The reason why some radio galaxies (RGs) grow to form so-called giant radio galaxies (GRGs) with sizes $>$ 700 kpc, is still unknown. In this study, we compare the radio, optical and environmental properties of GRGs with those of a control sample of smaller RGs we found in the three LOw-Frequency ARray (LOFAR) deep fields, namely the Bo\"otes, ELAIS-N1, Lockman Hole, for a total area of approx 95 deg$^2$. We inspected the LOFAR deep fields and created a catalogue of 1609 extended radio galaxies (ERGs). By visual inspection, we identified their host galaxies and spectroscopically or photometrically classified 280 of these as GRGs. We studied their properties, such as their accretion state, stellar mass and star formation rate (SFR) using deep optical and infrared survey data. Moreover, we explored the environment in terms of the surface number density of neighbouring galaxies within these surveys. Integrated flux densities and radio luminosities were also determined for a subset of ERGs through available survey images at 50, 150, 610, and 1400 MHz to compute integrated spectral indices. Considering the fraction of GRGs displaying an FRII morphology alongside the host galaxy properties, we suggest that GRGs consistently possess sufficient power to overcome jet frustration caused by the interstellar medium. Moreover, clear differences emerge in the environmental densities between GRGs and smaller RGs, using the number of neighbouring galaxies within 10 Mpc from the host galaxy as a proxy. GRGs preferentially reside in sparser environments compared to their smaller counterparts. In particular, only 3.6<!PCT!> of the GRGs reside within a 3D comoving distance of 5 Mpc from a previously reported galaxy cluster. We found that larger sources exhibit steeper integrated spectral indices, suggesting that GRGs are late-stage versions of RGs. These results suggest that GRGs are amongst the oldest radio sources with the most stable nuclear activity that reside in sparse environments.