Digitizing clinical trials.

Clinical trials are a fundamental tool used to evaluate the efficacy and safety of new drugs and medical devices and other health system interventions. The traditional clinical trials system acts as a quality funnel for the development and implementation of new drugs, devices and health system interventions. The concept of a "digital clinical trial" involves leveraging digital technology to improve participant access, engagement, trial-related measurements, and/or interventions, enable concealed randomized intervention allocation, and has the potential to transform clinical trials and to lower their cost. In April 2019, the US National Institutes of Health (NIH) and the National Science Foundation (NSF) held a workshop bringing together experts in clinical trials, digital technology, and digital analytics to discuss strategies to implement the use of digital technologies in clinical trials while considering potential challenges. This position paper builds on this workshop to describe the current state of the art for digital clinical trials including (1) defining and outlining the composition and elements of digital trials; (2) describing recruitment and retention using digital technology; (3) outlining data collection elements including mobile health, wearable technologies, application programming interfaces (APIs), digital transmission of data, and consideration of regulatory oversight and guidance for data security, privacy, and remotely provided informed consent; (4) elucidating digital analytics and data science approaches leveraging artificial intelligence and machine learning algorithms; and (5) setting future priorities and strategies that should be addressed to successfully harness digital methods and the myriad benefits of such technologies for clinical research.

Physical activity as a determinant of fasting and 2-h post-challenge glucose: a prospective cohort analysis of the NAVIGATOR trial.

AIM: To investigate whether previous physical activity levels are associated with blood glucose levels in individuals with impaired glucose tolerance in the context of an international pharmaceutical trial. METHODS: Data were analysed from the NAVIGATOR trial, which involved 9306 individuals with impaired glucose tolerance and high cardiovascular risk from 40 different countries, recruited in the period 2002-2004. Fasting glucose, 2-h post-challenge glucose and physical activity (pedometer) were assessed annually. A longitudinal regression analysis was used to determine whether physical activity levels 2 years (t-2 ) and 1 year (t-1 ) previously were associated with levels of glucose, after adjusting for previous glucose levels and other patient characteristics. Those participants with four consecutive annual measures of glucose and two consecutive measures of physical activity were included in the analysis. RESULTS: The analysis included 3964 individuals. Change in physical activity from t-2 to t-1 and activity levels at t-2 were both associated with 2-h glucose levels after adjustment for previous glucose levels and baseline characteristics; however, the associations were weak: a 100% increase in physical activity was associated with a 0.9% reduction in 2-h glucose levels. In addition, previous physical activity only explained an additional 0.05% of the variance in 2-h glucose over the variance explained by the history of 2-h glucose alone (R(2)  = 0.3473 vs. 0.3468). There was no association with fasting glucose. CONCLUSIONS: In the context of a large international clinical trial, previous physical activity levels did not meaningfully influence glucose levels in those with a high risk of chronic disease, after taking into account participants' previous trajectory of glucose control.

Regional, age and sex differences in baseline characteristics of patients enrolled in the Trial Evaluating Cardiovascular Outcomes with Sitagliptin (TECOS).

AIMS: To report baseline characteristics and cardiovascular (CV) risk management by region, age, sex and CV event type for 14 724 participants in the Trial Evaluating Cardiovascular Outcomes with Sitagliptin (TECOS), a randomized, double-blind, placebo-controlled trial exploring whether sitagliptin added to usual type 2 diabetes (T2DM) care affects time to first event in the composite endpoint of CV death, non-fatal myocardial infarction (MI), non-fatal stroke or unstable angina hospitalization. METHODS: TECOS enrolled patients aged ≥50 years, with T2DM and CV disease from 38 countries in five regions: North America, Eastern Europe, Western Europe, Asia Pacific and Latin America. Participants had a glycated haemoglobin concentration of 6.5-8.0% (48-64 mmol/mol) and were receiving oral and/or insulin-based antihyperglycaemic therapy. Analysis of variance or logistic regression was used to compare regional CV risk factors and treatments, referenced to North America. RESULTS: Patients had a mean [1 standard deviation (SD)] age of 66 (8) years, a median (interquartile range) diabetes duration of 9.4 (4.9, 15.3) years, and a mean (SD) body mass index 30.2 (5.7) kg/m² . Compared with North America, blood pressure and lipids were higher in all regions. Statin use was lowest in Latin America (68%) and Eastern Europe (70%) and aspirin use was lower compared with North America in all regions except Asia Pacific. Achievement of treatment targets did not differ by age group or insulin usage, but men and participants with previous MI were more likely than women or those with previous stroke or peripheral arterial disease to reach most treatment goals. CONCLUSION: The CV risk factors of participants in TECOS are reasonably controlled, but differences in CV risk management according to region, sex and history of disease exist. This diversity will enhance the generalizability of the trial results.

