Conditional power and friends: The why and how of (un)planned, unblinded sample size recalculations in confirmatory trials.

Adapting the final sample size of a trial to the evidence accruing during the trial is a natural way to address planning uncertainty. Since the sample size is usually determined by an argument based on the power of the trial, an interim analysis raises the question of how the final sample size should be determined conditional on the accrued information. To this end, we first review and compare common approaches to estimating conditional power, which is often used in heuristic sample size recalculation rules. We then discuss the connection of heuristic sample size recalculation and optimal two-stage designs, demonstrating that the latter is the superior approach in a fully preplanned setting. Hence, unplanned design adaptations should only be conducted as reaction to trial-external new evidence, operational needs to violate the originally chosen design, or post hoc changes in the optimality criterion but not as a reaction to trial-internal data. We are able to show that commonly discussed sample size recalculation rules lead to paradoxical adaptations where an initially planned optimal design is not invariant under the adaptation rule even if the planning assumptions do not change. Finally, we propose two alternative ways of reacting to newly emerging trial-external evidence in ways that are consistent with the originally planned design to avoid such inconsistencies.

Optimal planning of adaptive two-stage designs.

Adaptive designs are playing an increasingly important role in the planning of clinical trials. While there exists various research on the optimal determination of a two-stage design, non-optimal versions still are frequently applied in clinical research. In this article, we strive to motivate the application of optimal adaptive designs and give guidance on how to determine them. It is demonstrated that optimizing a trial design with respect to particular objective criteria can have a substantial benefit over the application of conventional adaptive sample size recalculation rules. Furthermore, we show that in many practical situations, optimal group-sequential designs show an almost negligible performance loss compared to optimal adaptive designs. Finally, we illustrate how optimal designs can be tailored to specific operational requirements by customizing the underlying optimization problem.

Risk factors associated with severe hospital burden of COVID-19 disease in Regione Lombardia: a cohort study.

BACKGROUND: Understanding the risk factors associated with hospital burden of COVID-19 is crucial for healthcare planning for any future waves of infection. METHODS: An observational cohort study is performed, using data on all PCR-confirmed cases of COVID-19 in Regione Lombardia, Italy, during the first wave of infection from February-June 2020. A multi-state modelling approach is used to simultaneously estimate risks of progression through hospital to final outcomes of either death or discharge, by pathway (via critical care or not) and the times to final events (lengths of stay). Logistic and time-to-event regressions are used to quantify the association of patient and population characteristics with the risks of hospital outcomes and lengths of stay respectively. RESULTS: Risks of severe outcomes such as ICU admission and mortality have decreased with month of admission (for example, the odds ratio of ICU admission in June vs March is 0.247 [0.120-0.508]) and increased with age (odds ratio of ICU admission in 45-65 vs 65 + age group is 0.286 [0.201-0.406]). Care home residents aged 65 + are associated with increased risk of hospital mortality and decreased risk of ICU admission. Being a healthcare worker appears to have a protective association with mortality risk (odds ratio of ICU mortality is 0.254 [0.143-0.453] relative to non-healthcare workers) and length of stay. Lengths of stay decrease with month of admission for survivors, but do not appear to vary with month for non-survivors. CONCLUSIONS: Improvements in clinical knowledge, treatment, patient and hospital management and public health surveillance, together with the waning of the first wave after the first lockdown, are hypothesised to have contributed to the reduced risks and lengths of stay over time.

Decreasing hospital burden of COVID-19 during the first wave in Regione Lombardia: an emergency measures context.

