Studies are extremely difficult to get right. I'm generally a little bit skeptical of data for this reason.

A bit of a tangent but still on the subject of environmental pollution; the other day I found out that CO2 sensor sensitivity naturally drifts over time... So when a CO2 sensor is replaced for long term climate research, if they try to calibrate the new sensor to the old one at the time of replacement, the drift would be carried over into the new sensor even if no actual real change of CO2 occurred... Apparently there are standards to prevent this but mistakes have been identified multiple times with the methodology for setting that standard... Anyway measuring data accurately is really hard.

People do not appreciate correctness enough.