A fairly powerful technique when you have square roots is to use trig substitutions, or alternatively, hyperbolic trig substitutions.
First, combine the terms inside the square root, so you have the square root of (4x+1)/(4x)
If you break it into a two square roots, and simplify the root in the denominator, you have
sqrt(4x+1) / 2sqrt(x)
Let 2sqrt(x) = tan(theta) *(I have some reservations with this step; can explain later, but haven't thought domain issues through completely)
Then,
4x = tan²(theta)
1/sqrt(x) dx = sec²(theta) dtheta
How convenient, since the integral has a sqrt(x) in the denominator.
So, using these substitutions, you end up with
1/2 integral sqrt(tan²(theta)+1) sec²(theta) dtheta
A quick trig simplification and it's
1/2 integral (sec³(theta)) dtheta
It's fairly simple** to integrate that using integration by parts. Alternately, you can use a reduction formula for integrating powers of sec(theta).
So, you get 1/2[ 1/2 sec(theta)tan(theta) + 1/2 ln (abs (sec(theta)+tan(theta)))]
Factor out the 1/2s and 1/4 [sec(theta)tan(theta) + ln|sec(theta)+tan(theta)|
We already know that tan(theta) = 2sqrt(x) (That was our original substitution.)
And, by either applying a trig identity, else employing some pythagorean theorem, we can deduce that sec(theta)= sqrt(1+4x)
So, finally, 1/4[2sqrt(1+4x)sqrt(x) + ln( sqrt(1+4x)+2sqrt(x) )
I verified this answer numerically for one definite integral - I used a TI 83+ to calculate the integral from 1 to 8. (=7.253)
And, using this formula for the definite integral yielded 8.73245-1.47353 = 7.253
**Note: on the integral of secant cubed, wikipedia notes "is a frequent and challenging indefinite integral of elementary calculus." I say bullshit, but they have the footnote for it.