BACKGROUND: The aim of this study is to quantify the hospital burden of COVID-19 during the first wave and how it changed over calendar time; to interpret the results in light of the emergency measures introduced to manage the strain on secondary healthcare. METHODS: This is a cohort study of hospitalised confirmed cases of COVID-19 admitted from February-June 2020 and followed up till 17th July 2020, analysed using a mixture multi-state model. All hospital patients with confirmed COVID-19 disease in Regione Lombardia were involved, admitted from February-June 2020, with non-missing hospital of admission and non-missing admission date. RESULTS: The cohort consists of 40,550 patients hospitalised during the first wave. These patients had a median age of 69 (interquartile range 56-80) and were more likely to be men (60%) than women (40%). The hospital-fatality risk, averaged over all pathways through hospital, was 27.5% (95% CI 27.1-28.0%); and steadily decreased from 34.6% (32.5-36.6%) in February to 7.6% (6.3-10.6%) in June. Among surviving patients, median length of stay in hospital was 11.8 (11.6-12.3) days, compared to 8.1 (7.8-8.5) days in non-survivors. Averaged over final outcomes, median length of stay in hospital decreased from 21.4 (20.5-22.8) days in February to 5.2 (4.7-5.8) days in June. CONCLUSIONS: The hospital burden, in terms of both risks of poor outcomes and lengths of stay in hospital, has been demonstrated to have decreased over the months of the first wave, perhaps reflecting improved treatment and management of COVID-19 cases, as well as reduced burden as the first wave waned. The quantified burden allows for planning of hospital beds needed for current and future waves of SARS-CoV-2 i.

Physostigmine for prevention of postoperative delirium and long-term cognitive dysfunction in liver surgery: A double-blinded randomised controlled trial.

BACKGROUND: Anecdotally, cholinergic stimulation has been used to treat delirium and reduce cognitive dysfunction. OBJECTIVE: The aim of this investigation was to evaluate whether physostigmine reduced the incidence of postoperative delirium (POD) and postoperative cognitive dysfunction (POCD) in patients undergoing liver resection. DESIGN: This was a double-blind, randomised, placebo-controlled trial. Between 11 August 2009 and 3 March 2016, patients were recruited at the Charité - Universitätsmedizin Berlin in Germany. Follow-ups took place at 1 week (T1), 90 days (T2) and 365 days (T3) after surgery. SETTING: This single-centre study was conducted at an academic medical centre. PARTICIPANTS: In total, 261 participants aged at least 18 years scheduled for elective liver surgery were randomised. The protocol also included 45 non-surgical matched controls to provide normative data for POCD and neurocognitive deficit (NCD). INTERVENTION: Participants were allocated to receive either intravenous physostigmine, as a bolus of 0.02 mg kg-1 body weight followed by 0.01 mg kg-1 body weight per hour (n = 130), or placebo (n = 131), for 24 h after induction of anaesthesia. MAIN OUTCOMES AND MEASURES: Primary outcomes were POD, assessed using the Diagnostic and Statistical Manual of Mental Disorders (DSM-4-TR) twice daily up to day 7 after surgery, and POCD assessed via the CANTAB neuropsychological test battery, and two paper pencil tests on the day before surgery, and on postoperative days 7, 90 and 365. RESULTS: In total, 261 patients were randomised, 130 to the physostigmine and 131 to the placebo group. The incidence of POD did not differ significantly between the physostigmine and placebo groups (20 versus 15%; P = 0.334). Preoperative cognitive impairment and POCD frequencies did not differ significantly between the physostigmine and placebo groups at any time. Lower mortality rates were found in the physostigmine group compared with placebo at 3 months [2% (95% confidence interval (CI), 0 to 4) versus 11% (95% CI, 6 to 16), P = 0.002], and 6 months [7% (95% CI, 3 to 12) versus 16% (95% CI, 10 to 23), P = 0.012] after surgery. CONCLUSION: Physostigmine had no effect on POD and POCD when applied after induction of anaesthesia up to 24 h. TRIAL REGISTRATION: DOI 10.1186/ISRCTN18978802, EudraCT 2008-007237-47, Ethics approval ZS EK 11 618/08 (15 January 2009).

Understanding the relationship between cognitive performance and function in daily life after traumatic brain injury.

