Jason P. Fine

A Proportional Hazards Model for the Subdistribution of a Competing Risk

Abstract With explanatory covariates, the standard analysis for competing risks data involves modeling the cause-specific hazard functions via a proportional hazards assumption. Unfortunately, the cause-specific hazard function does not have a direct interpretation in terms of survival probabilities for the particular failure type. In recent years many clinicians have begun using the cumulative incidence function, the marginal failure probabilities for a particular cause, which is intuitively appealing and more easily explained to the nonstatistician. The cumulative incidence is especially relevant in cost-effectiveness analyses in which the survival probabilities are needed to determine treatment utility. Previously, authors have considered methods for combining estimates of the cause-specific hazard functions under the proportional hazards formulation. However, these methods do not allow the analyst to directly assess the effect of a covariate on the marginal probability function. In this article we pro...

Diagnosis of invasive aspergillosis using a galactomannan assay:a meta-analysis

BACKGROUND A double-sandwich enzyme-linked immunosorbent galactomannan assay has been approved for surveillance for invasive aspergillosis in immunocompromised patients. We undertook a meta-analysis to assess the accuracy of a galactomannan assay for diagnosing invasive aspergillosis. METHODS Studies of the galactomannan assay that used the European Organization for Research and Treatment of Cancer or similar criteria as a reference standard and provided data to calculate sensitivity and specificity were included. Pooled sensitivity and specificity and summary measures of accuracy, Q* (the upper left-most point on the summary receiver-operating characteristic curve), mean D (a log odds ratio), and Youden index were calculated. Subgroup analyses were performed to explore heterogeneity. RESULTS Twenty-seven studies from 1966 to 28 February 2005 were included. Overall, the galactomannan assay had a sensitivity of 0.71 (95% confidence interval [CI], 0.68-0.74) and specificity of 0.89 (95% CI, 0.88-0.90) for proven cases of invasive aspergillosis. The Youden index, mean D, and Q* were 0.54 (95% CI, 0.41-0.65), 2.74 (95% CI, 21.12-3.36), and 0.80 (95% CI, 0.74-0.86), respectively, indicating moderate accuracy. Subgroup analyses showed that the performance of the test differed by patient population and type of reference standard used. Significant heterogeneity was present. CONCLUSIONS The galactomannan assay has moderate accuracy for diagnosis of invasive aspergillosis in immunocompromised patients. The test is more useful in patients who have hematological malignancy or who have undergone hematopoietic cell transplantation than in solid-organ transplant recipients. Further studies with attention to the impact of antifungal therapy, rigorous assessment of false-positive test results, and assessment of the utility of the test under nonsurveillance conditions are needed.

Meta-Analysis: Methods for Diagnosing Intravascular Device–Related Bloodstream Infection

Context Several methods exist for diagnosing intravascular devicerelated bloodstream infection. Contribution This meta-analysis summarized data from 51 studies for 8 diagnostic methods. Paired quantitative cultures drawn concomitantly from a peripheral vein and the suspected catheter appeared most accurate, but several methods, including quantitative and semi-quantitative catheter segment cultures, acridine orange leukocyte cytospin tests, and quantitative blood cultures drawn through the catheter, had reasonable accuracy. Cautions Studies used different diagnostic reference standards. Some methods were evaluated in few studies. The Editors Safe and reliable vascular access is essential to modern medical practice. Nearly 200 million intravascular devices (IVDs) are sold in the United States every year (1). Noncuffed percutaneously inserted catheters placed in the femoral, internal jugular, or subclavian vein are the most common centrally placed devices for short-term use, with more than 7 million sold each year (2). Devices for intermediate- and long-term venous access include cuffed and tunneled surgically implanted catheters; totally implantable subcutaneous ports; and, most recently, peripherally inserted central venous catheters (3-7). The most common life-threatening complication of vascular access is bloodstream infection caused by colonization of the implanted IVD or contamination of the catheter hub or infusate administered through the device (2, 8). Central venous catheters of all types are the most frequent cause of nosocomial bloodstream infection (2, 9-12), and an estimated 250000 to 500000 episodes of IVD-related bloodstream infection occur in the United States annually (9-14). These episodes are associated with an attributable mortality rate of 12% to 25% (15, 16), prolongation of hospitalization by 10 to 40 days (15, 17), and marginal cost to the health care system of up to

Introduction to the Analysis of Survival Data in the Presence of Competing Risks.

