External control arms in oncology: current use and future directions.

Although randomized control trials allow for a comparison of treatment arms with minimal concern for confounding by known and unknown factors, a randomized study is not feasible in certain disease settings. When a randomized design is not possible, incorporating external control data into the study design can be an effective way to expand the interpretability of the results of an experimental arm by introducing the ability to carry out a formal or an informal comparative analysis. This paper provides an introduction to the concepts of external controls in oncology trials, followed by a review of relevant and current research on this topic. The paper also focuses on general considerations for designing a trial that may incorporate external control data, followed by case studies of the marketing applications submitted to the Food and Drug Administration that included external control data.

Renal artery revascularization in patients with atherosclerotic renal artery stenosis and impaired renal function: conservative management versus renal artery stenting.

BACKGROUND: The impact of percutaneous renal artery angioplasty and stenting (PTRAS) for treatment of atherosclerotic renal artery stenosis (ARAS) is not fully understood, especially in patients with chronic kidney disease (CKD). The goal of this study was to compare renal outcomes in patients treated with PTRAS with those managed conservatively. METHODS: Retrospective cohort study of 118 prevalent patients with significant ARAS and moderate-to-severe CKD who were treated medically (n = 71) or with PTRAS (n = 47). The primary endpoint was change in renal function over the first year after diagnosis/treatment. RESULTS: Average age was 73 +/- 8 years, baseline glomerular filtration rate was 37 +/- 15 ml/min/1.73 m2, and average follow-up was 34 months. Baseline characteristics were similar between the two groups, with the exception of higher diastolic blood pressure in the PTRAS group (75 versus 70 mmHg, p = 0.028). There were no significant differences between the two groups during follow-up. The decline in glomerular filtration rate was similar in both groups (-1.6 ml/min/1.73 m2 in the medical group versus -1.4 ml/min/1.73 m2 in the PTRAS group, p = 0.938). Multivariate models did not indicate an association between treatment modality and changes in renal function or rates of ESRD or death. CONCLUSION: In patients with advanced kidney disease, medical therapy and renal artery stenting appear comparable in stabilizing renal function for ARAS.

Biomarkers for the diagnosis and risk stratification of acute kidney injury: a systematic review.

The diagnosis of acute kidney injury (AKI) is usually based on changes in serum creatinine, but such measurements are a poor marker of acute deterioration in kidney function. We performed a systematic review of publications that evaluated the accuracy and reliability of serum and urinary biomarkers in human subjects when used for the diagnosis of established AKI or early AKI, or to risk stratify patients with AKI. Two reviewers independently searched the MEDLINE and EMBASE databases (January 2000-March 2007) for studies pertaining to biomarkers for AKI. Studies were assessed for methodologic quality. In total, 31 studies evaluated 21 unique serum and urine biomarkers. Twenty-five of the 31 studies were scored as having 'good' quality. The results of the studies indicated that serum cystatin C, urine interleukin-18 (IL-18), and urine kidney injury molecule-1 (KIM-1) performed best for the differential diagnosis of established AKI. Serum cystatin C and urine neutrophil gelatinase-associated lipocalin, IL-18, glutathione-S-transferase-pi, and gamma-glutathione-S-transferase performed best for early diagnosis of AKI. Urine N-acetyl-beta-D-glucosaminidase, KIM-1, and IL-18 performed the best for mortality risk prediction after AKI. In conclusion, published data from studies of serum and urinary biomarkers suggest that biomarkers may have great potential to advance the fields of nephrology and critical care. These biomarkers need validation in larger studies, and the generalizability of biomarkers to different types of AKI as well as the incremental prognostic value over traditional clinical variables needs to be determined.

Effect of non-insulin-dependent diabetes mellitus on myocardial insulin responsiveness in patients with ischemic heart disease.

