Christopher H. Schmid

Quantitative synthesis in systematic reviews.

A quantitative systematic review, or meta-analysis, uses statistical methods to combine the results of multiple studies. Meta-analyses have been done for systematic reviews of therapeutic trials, diagnostic test evaluations, and epidemiologic studies. Although the statistical methods involved may at first appear to be mathematically complex, their purpose is simple: They are trying to answer four basic questions. Are the results of the different studies similar? To the extent that they are similar, what is the best overall estimate? How precise and robust is this estimate? Finally, can dissimilarities be explained? This article provides some guidance in understanding the key technical aspects of the quantitative approach to these questions. We have avoided using equations and statistical notations; interested readers will find implementations of the described methods in the listed references. We focus here on the quantitative synthesis of reports of randomized, controlled, therapeutic trials because far more meta-analyses on therapeutic studies than on other types of studies have been published. For practical reasons, we present a stepwise description of the tasks that are performed when statistical methods are used to combine data. These tasks are 1) deciding whether to combine data and defining what to combine, 2) evaluating the statistical heterogeneity of the data, 3) estimating a common effect, 4) exploring and explaining heterogeneity, 5) assessing the potential for bias, and 6) presenting the results. Deciding Whether To Combine Data and Defining What To Combine By the time one performs a quantitative synthesis, certain decisions should already have been made about the formulation of the question and the selection of included studies. These topics were discussed in two previous articles in this series [1, 2]. Statistical tests cannot compensate for lack of common sense, clinical acumen, and biological plausibility in the design of the protocol of a meta-analysis. Thus, a reader of a systematic review should always address these issues before evaluating the statistical methods that have been used and the results that have been generated. Combining poor-quality data, overly biased data, or data that do not make sense can easily produce unreliable results. The data to be combined in a meta-analysis are usually either binary or continuous. Binary data involve a yes/no categorization (for example, death or survival). Continuous data take a range of values (for example, change in diastolic blood pressure after antihypertensive treatment, measured in mm Hg). When one is comparing groups of patients, binary data can be summarized by using several measures of treatment effect that were discussed earlier in this series [3]. These measures include the risk ratio; the odds ratio; the risk difference; and, when study duration is important, the incidence rate. Another useful clinical measure, the number needed to treat (NNT), is derived from the inverse of the risk difference [3]. Treatment effect measures, such as the risk ratio and the odds ratio, provide an estimate of the relative efficacy of an intervention, whereas the risk difference describes the interventions absolute benefit. The various measures of treatment effect offer complementary information, and all should be examined [4]. Continuous data can be summarized by the raw mean difference between the treatment and control groups when the treatment effect is measured on the same scale (for example, diastolic blood pressure in mm Hg), by the standardized mean difference when different scales are used to measure the same treatment effect (for example, different pain scales being combined), or by the correlation coefficients between two continuous variables [5]. The standardized mean difference, also called the effect size, is obtained by dividing the difference between the mean in the treatment group and the mean in the control group by the SD in the control group. Evaluating the Statistical Heterogeneity of the Data This step is intended to answer the question, Are the results of the different studies similar (homogeneous)? It is important to answer this question before combining any data. To do this, one must calculate the magnitude of the statistical diversity (heterogeneity) of the treatment effect that exists among the different sets of data. Statistical diversity can be thought of as attributable to one or both of two causes. First, study results can differ because of random sampling error. Even if the true effect is the same in each study, the results of different studies would be expected to vary randomly around the true common fixed effect. This diversity is called the within-study variance. Second, each study may have been drawn from a different population, depending on the particular patients chosen and the interventions and conditions unique to the study. Therefore, even if each study enrolled a large patient