Oliver Kuss

Methods to estimate the between-study variance and its uncertainty in meta-analysis

Meta‐analyses are typically used to estimate the overall/mean of an outcome of interest. However, inference about between‐study variability, which is typically modelled using a between‐study variance parameter, is usually an additional aim. The DerSimonian and Laird method, currently widely used by default to estimate the between‐study variance, has been long challenged. Our aim is to identify known methods for estimation of the between‐study variance and its corresponding uncertainty, and to summarise the simulation and empirical evidence that compares them. We identified 16 estimators for the between‐study variance, seven methods to calculate confidence intervals, and several comparative studies. Simulation studies suggest that for both dichotomous and continuous data the estimator proposed by Paule and Mandel and for continuous data the restricted maximum likelihood estimator are better alternatives to estimate the between‐study variance. Based on the scenarios and results presented in the published studies, we recommend the Q‐profile method and the alternative approach based on a ‘generalised Cochran between‐study variance statistic’ to compute corresponding confidence intervals around the resulting estimates. Our recommendations are based on a qualitative evaluation of the existing literature and expert consensus. Evidence‐based recommendations require an extensive simulation study where all methods would be compared under the same scenarios.

Statistical methods for meta-analyses including information from studies without any events—add nothing to nothing and succeed nevertheless

Meta-analyses with rare events, especially those that include studies with no event in one (single-zero) or even both (double-zero) treatment arms, are still a statistical challenge. In the case of double-zero studies, researchers in general delete these studies or use continuity corrections to avoid them. A number of arguments against both options has been given, and statistical methods that use the information from double-zero studies without using continuity corrections have been proposed. In this paper, we collect them and compare them by simulation. This simulation study tries to mirror real-life situations as completely as possible by deriving true underlying parameters from empirical data on actually performed meta-analyses. It is shown that for each of the commonly encountered effect estimators valid statistical methods are available that use the information from double-zero studies without using continuity corrections. Interestingly, all of them are truly random effects models, and so also the current standard method for very sparse data as recommended from the Cochrane collaboration, the Yusuf-Peto odds ratio, can be improved on. For actual analysis, we recommend to use beta-binomial regression methods to arrive at summary estimates for the odds ratio, the relative risk, or the risk difference. Methods that ignore information from double-zero studies or use continuity corrections should no longer be used. We illustrate the situation with an example where the original analysis ignores 35 double-zero studies, and a superior analysis discovers a clinically relevant advantage of off-pump surgery in coronary artery bypass grafting.

Ministernotomy versus conventional sternotomy for aortic valve replacement: matched propensity score analysis of 808 patients

OBJECTIVESnThe proportion of minimally invasive approaches is rising in cardiac surgery, in part driven by increasing patient demand. This study aimed to perform a risk-adjusted comparison of mortality, rate of stroke and perioperative morbidity of aortic valve replacement (AVR) conducted through either partial mini-sternotomy or conventional sternotomy.nnnMETHODSnBetween July 2009 and July 2012, data from 984 consecutive patients undergoing isolated AVR were prospectively recorded. In 44.3% (n = 436), the less invasive partial mini-sternotomy was used. Propensity score matching was performed based on 15 preoperative risk factors to correct for selection bias. In-hospital mortality, stroke rate as well as other major complications in the minimally invasive group and conventional sternotomy group were compared in 404 matched patient pairs (total 808).nnnRESULTSnIn-hospital mortality and rate of postoperative intra-aortic balloon pump use were identical for propensity-matched patients, 1.0% (4 in each group). The rate of stroke [OR (95% confidence interval (CI)): 0.80 (0.22-2.98)], perioperative myocardial infarction [OR (95% CI): 2.00 (0.18-22.06)], low-output syndrome [OR (95% CI): 0.90 (0.37-2.22)], new onset of dialysis [OR (95% CI): 1.25 (0.49-3.17)] and re-exploration for bleeding [OR (95% CI): 0.88 (0.50-1.56)] were similar. Likewise, resource utilization (operation time, duration of stay in the intensive care unit and in-hospital stay) and valve selection (type and size) was not affected by the surgical approach either.nnnCONCLUSIONSnAVR can be safely conducted through a partial mini-sternotomy. This approach is not associated with an increased rate of complications. However, wide CIs reflect the still prevailing statistical uncertainty in estimates, not excluding patient-relevant differences between approaches. Large trials, which also address end points, such as postoperative pain, duration of postoperative recovery and quality of life, are needed to clarify the role of minimally invasive AVR.