BACKGROUND: Patients with non-insulin-dependent diabetes mellitus (NIDDM) exhibit poor clinical outcomes from myocardial ischemia. This may reflect an impairment in their cardiac insulin-response system. METHODS AND RESULTS: We used AV balance and intracoronary infusion techniques to compare the intrinsic cardiac responsiveness to insulin in 26 coronary disease patients with (n=13) and without (n=13) NIDDM. During fasting, NIDDM hearts demonstrated lower fractional extraction of glucose from arterial plasma than controls (1.0+/-0.5% versus 2.1+/-0.5%, P<0.05) despite higher circulating insulin levels (26+/-5 versus 13+/-4 microU. mL, P<0.05). This was compensated for by higher circulating glucose levels, so that net cardiac glucose uptake in the 2 groups was equivalent (5.2+/-1.1 versus 5.3+/-1.1 micromol. min). Intracoronary insulin infusion produced an approximately 3-fold increase in fractional extraction and net uptake of glucose across the heart in both groups (to 3.7+/-0.4% and 18.3+/-3.5 micromol. min in NIDDM and to 5.4+/-0.7% and 17.7+/-4.3 micromol. min in controls) accompanied by an approximately 30% increase in net lactate uptake, suggesting preserved insulin action on both glucose uptake and glucose oxidation in the NIDDM heart. In nondiabetics, insulin consistently increased coronary blood flow, but this effect was absent in NIDDM. CONCLUSIONS: In contrast to their peripheral tissues and coronary vasculature, the myocardium of patients with NIDDM expresses a competent insulin-response system with respect to glucose metabolism. This suggests that insulin resistance is mediated at the level of individual organs and that different mechanisms are involved in muscle and vascular tissue.

Predicting which patients can resume oral nutrition after percutaneous endoscopic gastrostomy tube placement.

BACKGROUND: Percutaneous endoscopic gastrostomy tubes are placed with high frequency and relative safety for a variety of indications. One of these indications is temporary nutritional support for patients expected to resume oral nutrition. AIMS: To determine if baseline clinical characteristics can predict which patients attain the clinical goal of resuming oral nutrition with consequent tube removal. METHODS: We conducted a single site observational cohort study from December 1999 to April 2001, enrolling all patients scheduled for percutaneous endoscopic gastrostomy placement. Standard descriptive and bivariate analyses were performed. Cox proportional hazard models were constructed to identify patient characteristics prior to percutaneous endoscopic gastrostomy placement that might predict resumption of oral nutrition with tube removal. RESULTS: Bivariate analyses revealed four potential clinical predictors: age < 65 years, localized head and neck cancer, serum albumin > or = 3.75 g/dL, and serum creatinine < or = 1.1 mg/dL. In multivariable analysis, age < 65 years (HR = 3.7, 95% CI: 1.0-14.3) and a diagnosis of localized head and neck cancer (HR = 4.6, 95% CI: 1.4-15.0) predicted resumption of oral nutrition with percutaneous endoscopic gastrostomy removal. CONCLUSIONS: When discussing percutaneous endoscopic gastrostomy placement, doctors should consider the likelihood of achieving clinically important outcomes such as the resumption of oral nutrition with tube removal. This clinical goal is unlikely for older patients with diagnoses other than localized head and neck cancer.

Endemic tuberculosis among homeless men in New York City.

OBJECTIVES: The purpose of the study was to describe demographic and clinical characteristics of patients at the only long-term care facility for homeless men with tuberculosis in New York City, and to evaluate the outcome of a directly observed therapy program for these men. METHODS: The study population included residents at the "tuberculosis unit" for men in the New York City municipal shelter system. A cross-sectional survey described the characteristics of 76 men in the unit during November 1991. A retrospective cohort study evaluated 104 consecutive admissions to the facility from October 1, 1990, through March 30, 1991, and determined the outcome of directly observed therapy. RESULTS: Cross-sectional survey (n = 76). The median age was 43 years (range, 25 to 60 years); 67 patients (88%) had pulmonary tuberculosis. Among 58 isolates of Mycobacterium tuberculosis, eight were resistant to one drug (14%) and an additional nine were resistant to at least two drugs (16%). A history of previous treatment was associated with an odds ratio of 5.1 for having multiple drug-resistant tuberculosis (exact 95% confidence interval, 0.8 to 53.5). Retrospective cohort (n = 104). Excluding 21 men whose care was transferred to other agencies or institutions, 39 (47%) of 83 subjects completed or were still receiving treatment after 12 months and 44 (53%) of 83 subjects failed to complete the program. CONCLUSIONS: As expected, previous treatment for tuberculosis among homeless men is associated with an increased risk of having multiple-drug resistance. A directly observed therapy program successfully treated less than half of the enrolled subjects. Increased efforts are needed to control the spread of tuberculosis among homeless individuals.