sample, the treatment effect would be expected to differ. These differences, called random effects, describe the between-study variation with regard to an overall mean of the effects of all of the studies that could be undertaken. The test most commonly used to assess the statistical significance of between-study heterogeneity is based on the chi-square distribution [6]. It provides a measure of the sum of the squared differences between the results observed and the results expected in each study, under the assumption that each study estimates the same common treatment effect. A large total deviation indicates that a single common treatment effect is unlikely. Any pooled estimate calculated must account for the between-study heterogeneity. In practice, this test has low sensitivity for detecting heterogeneity, and it has been suggested that a liberal significance level, such as 0.1, should be used [6]. Estimating a Common Effect The questions that this step tries to answers are, 1) To the extent that data are similar, what is their best common point estimate of a therapeutic effect, and 2) how precise is this estimate? The mathematical process involved in this step generally involves combining (pooling) the results of different studies into an overall estimate. Compared with the results of individual studies, pooled results can increase statistical power and lead to more precise estimates of treatment effect. Each study is given a weight according to the precision of its results. The rationale is that studies with narrow CIs should be weighted more heavily than studies with greater uncertainty. The precision is generally expressed by the inverse of the variance of the estimate of each study. The variance has two components: the variance of the individual study and the variance between different studies. When the between-study variance is found to be or assumed to be zero, each study is simply weighted by the inverse of its own variance, which is a function of the study size and the number of events in the study. This approach characterizes a fixed-effects model, as exemplified by the Mantel-Haenszel method [7, 8] or the Peto method [9] for dichotomous data. The Peto method has been particularly popular in the past. It has the advantage of simple calculation; however, although it is appropriate in most cases, it may introduce large biases if the data are unbalanced [10, 11]. On the other hand, random-effects models also add the between-study variance to the within-study variance of each individual study when the pooled mean of the random effects is calculated. The random-effects model most commonly used for dichotomous data is the DerSimonian and Laird estimate of the between-study variance [12]. Fixed- and random-effects models for continuous data have also been described [13]. Pooled results are generally reported as a point estimate and CI, typically a 95% CI. Other quantitative techniques for combining data, such as the Confidence Profile Method [14], use Bayesian methods to calculate posterior probability distributions for effects of interest. Bayesian statistics are based on the principle that each observation or set of observations should be viewed in conjunction with a prior probability describing the prior knowledge about the phenomenon of interest [15]. The new observations alter this prior probability to generate a posterior probability. Traditional meta-analysis assumes that nothing is known about the magnitude of the treatment effect before randomized trials are performed. In Bayesian terms, the prior probability distribution is noninformative. Bayesian approaches may also allow the incorporation of indirect evidence in generating prior distributions [14] and may be particularly helpful in situations in which few data from randomized studies exist [16]. Bayesian analyses may also be used to account for the uncertainty introduced by estimating the between-study variance in the random-effects model, leading to more appropriate estimates and predictions of treatment efficacy [17]. Exploring and Explaining Heterogeneity The next important issue is whether the common estimate obtained in the previous step is robust. Sensitivity analyses determine whether the common estimate is influenced by changes in the assumptions and in the protocol for combining the data. A comparison of the results of fixed- and random-effects models is one such sensitivity analysis [18]. Generally, the random-effects model produces wider CIs than does the fixed-effects model, and the level of statistical significance may therefore be different depending on the model used. The pooled point estimate per se is less likely to be affected, although exceptions are possible [19]. Other sensitivity analyses may include the examination of the residuals and the chi-square components [13] and assessment of the effect of deleting each study in turn. Statistically significant results that depend on a single study may require further exploration. Cumulative Meta-Analysis Cu