Cohort profile: the German Diabetes Study (GDS)

BackgroundThe German Diabetes Study (GDS) is a prospective longitudinal cohort study describing the impact of subphenotypes on the course of the disease. GDS aims at identifying prognostic factors and mechanisms underlying the development of related comorbidities.Study design and methodsThe study comprises intensive phenotyping within 12xa0months after clinical diagnosis, at 5-year intervals for 20xa0years and annual telephone interviews in between. Dynamic tests, including glucagon, mixed meal, intravenous glucose tolerance and hyperinsulinemic clamp tests, serve to assess beta-cell function and tissue-specific insulin sensitivity. Magnetic resonance imaging and multinuclei spectroscopy allow quantifying whole-body fat distribution, tissue-specific lipid deposition and energy metabolism. Comprehensive analyses of microvascular (nerve, eye, kidney) and macrovascular (endothelial, cardiorespiratory) morphology and function enable identification and monitoring of comorbidities. The GDS biobank stores specimens from blood, stool, skeletal muscle, subcutaneous adipose tissue and skin for future analyses including multiomics, expression profiles and histology. Repeated questionnaires on socioeconomic conditions, patient-reported outcomes as quality of life, health-related behavior as physical activity and nutritional habits are a specific asset of GDS. This study will recruit 3000 patientsxa0and a group of humans without familiy history of diabetes. 237 type 1 and 456 type 2 diabetes patients have been already included.

Improvement of pain-related self-management for cancer patients through a modular transitional nursing intervention: A cluster-randomized multicenter trial

Summary This trial reveals the positive impact of a nursing intervention to improve patients self‐management of cancer pain and reduce care transition‐related problems. ABSTRACT Patients self‐management skills are affected by their knowledge, activities, and attitudes toward pain management. This trial aimed to test the Self Care Improvement through Oncology Nursing (SCION)‐PAIN program, a multimodular structured intervention to reduce patients barriers to self‐management of cancer pain. Two hundred sixty‐three patients with diagnosed malignancy, pain > 3 days, and average pain ≥ 3/10 participated in a cluster‐randomized trial on 18 wards in 2 German university hospitals. Patients on the intervention wards received, in addition to standard pain treatment, the SCION‐PAIN program consisting of 3 modules: pharmacologic, nonpharmacologic pain management, and discharge management. The intervention was conducted by specially trained cancer nurses and included components of patient education, skills training, and counseling. Starting with admission, patients received booster sessions every third day and one follow‐up telephone counseling session within 2 to 3 days after discharge. Patients in the control group received standard care. Primary end point was the group difference in patient‐related barriers to self‐management of cancer pain (Barriers Questionnaire – BQ II) 7 days after discharge. The SCION‐PAIN program resulted in a significant reduction of patient‐related barriers to pain management 1 week after discharge from the hospital: mean difference on BQ II was −0.49 points (95% confidence interval −0.87 points to −0.12 points; P = 0.02). Furthermore, patients showed improved adherence to pain medication; odds ratio 8.58 (95% confidence interval 1.66–44.40; P = 0.02). A post hoc analysis indicated reduced average and worst pain intensity as well as improved quality of life. This trial reveals the positive impact of a nursing intervention to improve patients self‐management of cancer pain.

