If by "integration coefficient" you mean "integrating factor", that technique only works on first-orderlinearequations, which this obviously is not. Why not try the Bernoulli approach? Standard form would be

So the substitution would be what?