Granulocyte-colony stimulating factor therapy in drug-induced agranulocytosis.

Drug-induced agranulocytosis in the outpatient setting is a rare but potentially fatal adverse effect of many classes of medications. Five patients with this disorder presented to Yale-New Haven (Conn) Hospital during 1990 through 1992. Three patients treated with granulocyte-colony stimulating factor and two patients treated with routine care were studied for relevant clinical outcomes. Treatment with granulocyte-colony stimulating factor was associated with a shorter duration of neutropenia and a decreased length of hospital stay, consistent with recent case reports. Despite the high cost of the drug, treatment with granulocyte-colony stimulating factor was found to be cost-effective for patients with uncomplicated drug-induced agranulocytosis.

Asking patients what they like: overlooked attributes of patient satisfaction with primary care.

PURPOSE: Ask patients to describe important attributes of their primary health care; and use the responses to develop a taxonomy for classifying patient satisfaction. DESIGN: Open-ended questions were administered to patients immediately after a clinic visit. SETTING: Primary care clinics at an academically affiliated Veterans Affairs Medical Center in New England. PATIENTS: Two hundred two of 204 randomly selected English-speaking patients who agreed (and were able) to participate. INTERVENTIONS: Clinimetric methods were used to obtain responses to three open-ended questions about what patients liked, disliked, and would like to see changed about their care. These "raw" descriptions were then combined into pertinent groups and arranged as a taxonomy of patient satisfaction. RESULTS: The taxonomy was divided into five main axes referring to physician staff, nonphysician staff, attributes of the clinic, related services, and the institution. The axes contained a total of 34 items related to patient satisfaction. The items have demonstrable face validity, and are likely to be "transparently" sensible to clinicians and policy makers, but many of the items-such as problems with parking-were not included in either of two existing psychometric instruments used to measure patient satisfaction in the same clinics. CONCLUSIONS: The clinimetric strategy leads to a simple, clinically relevant, and easily understood assessment of patient satisfaction with health care services. The assessment can be done with three simple questions; and the responses can be catalogued, when desired, in a suitable taxonomy.

The quest for "power": contradictory hypotheses and inflated sample sizes.

To have the "power" of avoiding undersized clinical trials, the customary statistical strategy used in the past few decades is aimed at rejecting both a null stochastic hypothesis and a contradictory alternative hypothesis. This approach gives a trial the "power" to confirm the "insignificance" of differences much smaller than the large value of delta desired in trials done to show efficacy. In many instances, however, a prime problem is that the current "double-significance" approach produces sample sizes 2-3 times larger than needed for stochastic confirmation of large differences (> or =delta). The inflated sample sizes and consequent problems can be avoided if a realistic value for delta is chosen and maintained thereafter, and if an adequate "capacity" is calculated for "single significance."

Importance of events per independent variable in proportional hazards analysis. I. Background, goals, and general strategy.

Multivariable methods of analysis can yield problematic results if methodological guidelines and mathematical assumptions are ignored. A problem arising from a too-small ratio of events per variable (EPV) can affect the accuracy and precision of regression coefficients and their tests of statistical significance. The problem occurs when a proportional hazards analysis contains too few "failure" events (e.g., deaths) in relation to the number of included independent variables. In the current research, the impact of EPV was assessed for results of proportional hazards analysis done with Monte Carlo simulations in an empirical data set of 673 subjects enrolled in a multicenter trial of coronary artery bypass surgery. The research is presented in two parts: Part I describes the data set and strategy used for the analyses, including the Monte Carlo simulation studies done to determine and compare the impact of various values of EPV in proportional hazards analytical results. Part II compares the output of regression models obtained from the simulations, and discusses the implication of the findings.

Importance of events per independent variable in proportional hazards regression analysis. II. Accuracy and precision of regression estimates.

