Holczer, Mazeh, and collaborators (Holczer et al. 2016ApJS..225....9H, Cat. J/ApJS/225/9) used the Kepler 4-yr observations to derive a transit-timing catalog, identifying 260 Kepler objects of interest (KOI) with significant transit timing variations (TTV). For KOIs with high enough S/Ns, HM+16 also derived the duration and depth of their transits. In this work, we use the duration measurements of HM+16 to systematically study the duration changes of 561 KOIs and identify 15 KOIs with a significant long-term linear change of transit durations and another 16 KOIs with an intermediate significance. We show that the observed linear trend is probably caused by a precession of the orbital plane of the transiting planet, induced in most cases by another planet. The leading term of the precession rate depends on the mass and relative inclination of the perturber, and the period ratio between the two orbits, but not on the mass and period of the transiting planet itself. Interestingly, our findings indicate that, as a sample, the detected time derivatives of the durations get larger as a function of the planetary orbital period, probably because short-period planetary systems display small relative inclinations. The results might indicate that short-period planets reside in relatively flattened planetary systems, suggesting these systems experienced stronger dissipation either when formed or when migrated to short orbits. This should be used as a possible clue for the formation of such systems.