ABSTRACT
The direct epifluorescent filter technique (DEFT) for rapid estimation of microbial numbers was evaluated by comparison with the plate count on a variety of uncooked red meat and poultry samples. Good agreement [correlation coefficient (r) = 0.95-0.96] was obtained from samples with plate counts of 5 × 103/g or /cm2 and above from red meat carcasses (surface swabbed), aerobic or vacuum packed chill-stored joints (surface sampled - stomachered) and frozen beef (thawed stomachered). For stored and unstored raw poultry sampled by skin scraping or stomachering of muscle and skin good overall correlation (r = 0.88-0.89) was obtained between the DEFT count and the plate count in the ranges 1.1 × 103 to 1.3 × 107/cm2 (skin scraping) and 1 × 104 to 9.5 × 106/g (muscle and skin) even though the DEFT always overestimated counts on samples on which no growth had occurred (plate count <7×104/cm2 or <1×105/g). However, good linearity between DEFT and plate counts allowed use of the regression equation to obtain a good estimate of the plate count on these samples. The DEFT was unsuitable for application to poultry neck skin sampled by shaking because particulate material interfered with counting. This was also a problem with Mechanically Recovered Meat although the DEFT gave a fair estimate (r = 0.72) of the plate count on certain types (beef and veal) of this product. The DEFT was capable of providing counts within 35 to 45 min and its applicability to the rapid estimation of bacterial numbers in meat and poultry is discussed.