35000 per episode (13-18). Accurate and early diagnosis is essential to guide management of IVD-related bloodstream infection. A variety of diagnostic tests that are based on current understanding of the pathogenesis of IVD-related bloodstream infection (12) have been developed (19-23). They can be broadly categorized as methods that necessitate removal of the IVD and those that do not require removal of the IVD (Table 1, Appendix). Table 1. Major Diagnostic Methods for Intravascular DeviceRelated Bloodstream Infection We performed a meta-analysis to determine the most accurate diagnostic methods for IVD-related bloodstream infection. Methods Search and Selection Processes We searched the MEDLINE database (1966 to 31 July 2004), Current Contents (1993 to 31 July 2004), PubMed (1966 to 31 July 2004), and the Cochrane Network by using the search terms intravascular device, vascular catheter, bloodstream infection, diagnosis, blood cultures, and infection, and combinations of these terms. Abstracts of meetings of the InterScience Conference on Antimicrobial Agents and Chemotherapy, the American Society of Microbiology, the Infectious Diseases Society of America, the Society for Healthcare Epidemiology of America, and the Association for Professionals in Infection Control were also reviewed. References from recent published reviews (1-3, 7, 12, 13, 19-23, 38-43) and a previous meta-analysis (30) were also searched. Included studies had to evaluate a diagnostic method for IVD-related bloodstream infection compared with a reference standard and provide sufficient data to calculate the sensitivity and specificity of the test. We excluded case reports, review articles, and nonEnglish-language articles. Studies that assessed the utility of blood cultures drawn from venous or arterial catheters to test for true bacteremia as opposed to contamination were also excluded (44, 45), as were studies of IVD colonization rather than IVD-related bloodstream infection. Data Extraction We used a standard form to extract data on study quality, diagnostic methods studied, reference standard used, patient characteristics, duration of catheterization, antibiotic use, prevalence, sensitivity, and specificity. The Standards for Reporting of Diagnostic Accuracy statement and other published criteria were used to assess study quality (46-48). We evaluated studies for description of the sample; setting; type of IVD studied; method of participant recruitment (all patients with IVDs as opposed to only those with suspected IVD-related bloodstream infection); design (retrospective or prospective); reference standard; definition of cut-off values for positivity; whether evaluators of the test were blinded to the results; statistical methods used to compare diagnostic accuracy and precision; description of indeterminate results; subgroup analyses; and presence of biases that may affect study results, such as incorporation bias (in which the test being studied is part of the reference standard) and work-up bias (46). Data Synthesis We studied the 8 diagnostic methods that are most frequently used in clinical practice and for which performance data have been published: qualitative catheter segment culture, semi-quantitative catheter segment culture (roll-plate method), or quantitative catheter segment culture, each combined with demonstrated concordance with results of concomitant blood cultures; qualitative blood culture drawn through an IVD; quantitative blood culture drawn through an IVD; paired quantitative peripheral and IVD-drawn blood cultures; acridine orange leukocyte cytospin testing of IVD-drawn blood; and differential time to positivity of concomitant qualitative IVD-drawn and peripheral blood cultures (>2 hours). We did not include endoluminal brushing in the meta-analysis because few studies have assessed the test. Four of the 5 studies identified (37, 49-52) were performed by the same group of investigators, and 1 study did not define IVD-related bloodstream infection (51). We also excluded