Bayesian central statistical monitoring using finite mixture models in multicenter clinical trials

Assessing the performance of methods for central statistical monitoring of a binary or continuous outcome in multi-center trials: A simulation study

BACKGROUND

Quality study monitoring is fundamental to patient safety and data integrity. Regulators and industry consortia have increasingly advocated for risk-based monitoring (RBM) and central statistical monitoring (CSM) for more effective and efficient monitoring. Assessing which statistical methods underpin these approaches can best identify unusual data patterns in multi-center clinical trials that may be driven by potential systematic errors is important.

METHODS

We assessed various CSM techniques, including cross-tests, fixed-effects, mixed-effects, and finite mixture models, across scenarios with different sample sizes, contamination rates, and overdispersion via simulation. Our evaluation utilized threshold-independent metrics such as the area under the curve (AUC) and average precision (AP), offering a fuller picture of CSM performance.

RESULTS

All CSM methods showed consistent characteristics across center sizes or overdispersion. The adaptive finite mixture model outperformed others in AUC and AP, especially at 30% contamination, upholding high specificity unless converging to a single-component model due to low contamination or deviation. The mixed-effects model performed well at lower contamination rates. However, it became conservative in specificity and exhibited declined performance for binary outcomes under high deviation. Cross-tests and fixed-effects methods underperformed, especially when deviation increased.

CONCLUSION

Our evaluation explored the merits and drawbacks of multiple CSM methods, and found that relying on sensitivity and specificity alone is likely insufficient to fully measure predictive performance. The finite mixture method demonstrated more consistent performance across scenarios by mitigating the influence of outliers. In practice, considering the study-specific costs of false positives/negatives with available resources for monitoring is important.

Assessing the impact of risk-based data monitoring on outcomes for a paediatric multicentre randomised controlled trial

BACKGROUND/AIMS

Regulatory guidelines recommend that sponsors develop a risk-based approach to monitoring clinical trials. However, there is a lack of evidence to guide the effective implementation of monitoring activities encompassed in this approach. The aim of this study was to assess the efficiency and impact of the risk-based monitoring approach used for a multicentre randomised controlled trial comparing treatments in paediatric patients undergoing cardiac bypass surgery.

METHODS

This is a secondary analysis of data from a randomised controlled trial that implemented targeted source data verification as part of the risk-based monitoring approach. Monitoring duration and source to database error rates were calculated across the monitored trial dataset. The monitored and unmonitored trial dataset, and simulated trial datasets with differing degrees of source data verification and cohort sizes were compared for their effect on trial outcomes.

RESULTS

In total, 106,749 critical data points across 1,282 participants were verified from source data either remotely or on-site during the trial. The total time spent monitoring was 365 hours, with a median (interquartile range) of 10 (7, 16) minutes per participant. An overall source to database error rate of 3.1% was found, and this did not differ between treatment groups. A low rate of error was found for all outcomes undergoing 100% source data verification, with the exception of two secondary outcomes with error rates >10%. Minimal variation in trial outcomes were found between the unmonitored and monitored datasets. Reduced degrees of source data verification and reduced cohort sizes assessed using simulated trial datasets had minimal impact on trial outcomes.

CONCLUSIONS

Targeted source data verification of data critical to trial outcomes, which carried with it a substantial time investment, did not have an impact on study outcomes in this trial. This evaluation of the cost-effectiveness of targeted source data verification contributes to the evidence-base regarding the context where reduced emphasis should be placed on source data verification as the foremost monitoring activity.

Performance of four centralized statistical monitoring methods for early detection of an atypical center in a multicenter study

Background

Ensuring the quality of data is essential for the credibility of a multicenter clinical trial. Centralized Statistical Monitoring (CSM) of data allows the detection of a center in which the distribution of a specific variable is atypical compared to other centers. The ideal CSM method should allow early detection of problem and therefore involve the fewest possible participants.

Methods

We simulated clinical trials and compared the performance of four CSM methods (Student, Hatayama, Desmet, Distance) to detect whether the distribution of a quantitative variable was atypical in one center in relation to the others, with different numbers of participants and different mean deviation amplitudes.

Results

The Student and Hatayama methods had good sensitivity but poor specificity, which disqualifies them for practical use in CSM. The Desmet and Distance methods had very high specificity for detecting all the mean deviations tested (including small values) but low sensitivity with mean deviations less than 50%.

Conclusion

Although the Student and Hatayama methods are more sensitive, their low specificity would lead to too many alerts being triggered, which would result in additional unnecessary control work to ensure data quality. The Desmet and Distance methods have low sensitivity when the deviation from the mean is low, suggesting that the CSM should be used alongside other conventional monitoring procedures rather than replacing them. However, they have excellent specificity, which suggests they can be applied routinely, since using them takes up no time at central level and does not cause any unnecessary workload in investigating centers.

Remote Health Monitoring in Clinical Trial using Machine Learning Techniques: A Conceptual Framework

Monitoring any process is crucial and very necessary, this is to ensure that standard protocols and procedures are strictly adhered to, monitoring clinical trials is not an exception. It is one of the most crucial processes that should be monitored because human subjects are involved. In trying to monitor clinical trial, information and communication technology techniques can be deployed to facilitate the process and hence improve accuracy. This research formulates a new conceptual framework for monitoring clinical trial using Support Vector Machine and Artificial Neural Network classifiers with physiological datasets from a wearable device. The proposed framework prototype consists of data collection module, data transmission module, and data analysis and prediction module. The data analytic and prediction module is the core section of the proposed framework tailored with data analysis. These datasets are preprocessed and transformed and then used to train and test the system, through different experimental analysis including bagging Support Vector Machine (SVM) and Artificial Neural Network (ANN). The outcome of the analysis presents classification into three different categories, such as fit, unfit, and undecided participants. These various classifications are used to determine if a participant should be allowed to continue in the trial or not. This research provides a framework that is useful in monitoring clinical trial remotely, thereby informing the decision-making process of the research team.