OBJECTIVE: Cognitive impairment is a key cause of disability after traumatic brain injury (TBI) but relationships with overall functioning in daily life are often modest. The aim is to examine cognition at different levels of function and identify domains associated with disability. METHODS: 1554 patients with mild-to-severe TBI were assessed at 6 months post injury on the Glasgow Outcome Scale-Extended (GOSE), the Short Form-12v2 and a battery of cognitive tests. Outcomes across GOSE categories were compared using analysis of covariance adjusting for age, sex and education. RESULTS: Overall effect sizes were small to medium, and greatest for tests involving processing speed (ηp 2 0.057-0.067) and learning and memory (ηp 2 0.048-0.052). Deficits in cognitive performance were particularly evident in patients who were dependent (GOSE 3 or 4) or who were unable to participate in one or more major life activities (GOSE 5). At higher levels of function (GOSE 6-8), cognitive performance was surprisingly similar across categories. There were decreases in performance even in patients reporting complete recovery without significant symptoms. Medium to large effect sizes were present for summary measures of cognition (ηp 2 0.111), mental health (ηp 2 0.131) and physical health (ηp 2 0.252). CONCLUSIONS: This large-scale study provides novel insights into cognitive performance at different levels of disability and highlights the importance of processing speed in function in daily life. At upper levels of outcome, any influence of cognition on overall function is markedly attenuated and differences in mental health are salient.

Optimal adaptive single-arm phase II trials under quantified uncertainty.

Single-arm trials with binary endpoint are firmly established in e.g., early clinical oncology. Here, two-stage designs are often employed to allow early termination of the trial in the case of an unexpectedly large or small response rate to the new treatment. Various designs have been proposed over the last few years which usually require strong assumptions about the true response rate during planning. Often, these designs are not robust to deviations from the planning assumptions. In this paper, we define a Bayesian framework for scoring two-stage designs under uncertainty and investigate the characteristics of designs optimizing a commonly employed performance score of Liu et al. The resulting optimal designs are compared with an alternative, utility-based approach incorporating expected power and sample size. We provide insights in the underlying implicit assumptions of using expected power for scoring adaptive designs and relate the global score function to the practice of sample size recalculation based on conditional power. An in-depth comparison of the features of the different performance scores and their respective optimizing designs provides the guidance for practitioners who face the problem of choosing between the various options. A software implementation of the proposed methods is publicly available online.

A variational approach to optimal two-stage designs.

Recalculating the sample size in adaptive two-stage designs is a well-established method to gain flexibility in a clinical trial. Jennison and Turnbull (2015) proposed an "optimal" adaptive two-stage design based on the inverse normal combination test, which minimizes a mixed criterion of expected sample size under the alternative and conditional power. We demonstrate that the use of a combination test is not necessary to control the type one error rate and use variational techniques to develop a general adaptive design that is globally optimal under predefined optimality criteria. This approach yields to more efficient designs and furthermore allows to investigate the efficiency of the inverse normal method and the relation between local (interim-based) recalculation rules and global (unconditional) optimality of adaptive two-stage designs.

Lasting s-ketamine block of spreading depolarizations in subarachnoid hemorrhage: a retrospective cohort study.

OBJECTIVE: Spreading depolarizations (SD) are characterized by breakdown of transmembrane ion gradients and excitotoxicity. Experimentally, N-methyl-D-aspartate receptor (NMDAR) antagonists block a majority of SDs. In many hospitals, the NMDAR antagonist s-ketamine and the GABAA agonist midazolam represent the current second-line combination treatment to sedate patients with devastating cerebral injuries. A pressing clinical question is whether this option should become first-line in sedation-requiring individuals in whom SDs are detected, yet the s-ketamine dose necessary to adequately inhibit SDs is unknown. Moreover, use-dependent tolerance could be a problem for SD inhibition in the clinic. METHODS: We performed a retrospective cohort study of 66 patients with aneurysmal subarachnoid hemorrhage (aSAH) from a prospectively collected database. Thirty-three of 66 patients received s-ketamine during electrocorticographic neuromonitoring of SDs in neurointensive care. The decision to give s-ketamine was dependent on the need for stronger sedation, so it was expected that patients receiving s-ketamine would have a worse clinical outcome. RESULTS: S-ketamine application started 4.2 ± 3.5 days after aSAH. The mean dose was 2.8 ± 1.4 mg/kg body weight (BW)/h and thus higher than the dose recommended for sedation. First, patients were divided according to whether they received s-ketamine at any time or not. No significant difference in SD counts was found between groups (negative binomial model using the SD count per patient as outcome variable, p = 0.288). This most likely resulted from the fact that 368 SDs had already occurred in the s-ketamine group before s-ketamine was given. However, in patients receiving s-ketamine, we found a significant decrease in SD incidence when s-ketamine was started (Poisson model with a random intercept for patient, coefficient - 1.83 (95% confidence intervals - 2.17; - 1.50), p < 0.001; logistic regression model, odds ratio (OR) 0.13 (0.08; 0.19), p < 0.001). Thereafter, data was further divided into low-dose (0.1-2.0 mg/kg BW/h) and high-dose (2.1-7.0 mg/kg/h) segments. High-dose s-ketamine resulted in further significant decrease in SD incidence (Poisson model, - 1.10 (- 1.71; - 0.49), p < 0.001; logistic regression model, OR 0.33 (0.17; 0.63), p < 0.001). There was little evidence of SD tolerance to long-term s-ketamine sedation through 5 days. CONCLUSIONS: These results provide a foundation for a multicenter, neuromonitoring-guided, proof-of-concept trial of ketamine and midazolam as a first-line sedative regime.