studies of cultures of catheter insertion sites or hubs because of methodologic differences among the studies and a wide range of cut-points for positivity. Statistical Analysis We calculated pooled sensitivities and specificities and 95% CIs for each category of diagnostic tests and an estimate of overall sensitivity and specificity by using a random-effects model and estimating equations similar to those proposed by Zhou and colleagues (53). Heterogeneity in the estimates of sensitivity and specificity was assessed by using the Pearson chi-square test or the Fisher exact test. To combine sensitivity and specificity, we used the approach of Moses and coworkers (54) and calculated D= logit (TPR)logit (FPR) and S= logit (TPR)+logit (FPR), where TPR is the true-positive rate or sensitivity and FPR is the false-positive rate (1specificity). D is interpreted as the log odds ratio, that is, the ratio of the odds that a person who has IVD-related bloodstream infection tests positive to the odds that a person who does not have the disease tests positive for it. We calculated the mean and median values of D by using the values computed within each study. Using the summary receiver-operating characteristic (ROC) curve method of Moses and coworkers (54), we also calculated Q*, which corresponds to the upper leftmost point on the summary ROC curve, where sensitivity equals specificity. The summary measure Q* has been advocated over area under the curve because it is meaningful in the ROC region of greatest interest (54, 55). The ROC curves were derived from linear regressions of D on S and account for random thresholds across studies, as discussed by Moses and coworkers (54). Because the tests for homogeneity were significant, the measure Q* may be better suited to comparing tests than are measures that do not adjust for these differences, since it accounts for random thresholds. The regression model was fit by using equally weighted least squares with the function *1m* in S-PLUS software, version 3.4 (MathSoft, Inc., Seattle, Washington), and a robust resistant method using median regression implemented in *11fit* in S-PLUS software (54). The 95% CIs were reported for mean D and for Q* based on the equally weighted least-squares method. Differences in mean D across all tests were evaluated by using analysis of variance of D computed within individual studies. We also assessed whether increasing degrees of quantitation for methods of catheter segment culture and blood culture would improve the accuracy of the tests. For mean D, separate linear regression analyses were performed for each set of tests, with a covariate for level of quantitation that was coded as an ordinal variable. The same analysis was also done for summary ROC curves (54). A difficulty with mean and median D and with Q* based on summary ROC curves is that these measures do not account for the prevalence of the disease in the group of interest (54). In selecting a test for clinical use, its practical utility will depend not only on its operating characteristics (sensitivity and specificity) but also the patients in which it is being used. The relevant quantities for decision making in this setting are positive predictive value and negative predictive value. We determined positive predictive value and negative predictive value over a wide range of prevalences for each of the tests, on the basis of prevalences from the studies in this meta-analysis. Pooled estimates of sensitivity and specificity were used in these calculations. Heterogeneity was assessed by using 2 subgroup analyses. One subgroup analysis was done to determine whether duration of IVD implantation affected the diagnostic accuracy of the various tests. Studies that did not report the type of IVD studied or that used a mix of short- and long-term catheterization were excluded from this analysis. For each diagnostic test category, pooled sensitivity, specificity, and mean D were calculated separately for short- and long-term catheter placemen