Recommendations for examining and interpreting funnel plot asymmetry in meta-analyses of randomised controlled trials.

Funnel plots, and tests for funnel plot asymmetry, have been widely used to examine bias in the results of meta-analyses. Funnel plot asymmetry should not be equated with publication bias, because it has a number of other possible causes. This article describes how to interpret funnel plot asymmetry, recommends appropriate tests, and explains the implications for choice of meta-analysis model

Estimating Glomerular Filtration Rate from Serum Creatinine and Cystatin C

BACKGROUND Estimates of glomerular filtration rate (GFR) that are based on serum creatinine are routinely used; however, they are imprecise, potentially leading to the overdiagnosis of chronic kidney disease. Cystatin C is an alternative filtration marker for estimating GFR. METHODS Using cross-sectional analyses, we developed estimating equations based on cystatin C alone and in combination with creatinine in diverse populations totaling 5352 participants from 13 studies. These equations were then validated in 1119 participants from 5 different studies in which GFR had been measured. Cystatin and creatinine assays were traceable to primary reference materials. RESULTS Mean measured GFRs were 68 and 70 ml per minute per 1.73 m(2) of body-surface area in the development and validation data sets, respectively. In the validation data set, the creatinine-cystatin C equation performed better than equations that used creatinine or cystatin C alone. Bias was similar among the three equations, with a median difference between measured and estimated GFR of 3.9 ml per minute per 1.73 m(2) with the combined equation, as compared with 3.7 and 3.4 ml per minute per 1.73 m(2) with the creatinine equation and the cystatin C equation (P=0.07 and P=0.05), respectively. Precision was improved with the combined equation (interquartile range of the difference, 13.4 vs. 15.4 and 16.4 ml per minute per 1.73 m(2), respectively [P=0.001 and P<0.001]), and the results were more accurate (percentage of estimates that were >30% of measured GFR, 8.5 vs. 12.8 and 14.1, respectively [P<0.001 for both comparisons]). In participants whose estimated GFR based on creatinine was 45 to 74 ml per minute per 1.73 m(2), the combined equation improved the classification of measured GFR as either less than 60 ml per minute per 1.73 m(2) or greater than or equal to 60 ml per minute per 1.73 m(2) (net reclassification index, 19.4% [P<0.001]) and correctly reclassified 16.9% of those with an estimated GFR of 45 to 59 ml per minute per 1.73 m(2) as having a GFR of 60 ml or higher per minute per 1.73 m(2). CONCLUSIONS The combined creatinine-cystatin C equation performed better than equations based on either of these markers alone and may be useful as a confirmatory test for chronic kidney disease. (Funded by the National Institute of Diabetes and Digestive and Kidney Diseases.).

Progression of Chronic Kidney Disease: The Role of Blood Pressure Control, Proteinuria, and Angiotensin-Converting Enzyme Inhibition: A Patient-Level Meta-Analysis

