Context: In the laboratory, hydrogen peroxide (HOOH) was proven to be an intermediate product in the solid-state reaction scheme that leads to the formation of water on icy dust grains. When HOOH desorbs from the icy grains, it can be detected in the gas phase. In combination with water detections, it may provide additional information on the water reaction network. Hydrogen peroxide has previously been found toward $\rho$ Oph A. However, further searches for this molecule in other sources failed. Hydrogen peroxide plays a fundamental role in the understanding of solid-state water formation and the overall water reservoir in young stellar objects (YSOs). Without further HOOH detections, it is difficult to assess and develop suitable chemical models that properly take into account the formation of water on icy surfaces. Aims: The objective of this work is to identify HOOH in YSOs and thereby constrain the grain surface water formation hypothesis. Methods: Using an astrochemical model based on previous work in combination with a physical model of YSOs, the sources R CrA-IRS\,5A, NGCC1333-IRAS\,2A, L1551-IRS\,5, and L1544 were identified as suitable candidates for an HOOH detection. Long integration times on the APEX 12m and IRAM 30m telescopes were applied to search for HOOH signatures in these sources. Results: None of the four sources under investigation showed convincing spectral signatures of HOOH. The upper limit for HOOH abundance based on the noise level at the frequency positions of this molecule for the source R CrA-IRS\,5A was close to the predicted value. For NGC1333-IRAS 2A, L1544, and L1551-IRS\,5, the model overestimated the hydrogen peroxide abundances.