Natural killer cell activation enhances immune pathology and promotes chronic infection by limiting CD8+ T-cell immunity

Competing risks occur frequently in the analysis of survival data. A competing risk is an event whose occurrence precludes the occurrence of the primary event of interest. In a study examining time to death attributable to cardiovascular causes, death attributable to noncardiovascular causes is a competing risk. When estimating the crude incidence of outcomes, analysts should use the cumulative incidence function, rather than the complement of the Kaplan-Meier survival function. The use of the Kaplan-Meier survival function results in estimates of incidence that are biased upward, regardless of whether the competing events are independent of one another. When fitting regression models in the presence of competing risks, researchers can choose from 2 different families of models: modeling the effect of covariates on the cause-specific hazard of the outcome or modeling the effect of covariates on the cumulative incidence function. The former allows one to estimate the effect of the covariates on the rate of occurrence of the outcome in those subjects who are currently event free. The latter allows one to estimate the effect of covariates on the absolute risk of the outcome over time. The former family of models may be better suited for addressing etiologic questions, whereas the latter model may be better suited for estimating a patient’s clinical prognosis. We illustrate the application of these methods by examining cause-specific mortality in patients hospitalized with heart failure. Statistical software code in both R and SAS is provided.

Bone-Density Testing Interval and Transition to Osteoporosis in Older Women

Infections with HIV, hepatitis B virus, and hepatitis C virus can turn into chronic infections, which currently affect more than 500 million patients worldwide. It is generally thought that virus-mediated T-cell exhaustion limits T-cell function, thus promoting chronic disease. Here we demonstrate that natural killer (NK) cells have a negative impact on the development of T-cell immunity by using the murine lymphocytic choriomeningitis virus. NK cell-deficient (Nfil3−/−, E4BP4−/−) mice exhibited a higher virus-specific T-cell response. In addition, NK cell depletion caused enhanced T-cell immunity in WT mice, which led to rapid virus control and prevented chronic infection in lymphocytic choriomeningitis virus clone 13- and reduced viral load in DOCILE-infected animals. Further experiments showed that NKG2D triggered regulatory NK cell functions, which were mediated by perforin, and limited T-cell responses. Therefore, we identified an important role of regulatory NK cells in limiting T-cell immunity during virus infection.

Effect of Adjuvant Chemotherapy on Survival of Patients With Stage III Colon Cancer Diagnosed After Age 75 Years

BACKGROUND Although bone mineral density (BMD) testing to screen for osteoporosis (BMD T score, -2.50 or lower) is recommended for women 65 years of age or older, there are few data to guide decisions about the interval between BMD tests. METHODS We studied 4957 women, 67 years of age or older, with normal BMD (T score at the femoral neck and total hip, -1.00 or higher) or osteopenia (T score, -1.01 to -2.49) and with no history of hip or clinical vertebral fracture or of treatment for osteoporosis, followed prospectively for up to 15 years. The BMD testing interval was defined as the estimated time for 10% of women to make the transition to osteoporosis before having a hip or clinical vertebral fracture, with adjustment for estrogen use and clinical risk factors. Transitions from normal BMD and from three subgroups of osteopenia (mild, moderate, and advanced) were analyzed with the use of parametric cumulative incidence models. Incident hip and clinical vertebral fractures and initiation of treatment with bisphosphonates, calcitonin, or raloxifene were treated as competing risks. RESULTS The estimated BMD testing interval was 16.8 years (95% confidence interval [CI], 11.5 to 24.6) for women with normal BMD, 17.3 years (95% CI, 13.9 to 21.5) for women with mild osteopenia, 4.7 years (95% CI, 4.2 to 5.2) for women with moderate osteopenia, and 1.1 years (95% CI, 1.0 to 1.3) for women with advanced osteopenia. CONCLUSIONS Our data indicate that osteoporosis would develop in less than 10% of older, postmenopausal women during rescreening intervals of approximately 15 years for women with normal bone density or mild osteopenia, 5 years for women with moderate osteopenia, and 1 year for women with advanced osteopenia. (Funded by the National Institutes of Health.).

Background Fluctuation of Kidney Function Versus Contrast-Induced Nephrotoxicity

PURPOSE Few patients 75 years of age and older participate in clinical trials, thus whether adjuvant chemotherapy for stage III colon cancer (CC) benefits this group is unknown. METHODS A total of 5,489 patients ≥ 75 years of age with resected stage III CC, diagnosed between 2004 and 2007, were selected from four data sets containing demographic, stage, treatment, and survival information. These data sets included SEER-Medicare, a linkage between the New York State Cancer Registry (NYSCR) and its Medicare programs, and prospective cohort studies Cancer Care Outcomes Research and Surveillance Consortium (CanCORS) and the National Comprehensive Cancer Network. Data sets were analyzed in parallel using covariate adjusted and propensity score (PS) matched proportional hazards models to evaluate the effect of treatment on survival. PS trimming was used to mitigate the effects of selection bias. RESULTS Use of adjuvant therapy declined with age and comorbidity. Chemotherapy receipt was associated with a survival benefit of comparable magnitude to clinical trials results (SEER-Medicare PS-matched mortality, hazard ratio [HR], 0.60; 95% CI, 0.53 to 0.68). The incremental benefit of oxaliplatin over non-oxaliplatin-containing regimens was also of similar magnitude to clinical trial results (SEER-Medicare, HR, 0.84; 95% CI, 0.69 to 1.04; NYSCR-Medicare, HR, 0.82, 95% CI, 0.51 to 1.33) in two of three examined data sources. However, statistical significance was inconsistent. The beneficial effect of chemotherapy and oxaliplatin did not seem solely attributable to confounding. CONCLUSION The noninvestigational experience suggests patients with stage III CC ≥ 75 years of age may anticipate a survival benefit from adjuvant chemotherapy. Oxaliplatin offers no more than a small incremental benefit. Use of adjuvant chemotherapy after the age of 75 years merits consideration in discussions that weigh individual risks and preferences.

