify factors associated with discrepancies between patient-predicted and model-predicted prognosis. DESIGN, SETTING, AND PARTICIPANTS Prospective face-to-face survey of patients from the single-center Duke Heart Failure Disease Management Program between July and December 2004, with follow-up through February 2008. Patient-predicted life expectancy was obtained using a visual analog scale. Model-predicted life expectancy was calculated using the Seattle Heart Failure Model. Actuarial-predicted life expectancy, based on age and sex alone, was calculated using life tables. Observed survival was determined from review of medical records and search of the Social Security Death Index. MAIN OUTCOME MEASURE Life expectancy ratio (LER), defined as the ratio of patient-predicted to model-predicted life expectancy. RESULTS The cohort consisted of 122 patients (mean age, 62 years; 47% African American, 42% New York Heart Association [NYHA] class III or IV). On average, patients overestimated their life expectancy relative to model-predicted life expectancy (median patient-predicted life expectancy, 13.0 years; model-predicted expectancy, 10.0 years). Median LER was 1.4 (interquartile range, 0.8-2.5). Younger age, increased NYHA class, lower ejection fraction, and less depression were the most significant predictors of higher LER. During a median follow-up of 3.1 years, 29% of the original cohort died. There was no association between higher LER and improved survival (adjusted hazard ratio for overestimated compared with concordant LER, 1.05; 95% confidence interval, 0.46-2.42). CONCLUSIONS Ambulatory patients with heart failure tended to substantially overestimate their life expectancy compared with model-based predictions for survival. Because differences in perceived survival could affect decision making regarding advanced therapies and end-of-life planning, the causes of these discordant predictions warrant further study.

Doubly Robust Estimation of Causal Effects

Doubly robust estimation combines a form of outcome regression with a model for the exposure (i.e., the propensity score) to estimate the causal effect of an exposure on an outcome. When used individually to estimate a causal effect, both outcome regression and propensity score methods are unbiased only if the statistical model is correctly specified. The doubly robust estimator combines these 2 approaches such that only 1 of the 2 models need be correctly specified to obtain an unbiased effect estimator. In this introduction to doubly robust estimators, the authors present a conceptual overview of doubly robust estimation, a simple worked example, results from a simulation study examining performance of estimated and bootstrapped standard errors, and a discussion of the potential advantages and limitations of this method. The supplementary material for this paper, which is posted on the Journals Web site (http://aje.oupjournals.org/), includes a demonstration of the doubly robust property (Web Appendix 1) and a description of a SAS macro (SAS Institute, Inc., Cary, North Carolina) for doubly robust estimation, available for download at http://www.unc.edu/~mfunk/dr/.

Propensity score estimation: neural networks, support vector machines, decision trees (CART), and meta-classifiers as alternatives to logistic regression.

OBJECTIVE Propensity scores for the analysis of observational data are typically estimated using logistic regression. Our objective in this review was to assess machine learning alternatives to logistic regression, which may accomplish the same goals but with fewer assumptions or greater accuracy. STUDY DESIGN AND SETTING We identified alternative methods for propensity score estimation and/or classification from the public health, biostatistics, discrete mathematics, and computer science literature, and evaluated these algorithms for applicability to the problem of propensity score estimation, potential advantages over logistic regression, and ease of use. RESULTS We identified four techniques as alternatives to logistic regression: neural networks, support vector machines, decision trees (classification and regression trees [CART]), and meta-classifiers (in particular, boosting). CONCLUSION Although the assumptions of logistic regression are well understood, those assumptions are frequently ignored. All four alternatives have advantages and disadvantages compared with logistic regression. Boosting (meta-classifiers) and, to a lesser extent, decision trees (particularly CART), appear to be most promising for use in the context of propensity score analysis, but extensive simulation studies are needed to establish their utility in practice.

