The effect of patient inhomogeneities surrounding the oesophagus on the dosimetry planning of an upper thoracic oesophageal 192Ir HDR brachytherapy treatment is studied. The MCNPX Monte Carlo code is used for dosimetry in a patient-equivalent phantom geometry and results are compared in terms of isodose contours as well as dose volume histograms with corresponding calculations by a contemporary treatment planning system software featuring a full TG-43 dose calculation algorithm (PLATO BPS version 14.2.4). It is found that the presence of patient inhomogeneities does not alter the delivery of the planned dose distribution to the planning treatment volume. Regarding the organs at risk, the common practice of current treatment planning systems (TPSs) to consider the patient geometry as a homogeneous water medium leads to a dose overestimation of up to 13% to the spinal cord and an underestimation of up to 15% to the sternum bone. These findings which correspond to the dose region of about 5-10% of the prescribed dose could only be of significance when brachytherapy is used as a boost to external beam therapy. Additionally, an analytical dosimetry model, which is efficient in calculating dose in mathematical phantoms containing inhomogeneity shells of materials of radiobiological interest, is utilized for dosimetry in the patient-equivalent inhomogeneous phantom geometry. Analytical calculations in this work are in good agreement with corresponding Monte Carlo results within the bone inhomogeneities of spinal cord and sternum bone but, like treatment planning system calculations, the model fails to predict the dose distribution in the proximal lung surface as well as within the lungs just as the TPS does, due to its inherent limitation in treating lateral scatter and backscatter radiation.