Correlation of MRI findings with clinical findings of trochanteric pain syndrome

OBJECTIVE The reported incidence of contrast-induced acute kidney injury varies widely. Almost no studies have been conducted to quantify the background fluctuation of kidney function of patients receiving iodinated contrast medium. The purpose of this study was a retrospective comparison of the incidence of acute kidney injury among patients undergoing CT with low-osmolar (iohexol) or isoosmolar (iodixanol) contrast medium with the incidence among patients undergoing CT without contrast administration. MATERIALS AND METHODS Creatinine concentration and estimated glomerular filtration rate were evaluated for 11,588 patients. Rates of acute kidney injury (defined as a 0.5 mg/dL increase in serum creatinine concentration or a 25% or greater decrease in estimated glomerular filtration rate within 3 days after CT) were compared among groups and stratified according to creatinine concentration and estimated glomerular filtration rate before the imaging examination. RESULTS In all groups, the incidence of acute kidney injury increased with increasing baseline creatinine concentration. No significant difference in incidence of presumed contrast-induced kidney injury was identified between the isoosmolar contrast medium and the control groups. The incidence of acute kidney injury in the low-osmolar contrast medium cohort paralleled that of the control cohort up to a creatinine level of 1.8 mg/dL, but increases above this level were associated with a higher incidence of acute kidney injury. CONCLUSION We identified a high incidence of acute kidney injury among control subjects undergoing unenhanced CT. The incidence of creatinine elevation in this group was statistically similar to that in the isoosmolar contrast medium group for all baseline creatinine values and all stages of chronic kidney disease. These findings suggest that the additional risk of acute kidney injury accompanying administration of contrast medium (contrast-induced nephrotoxicity) may be overstated and that much of the creatinine elevation in these patients is attributable to background fluctuation, underlying disease, or treatment.

Prediction of cumulative incidence function under the proportional hazards model.

ObjectiveGreater trochanter pain syndrome due to tendinopathy or bursitis is a common cause of hip pain. The previously reported magnetic resonance (MR) findings of trochanteric tendinopathy and bursitis are peritrochanteric fluid and abductor tendon abnormality. We have often noted peritrochanteric high T2 signal in patients without trochanteric symptoms. The purpose of this study was to determine whether the MR findings of peritrochanteric fluid or hip abductor tendon pathology correlate with trochanteric pain.Materials and methodsWe retrospectively reviewed 131 consecutive MR examinations of the pelvis (256 hips) for T2 peritrochanteric signal and abductor tendon abnormalities without knowledge of the clinical symptoms. Any T2 peritrochanteric abnormality was characterized by size as tiny, small, medium, or large; by morphology as feathery, crescentic, or round; and by location as bursal or intratendinous. The clinical symptoms of hip pain and trochanteric pain were compared to the MR findings on coronal, sagittal, and axial T2 sequences using chi-square or Fisher’s exact test with significance assigned as p < 0.05.ResultsClinical symptoms of trochanteric pain syndrome were present in only 16 of the 256 hips. All 16 hips with trochanteric pain and 212 (88%) of 240 without trochanteric pain had peritrochanteric abnormalities (p = 0.15). Eighty-eight percent of hips with trochanteric symptoms had gluteus tendinopathy while 50% of those without symptoms had such findings (p = 0.004). Other than tendinopathy, there was no statistically significant difference between hips with or without trochanteric symptoms and the presence of peritrochanteric T2 abnormality, its size or shape, and the presence of gluteus medius or minimus partial thickness tears.ConclusionsPatients with trochanteric pain syndrome always have peritrochanteric T2 abnormalities and are significantly more likely to have abductor tendinopathy on magnetic resonance imaging (MRI). However, although the absence of peritrochanteric T2 MR abnormalities makes trochanteric pain syndrome unlikely, detection of these abnormalities on MRI is a poor predictor of trochanteric pain syndrome as these findings are present in a high percentage of patients without trochanteric pain.