DOE Software for Quality by Design (QbD) and Process Optimization

The typical alternative to using DOE to determine which processes or formulas are contenders is the classic “COST” approach.  The COST approach, which stands for “Change One Separate factor at a Time” is a logical way to approach an experimentation, but it requires a lot of time and effort, and may lead to the wrong conclusions.  

COST doesn’t let you consider the interaction between investigated factors and may give you different results depending on your starting point. DOE helps shorten that process and uses advanced data analytics to help inform your decisions about which factors need to be tested.  

Another weakness of the COST approach is that it can’t define the number of runs that will be needed to find the maximum. You can only test two factors at a time and can’t define how many runs you’ll need to be sure you’ve covered all the options. 

Why using DOE is better than using a COST approach: 

1. DOE suggests the correct number of runs needed (often fewer than used by the COST approach) 
2. DOE provides an unbiased model for the direction to follow that is independent of start settings 
3. Many factors can be used (not just two) and their interactions taken into account 

Step 1: Choose the Right Measurement Factors  

Ensure that the factors selected to study can be used to predict an acceptable formulation parameter range where all the values for the assessed quality attributes will be inside the specified limits. 

Step 2: Design a Statistically Valid Study 

Consider how the factors being investigated fit into a full factorial design. For pharma companies, for example, robustness studies must be able to prove that specific critical quality attributes stay within the acceptable ranges for the entire shelf-life period. In addition: 

• The study must result in a regression model that is statistically significant 

• The study must provide output parameters (quality attributes) that are within predefined limits 

Step 3. Analyze the Data Using Multiple Linear Regression 

One important way to produce a valid testing model is to use a tool that makes Design of Experiments easier. For example, MODDE^® Design of Experiments Software, can help you set up multivariate formulation robustness studies that demonstrate the acceptable ranges of quality for a target composition, define the allowable edges of the composition range, and predict the stability requirements needed to reach the end of shelf life. 

Scientific investigations involve changing a number of controlled variables to direct the response in question towards a desired level. Design of Experiments (DOE) is a rational and cost-effective approach to practical experimentation that allows the effect of variables to be assessed using only the minimum of resources. DOE is the backbone for efficient QbD implementation strategies. The final specifications for a region where all specifications are fulfilled to a defined risk level is called Design Space. Determining the optimal Design Space is a key advantage of using MODDE^® DOE Software. 

Find a Course