To some extent, errors are an inevitable part of the drug discovery and development process. In particular, working with biologics can present unique challenges, due to their complexity and sensitivity to external conditions.
But errors can be incredibly costly, wasting time and money, lowering product quality, and even resulting in noncompliance with various laws and regulations, if unaddressed. It’s estimated that errors cost an average U.S. lab about $180,000 each year in the pre- and post-analytical stages. For that reason, it’s critical to anticipate sources of human error in biologics experiments—including antibody discovery, cell-based therapies, and RNA therapies—and take steps to avoid them.
Systematic errors are flaws in an experiment’s design or procedures that shift all measurements in the same direction. As a result, they reduce the accuracy of the experiment. Examples of systematic errors include:
Calibration errors: A measurement instrument is improperly calibrated or the experimenter forgets to calibrate it. Calibration errors can also occur if equipment is not serviced periodically and maintained to a high standard.
Estimation errors: On some instruments, reading a measurement is subject to errors in human estimation. For example, viewing a meniscus from slightly different angles can lead to different recorded measurements.
Instrument drift: Instrument accuracy can change over time. For example, measurements collected from an electronic instrument may change as it warms up. Hysteresis, where a physically observable effect lags behind its cause, can also occur and should be accounted for.
Using insensitive or faulty equipment: Some measurements require more sensitivity than others. Relying on instruments with poor or inadequate sensitivity can cause experimenters to miss low-level samples.
A few key steps can help mitigate systematic errors in the lab. First, equipment should be regularly maintained and calibrated. Second, staff should be properly trained and supervised in operating instruments and recording data to minimize deviation from experimental protocols. Finally, automating as many lab activities as possible can reduce opportunities for human error (more on this below).
Random errors are caused by fluctuations in the experimental or measurement conditions. They tend to be small, but can still impact experimental outcomes. Examples of random errors include:
Environmental factors: Changes in temperature, light levels, and electrical or magnetic noise can all affect observation and measurements.
Transcriptional error: Transcriptional error occurs when data is recorded incorrectly.
Experimenter fatigue or inexperience: Lack of experience with equipment can cause measurements to be inaccurate or unreliable. Similarly, waning attentional resources can cause experimenter observations to decrease in accuracy over time.
Many of the human errors in this category can be reduced or eliminated by delivering proper training to staff, as well as providing sufficient breaks or alternating tasks to maintain vigilance. Conditions in the lab, including lighting and temperature, should also be monitored for consistency at regular intervals.
Errors in human decision-making can jeopardize an experiment from the start or introduce bias at any stage. Some common decision-making errors include:
Confirmation bias: Experimenters are less likely to detect errors in measurement if the error causes data to align with a hypothesis or desired result. They are also less likely to double-check the results of an analysis if those results support a hypothesis. Implementing extra checks, including those conducted by a disinterested party, can help experimenters avoid confirmation bias.
Experimenter bias: In any situation that involves human judgment, experimenters who are aware of the condition they are observing (control vs. treatment) can introduce unconscious bias into their measurements and decisions. Blinding a study eliminates this source of error.
The Power of Lab Automation
Lab automation offers a powerful solution for mitigating human error in biologics experiments. Relying on robotic equipment to perform various lab tasks, such as the sorting, loading, and centrifugation of specimens, can greatly reduce errors in measurement or breaches of experimental protocols.
Utilizing an ELN or informatics platform to automate various aspects of the experimental workflow can also help eliminate the bulk of human errors. Some of the ways ELNs reduce human error include:
Calibration management: An ELN with built-in calibration management software enables labs to track the status of equipment and ensure that calibration is performed in a timely manner.
Structured data entry: Allowing experimenters to enter data without predetermined parameters can result in manual data entry errors. With an ELN, predefining options for data entry cuts down on transcriptional errors.
Barcode labelling: Barcodes enable automated sample tracking and inventory management. This feature can reduce transcriptional error, as well as delays or errors in experiments resulting from depleted supplies.
Automation of lab workflows: Automating various lab activities, from note-taking to analysis, cuts down on bias and other forms of human error.