Longitudinal Trajectories of Metabolic Control From Childhood to Young Adulthood in Type 1 Diabetes From a Large German/Austrian Registry: A Group-Based Modeling Approach

OBJECTIVE Worsening of glycemic control in type 1 diabetes during puberty is a common observation. However, HbA1c remains stable or even improves for some youths. The aim is to identify distinct patterns of glycemic control in type 1 diabetes from childhood to young adulthood. RESEARCH DESIGN AND METHODS A total of 6,433 patients with type 1 diabetes were selected from the prospective, multicenter diabetes patient registry Diabetes-Patienten-Verlaufsdokumentation (DPV) (follow-up from age 8 to 19 years, baseline diabetes duration ≥2 years, HbA1c aggregated per year of life). We used latent class growth modeling as the trajectory approach to determine distinct subgroups following a similar trajectory for HbA1c over time. RESULTS Five distinct longitudinal trajectories of HbA1c were determined, comprising group 1 = 40%, group 2 = 27%, group 3 = 15%, group 4 = 13%, and group 5 = 5% of patients. Groups 1–3 indicated stable glycemic control at different HbA1c levels. At baseline, similar HbA1c was observed in group 1 and group 4, but HbA1c deteriorated in group 4 from age 8 to 19 years. Similar patterns were present in group 3 and group 5. We observed differences in self-monitoring of blood glucose, insulin therapy, daily insulin dose, physical activity, BMI SD score, body-height SD score, and migration background across all HbA1c trajectories (all P ≤ 0.001). No sex differences were present. Comparing groups with similar initial HbA1c but different patterns, groups with higher HbA1c increase were characterized by lower frequency of self-monitoring of blood glucose and physical activity and reduced height (all P < 0.01). CONCLUSIONS Using a trajectory approach, we determined five distinct longitudinal patterns of glycemic control from childhood to early adulthood. Diabetes self-care, treatment differences, and demographics were related to different HbA1c courses.

A modified Wald interval for the area under the ROC curve (AUC) in diagnostic case-control studies

BackgroundThe area under the receiver operating characteristic (ROC) curve, referred to as the AUC, is an appropriate measure for describing the overall accuracy of a diagnostic test or a biomarker in early phase trials without having to choose a threshold. There are many approaches for estimating the confidence interval for the AUC. However, all are relatively complicated to implement. Furthermore, many approaches perform poorly for large AUC values or small sample sizes.MethodsThe AUC is actually a probability. So we propose a modified Wald interval for a single proportion, which can be calculated on a pocket calculator. We performed a simulation study to compare this modified Wald interval (without and with continuity correction) with other intervals regarding coverage probability and statistical power.ResultsThe main result is that the proposed modified Wald intervals maintain and exploit the type I error much better than the intervals of Agresti-Coull, Wilson, and Clopper-Pearson. The interval suggested by Bamber, the Mann-Whitney interval without transformation and also the interval of the binormal AUC are very liberal. For small sample sizes the Wald interval with continuity has a comparable coverage probability as the LT interval and higher power. For large sample sizes the results of the LT interval and of the Wald interval without continuity correction are comparable.ConclusionsIf individual patient data is not available, but only the estimated AUC and the total sample size, the modified Wald intervals can be recommended as confidence intervals for the AUC. For small sample sizes the continuity correction should be used.

Burden of Mortality Attributable to Diagnosed Diabetes: A Nationwide Analysis Based on Claims Data From 65 Million People in Germany

OBJECTIVE In Germany, as in many other countries, nationwide data on mortality attributable to diagnosed diabetes are not available. This study estimated the absolute number of excess deaths associated with diabetes (all types) and type 2 diabetes in Germany. RESEARCH DESIGN AND METHODS A prevalence approach that included nationwide routine data from 64.9 million people insured in the German statutory health insurance system in 2010 was used for the calculation. Because nationwide data on diabetes mortality are lacking in Germany, the mortality rate ratio from the Danish National Diabetes Register was used. The absolute number of excess deaths associated with diabetes was calculated as the number of deaths due to diabetes minus the number of deaths due to diabetes with a mortality that was as high as in the population without diabetes. Furthermore, the mortality population-attributable fraction was calculated. RESULTS A total of 174,627 excess deaths were due to diabetes in 2010, including 137,950 due to type 2 diabetes. Overall, 21% of all deaths in Germany were attributable to diabetes and 16% were attributable to type 2 diabetes. Most of the excess deaths (34% each) occurred in the 70- to 89-year-old age-group. CONCLUSIONS In this first nationwide calculation of excess deaths related to diabetes in Germany, the results suggest that the official German estimates that rely on information from death certificates are grossly underestimated. Countries without national cohorts or diabetes registries could easily use this method to estimate the number of excess deaths due to diabetes.

