This study was aimed at understanding the effect of near infrared (NIR) spectrometer wavelength range and calibration approaches on the performance of pharmaceutical tablet crushing strength prediction methods. The study utilised a six-component tablet formulation, created with variability in excipient ratio and compression force. Method development was tested with two NIR spectrometers and three calibration approaches: spectral slope, partial least squares (PLS) and multiple linear regression (MLR). The sensors used in the study were based on diffuse reflectance sensing geometry with wavelength ranges of 1100–2498 nm and 1350–1800 nm. The three calibration approaches demonstrated an improved prediction of crushing strength with the longer wavelength range, indicative of its significant impact on method performance. It was established that the PLS approach reduced the impact of wavelength range on performance by maximising the covariance between crushing strength and the spectral features associated with baseline shift/slope change and excipient ratio. The MLR approach incorporated variables of physical/chemical nature to mitigate risks to method performance.
© 2016 The Author(s)PDF Article