Risk Factors for Early Dialysis Dependency in Autosomal Recessive Polycystic Kidney Disease.

OBJECTIVE: To identify prenatal, perinatal, and postnatal risk factors for dialysis within the first year of life in children with autosomal recessive polycystic kidney disease (ARPKD) as a basis for parental counseling after prenatal and perinatal diagnosis. STUDY DESIGN: A dataset comprising 385 patients from the ARegPKD international registry study was analyzed for potential risk markers for dialysis during the first year of life. RESULTS: Thirty-six out of 385 children (9.4%) commenced dialysis in the first year of life. According to multivariable Cox regression analysis, the presence of oligohydramnios or anhydramnios, prenatal kidney enlargement, a low Apgar score, and the need for postnatal breathing support were independently associated with an increased hazard ratio for requiring dialysis within the first year of life. The increased risk associated with Apgar score and perinatal assisted breathing was time-dependent and vanished after 5 and 8 months of life, respectively. The predicted probabilities for early dialysis varied from 1.5% (95% CI, 0.5%-4.1%) for patients with ARPKD with no prenatal sonographic abnormalities to 32.3% (95% CI, 22.2%-44.5%) in cases of documented oligohydramnios or anhydramnios, renal cysts, and enlarged kidneys. CONCLUSIONS: This study, which identified risk factors associated with onset of dialysis in ARPKD in the first year of life, may be helpful in prenatal parental counseling in cases of suspected ARPKD.

A weighted combined effect measure for the analysis of a composite time-to-first-event endpoint with components of different clinical relevance.

Composite endpoints combine several events within a single variable, which increases the number of expected events and is thereby meant to increase the power. However, the interpretation of results can be difficult as the observed effect for the composite does not necessarily reflect the effects for the components, which may be of different magnitude or even point in adverse directions. Moreover, in clinical applications, the event types are often of different clinical relevance, which also complicates the interpretation of the composite effect. The common effect measure for composite endpoints is the all-cause hazard ratio, which gives equal weight to all events irrespective of their type and clinical relevance. Thereby, the all-cause hazard within each group is given by the sum of the cause-specific hazards corresponding to the individual components. A natural extension of the standard all-cause hazard ratio can be defined by a "weighted all-cause hazard ratio" where the individual hazards for each component are multiplied with predefined relevance weighting factors. For the special case of equal weights across the components, the weighted all-cause hazard ratio then corresponds to the standard all-cause hazard ratio. To identify the cause-specific hazard of the individual components, any parametric survival model might be applied. The new weighted effect measure can be tested for deviations from the null hypothesis by means of a permutation test. In this work, we systematically compare the new weighted approach to the standard all-cause hazard ratio by theoretical considerations, Monte-Carlo simulations, and by means of a real clinical trial example.

Assessing abrasion of orthodontic surface sealants using a modified ophthalmic optical coherence tomography device.