Context Guidelines recommend a blood pressure of less than 130/80 mm Hg for patients with chronic kidney disease. Contribution This meta-analysis showed that systolic blood pressure and urinary protein excretion were related to the risk for renal disease progression in patients with nondiabetic kidney disease. Systolic pressures of 110 to 129 mm Hg were associated with the lowest risks. Higher risks with higher pressures were marked in patients with protein excretion greater than 1.0 g/d and were not apparent in those with lower urinary protein excretion. Implications In patients with urinary protein excretion greater than 1.0 g/d, systolic blood pressure of 110 to 129 mm Hg is associated with the lowest risk for progression of renal disease. The Editors Chronic kidney disease is a major public health problem in the United States. The prevalence of kidney failure (recorded as end-stage renal disease) has risen steadily since Medicare assumed funding for the condition in 1973. By 2010, it is estimated that the prevalence will be greater than 650 000 (1). The prevalence of earlier stages of chronic kidney disease is even higher. The Third National Health and Nutrition Examination Survey (NHANES III), conducted from 1988 to 1994, estimates that 5.6 million persons 17 years of age or older had decreased kidney function, as defined by an elevated serum creatinine concentration ( 141 mol/L [ 1.6 mg/dL] in men and 124 mol/L [ 1.4 mg/dL] in women) (2). Hypertension and proteinuria occur in most patients with chronic kidney disease and are risk factors for faster progression of kidney disease. Antihypertensive agents reduce blood pressure and urine protein excretion and slow the progression of kidney disease. The sixth report of the Joint National Committee for the Prevention, Detection, Evaluation, and Treatment of High Blood Pressure (JNC-VI) recommends a lower blood pressure goal for patients with decreased kidney function (<130/85 mm Hg if urine protein excretion is <1 g/d and <125/75 mm Hg if urine protein excretion is >1 g/d) than for patients without target organ damage (<140/90 mm Hg) (3). It is not known whether even lower blood pressure might provide additional benefit. On the other hand, there is concern about excessive lowering of blood pressure because it may be associated with a higher risk for cardiovascular disease (4, 5). Additional lowering of blood pressure might also have a detrimental effect on kidney disease. The recommendations in JNC-VI are based principally on the results of the Modification of Diet in Renal Disease (MDRD) Study (6, 7), a study of nondiabetic kidney disease that did not evaluate the effect of angiotensin-converting enzyme (ACE) inhibitors or angiotensin-receptor blockers. Since publication of the JNC-VI, other large studies and meta-analyses have shown that antihypertensive regimens containing ACE inhibitors or angiotensin-receptor blockers seem to be more effective than other regimens in slowing the progression of chronic kidney disease (8-17). In some studies, the beneficial effect of these agents seemed to be greater in patients with proteinuria (8-11, 13) and was mediated in part by their effects to lower blood pressure and urine protein excretion (13). Of these studies, only the African American Study of Kidney Disease and Hypertension (AASK) compared two levels of blood pressure in patients treated with an ACE inhibitor (11). In that study of patients with hypertensive nephrosclerosis, a type of kidney disease associated with low levels of proteinuria, a lower blood pressure goal did not reduce the risk for progression of kidney disease when compared with a usual blood pressure goal. However, the AASK does not address the relationships of blood pressure and urine protein excretion with the progression of kidney disease in patients with higher levels of urine protein excretion. We performed a patient-level meta-analysis using data from the ACE Inhibition in Progressive Renal Disease (AIPRD) Study Group database (13) to assess these relationships among patients with nondiabetic kidney disease across a wide range of urine protein excretion values during antihypertensive therapy with and without ACE inhibitors. Methods Study Design The AIPRD Study Group database includes 1860 patients with nondiabetic kidney disease enrolled in 11 randomized, controlled trials of ACE inhibitors to slow the progression of kidney disease. The database contains information on blood pressure, urine protein excretion, serum creatinine concentration, and onset of kidney failure during 22 610 visits. Inclusion and exclusion criteria, search strategies for identifying clinical trials, and details of database formulation have been previously described (13, 18). The AIPRD Study Group was formed in 1997. Briefly, we identified studies by searching the MEDLINE database for English-language reports evaluating the effect of ACE inhibitors or kidney disease in humans between 1977 (when ACE inhibitors were