Monitoring strategies for clinical intervention studies

BACKGROUND

Trial monitoring is an important component of good clinical practice to ensure the safety and rights of study participants, confidentiality of personal information, and quality of data. However, the effectiveness of various existing monitoring approaches is unclear. Information to guide the choice of monitoring methods in clinical intervention studies may help trialists, support units, and monitors to effectively adjust their approaches to current knowledge and evidence.

OBJECTIVES

To evaluate the advantages and disadvantages of different monitoring strategies (including risk-based strategies and others) for clinical intervention studies examined in prospective comparative studies of monitoring interventions.

SEARCH METHODS

We systematically searched CENTRAL, PubMed, and Embase via Ovid for relevant published literature up to March 2021. We searched the online 'Studies within A Trial' (SWAT) repository, grey literature, and trial registries for ongoing or unpublished studies.

SELECTION CRITERIA

We included randomized or non-randomized prospective, empirical evaluation studies of different monitoring strategies in one or more clinical intervention studies. We applied no restrictions for language or date of publication.

DATA COLLECTION AND ANALYSIS

We extracted data on the evaluated monitoring methods, countries involved, study population, study setting, randomization method, and numbers and proportions in each intervention group. Our primary outcome was critical and major monitoring findings in prospective intervention studies. Monitoring findings were classified according to different error domains (e.g. major eligibility violations) and the primary outcome measure was a composite of these domains. Secondary outcomes were individual error domains, participant recruitment and follow-up, and resource use. If we identified more than one study for a comparison and outcome definitions were similar across identified studies, we quantitatively summarized effects in a meta-analysis using a random-effects model. Otherwise, we qualitatively summarized the results of eligible studies stratified by different comparisons of monitoring strategies. We used the GRADE approach to assess the certainty of the evidence for different groups of comparisons.

MAIN RESULTS

We identified eight eligible studies, which we grouped into five comparisons. 1. Risk-based versus extensive on-site monitoring: based on two large studies, we found moderate certainty of evidence for the combined primary outcome of major or critical findings that risk-based monitoring is not inferior to extensive on-site monitoring. Although the risk ratio was close to 'no difference' (1.03 with a 95% confidence interval [CI] of 0.81 to 1.33, below 1.0 in favor of the risk-based strategy), the high imprecision in one study and the small number of eligible studies resulted in a wide CI of the summary estimate. Low certainty of evidence suggested that monitoring strategies with extensive on-site monitoring were associated with considerably higher resource use and costs (up to a factor of 3.4). Data on recruitment or retention of trial participants were not available. 2. Central monitoring with triggered on-site visits versus regular on-site visits: combining the results of two eligible studies yielded low certainty of evidence with a risk ratio of 1.83 (95% CI 0.51 to 6.55) in favor of triggered monitoring intervention. Data on recruitment, retention, and resource use were not available. 3. Central statistical monitoring and local monitoring performed by site staff with annual on-site visits versus central statistical monitoring and local monitoring only: based on one study, there was moderate certainty of evidence that a small number of major and critical findings were missed with the central monitoring approach without on-site visits: 3.8% of participants in the group without on-site visits and 6.4% in the group with on-site visits had a major or critical monitoring finding (odds ratio 1.7, 95% CI 1.1 to 2.7; P = 0.03). The absolute number of monitoring findings was very low, probably because defined major and critical findings were very study specific and central monitoring was present in both intervention groups. Very low certainty of evidence did not suggest a relevant effect on participant retention, and very low certainty evidence indicated an extra cost for on-site visits of USD 2,035,392. There were no data on recruitment. 4. Traditional 100% source data verification (SDV) versus targeted or remote SDV: the two studies assessing targeted and remote SDV reported findings only related to source documents. Compared to the final database obtained using the full SDV monitoring process, only a small proportion of remaining errors on overall data were identified using the targeted SDV process in the MONITORING study (absolute difference 1.47%, 95% CI 1.41% to 1.53%). Targeted SDV was effective in the verification of source documents, but increased the workload on data management. The other included study was a pilot study, which compared traditional on-site SDV versus remote SDV and found little difference in monitoring findings and the ability to locate data values despite marked differences in remote access in two clinical trial networks. There were no data on recruitment or retention. 5. Systematic on-site initiation visit versus on-site initiation visit upon request: very low certainty of evidence suggested no difference in retention and recruitment between the two approaches. There were no data on critical and major findings or on resource use.

AUTHORS' CONCLUSIONS

The evidence base is limited in terms of quantity and quality. Ideally, for each of the five identified comparisons, more prospective, comparative monitoring studies nested in clinical trials and measuring effects on all outcomes specified in this review are necessary to draw more reliable conclusions. However, the results suggesting risk-based, targeted, and mainly central monitoring as an efficient strategy are promising. The development of reliable triggers for on-site visits is ongoing; different triggers might be used in different settings. More evidence on risk indicators that identify sites with problems or the prognostic value of triggers is needed to further optimize central monitoring strategies. In particular, approaches with an initial assessment of trial-specific risks that need to be closely monitored centrally during trial conduct with triggered on-site visits should be evaluated in future research.