Change in levels of physical activity after diagnosis of type 2 diabetes: an observational analysis from the NAVIGATOR study.

Increased physical activity is known to be beneficial in people with type 2 diabetes mellitus (T2DM), but it is not known whether individuals change their activity levels after T2DM diagnosis. The present Nateglinide and Valsartan in Impaired Glucose Tolerance Outcomes Research (NAVIGATOR) trial, conducted in participants with impaired glucose tolerance at high cardiovascular risk, assessed ambulatory activity annually using research-grade pedometers. Oral glucose tolerance tests were performed annually and repeated to confirm T2DM diagnosis. This observational analysis used general linear models to compare step counts before and after T2DM diagnosis in the 2816 participants with the requisite data. Participants were relatively inactive at baseline, taking a median (interquartile range) of 5488 (3258-8361) steps/day, which decreased after T2DM diagnosis by a mean (s.e.) of 258 (64) steps/day (p < 0.0001); however, after adjusting for background trend for activity, step count after T2DM diagnosis was unchanged [mean (s.e.) of 103 (87) fewer steps/day; p = 0.23]. Awareness of T2DM diagnosis had no impact on the trajectory of activity established before the diagnosis.

Population pharmacokinetics and pharmacodynamics of rivaroxaban in patients with non-valvular atrial fibrillation: results from ROCKET AF.

Two once-daily rivaroxaban dosing regimens were compared with warfarin for stroke prevention in patients with non-valvular atrial fibrillation in ROCKET AF: 20 mg for patients with normal/mildly impaired renal function and 15 mg for patients with moderate renal impairment. Rivaroxaban population pharmacokinetic (PK)/pharmacodynamic (PD) modeling data from ROCKET AF patients (n = 161) are reported and are used to confirm established rivaroxaban PK and PK/PD models and to re-estimate values of the models' parameters for the current AF population. An oral one-compartment model with first-order absorption adequately described rivaroxaban PK. Age, renal function, and lean body mass influenced the PK model. Prothrombin time and prothrombinase-induced clotting time exhibited a near-linear relationship with rivaroxaban plasma concentration; inhibitory effects were observed through to 24 hours post-dose. Rivaroxaban plasma concentration and factor Xa activity had an inhibitory maximum-effect (Emax ) relationship. Renal function (on prothrombin time; prothrombinase-induced clotting time) and age (on factor Xa activity) had moderate effects on PK/PD models. PK and PK/PD models were shown to be adequate for describing the current dataset. These findings confirm the modeling and empirical results that led to the selection of doses tested against warfarin in ROCKET AF.

Integrated efficacy to effectiveness trials.

Experts in clinical research, therapeutic development, and comparative effectiveness are continually frustrated in their attempts to fit the square peg of therapeutic development into the round hole of clinical trials. Trials can be optimized to provide signals in highly controlled experiments or to estimate an intervention's effect in poorly controlled real-world settings, but not both simultaneously. Selker and colleagues propose a continuum that creates a smooth transition from controlled experiments to real-world, real-time studies within a single mechanism.

Impediments to clinical research in the United States.

Clinical trials are essential to the evaluation of promising scientific discoveries, but they are becoming unsustainably burdensome, threatening to deprive patients and health-care providers of new therapies and new evidence to guide the use of existing treatments. Regulations are often blamed for impeding clinical research, but there are other elements of the clinical trials enterprise that also have the potential to add burdens, through either imposed requirements or incentives that do not favor clinical research (Figure 1).

Departures from the protocol during conduct of a clinical trial: a pattern from the data record consistent with a learning curve.