approved for trials in humans) and 1999 (when the database was closed). We included only randomized trials (with a minimum 1-year follow-up) that compared the effects of antihypertensive regimens that included ACE inhibitors with the effects of regimens that did not include ACE inhibitors. Hypertension or decreased kidney function was required for entry into all studies. Exclusion criteria common to all studies were acute kidney failure, treatment with immunosuppressive medications, clinically significant congestive heart failure, obstructive uropathy, renal artery stenosis, active systemic disease, type 1 diabetes mellitus, history of transplantation, history of allergy to ACE inhibitors, and pregnancy. The institutional review board at each participating center approved the study, and all patients gave informed consent. Patients were enrolled between March 1986 and April 1996. All patients were randomly assigned to antihypertensive regimens either with or without ACE inhibitors. The ACE inhibitors included captopril, enalapril, cilazapril, benazepril, and ramipril. Concomitant antihypertensive medications were used in both groups to achieve a target blood pressure less than 140/90 mm Hg in all studies. All patients were followed at least once every 3 months for the first year and at least once every 6 months thereafter. Justification for pooling the 11 clinical trials is based on the similarity of study designs and patient characteristics. Justification for pooling placebo-controlled and active-drugcontrolled trials is based on the presence of preexisting hypertension and the use of antihypertensive agents in most patients in the control groups in each clinical trial. Thus, the pooled analysis addresses the clinically relevant question of the relationship of the level of blood pressure and urine protein excretion with the kidney disease progression during antihypertensive therapy, either with or without ACE inhibitors. Definition and Ascertainment of Blood Pressure and Urine Protein Excretion Clinical trial protocols stipulated measurement of blood pressure more frequently than urine protein excretion. In our database, visit was defined as any contact with patients during which study-related information was recorded or clinical variables were measured. Blood pressure was recorded on the same day as the visit in 94% of the visits and within 3 months before the visit in 99% of the visits. Urine protein excretion was recorded on the same day as the visit in 62% of the visits and within 6 months before the visit in 97% of the visits. Blood pressure and urine protein excretion levels at follow-up visits are defined as the current levels. We used current as well as baseline levels as the variables of interest for these analyses because guidelines for blood pressure target current values (3) and our previous studies have demonstrated that the current level of urine protein excretion is a stronger predictor of kidney disease progression than is the baseline level (19). Blood pressure was measured by using a mercury sphygmomanometer in nine studies (8-10, 20-24; Brenner BM. Personal communication) (93% of visits) and calibrated automatic device in two studies (25, 26). Systolic and diastolic blood pressure were measured after 5 to 10 minutes of rest in the supine position in 10 studies (8-10, 20, 22-26; Brenner BM. Personal communication) and in the sitting position in 1 study (21). Urine protein excretion was reported as total urine protein excretion in a 24-hour urine sample in 10 studies (8-10, 20-22, 24-26; Brenner BM. Personal communication) (95% of visits). One study performed a dipstick assessment in an untimed urine sample and reported quantitative measurement only if the dipstick result was positive (23). For that study, all values of dipstick negative were assigned a value of 0.1 g/d. In all studies, results for urine protein excretion of 0.1 g/d or lower were also assigned a value of 0.1 g/d. Values greater than 0.1 g/d were recorded as the exact values reported in the study and rounded to the nearest 0.1 g/d. Outcomes Serum creatinine concentration was recorded on the same day as the visit in 78% of visits and within 3 months after the visit in 96% of the visits. The primary outcome for the pooled analysis was kidney disease progression, defined as a combined end point of a twofold increase (doubling) in serum creatinine concentration from baseline values or development of kidney failure, defined as the initiation of long-term dialysis therapy. Statistical Analyses We used S-Plus 2000 (Insightful Corp., Seattle, Washington) and SAS software, version 8.2 (SAS Institute, Inc., Cary, North Carolina), software programs for statistical analyses. Cox proportional-hazards regression analysis was performed to detect associations between the covariates and outcomes. Baseline patient characteristics were treatment assignment (ACE inhibitor vs. control, using the intention-to-treat principle), age (logarithmic