OBJECTIVE: Optical coherence tomography (OCT) is a clinical standard in ophthalmology. Currently, its application in dentistry is gaining increasing interest. In this study, we tested the possibility to use a modified commercially available spectral domain OCT (SD-OCT) to assess the layer thickness of orthodontic surface sealants. MATERIALS AND METHODS: Reference samples of surface sealants for calibration and repeatability testing were measured using a micrometer screw. SD-OCT measurements were compared with micro-CT and light microscopic analyses. After validating the calibration of the SD-OCT, surface sealant layer thickness after aging (thermo cycling) and simulation of professional tooth cleaning (PTC) was assessed using the SD-OCT on 45 extracted teeth assigned to three test groups (n = 15 each): Light Bond™ Sealant, Pro Seal®, and Opal® Seal. RESULTS: SD-OCT showed excellent repeatability and accuracy for measurements of surface sealant layer thickness. Compared with micro-CT, SD-OCT showed better accordance with the reference measurements. The analysis of surface sealants after thermo cycling and PTC revealed poor resistance of Light Bond after only aging and demonstrated substantial wear of all sealants after aging and PTC. CONCLUSION: Imaging using commercially available ophthalmic SD-OCT might represent a suitable non-invasive methodology for longitudinal assessments of surface sealant layer thickness in vitro and in vivo. CLINICAL RELEVANCE: SD-OCT might be a suitable non-invasive method for longitudinal assessments of surface sealant durability in clinical trials.

Test-compatible confidence intervals for adaptive two-stage single-arm designs with binary endpoint.

Inference after two-stage single-arm designs with binary endpoint is challenging due to the nonunique ordering of the sampling space in multistage designs. We illustrate the problem of specifying test-compatible confidence intervals for designs with nonconstant second-stage sample size and present two approaches that guarantee confidence intervals consistent with the test decision. Firstly, we extend the well-known Clopper-Pearson approach of inverting a family of two-sided hypothesis tests from the group-sequential case to designs with fully adaptive sample size. Test compatibility is achieved by using a sample space ordering that is derived from a test-compatible estimator. The resulting confidence intervals tend to be conservative but assure the nominal coverage probability. In order to assess the possibility of further improving these confidence intervals, we pursue a direct optimization approach minimizing the mean width of the confidence intervals. While the latter approach produces more stable coverage probabilities, it is also slightly anti-conservative and yields only negligible improvements in mean width. We conclude that the Clopper-Pearson-type confidence intervals based on a test-compatible estimator are the best choice if the nominal coverage probability is not to be undershot and compatibility of test decision and confidence interval is to be preserved.

Metabolic acidosis is common and associates with disease progression in children with chronic kidney disease.

Recent studies in adult chronic kidney disease (CKD) suggest that metabolic acidosis is associated with faster decline in estimated glomerular filtration rate (eGFR). Alkali therapies improve the course of kidney disease. Here we investigated the prevalence and determinants of abnormal serum bicarbonate values and whether metabolic acidosis may be deleterious to children with CKD. Associations between follow-up serum bicarbonate levels categorized as under 18, 18 to under 22, and 22 or more mmol/l and CKD outcomes in 704 children in the Cardiovascular Comorbidity in Children with CKD Study, a prospective cohort of pediatric patients with CKD stages 3-5, were studied. The eGFR and serum bicarbonate were measured every six months. At baseline, the median eGFR was 27 ml/min/1.73m2 and median serum bicarbonate level 21 mmol/l. During a median follow-up of 3.3 years, the prevalence of metabolic acidosis (serum bicarbonate under 22 mmol/l) was 43%, 60%, and 45% in CKD stages 3, 4, and 5, respectively. In multivariable analysis, the presence of metabolic acidosis as a time-varying covariate was significantly associated with log serum parathyroid hormone through the entire follow-up, but no association with longitudinal growth was found. A total of 211 patients reached the composite endpoint (ESRD or 50% decline in eGFR). In a multivariable Cox model, children with time-varying serum bicarbonate under 18 mmol/l had a significantly higher risk of CKD progression compared to those with a serum bicarbonate of 22 or more mmol/l (adjusted hazard ratio 2.44; 95% confidence interval 1.43-4.15). Thus, metabolic acidosis is a common complication in pediatric patients with CKD and may be a risk factor for secondary hyperparathyroidism and kidney disease progression.