OBJECTIVE: Recognition of learning curves in medical skill acquisition has enhanced patient safety through improved training techniques. Clinical trials research has not been similarly scrutinised. The VALsartan In Acute myocardial iNfarcTion, a large multinational, pragmatic, randomised, double-blind, multicentre trial, was retrospectively evaluated for evidence of research conduct consistent with a performance "learning curve". DESIGN: Records provided protocol departure (deviations/violations) and documentation query data. For each site, analysis included patient order (eg, first, second), recruitment rate and first enrollment relative to study start date. SETTING: Computerised data from a trial coordinated by an academic research organisation collaborating with 10 academic and 2 commercial research organisations and an industry sponsor. Interventions 931 sites enrolled 14,703 patients. Departures were restricted to the first year. Exclusions included patient's death or loss to follow-up within 12 months and subjects enrolled 80th or higher at a site. Departures were assessed for variance with higher patient rank, more frequent recruitment and later start date. METHODS AND RESULTS: 12,367 patients at 931 sites were analysed. Departures were more common for patients enrolled earlier at a site (p<0.0001). For example, compared with the 30th patient, the first had 47% more departures. Departures were also more common with slower enrollment and site start closer to the trial start date (p<0.0001). Similar patterns existed for queries. CONCLUSIONS: Research performance improved during the VALsartan In Acute myocardial iNfarcTion consistent with a "learning curve". Although effects were not related to a change in outcome (mortality), learning curves in clinical research may have important safety, ethical, research quality and economic implications for trial conduct.

Cough in pediatric patients receiving angiotensin-converting enzyme inhibitor therapy or angiotensin receptor blocker therapy in randomized controlled trials.

The incidence of cough in children receiving antihypertension therapy with angiotensin-converting enzyme inhibitors (ACEis) or angiotensin receptor blockers (ARBs) is unknown. We analyzed patient-level data from eight randomized trials for the treatment of pediatric hypertension, six of them involving ACEis and two involving ARBs. The incidence of cough in children receiving ACEis (reported cough, 3.2%) was similar to that in children receiving ARBs (reported cough, 1.8%) (P = 0.34). Reports of cough were lower in children than in adults.

Metabolic syndrome and alanine aminotransferase: a global perspective from the NAVIGATOR screening population.

AIMS: Non-alcoholic fatty liver disease (NAFLD) is associated with features of the metabolic syndrome (MetS) and may be an expression of the syndrome within the liver. Using screening data from the Nateglinide And Valsartan in Impaired Glucose Tolerance Outcomes Research (NAVIGATOR) study (n = 42 149), we examined whether alanine aminotransferase (ALT), a biomarker for NAFLD, clustered with features of MetS and whether the clusters differed across global geographic regions. METHODS: Exploratory factor analysis using principle components analysis was applied to data drawn from the NAVIGATOR screening population (n = 41 111). Demographic data, anthropomorphic measurements and blood pressure (BP) collected during the screening visit, as well as blood samples analysed for ALT, total cholesterol, triglycerides, high-density lipoprotein, low-density lipoprotein, and fasting and 2-h glucose measures after an oral glucose tolerance test were used for our analysis. RESULTS: Two factors, interpreted as lipid (Factor 1), and BP/obesity (Factor 2) were identified, explaining approximately 50% of the variance in the overall population. Similar patterns of aggregation were reproducible across all geographic regions except Asia, where fasting glucose loaded more consistently on Factor 1. ALT loaded with mean arterial pressure, fasting glucose and waist circumference except in Asia, where it loaded only with mean arterial pressure and waist circumference. CONCLUSIONS: ALT aggregated with components of MetS, and the pattern of aggregation of ALT with other features of MetS was similar across regions except Asia, possibly indicating a different pathophysiology for NAFLD in Asia. Predictive models of NAFLD may need to be adjusted for regional and ethnic differences.

Racial differences in blood pressure response to angiotensin-converting enzyme inhibitors in children: a meta-analysis.

Angiotensin-converting enzyme (ACE) inhibitors are frequently used to treat hypertension in children.(1) ACE inhibitors alter the balance between the vasoconstrictive, salt-retentive, and cardiac hypertrophic properties of angiotensin II and the vasodilatory and natriuretic properties of bradykinin; they also alter the metabolism of other vasoactive substances.(2) Through these mechanisms, ACE inhibitors decrease systemic vascular resistance and promote natriuresis without increasing heart rate. This study evaluated the results of six trials of ACE inhibitors in children, using meta-analytic techniques to estimate the effect of race on blood pressure response.

Role of the centers for education and research on therapeutics (CERTs) in pharmacovigilance and proper use of therapeutics.

According to the Institute of Medicine (IOM) Committee on the Assessment of the U.S. Drug Safety System, "The recent highly publicized controversies surrounding the safety of some drugs have contributed to a public perception that the drug safety system is in crisis. It seems fair to say that this perception has created an opportunity for a thorough evaluation of the U.S. drug safety system." The evaluation was focused on the U.S. Food and Drug Administration (FDA). To improve the FDA and its function in the public health system to improve therapeutics, it is critical to understand the contributions of other components.

