The use of small field-of-view (FOV) detectors in cardiac SPECT imaging leads to severe truncation of the non-cardiac regions in the emission data. Such truncation can potentially undermine the effect of attenuation correction (AC) on cardiac imaging. In this work, we use simulation studies to systematically address if accurate AC can be achieved using an iterative image reconstruction algorithm. We first generate truncation-free data using an MCAT phantom, from which we then generate data with different degree of truncation (to simulate detectors of different FOV). Later we use an OSEM algorithm with different reconstruction supports for image reconstruction and use different attenuation maps for AC. In all the simulation studies, the cardiac insert of the MCAT phantom is at the center of imaging FOV and it is not truncated at any projection views. Two liver positions (up and down) are simulated to mimic different levels of background activity. The results show that (1) the use of enlarged reconstruction supports (larger than the imaging FOV of the small FOV detectors) is critical to avoid reconstruction artifacts, especially when the background activity level is high; (2) There is no observable difference between images without AC from small and large FOV detectors when enlarged reconstruction supports are used; (3) The use of non-truncated attenuation maps is necessary for accurate AC; and (4) Quantitatively accurate AC images can be obtained through the use of enlarged reconstruction supports and AC with non-truncated attenuation maps.