Timing of HAART initiation and clinical outcomes in human immunodeficiency virus type 1 seroconverters

BACKGROUND To estimate the clinical benefit of highly active antiretroviral therapy (HAART) initiation vs deferral in a given month in patients with CD4 cell counts less than 800/μL. METHODS In this observational cohort study of human immunodeficiency virus type 1 seroconverters from CASCADE (Concerted Action on SeroConversion to AIDS and Death in Europe), we constructed monthly sequential nested subcohorts between January 1996 and May 2009, including all eligible HAART-naive, AIDS-free individuals with a CD4 cell count less than 800/μL. The primary outcome was time to AIDS or death in those who initiated HAART in the baseline month compared with those who did not, pooled across subcohorts and stratified by CD4 cell count. Using inverse probability-of-treatment weighted survival curves and Cox proportional hazards regression models, we estimated the absolute and relative effects of treatment with robust 95% confidence intervals (CIs). RESULTS Of 9455 patients with 52,268 person-years of follow-up, 812 (8.6%) developed AIDS and 544 (5.8%) died. In CD4 cell count strata of 200 to 349, 350 to 499, and 500 to 799/μL, HAART initiation was associated with adjusted hazard ratios (95% CIs) for AIDS/death of 0.59 (0.43-0.81), 0.75 (0.49-1.14), and 1.10 (0.67-1.79), respectively. In the analysis of all-cause mortality, HAART initiation was associated with adjusted hazard ratios (95% CIs) of 0.71 (0.44-1.15), 0.51 (0.33-0.80), and 1.02 (0.49-2.12), respectively. Numbers needed to treat (95% CIs) to prevent 1 AIDS event or death within 3 years were 21 (14-38) and 34 (20-115) in CD4 cell count strata of 200 to 349 and 350 to 499/μL, respectively. CONCLUSION Compared with deferring in a given month, HAART initiation at CD4 cell counts less than 500/μL (but not 500-799/μL) was associated with slower disease progression.

The role of the c-statistic in variable selection for propensity score models

The applied literature on propensity scores has often cited the c‐statistic as a measure of the ability of the propensity score to control confounding. However, a high c‐statistic in the propensity model is neither necessary nor sufficient for control of confounding. Moreover, use of the c‐statistic as a guide in constructing propensity scores may result in less overlap in propensity scores between treated and untreated subjects; this may require the analyst to restrict populations for inference. Such restrictions may reduce precision of estimates and change the population to which the estimate applies. Variable selection based on prior subject matter knowledge, empirical observation, and sensitivity analysis is preferable and avoids many of these problems. Copyright

Cesarean delivery on maternal : Request maternal and neonatal outcomes

OBJECTIVE: To review systematically the evidence about maternal and infant outcomes of cesarean delivery on maternal request and planned vaginal delivery. DATA SOURCES: We searched MEDLINE, Cochrane Collaboration resources, and Embase and identified 1,406 articles through dual review using a priori inclusion criteria. METHODS OF STUDY SELECTION: We included English language studies published from 1990 to June 2005 that compared the key reference group (cesarean delivery on maternal request or proxies) and planned vaginal delivery. TABULATION, INTEGRATION, AND RESULTS: We identified 54 articles for maternal and infant outcomes. Virtually no studies exist on cesarean delivery on maternal request, so the knowledge base rests on indirect evidence from proxies with unique and significant limitations. Most studies compared outcomes by actual routes of delivery, resulting in variable relevance to planned routes of delivery. Primary cesarean delivery on maternal request and planned vaginal delivery likely differ with respect to individual outcomes; for instance, risks of urinary incontinence and maternal hemorrhage were lower with planned cesarean, whereas the risk of neonatal respiratory morbidity was higher and maternal length of stay was longer with planned cesarean delivery. However, our comprehensive assessment, across many outcomes, suggests no major differences between primary cesarean delivery on maternal request and planned vaginal delivery, but the evidence is too weak to conclude definitively that differences are completely absent. If a woman chooses to have a cesarean delivery in her first delivery, she is more likely to have subsequent deliveries by cesarean. With increasing numbers of cesarean delivery, risks occur with increasing frequency. CONCLUSION: The evidence is significantly limited by its minimal relevance to primary cesarean delivery on maternal request. Future research requires developing consensus about terminology, creating a minimum data set for cesarean delivery on maternal request, improving study design and statistical analyses, attending to major outcomes and their special measurement issues, assessing both short- and long-term outcomes with better measurement strategies, dealing better with confounders, and considering the value or utility of different outcomes.