Point estimation in adaptive enrichment designs.

Adaptive enrichment designs are an attractive option for clinical trials that aim at demonstrating efficacy of therapies, which may show different benefit for the full patient population and a prespecified subgroup. In these designs, based on interim data, either the subgroup or the full population is selected for further exploration. When selection is based on efficacy data, this introduces bias to the commonly used maximum likelihood estimator. For the situation of two-stage designs with a single prespecified subgroup, we present six alternative estimators and investigate their performance in a simulation study. The most consistent reduction of bias over the range of scenarios considered was achieved by a method combining the uniformly minimum variance conditionally unbiased estimator with a conditional moment estimator. Application of the methods is illustrated by a clinical trial example.

Cryoballoon vs. open irrigated radiofrequency ablation for paroxysmal atrial fibrillation: long-term FreezeAF outcomes.

BACKGROUND: Effective treatment of paroxysmal atrial fibrillation (AF) is essential for reducing the risk of stroke and heart failure. Cryoballoon (CB) ablation has been developed as an alternative to the use of radiofrequency (RF) energy for electrical isolation of the pulmonary veins. Herein, we provide long-term data regarding the efficacy of CB ablation in comparison to RF. METHODS: FreezeAF was a randomised non-inferiority study comparing CB ablation with RF ablation for the treatment of patients with drug-refractory paroxysmal AF. Procedural success for the long-term follow-up (30 months) was defined as freedom from AF with an absence of persistent complications. RESULTS: Of the 315 patients that were randomised and received catheter ablation, 292 (92.7%) completed the 30-month follow-up (147 in the RF group and 145 in the CB group). The baseline characteristics of the RF and CB groups were similar. Single-procedure success was achieved by 40% of patients in the RF group and 42% of the CB group (p < 0.001 for non-inferiority). When including re-do procedures in the analysis, the multiple procedure success rate was 72% in the RF group and 76% in the CB group. CONCLUSION: The data provide long-term evidence that CB ablation is non-inferior to RF ablation, with high proportions of patients reporting freedom from AF 30 months after the index procedure. TRIAL REGISTRATION: ClinicalTrials.gov Identifier: NCT00774566 ; first registered October 16, 2008; first patient included October 20, 2008.

Cryoballoon Versus Open Irrigated Radiofrequency Ablation in Patients With Paroxysmal Atrial Fibrillation: The Prospective, Randomized, Controlled, Noninferiority FreezeAF Study.

BACKGROUND: There is a lack of data on the comparative efficacy and procedural safety of open irrigated radiofrequency (RF) and cryoballoon catheter (CB) ablation for pulmonary vein isolation in patients with paroxysmal atrial fibrillation. METHODS AND RESULTS: In a prospective, noninferiority study, 315 patients were randomly assigned to RF (n=159) or CB (n=156) ablation. The primary end point was freedom from atrial arrhythmia with absence of persistent complications. Patients were largely comparable between groups with more vascular disease in the RF group (8.2% versus 2.6% for CB; P=0.028). The primary end point at 12 months was achieved by 70.7% with RF and 73.6% with CB (multiple procedure success), including 31 redo procedures in each group (19.5% of RF versus 19.9% of CB; P=0.933). For the intention-to-treat population, noninferiority of CB was revealed for the predefined inferiority margin (risk difference, 0.029; 95% confidence interval, -0.074 to 0.132; P<0.001). Rates at 6 months were 63.1% and 64.1% for the RF and CB groups (single procedure success), and noninferiority was confirmed (risk difference, 0.010; 95% confidence interval, -0.097 to 0.116; P=0.002). Periprocedural complications for the index procedure were more frequent in the CB group (5.0% RF, 12.2% CB; P=0.022) with a significant difference in phrenic nerve palsies (0% RF, 5.8% CB; P=0.002). CONCLUSION: This large, prospective, randomized, controlled study demonstrates noninferiority of CB ablation versus RF ablation for treating patients with paroxysmal atrial fibrillation. CLINICAL TRIAL REGISTRATION: URL: http://www.clinicaltrials.gov. Unique identifier: NCT00774566.