The analytical effect of the number of events per variable (EPV) in a proportional hazards regression analysis was evaluated using Monte Carlo simulation techniques for data from a randomized trial containing 673 patients and 252 deaths, in which seven predictor variables had an original significance level of p < 0.10. The 252 deaths and 7 variables correspond to 36 events per variable analyzed in the full data set. Five hundred simulated analyses were conducted for these seven variables at EPVs of 2, 5, 10, 15, 20, and 25. For each simulation, a random exponential survival time was generated for each of the 673 patients, and the simulated results were compared with their original counterparts. As EPV decreased, the regression coefficients became more biased relative to the true value; the 90% confidence limits about the simulated values did not have a coverage of 90% for the original value; large sample properties did not hold for variance estimates from the proportional hazards model, and the Z statistics used to test the significance of the regression coefficients lost validity under the null hypothesis. Although a single boundary level for avoiding problems is not easy to choose, the value of EPV = 10 seems most prudent. Below this value for EPV, the results of proportional hazards regression analyses should be interpreted with caution because the statistical model may not be valid.

A nested case-control study of the effectiveness of screening for prostate cancer: research design.

Although prostate-specific antigen (PSA) and digital rectal examination (DRE) are commonly used to screen for prostate cancer, available data do not confirm that either test improves survival. This report describes the methodological aspects of a nested case-control study addressing the question of whether PSA screening, with or without DRE, is effective in increasing survival. Potential sources of bias are discussed, as well as corresponding strategies used to avoid them. Results are expected in the year 2002.

A simulation study of the number of events per variable in logistic regression analysis.

We performed a Monte Carlo study to evaluate the effect of the number of events per variable (EPV) analyzed in logistic regression analysis. The simulations were based on data from a cardiac trial of 673 patients in which 252 deaths occurred and seven variables were cogent predictors of mortality; the number of events per predictive variable was (252/7 =) 36 for the full sample. For the simulations, at values of EPV = 2, 5, 10, 15, 20, and 25, we randomly generated 500 samples of the 673 patients, chosen with replacement, according to a logistic model derived from the full sample. Simulation results for the regression coefficients for each variable in each group of 500 samples were compared for bias, precision, and significance testing against the results of the model fitted to the original sample. For EPV values of 10 or greater, no major problems occurred. For EPV values less than 10, however, the regression coefficients were biased in both positive and negative directions; the large sample variance estimates from the logistic model both overestimated and underestimated the sample variance of the regression coefficients; the 90% confidence limits about the estimated values did not have proper coverage; the Wald statistic was conservative under the null hypothesis; and paradoxical associations (significance in the wrong direction) were increased. Although other factors (such as the total number of events, or sample size) may influence the validity of the logistic model, our findings indicate that low EPV can lead to major problems.

Alcohol exposure and health services utilization in older veterans.

The objective of this study was to determine if increased alcohol exposure is associated with greater use of health services among older veterans. A total of 129 older veterans (> or =65 years old), receiving longitudinal care in a Veterans' Administration primary care clinic, were followed retrospectively for up to 42 months. Subjects were screened at baseline for problem drinking with the CAGE or the quantity-frequency questions from the Alcohol Use Disorders Identification Test (QF-AUDIT), and stratified by exposure into three categories: abstainers, social drinkers, and problem drinkers. Outcomes included total outpatient clinic visits, laboratory tests, radiologic and other technologic procedures, as well as acute care hospitalizations. For all subjects (N = 129), no association was found between alcohol exposure and use of any outpatient services. Among CAGE-screened (n = 62) abstainers, social drinkers, and problem drinkers, significant differences were found in the median number of laboratory tests (7.3 vs. 3.4 vs. 7.1, P = 0.004) and hospitalizations (0.3 vs. 0.0 vs. 0.3, P = 0.001) per patient year of follow-up. No exposure-outcome associations were present, however, among QF-AUDIT-screened subjects (n = 67). We were unable to demonstrate a consistent relationship between alcohol exposure and health services utilization. The effects of alcohol on older veterans' use of health services varied with the method used to measure alcohol exposure. Additional studies are needed to determine whether multiple, or possibly new, measures can more precisely define the health effects of alcohol in older populations.

Risk adjustment for older hospitalized persons: a comparison of two methods of data collection for the Charlson index.