Creatine kinase-MB elevation after percutaneous coronary intervention predicts adverse outcomes in patients with acute coronary syndromes.

AIM: To study the relationship between outcomes and peak creatine kinase (CK)-MB levels after percutaneous coronary intervention (PCI) in patients with non-ST-segment elevation acute coronary syndromes (NSTE ACS). METHODS AND RESULTS: Peak CK-MB ratios (peak CK-MB level/upper limit of normal [ULN]) after PCI were analysed in 6164 patients with NSTE ACS from four randomized trials who underwent in-hospital PCI. We excluded 696 patients with elevated CK or CK-MB levels <24h before PCI; the primary analysis included 2384 of the remaining 5468 patients (43.6%) with CK-MB levels measured <==24h after PCI. The incidence of in-hospital heart failure (0.1%, 0.8%, 3.4%, 4.1%, and 6.1%; P<0.001), arrhythmias (0.8%, 1.9%, 6.9%, 4.1%, and 7.9%; P<0.001), cardiogenic shock (0.1%, 1.3%, 2.0%, 2.3%, and 2.6%; P=0.004), and mortality through 6 months (2.1%, 2.4%, 4.9%, 4.1%, and 5.7%, P=0.005) was increased with peak CK-MB ratios of 0-1, 1-3, 3-5, 5-10, and >10xULN, respectively. The continuous peak CK-MB ratio after PCI significantly predicted adjusted 6-month mortality (risk ratio, 1.06 per unit increase above ULN; 95% confidence interval, 1.01-1.11; P=0.017). CONCLUSIONS: Greater CK-MB elevation after PCI is independently associated with adverse outcomes in NSTE ACS. These results underscore the adverse implications of elevated CK-MB levels after PCI in this high-risk population.

Risk of intracranial haemorrhage with combined fibrinolytic and glycoprotein IIb/IIIa inhibitor therapy in acute myocardial infarction. Dichotomous response as a function of age in the GUSTO V trial.

BACKGROUND: Intracranial haemorrhage is an important limitation to pharmacologic reperfusion therapy for acute myocardial infarction. The combination of a glycoprotein IIb/IIIa inhibitor, half-dose plasminogen activator and reduced-dose heparin has been evaluated as an alternative to standard fibrinolytic therapy in this setting. METHODS AND RESULTS: We evaluated the relation between univariate and multivariate predictors of intracranial haemorrhage and the effect of treatment with either reteplase alone (10 U bolus twice, 30 min apart) with standard-dose heparin (5000 U bolus followed by an infusion of 1000 Uh(-1)for patients > or =80 kg and 800 Uh(-1)for those <80 kg) or combination therapy with abciximab (0.25mg/kg bolus and 0.125 microg/kg/min for 12h) and half-dose reteplase (two boluses of 5U 30 min apart) with reduced-dose heparin (60 Ukg(-1)bolus, maximum 5000 U, followed by an infusion of 7 Ukg(-1)h(-1)) in the 16 588 patients randomized in the GUSTO V trial. Overall, the incidence of intracranial haemorrhage was similar in the two groups (0.6% vs 0.6%; OR 1.05, 95% CI 0.71, 1.56). The median (25th-75th) time from drug administration to intracranial haemorrhage was 5.5 (3.4-11) hours with combination therapy and 9.2 (5.9-22) hours with reteplase (P=0.048). Among the multivariable predictors of intracranial haemorrhage, only age showed a significant interaction with treatment effect (age per treatment interaction chi-square 4.60, P=0.032) with a lower risk of combination therapy for younger patients and a higher risk for the elderly. CONCLUSIONS: Although no additional risk of intracranial haemorrhage has been observed with combination therapy in the whole population, a significant age pertreatment interaction exists, with a lower risk with combination therapy in younger patients, and a higher risk in the elderly.

Issues facing clinical trials of the future.

Diagnostic and therapeutic technology continues to advance rapidly, as does our knowledge of therapeutics and of clinical research methods. Unfortunately, these advances have been only poorly coupled with our knowledge of therapeutic principles, leading to increasing uncertainty about which technologies are truly effective and, amongst those that are effective, which are most effective for the cost. This article presents general principles derived from several investigators' experiences with clinical trials, and uses them to suggest how future clinical trials may differ from current approaches. A proposed organization for future trials also is elucidated.

