Prior to the launch of the Swift mission several X-ray line detections were reported in gamma-ray burst afterglow spectra. To date, these pre-Swift era results have not been conclusively confirmed. The most contentious issue in this area is the choice of statistical method used to evaluate the significance of these features. In this paper we compare three different methods already extant in the literature for assessing the significance of possible line features and discuss their relative advantages and disadvantages. The methods are demonstrated by application to observations of 40 bursts from the archive of Swift XRT at early times (<a few kiloseconds posttrigger in the rest frame of the burst). Based on this thorough analysis we found no strong evidence for emission lines. For each of the three methods we have determined detection limits for emission- line strengths in bursts with spectral parameters typical of the Swift- era sample. We also discuss the effects of the current calibration status on emission-line detection.