Background: Above the nucleon resonance region, the N (e,e π ± )N data cannot be explained by conventional hadronic models. For example, the observed magnitude of the transverse cross section is significantly underestimated in a framework with Reggeized background amplitudes. Purpose: Develop a phenomenological framework for the N (e,e π ± )N reaction at high invariant mass W and deep photon virtuality Q 2 . Method: Building on the work of Kaskulov and Mosel [Phys. Rev. C 81, 045202 (2010)], a gauged pion-exchange current is introduced with a running cutoff energy for the proton electromagnetic transition form factor. A new transition form factor is proposed. It respects the correct on-shell limit, has a simple physical interpretation, and reduces the number of free parameters by one. Results: A study of the W dependence of the N (e,e π ± )N lends support for the newly proposed transition form factor. In addition, an improved description of the separated and unseparated cross sections at −t 0.5 GeV 2 is obtained. The predictions overshoot the measured unseparated cross sections for −t > 0.5 GeV 2 . Introducing a strong hadronic form factor in the Reggeized background amplitudes brings the calculations considerably closer to the high −t data. Conclusions: Hadronic models corrected for resonance-parton duality describe the separated pion electroproduction cross sections above the resonance region reasonably well at low −t. In order to validate the applicability of these models at high −t, separated cross sections are needed. These are expected to provide a more profound insight into the relevant reaction mechanisms.