As R&D data becomes more complex, opaque, and expensive to generate, it’s increasingly vital for pharma labs to deliberately root out poor-quality data. In the past, scientists could identify outliers and anomalies during the final data analysis. Today, waiting that long can cost a lab tens of thousands of dollars or more from having to re-run failed experiments. So to mitigate this risk, many R&D pharma labs are adopting real-time audits and validation that flag data quality issues in time to fix the underlying problem or cut the experiment short.
This article explores how R&D teams at pharma organizations can implement real-time validation and audits that ensure data accuracy and reliability.
Audits in the Life Sciences Industry
Audits are a crucial process in various industries for evaluating performance and efficiency. The audit process involves a systematic examination of data, statistics, and other relevant information by experts with specialized knowledge and techniques.
The primary objective of an audit is to provide an accurate and unbiased evaluation of an organization’s performance, highlighting strengths and weaknesses, and recommending improvements. Audits help organizations reduce costs, improve productivity, comply with regulatory requirements, and enhance overall performance.
The Importance of Validation
Validation of research and evaluation processes ensures that the data and methods used are accurate and reliable. The validation process involves verifying the authenticity and credibility of information, using various techniques and methods to confirm the results. Although the process can be time consuming and resource intensive, it is a crucial step in ensuring the quality and integrity of research. The different methods, including data validation, process validation, and outcome validation, each have their strengths and limitations.
By rigorously validating data and processes, researchers can ensure that their findings are robust and trustworthy.
Setting Priorities
Whether real-time audits and validation are automated or manual, they all require something that’s rare in most R&D labs: consistency.
Particularly in the early stages of drug discovery, R&D labs deliberately build flexible processes and workflows that allow scientists to adapt their experiments as they improve their understanding of the underlying biology. As they get closer to clinical trials, they trade some of this flexibility for consistency, particularly when regulatory standards are involved. But this means that real-time audits and validation may not be feasible for every assay, particularly in early discovery.
So rather than trying to implement real-time audits and validation across all assays and experiments, it’s better to stratify them and focus on the areas where it’s both feasible and more important, particularly to ensure seamless collaboration between research and clinical teams.
Here’s an example classification:
- GXP assays: Processes or assays that require regulatory compliance or are otherwise impacted by regulatory standards. These are both highly consistent and the highest stakes when it comes to quality control.
- Frequent assays: Experiments that are run regularly, either by a single team or across multiple teams. A little effort to implement audits and validation will go a long way here. And if they don’t already have consistent SOPs, creating them will have additional benefits.
- High-stakes assays: Experiments that play an out-sized role in expensive decisions. It may be harder to create consistent SOPs, but flagging data issues here can save a lot of time and money later.
- Low-stakes assays: Infrequent experiments that mostly validate other assays or provide secondary insights. Here, flexibility is usually more important than consistency, so real-time audits may not be feasible.
Set Conventions and Expectations
Once you’ve identified which assays to address, the next step is to define what you’re going to audit. This can typically be split into two parts:
- Conventions define how the software or individuals doing the audit will find and access everything from raw data and electronic lab notebooks to structured data: naming conventions for files and folders, file formats, column names, and numerical units.
- Expectations define what kinds of errors or anomalies should be flagged: missing data, out-of-range values, unexpected distributions, etc.
Both conventions and expectations can be defined at different levels of detail, depending on the data classification. For GXP assays, naming conventions and data formats should be specified in as much detail as possible. For low-stakes, infrequent assays, just defining folder naming conventions might be enough. Wherever possible, conventions should be reinforced by your clinical research management software.
It is important to detail the specifics of naming conventions and data formats to avoid ambiguity and ensure consistency across the board. Data silos make this harder because you’ll have to negotiate this with multiple data owners across multiple systems.
Automate and Support Research Processes
To minimize the number of unnecessary alarms, any auditing systems should be implemented in parallel with measures to minimize the risk of human error.
The best way to do this is to take users out of the equation by automating the processes. But this isn’t always possible: Some data processing steps require human judgment such as setting cut-offs or defining outliers. Other steps are either fundamentally inconsistent or just too infrequent to justify automation.
So in cases where automation doesn’t make sense, the next best thing is to provide tools that help users do the right thing every time.
Either way, the goal is to have the data consistently follow the conventions so that the real-time audit and validation process can accurately verify the expectations.
Create Dashboards and Alerts
If you can appropriately stratify your data, define conventions and expectations, and then automate and support those conventions, the actual audits and validation will be the easy parts. The only trick that remains is to make sure that the right people pay attention to the reports.
Alerts should only be sent to team members who have the ability to act on them and know what they need to do. And there should be a single person who’s primarily responsible for each alert. Otherwise, everyone will assume someone else is dealing with it.
Every dashboard needs a regularly scheduled time for review. Whether that’s a meeting on multiple people’s calendars or just a single person’s responsibility, if it isn’t on the calendar it won’t happen. And a dashboard that no one looks at isn’t worth the time to create.
Maintaining Data Integrity
Real-time audits and validation can save research labs valuable resources by preventing failed experiments or fixing them before they happen. While they can be difficult to implement in the context of an R&D lab, robust data management practices will make the job easier.To overcome the challenges of the audit process, teams must be proactive and responsive to changing circumstances. It helps to use clinical research management software that makes it easy to define and enforce conventions, automate research processes, and create dashboards and alerts. With the right software tools, teams can ensure that their evaluations are accurate and actionable.