Angiotensin-Converting Enzyme Inhibitors and Progression of Nondiabetic Renal Disease: A Meta-Analysis of Patient-Level Data

Chronic renal disease is a major public health problem in the United States. According to the 1999 Annual Data Report of the U.S. Renal Data System, more than 357 000 people have end-stage renal disease (ESRD), and the annual cost of treatment with dialysis and renal transplantation exceeds

Estimating GFR Using Serum Cystatin C Alone and in Combination With Serum Creatinine: A Pooled Analysis of 3,418 Individuals With CKD

15.6 billion (1). Patients undergoing dialysis have reduced quality of life, a high morbidity rate, and an annual mortality rate of 20% to 25% (1). Identification of therapies to prevent ESRD is an important public health goal. Angiotensin-converting enzyme (ACE) inhibitors are highly effective in slowing the progression of renal disease due to type 1 diabetes (26), and evidence of their efficacy in type 2 diabetes is growing (712). However, although 14 randomized, controlled trials have been completed (1325; Brenner BM; Toto R. Personal communications), no consensus exists on the use of ACE inhibitors in nondiabetic renal disease (2628). In a previous meta-analysis of 11 randomized, controlled trials, we found that therapy with ACE inhibitors slowed the progression of nondiabetic renal disease (29). Since our meta-analysis was performed on group data rather than individual-patient data, we could not fully assess the relationship between the effect of ACE inhibitors and blood pressure, urinary protein excretion, or other patient characteristics (30). Thus, we could not determine whether an equal reduction in blood pressure or urinary protein excretion by using other antihypertensive agents would be as effective in slowing the progression of renal disease. Nor could we determine whether the baseline blood pressure, urinary protein excretion, or other patient characteristics modified the response to treatment. In the current report, we used pooled analysis of individual-patient data to answer these questions. We reasoned that the large number of patients in the pooled analysis would provide sufficient statistical power to detect relationships between patient characteristics and risk for progression of renal disease and interactions of patient characteristics with treatment effect. In principle, strong and consistent results from analysis of this large database would clarify the effects of ACE inhibitors for treatment of nondiabetic renal disease. Methods Study Design We obtained individual-patient data from nine published (1322) and two unpublished (Brenner BM; Toto R. Personal communications) randomized, controlled trials assessing the effects of ACE inhibitors on renal disease progression in predominantly nondiabetic patients. Search strategies used to identify clinical trials have been described elsewhere and are reviewed in Appendix 2. We included 11 randomized trials on progression of renal disease that compared the effects of antihypertensive regimens including ACE inhibitors to the effects of regimens without ACE inhibitors, with a follow-up of at least 1 year. In these studies, the institutional review board at each participating center approved the study, and all patients gave informed consent. Patients underwent randomization between March 1986 and April 1996. Hypertension or decreased renal function was required for entry into all studies. Exclusion criteria common to all studies were acute renal failure, treatment with immunosuppressive medications, clinically significant congestive heart failure, obstructive uropathy, renal artery stenosis, active systemic disease, insulin-dependent diabetes mellitus, history of transplantation, history of allergy to ACE inhibitors, and pregnancy. Table 1 shows characteristics of the patients in each study. Table 1. Study and Patient Characteristics in the Randomized, Controlled Trials Included in the Pooled Analysis Before randomization, patients already taking an ACE inhibitor were switched to alternative medications for at least 3 weeks. After randomization, the ACE inhibitor groups received enalapril in seven studies (1419; Brenner BM; Toto R. Personal communications) and captopril (13), benazepril (20), cilazapril (18), and ramipril (21, 22) in one study each. The control groups received placebo in five studies (1922; Brenner BM; Toto R. Personal communications), a specified medication in five studies (nifedipine in two studies [13, 17] and atenolol or acebutolol in three studies [15, 16, 18]), and no specified medication in one study (14). Other antihypertensive medications were used in both groups to reach the target blood pressure, which was less than 140/90 mm Hg in all studies. All patients were followed at least once every 6 months for the first year and at least once yearly thereafter. Blood pressure and laboratory variables were measured at each visit. Table 1 shows outcomes of each study. We pooled the 11 clinical trials on the basis of similarity of study designs and patient characteristics. In addition, the presence of preexisting