Performing clinical trial Centralized Monitoring (CM) involves evaluating risk signals such as site-specific Key Risk Indicators (KRI) metrics that fall outside set limits and monitoring their progress. Many systems use dashboards with traffic lights-type risk signals to highlight sites with computed KRI metrics values that fall outside their normal range. When these lights are triggered, they usually initiate a sequence of actions by the clinical team aimed at mitigating the risk they represent. Nevertheless, upon closer examination, it appears that most risk signals are the result of perfectly normal phenomena and actually requires no action by the clinical team. Here are some examples:
- Example 1: A low AE rate may be justified because the site is located in a location where the population is generally healthy. Sending CRAs on a hunt for unreported AE may not be resource-efficient in this case. Another site may have a high AE rate simply because one of the subjects was involved in a motor vehicle accident (MVA) and suffered several injuries. Labelling a site as “high-risk” because one subject had an accident unrelated with treatment is simply uncalled-for in this case.
- Example 2: A site may experience a significant increase in Query rate and Error rate which may indicate that a CRC is having a hard time with the EDC system and could use some additional training to make her job easier. Elevated Query rates and Error rates can also be the result of a CRA’s on-site visits. Typically, CRAs generate queries while performing Source Data Verification (SDV) and CRCs make changes in the EDC system which count as errors. No action needs to be performed in this case if the risk signals are actually the results of the clinical team’s own actions.
- Example 3: If a study uses ePRO questionnaires, a high level of missing data may be justified because a subject is located in a location that does not have good connectivity rather than being non-compliant. If off-line completion of ePRO questionnaire is possible, the risk signal coming from this specific subject should not trigger any additional actions.
When encountering such risk signals, no action may be required but gathering information and writing an analysis regarding the outlying trend is important to demonstrate proper study oversight. The evaluations of risk signals can be included in periodic central monitoring reports shared with the clinical team to keep stakeholders informed about what transpires from the data collected. It might appear like wasting resources as the analysis often leads to no concrete action but in fact, documenting the risk signal and its evaluation provides perspectives into the different realities at each site, gives confidence regarding the quality of a trial and justifies lowering the level of costly Source Data Verification (SDV). Moreover, documenting current risk-signals along with their evaluation in periodic reports gradually builds a repository of information that can be drawn from during subsequent risk evaluation and quality audit.
Of course, to reduce the amount of false signals, the limits that are related to each KRI need to be properly set. It might be a good idea to widen the tolerance window if too many sites generate false signals. Also, the way a KRI is calculated can have an impact on the amount of false signals. For example, considering only a subset of queries (e.g. reviewer generated vs. automatic system generated) in the calculation of Query rate may reduce the noise and provide more meaningful signals. As such, determining how KRIs are calculated and what limits are imposed on those KRIs while writing the centralized monitoring plan can reduce the amount of false signals.
Because false signals blur true signals, once a light goes off, the task of central monitors is to keep monitoring the risk signal and make sure that no true signal go unnoticed because of the noise from the false signal. For instance, if an AE rate is elevated because a subject was implicated in a MVA and the rate doesn’t decrease at the next review, it might be because of a real safety issue. As such, once a signal is detected, the task of centralized monitoring involves evaluating KRI metrics progress until the metrics value falls back into its normal range.
Translating risk signals into information relevant to the clinical trial team is the main function of central monitors and it is a task that computers can hardly achieve without a human brain. Also, because false risk signals are prevalent, broadcasting these signals without the filter of central monitoring can generate confusion and lead to ineffective actions by the clinical team. As such, central monitoring is a function that CROs require to reap benefits from the influx of clinical trial data. It requires a set of skills that is readily available. See the following article for an overview of the basic central monitoring skills: http://www.appliedclinicaltrialsonline.com/basics-clinical-trial-centralized-monitoring