Trends in the surgical management of stress urinary incontinence.

OBJECTIVES: To estimate the rates of stress urinary incontinence (SUI) surgery from 2000 to 2009 by type of procedure, year, age, and region of the country. METHODS: We used data between 2000 and 2009 from a database containing health care claims data from employer-based plans in the United States. We analyzed data for all women age 18–64 years, identifying all SUI procedures in this population. Rates per 100,000 person-years and 95% confidence intervals (CI) were calculated each year by procedure type, age, and region. RESULTS: The study population included 32.9 million women age 18–64 years observed for 74,007,937 person-years between 2000 and 2009. During that time, there were 182,110 SUI procedures for a rate of 246.1 per 100,000 person-years (95% CI 239.7–252.6). The most common SUI surgery was sling (198.3 per 100,000 person-years, 95% CI 192.8–203.9) followed by Burch (25.9 per 100,000 person-years, 95% CI 24.8–27.2). There was a dramatic increase in slings, with a corresponding decrease in Burch procedures from 2000 to 2009. Other SUI surgeries had lower rates. Although this trend was evident across all regions, the Northeast had the lowest rate of SUI surgery, whereas rates in the West, Midwest, and South were 1.44-times, 1.76-times, and 2.09-times higher, respectively. CONCLUSION: In a dramatic shift over the past decade, slings have become the dominant procedure for SUI among women age 18–64 years. Although this trend was seen across the United States, considerable variability exists in the SUI surgery rates by region. LEVEL OF EVIDENCE: III

Trends in use of surgical mesh for pelvic organ prolapse.

OBJECTIVE Limited data exist on the rates of pelvic organ prolapse procedures utilizing mesh. The objective of this study was to examine trends in vaginal mesh prolapse procedures (VMs), abdominal sacrocolpopexy (ASC), and minimally invasive sacrocolpopexy (MISC) from 2005 to 2010. STUDY DESIGN We utilized deidentified, adjudicated health care claims data from across the United States from 2005 to 2010. Among women 18 years old or older, we identified all mesh prolapse procedures based on current procedural terminology codes (57267 for VM, 57280 for ASC, and 57425 for MISC). VM procedures included all vaginal prolapse surgeries in which mesh was placed, whether in the anterior, apical, or posterior compartment. We estimated rates per 100,000 person-years (100,000 py) and 95% confidence intervals (CIs). RESULTS During 78.5 million person-years of observation, we identified 60,152 mesh prolapse procedures, for a rate of 76.0 per 100,000 py (95% CI, 73.6-78.5). Overall, VMs comprised 74.9% of these surgeries for an overall rate of 56.9 per 100,000 py (95% CI, 55.0-58.9). Rates of ASC and MISC were considerably lower at 12.0 per 100,000 py (95% CI, 11.6-12.5) and 9.5 per 100,000 py (95% CI, 9.2-9.9), respectively. Among sacrocolpopexies, ASC was more common than MISC in 2005-2007; however, since 2007, the rate of MISC has increased, whereas the rate of ASC has decreased. Regarding trends by age, VM was considerably more common than sacrocolpopexies at all ages, and ASC was more common than MISC in women older than 50 years. CONCLUSION From 2005 to 2010, the rate of mesh prolapse procedures has increased, with vaginal mesh surgeries constituting the vast majority.

Nonexperimental comparative effectiveness research using linked healthcare databases

Comparative Effectiveness Research (CER) has gained a great deal of attention over the past year through the new federal coordinating council,1 the recent Institute of Medicine (IOM) report,2 and the American Recovery & Reinvestment Act (ARRA) stimulus funding.3 CER has a broad scope as defined by the IOM, addressing “…the generation and synthesis of evidence that compares the benefits and harms of alternative methods to prevent, diagnose, treat and monitor a clinical condition or to improve the delivery of care. The purpose of CER is to assist consumers, clinicians, purchasers, and policymakers to make informed decisions that will improve health care at both the individual and population levels.”.2