hypertension and use of antihypertensive agents in most patients in the control groups in each clinical trial justified pooling data from placebo-controlled and active-controlled trials. Thus, the pooled analysis addresses the clinically relevant question of whether antihypertensive regimens including ACE inhibitors are more effective than anti-hypertensive regimens not including ACE inhibitors in slowing the progression of nondiabetic renal disease. Outcomes Two primary outcomes were defined: ESRD, defined as the initiation of long-term dialysis therapy, and a combined outcome of a twofold increase in serum creatinine concentration from baseline values or ESRD. Because ESRD is a clinically important outcome, we believed that definitive results of analyses using this outcome would be clinically relevant. However, because most chronic renal diseases progress slowly, few patients might reach this outcome during the relatively brief follow-up of these clinical trials, resulting in relatively low statistical power for these analyses. Doubling of baseline serum creatinine is a well-accepted surrogate outcome for progression of renal disease in studies of antihypertensive agents (2, 20) and would be expected to occur more frequently than ESRD, providing higher statistical power for analyses using this outcome. Doubling of baseline serum creatinine concentration was confirmed by repeated evaluation in only one study, which used this variable as the primary outcome. Therefore, we did not require confirmation of doubling for our analysis. Other outcomes included death and a composite outcome of ESRD and death. Withdrawal was defined as discontinuation of follow-up before the occurrence of an outcome or study end. Reasons for withdrawal were 1) nonfatal side effects possibly due to ACE inhibitors, including hyperkalemia, cough, angioedema, acute renal failure, or hypotension; 2) nonfatal cardiovascular disease events, including myo-cardial infarction, congestive heart failure, stroke, transient ischemic attack, or claudication; 3) other nonfatal events, such as malignant disease, pneumonia, cellulitis, headache, or gastrointestinal disturbance; and 4) other reasons, including loss to follow-up, protocol violation, or unknown. Statistical Analysis Five investigators participated in data cleaning. Summary tables were compiled from the individual-patient data from each study and checked against tables in published and unpublished reports. Discrepancies were resolved by contacting investigators at the clinical or data coordinating centers whenever possible. Because the studies followed different protocols, we had to standardize the variable definitions, follow-up intervals, and run-in periods; details of our approach are provided in Appendix 2. S-Plus (MathSoft, Inc., Seattle, Washington) and SAS (SAS Institute, Inc., Cary, North Carolina) software programs were used for all statistical analyses (31, 32). Univariate analysis was performed to detect associations between the covariates and outcomes. Baseline patient characteristics were treatment assignment (ACE inhibitor vs. control), age (logarithmic transformation), sex, ethnicity, systolic blood pressure, diastolic blood pressure, mean arterial pressure, serum creatinine concentration (reciprocal transformation), and urinary protein excretion. Study characteristics were blinding, type of antihypertensive regimen in the control group, planned duration of follow-up, whether dietary protein or sodium was restricted, and year of publication. Baseline patient characteristics and study characteristics were introduced as fixed covariates. Since renal biopsy was not performed in most cases and since criteria for classification of cause of renal disease were not defined, the cause of renal disease was not included as a variable in the analysis. Follow-up patient characteristics (blood pressure and urinary protein excretion) were adjusted as time-dependent covariates; the value recorded at the beginning of each time segment was used for that segment. This convention was used so that each outcome would be determined only by previous exposure. The intention-to-treat principle was followed for comparison of randomized groups. Cox proportional-hazards regression models were used to determine the effect of assignment to ACE inhibitors (treatment effect) and other covariates on risk for ESRD and the combined outcome (33, 34). Multivariable models were built by using candidate predictors that were associated with the outcome (P<0.2) in the univariate analysis. Each model was adjusted for study, but since some studies had no events, we could not include a dummy variable for each study. Rather, we adjusted models for studies that differed significantly from the rest (studies 2 [14], 5 [15], 10 [20], and 11 [21, 22]). We also performed tests for interactions between all covariates and treatment effect. All P values were based on two-sided tests, and significance was set at a P value less than 0.05. Results are expressed as relative risks with 95% CIs. Residual diagnostics were performed on these final models (33, 34)