Does the discharge ECG provide additional prognostic insight(s) in non-ST elevation ACS patients from that acquired on admission?

BACKGROUND: Although the prognostic value of admission ST changes in patients with non-ST elevation acute coronary syndrome (ACS) is established, the utility of the discharge ECG is unknown. Accordingly, using the PARAGON-B Troponin substudy, we assessed the prevalence of ST depression on both admission and discharge ECG, the likelihood of developing new Q-waves at discharge and the additional prognostic value of these changes. METHODS AND RESULTS: Nine hundred and eighteen patients were studied; 542 patients (59%) had admission ST downward arrow > or =1mm and 376 patients (41%) did not and their 6-month mortality was 4.4 vs 0.8%, P=0.002, respectively. Of patients with ST downward arrow on admission, 320 (59%) normalized their ST segment at discharge. Of patients without ST downward arrow on admission, 35 (9.3%) developed new ST downward arrow at discharge. Patients with persistent ST downward arrow on discharge had a higher 6-month mortality (6.0 vs 0.9%), (re)MI (16.3 vs 7.4%), and death/(re)MI (20.0 vs 8.3%) than those who never had ST downward arrow (all P< or =0.002). Two hundred and fifty-six patients had Q-waves on admission whereas by discharge 320 had Q-waves. Patients with Q-waves on discharge vs those without had a higher mortality (4.8 vs 1.9%), (re)MI (13.8 vs 8.3%), and death/(re)MI (16.4 vs 9.6%) at 6 months (all P< or =0.021). CONCLUSIONS: This study highlights that the dynamic ECG changes which occur between admission and discharge in non-ST elevation ACS patients allows further risk stratification in determining the likelihood of 6-month death and/or re(MI).

Predictors of 90-day outcome in patients stabilized after acute coronary syndromes.

AIMS: We investigated predictors of 90-day risk among patients surviving the early period after an acute coronary syndrome (ACS). METHODS AND RESULTS: The study population included 15 904 stabilized ST-segment elevation or non-ST-segment elevation ACS patients randomized in SYMPHONY and 2nd SYMPHONY. We developed risk models for death, death or myocardial infarction (MI), and death, MI, or severe recurrent ischaemia (SRI) using Cox proportional-hazards techniques. Demographic, history, and pre-randomization clinical and medication variables were tested. Validation techniques included development of individual trial models, backward elimination and bootstrapping. Of 118 variables, 17 independently predicted mortality. The strongest associations included greater age (chi(2)=31.1), higher randomization heart rate (chi(2)=27.4), and heart failure (HF) variables (HF between qualifying event and randomization, chi(2)=21.8; history of HF, chi(2)=12.2). Higher creatinine clearance (chi(2)=17.7) and percutaneous coronary intervention between qualifying event and randomization (chi(2)=11.1) most strongly predicted lower risk. Similar characteristics entered the double and triple composite models, but HF variables and age less strongly predicted these end-points. CONCLUSIONS: In patients stabilized after ACS, those at highest risk over the next 90 days can be identified. Typical clinical markers are better at identifying risk of death than non-fatal MI or SRI. Novel risk markers are needed for these outcomes.

Time-based risk assessment after myocardial infarction. Implications for timing of discharge and applications to medical decision-making.

AIMS: We evaluated timing of adverse cardiac events after thrombolysis to guide length of stay after ST-segment elevation myocardial infarction. METHODS AND RESULTS: Kaplan-Meier survival curves described timing of major postinfarction complications in 41021 fibrinolytic-treated patients in GUSTO-I. Using model-fitting, these data were best explained by a mixed-exponential survival model: an acute curve describing most adverse events and a chronic curve describing a lower background rate. We replicated this strategy in 15059 fibrinolytic-treated patients in GUSTO-III. From the relation between time and events described by the model's acute curve in GUSTO-III, we proposed times for hospital discharge. The acute curve explained 97% of deaths and 68%-96% of various event composites. Of complications within 10 days, 90% of deaths and 70% of acute curve death, stroke, shock, heart failure, or reinfarction occurred by 24 h. By 2.7 days, 95% of deaths, stroke, shock, heart failure, or reinfarction occurred. Most major ventricular arrhythmias occurred within 24 h, after which the hazard curve was flat. CONCLUSIONS: Mixed-exponential survival modelling describes timing of post-infarction complications and supports discharge 4 days after uncomplicated infarction. Such time-based risk assessment could guide decision-making in other settings in which randomized studies are impractical.