Mallinckrodt Pharmaceutical R&D
How to Use Design of Experiments Effectively to Develop Formulations for Pharmaceuticals Tablets by Richard Ruey-ching Hwang, PHD;MBA St. Louis, MO yr.2000

Doe Advantages
“…it provides an effective methodology to evaluate all potential factors systematically and results in a comprehensive understanding of the formulation system including the potential interaction among the critical factors.”
“…provides reasonable flexibility to the number of experiments required to develop formulations.  It is possible that a good formulation can be developed by conducting small number of experiments if appropriate DOE is used.”

DOE vs. Quantis
“It is a major task to evaluate all of the formulation and process variables to optimize a tablet formulation. However, there is usually a constraint on time, manpower, and materials when developing new pharmaceuticals products.  It is important for a pharmaceuticals scientist to utilize efficient methodologies to develop good products in a timely manner without sacrificing quality. A simple and effective way to develop tablet formulation efficiently is to use DOE.” …”This article will present the formulation optimization process for a IR tablet using various experimental design. In the meantime, a three factor two-level full factorial design, a three factor two-level fraction factorial design, and a Quantis design that requires only Np+1(the number of parameters plus one) experiments were also used to analyze the data. The formulations defined by different types of DOE were then evaluated and compared to the target formulation characteristics.”

Objective of the Experiment
“…..optimize these factors to develop a tablet formulation that had good compression characteristics…, good tablet characteristics…, and a disintegration time ….”

“……a three factor mixed level experimental design (Design A) that required 18 experiments…..was analyzed by computer software JMP.3.1.6 (SAS Institute Inc., Cary, NC). To illustrate the potential different results from the different types of DOE, the data were then analyzed by different methodologies.  These additional types of DOE were three factor two-level full factorial  design (Design B, 8 experiments), three factor two-level fractional factorial design (Design C, 4 experiments), and Quantis design (Design D, 4 experiments)…….Design D was analyzed by Quantisweb and the optimal formulation was defined directly by the Quantisweb program.  The formulations defined by different types of DOE were evaluated and compared to the target formulation characteristics.”

“For Design A … since this experimental design provides the most statistical resolution power, the results concluded from this design are considered optimal.”
“For Design B… the only significant different conclusion between Design B and Design A is the concentration of … in the formulation.”
“The results concluded from Design C are similar to the results concluded from Design A except for the concentration of ….in the formulation.”
“For Design D (Quantis design), is almost the same as the optimal formulation concluded from Design A except for the compression force. Quantis design appeared to be effective since it only required 4 experiments and concluded the same formulation as an 18-experiment study.”

“Designs A and D have almost the same results except for a slight difference in compression force.  To verify the accuracy of the predicted responses generated by Design D, the predicted responses for the formulation defined by Design D, and the actual ranges of responses from the experiments…..are reasonably close to the actual values.”