Is there evidence of potential overtreatment of glycaemia in elderly people with type 2 diabetes? Data from the GUIDANCE study.

AbstractAimsWe used data from the GUIDANCE Study to determine the care of people with type 2 diabetes according to age and accompanying cardiovascular diseases and to assess indicators of overtreatment of glycaemia.nMethodsThe GUIDANCE study was a retrospective, cross-sectional study from 2009–2010 based on the records of 7597 people in France, Belgium, Italy, the Netherlands, Sweden, UK, Ireland and Germany. We analysed the level of metabolic control achieved and blood glucose-lowering medication used in different age groups and in relation to accompanying diseases.nResults4.459 patients (59.1%) were 65xa0years or older. Their HbA1c levels were similar to those with <65xa0years. 44.7% of patients ≥65xa0years had an HbA1c ≤7% (53xa0mmol/mol) and were treated with insulin or sulfonylureas, and 27.1% of them had ischaemic heart disease or congestive heart failure. Significantly more patients with heart disease had HbA1c values ≤7% (53xa0mmol/mol) and were treated more often with insulin or sulfonylureas compared to patients of the same age without heart disease.nConclusionsMost patients were treated according to guidelines valid at the time this large international patient sample was surveyed. Older and younger patients were at a similar level of metabolic control, and almost half of the patients with an age of ≥65xa0years and treated with insulin or sulfonylurea had HbA1c levels below the target range (≤7%) for younger patients. However, these patients have an increased risk of severe hypoglycaemic events with potentially dangerous complications, particularly in those with cardiovascular diseases.

Anthropometric markers and their association with incident type 2 diabetes mellitus: which marker is best for prediction? Pooled analysis of four German population-based cohort studies and comparison with a nationwide cohort study

Objective To compare the association between different anthropometric measurements and incident type 2 diabetes mellitus (T2DM) and to assess their predictive ability in different regions of Germany. Methods Data of 10u2005258 participants from 4 prospective population-based cohorts were pooled to assess the association of body weight, body mass index (BMI), waist circumference (WC), waist-to-hip-ratio (WHR) and waist-to-height-ratio (WHtR) with incident T2DM by calculating HRs of the crude, adjusted and standardised markers, as well as providing receiver operator characteristic (ROC) curves. Differences between HRs and ROCs for the different anthropometric markers were calculated to compare their predictive ability. In addition, data of 3105 participants from the nationwide survey were analysed separately using the same methods to provide a nationally representative comparison. Results Strong associations were found for each anthropometric marker and incidence of T2DM. Among the standardised anthropometric measures, we found the strongest effect on incident T2DM for WC and WHtR in the pooled sample (HR for 1 SD difference in WC 1.97, 95% CI 1.75 to 2.22, HR for WHtR 1.93, 95% CI 1.71 to 2.17 in women) and in female DEGS participants (HR for WC 2.24, 95% CI 1.91 to 2.63, HR for WHtR 2.10, 95% CI 1.81 to 2.44), whereas the strongest association in men was found for WHR among DEGS participants (HR 2.29, 95% CI 1.89 to 2.78). ROC analysis showed WHtR to be the strongest predictor for incident T2DM. Differences in HR and ROCs between the different markers confirmed WC and WHtR to be the best predictors of incident T2DM. Findings were consistent across study regions and age groups (<65 vs ≥65u2005years). Conclusions We found stronger associations between anthropometric markers that reflect abdominal obesity (ie, WC and WHtR) and incident T2DM than for BMI and weight. The use of these measurements in risk prediction should be encouraged.