Navigating the High Stakes: An Intern’s Guide to Creating Short-Run Control Charts in Biotech Manufacturing

Illustrated technician reviewing control charts on a laptop beside lab equipment

October 2, 2025

Welcome to the cutting edge of biotech manufacturing! As an intern, you’re stepping into an environment where quality is paramount and the financial stakes are enormous, often exceeding $1 million per batch (for biotech drugs) meaning the tolerance for failure is incredibly low.

Traditional Statistical Process Control (SPC) methods, like standard $\overline{X}$ and $R$ charts, were designed for mass production, assuming long runs of identical products. However, in specialized biotech or continuous manufacturing (CM) lines, production cycles might be relatively short, resulting in a low volume of data for any single product or run.

When you need to monitor critical parameters but lack the 20 to 25 subgroups traditionally recommended for setting reliable control limits, you must turn to Short-Run SPC.

1. The Core Challenge: Lack of Historical Data

Short-run SPC is essential when you have insufficient data to estimate process parameters accurately, often occurring when a production process switches frequently between different products of a similar kind. The fundamental solution is standardization, which allows you to pool data from various products or small batches onto a single control chart.

2. Bare Minimum Requirements for Short-Run SPC

Given the high cost and low appetite for failed batches, implementing a control chart immediately (often called Phase II monitoring) is critical.

a. Essential Tools and Information

SPC Expertise: A person with adequate training in statistical process control techniques is recommended to develop the data collection plan and statistical procedures.
Calculation Tools: While specialized software like Minitab or JMP is ideal for running complex charts, simple computation can be done initially using software like Microsoft Excel.
Process Understanding: You need to confirm that, regardless of the product variation, the underlying variation of the process remains the same across all products being monitored (or is nearly the same).

b. Bare Minimum Data Points for Standardization

To create a standardized control chart for variables data (like Active Pharmaceutical Ingredient (API) concentration or tablet weight), you need to transform the measured value ( $X$ ) into a standardized statistic ( $Z$ or $\delta$ ).

For a single measurement or small sample group, you require:

The Measured Value ( $X$ ): The critical quality attribute (CQA) or critical process parameter (CPP) observed for the specific item or sample.
The Target/Nominal Value ( $T$ or $\mu$ ): The desired outcome for that product characteristic, established during process development (Quality by Design, QbD).
An Estimate of the Process Standard Deviation ( $\sigma$ ): This estimate must reflect the expected or historical variation ( $\overline{R}$ or $\sigma$ ) of the process itself, pooled from similar products or runs, rather than calculated solely from the single short run.

3. Creating the Short-Run Control Chart: Standardization

The fundamental method for short-run SPC is standardizing the measurements by relating the measured value to its specified target.

a. Step-by-Step: The $Z$ -Chart Method

The $Z$ -chart (or stabilized chart) is a popular and robust method because it generates a single plot independent of the unit of measure, allowing multiple different batches or products to be monitored simultaneously.

Calculate the Deviation ( $\delta$ ): Find the difference between the actual measurement ( $X$ ) and the target value ( $T$ or $\mu$ ): $\delta = X - T$
Standardize the Deviation (Calculate $Z$ -value): Divide the deviation by the pooled or estimated process standard deviation ( $\sigma$ ): $Z = \frac{X - \mu}{\sigma}$
Plot the $Z$ -Scores: The resulting $Z$ -value (or $Z$ -score) is plotted on the chart.
Set Control Limits: The standardized data should follow a distribution with a mean of 0 and a standard deviation of 1. Therefore, the center line (CL) is set at 0, and the upper and lower control limits (UCL/LCL) are typically set at $\pm 3$ $\sigma$ .
Monitor Moving Range ( $MR$ ): If you are plotting individual measurements, you should include a Moving Range (MR) sub-chart to assess process variability over time. The $Z-MR$ chart is common for short-run processes.

This resulting chart provides a continuous record of process performance, which is valuable in environments where runs are short and discontinuous.

4. Presenting to Leadership: Standardization Tradeoffs

When presenting your results and recommending a strategy for continued process monitoring (Stage 3 Continued Process Verification, CPV), your superiors will want to understand the tradeoffs between different advanced SPC standardization techniques, especially since maximizing yield and minimizing waste is crucial.

While standardized Shewhart charts (like the $\delta$ chart or $Z$ -chart) are foundational for mixed batches, memory-based charts offer superior detection capability for the inevitable, subtle drifts in complex processes.

Here are the key standardization techniques suitable for low-volume biotech manufacturing, along with their performance differences and tradeoffs:

Standardization Technique	Primary Strength	Tradeoffs & Performance Differences
$Z$ -Chart / $\delta$ Chart	Simplicity and Universality. Allows different parts/characteristics to be plotted on one chart. Clear limits ( $\pm 3$ $\sigma$ ) simplify understanding.	Insensitive to Small Shifts: Only plots the current point, failing to incorporate historical context (lacks “memory”). Requires variances to be equal across products if using the $\delta$ chart.
CUSUM (Cumulative Sum)	Excellent for Small Shifts. Incorporates previous data, making it a very efficient control chart for detecting small shifts or persistent changes. Provides a steady indication of control with minimal noise.	More Complex Interpretation: Requires calculating $C^+$ and $C^-$ values based on a reference value. While effective, the interpretation can be more complicated than simple Shewhart limits.
EWMA (Exponentially Weighted Moving Average)	Good for Small/Moderate Shifts. Also incorporates process history via a smoothing constant ( $\lambda$ ). Generally performs better than Shewhart charts for detecting shifts.	Slow Startup: The control limits may not approach their steady state value quickly when using small initial datasets. Complexity is high, involving six parameters for some variants.
Q-Charts (Quesenberry)	Immediate Start-up: Designed for start-up processes and short runs, enabling monitoring from the very first samples. Limits are independent of varying sample size.	Risk of Masking: Can be “biased,” meaning if a shift occurs early, the statistic used to standardize the data may be affected by the shift, leading to a poorer detection performance (masking). May fail to immediately detect a shift.

5. The High-Cost Context Recommendation

In an environment of highly expensive batches and low tolerance for failure, process control should prioritize the rapid and reliable detection of even the smallest deviations.

You should remind your superiors that:

Memory Charts (CUSUM/EWMA) are Statistically Superior for Stability: For detecting the small, sustained shifts often encountered in complex pharmaceutical processes, CUSUM charts and EWMA charts should be preferred over basic Shewhart charts (like the $Z$ -chart). CUSUM, in particular, tends to provide a stable, reliable indication of control with minimal noise, even with limited historical data.
Tradeoff: Detection Speed vs. Simplicity: While $Z$ -charts are simpler and easier for operators and engineers to immediately interpret, they are less sensitive to small shifts. The initial increase in complexity required for CUSUM/EWMA implementation is a worthwhile investment to gain increased sensitivity and reduce the risk of a high-value batch failure.
Q-Charts for Start-Up: If the objective is to monitor the very first few samples of a new run before sufficient process data exists, the Q-chart’s ability to “self-start” is valuable, but its reported susceptibility to masking makes it risky for long-term monitoring in this high-risk sector.

By selecting and implementing advanced, sensitive control charts—such as the standardized CUSUM or EWMA, you ensure that the company maintains continuous process verification (CPV) and maximizes compliance while safeguarding those expensive product batches.

Decoded, monthly

One email a month on what actually changed in pharma quality and regulatory practice, and what it means for a site. No sales sequence, no sales email.

← Back

Subscribed.

You are on the list. One email a month, nothing else. -Biostrategenix

One email a month. Unsubscribe any time. We do not share your address or use it for anything else — see our Privacy Policy.