The case of the misleading funnel plot

BACKGROUND Serum cystatin C was proposed as a potential replacement for serum creatinine in glomerular filtration rate (GFR) estimation. We report the development and evaluation of GFR-estimating equations using serum cystatin C alone and serum cystatin C, serum creatinine, or both with demographic variables. STUDY DESIGN Test of diagnostic accuracy. SETTING & PARTICIPANTS Participants screened for 3 chronic kidney disease (CKD) studies in the United States (n = 2,980) and a clinical population in Paris, France (n = 438). REFERENCE TEST Measured GFR (mGFR). INDEX TEST Estimated GFR using the 4 new equations based on serum cystatin C alone, serum cystatin C, serum creatinine, or both with age, sex, and race. New equations were developed by using linear regression with log GFR as the outcome in two thirds of data from US studies. Internal validation was performed in the remaining one third of data from US CKD studies; external validation was performed in the Paris study. MEASUREMENTS GFR was measured by using urinary clearance of iodine-125-iothalamate in the US studies and chromium-51-EDTA in the Paris study. Serum cystatin C was measured by using Dade-Behring assay, standardized serum creatinine values were used. RESULTS Mean mGFR, serum creatinine, and serum cystatin C values were 48 mL/min/1.73 m(2) (5th to 95th percentile, 15 to 95), 2.1 mg/dL, and 1.8 mg/L, respectively. For the new equations, coefficients for age, sex, and race were significant in the equation with serum cystatin C, but 2- to 4-fold smaller than in the equation with serum creatinine. Measures of performance in new equations were consistent across the development and internal and external validation data sets. Percentages of estimated GFR within 30% of mGFR for equations based on serum cystatin C alone, serum cystatin C, serum creatinine, or both levels with age, sex, and race were 81%, 83%, 85%, and 89%, respectively. The equation using serum cystatin C level alone yields estimates with small biases in age, sex, and race subgroups, which are improved in equations including these variables. LIMITATIONS Study population composed mainly of patients with CKD. CONCLUSIONS Serum cystatin C level alone provides GFR estimates that are nearly as accurate as serum creatinine level adjusted for age, sex, and race, thus providing an alternative GFR estimate that is not linked to muscle mass. An equation including serum cystatin C level in combination with serum creatinine level, age, sex, and race provides the most accurate estimates.

Vaccination against Lyme Disease with Recombinant Borrelia burgdorferi Outer-Surface Lipoprotein A with Adjuvant

Evidence based medicine insists on rigorous standards to appraise clinical interventions. Failure to apply the same rules to its own tools could be equally damaging

Summing up evidence : one answer is not always enough

BACKGROUND The risk of acquiring Lyme disease is high in areas in which the disease is endemic, and the development of a safe and effective vaccine is therefore important. METHODS We conducted a multicenter, double-blind, randomized trial involving 10,936 subjects who lived in areas of the United States in which Lyme disease is endemic. Participants received an injection of either recombinant Borrelia burgdorferi outer-surface lipoprotein A (OspA) with adjuvant or placebo at enrollment and 1 and 12 months later. In cases of suspected Lyme disease, culture of skin lesions, polymerase-chain-reaction testing, or serologic testing was done. Serologic testing was performed 12 and 20 months after study entry to detect asymptomatic infections. RESULTS In the first year, after two injections, 22 subjects in the vaccine group and 43 in the placebo group contracted definite Lyme disease (P=0.009); vaccine efficacy was 49 percent (95 percent confidence interval, 15 to 69 percent). In the second year, after the third injection, 16 vaccine recipients and 66 placebo recipients contracted definite Lyme disease (P<0.001); vaccine efficacy was 76 percent (95 percent confidence interval, 58 to 86 percent). The efficacy of the vaccine in preventing asymptomatic infection was 83 percent in the first year and 100 percent in the second year. Injection of the vaccine was associated with mild-to-moderate local or systemic reactions lasting a median of three days. CONCLUSIONS Three injections of vaccine prevented most definite cases of Lyme disease or asymptomatic B. burgdorferi infection.

Evaluation of the Modification of Diet in Renal Disease Study Equation in a Large Diverse Population

Are meta-analyses the brave new world, or are the critics of such combined analyses right to say that the biases inherent in clinical trials make them uncombinable? Negative trials are often unreported, and hence can be missed by meta-analysts. And how much heterogeneity between trials is acceptable? A recent major criticism is that large randomised trials do not always agree with a prior meta-analysis. Neither individual trials nor meta-analyses, reporting as they do on population effects, tell how to treat the individual patient. Here we take a more rounded approach to meta-analyses, arguing that their strengths outweigh their weaknesses, although the latter must not be brushed aside.