To compare Charlson indices based on chart data and ICD-9 data for agreement overall and on rating specific comorbid conditions, and to compare mortality risks associated with these indices. Prospective cohort study. Six general medicine wards at Yale-New Haven Hospital. 524 consecutive patients who had no clinical evidence of delirium at enrollment, admitted between November 6, 1989 and July 31, 1991, aged 70 years or older. Death within 1 year of the index hospital admission date. Scores using the chart-based data were significantly higher than those using ICD-9 data. About half of the individual conditions showed fair-to-good agreement between the two scores, whereas the other half showed poor agreement. A comparison of mortality prediction indicated that the weightings assigned to individual comorbidities differed substantially from those used in Charlson's original index. While mortality prediction of each individual index was comparable, the ICD-9 and chart indices contributed independently to mortality prediction in the presence of the other. Low agreement between Charlson scores based on the two methods of data collection and their cumulative contribution to mortality prediction suggest that these indices may include different information. Our results suggest that the original Charlson index may not provide optimal risk adjustment for elderly general medicine samples. We suggest development of an empirically-derived index of comorbid conditions and weights may be warranted for older general medical patients.

Alcohol use and functional disability among cognitively impaired adults.

BACKGROUND: The extent to which alcohol exposure increases risk for functional disability among older adults with cognitive impairment has not previously been assessed. OBJECTIVE: To examine the potential relationship between alcohol use and functional disability among older cognitively impaired adults. DESIGN: Retrospective medical record review. SETTING: Hospital-based geriatric assessment center. PARTICIPANTS: Two hundred forty-two consecutive participants with Mini-Mental Status Examination scores of < or = 24. MEASUREMENTS: Proxy-reported alcohol intake was classified in categories of never, former, light (< 1 drink/week), moderate (> or = 1 but < 14 drinks/week), and heavy (> or = 14 drinks/week) drinkers, and functional status was determined by proxy-reported performance in seven basic (BADL) and seven instrumental (IADL) activities of daily living (0 = poorest function and 14 = best function). RESULTS: Compared with never drinkers, moderate drinkers demonstrated higher mean BADL (12.2 vs 11.4, P = .033) and IADL scores (6.6 vs 5.6, P = .067), whereas heavy drinkers had higher BADL (12.8 vs 11.4, P = .019) but lower IADL scores (4.8 vs 5.6, P = .425). Former drinkers demonstrated both lower BADL (10.8 vs 11.4, P = .107) and IADL scores (3.9 vs 5.6, P = .011) compared with never drinkers. Evaluation of a potential dose-response effect was limited due to low numbers of light and heavy drinkers. CONCLUSIONS: Among cognitively impaired adults, moderate and heavy drinkers demonstrated better BADL function, whereas former drinkers had poorer IADL function, compared with never drinkers. Prospective studies that incorporate additional measures of exposure (e.g., cumulative lifetime consumption) and function (e.g., performance-based tests) may provide a more comprehensive understanding of alcohol's effects among older cognitively impaired adults.

Differences in cervical cancer mortality among black and white women.

OBJECTIVE: To determine whether stage of disease and treatment patterns account for mortality differences between black and white women with cervical cancer. METHODS: Using data obtained from the Surveillance, Epidemiology, and End Results (SEER) Program for 1988-1994, we determined the associations between race and stage, and race and treatment. Racial differences in survival for up to 7 years of follow-up were adjusted for age, marital status, SEER location, International Federation of Gynecology and Obstetrics (FIGO) stage of disease, lymph node status, grade, histology, and treatment. RESULTS: Cumulative mortality was 36% (366 deaths in 1029 women) for black women and 24% (1215 deaths in 5021 women) for white women; unadjusted hazard ratio was 1.60 (95% confidence interval [CI] 1.43, 1.80). Black women were more likely to present with advanced disease than white women (43.8% compared with 34.8%). In a model adjusting for demographics and FIGO stage, the hazard ratio for black women compared with white women decreased to 1.35 (95% CI 1.19, 1.54). Treatment varied by race, with black women receiving surgery less often (33.5% compared with 48.2%, respectively) and radiation therapy more often (35.3% and 25.2%, respectively) than white women. In a comprehensive model including demographic factors, FIGO stage, other tumor characteristics, and treatment, the adjusted hazard ratio for mortality remained high for black women at 1.30 (95% CI 1.14, 1.48). CONCLUSION: Race remains an independent predictor of cervical cancer survival after accounting for age, stage of disease, treatment patterns, and other factors. Future studies should assess racial differences in clinical severity of disease, comorbidity, and socioeconomic status.

