Intermittency is an important metric for determining the likelihood of extreme values of fluctuating quantities such as temperature gradients and vorticity in premixed turbulent flames. Such extreme values can lead to flow-altering events including extinction, auto- and re-ignition, and deflagration to detonation transition. In this study, we use a conditional probability density function (PDF) approach to analyze intermittency of enstrophy (i.e., vorticity magnitude) and temperature gradient magnitude fields. The analysis is based on data from direct numerical simulations (DNS) of stoichiometric premixed hydrogen-air flames in unconfined domains. The DNS are performed for a range of scenarios, including both single- and multi-step chemistry models, temperature-dependent and constant viscosities, different turbulence intensities, and two different resolutions. In each case, conditional PDFs are generated using a temperature-based progress variable in order to characterize intermittency at different locations within the flame. We show that intermittency in the scalar gradient magnitude is most strongly influenced by the turbulence intensity and the choice of chemistry model, while enstrophy intermittency is most strongly dependent on intensity and temperature dependence of the viscosity, with little dependence on chemistry.