Hyperglycaemia in patients with acute ischaemic stroke: how often do we screen for undiagnosed diabetes?

BACKGROUND: Hyperglycaemia is common among patients with acute ischaemic stroke, and may be due to the physiological stress of the acute stroke event or reflect underlying diabetes mellitus. The under-diagnosis of diabetes in the general population, combined with the association of diabetes and stroke, suggests a rationale for screening for diabetes among hyperglycaemic stroke patients. AIM: To determine how often clinicians screen for diabetes among hyperglycaemic stroke patients without a prior diagnosis of diabetes. DESIGN: Retrospective medical record review. METHODS: We reviewed the records of acute ischaemic stroke patients admitted at any of ten Connecticut hospitals from May 1996 through December 1998. RESULTS: We identified 90 acute stroke patients with no prior history of diabetes. The prevalence of hyperglycaemia varied from 31% down to 6%, depending on the maximum glucose cut-off used to define hyperglycaemia: from > or = 140 mg/dl (7.8 mmol/l) to > or = 200 mg/dl (11.1 mmol/l). Only one of the hyperglycaemic patients (1/90, 1%) had any evidence that a clinician screened or planned to screen for undiagnosed diabetes: one patient had a haemoglobin A1c measured during the hospitalization, none received oral glucose tolerance testing while hospitalized, and no discharge summary included a plan to screen for diabetes as an out-patient. DISCUSSION: Hyperglycaemic stroke patients without a previous diagnosis of diabetes are not routinely screened for diabetes. This situation represents an opportunity, currently unused, to identify an important and modifiable condition.

Screening for skin cancer in primary care settings.

OBJECTIVE: To estimate the frequency of recorded screening for skin cancer in primary care settings. DESIGN: Retrospective observational cohort study. SETTING: Two academically affiliated Department of Veterans Affairs Medical Centers. SUBJECTS: Two hundred randomly selected patients at least 50 years old and receiving care at outpatient medical clinics. MAIN OUTCOME MEASURE: Frequency of documented skin examinations, in comparison with other tests routinely done as screening, during a 2-year period. METHODS: Medical record review to identify how often selected components of the physical examination and specific procedures were documented during ambulatory visits. RESULTS: Among the 200 subjects, the frequency of documented examinations and procedures included fecal occult blood testing in 120 (60%), rectal examination in 128 (64%), and sigmoidoscopy in 93 (47%), prostate examination was performed in 114 (59%) of 193 men. In contrast, skin examination was documented in only 56 (28%) of 200 subjects (P < .001 for each comparison with other tests). As an estimate of the "true" frequency of screening for skin cancer, 35 (18%) of 165 patients without skin-related complaints had a documented skin examination. CONCLUSION: Skin cancer screening is infrequently documented and therefore possibly omitted in the context of primary care visits.

Monte Carlo methods in clinical research: applications in multivariable analysis.

BACKGROUND: Monte Carlo methods use "simulated" analyses with random numbers for solving problems, particularly those that defy solutions using mathematical theory alone. Research using Monte Carlo simulations is very popular in many branches of science and is sometimes done in clinical investigation. The origins and basic strategy of the technique, however, may not be well known to clinical researchers. The purpose of this paper is to describe the history and general principles of Monte Carlo methods and to demonstrate how Monte Carlo simulations were recently applied to examine a phenomenon in multivariable statistical analysis called the number of outcome events per independent variable (EPV). For example, in a cohort of 200 people, with 50 deaths and 5 independent (predictor) variables, EPV = 50/5 = 10. METHODS: The "real-world" data came from a clinical trial of 673 patients in which 7 variables were cogent predictors of 252 deaths, so that EPV = 252/7 = 36. For the Monte Carlo simulations, special models were used while allowing simulations of proportional hazards and logistic regression to maintain the basic relationship of variables and the same size of the original population, at EPV values of 2, 5, 10, 15, 20, and 25. RESULTS: The Monte Carlo simulations confirmed a previously undocumented "rule of thumb" stating that when the EPV is less than 10-20, the algebraic models used in logistic regression and proportional hazards regression may be unreliable, leading to imprecise or spurious results. CONCLUSION: Monte Carlo techniques offer attractive methods for clinical investigators to use in solving problems that are not amenable to customary mathematical approaches.