If you're just looking to measure a voltage that is negative with respect to the PIC's circuit ground, then isolation isn't necessary. In that case, it's possible to use an opamp circuit to shift the voltage to within the range (0 to Vdd, often 0 to 5 V) of the PIC's ADC, or even just resistors, as in

http://cq.cx/interface.pl#6

That would be suitable in the case where your voltage to be measured is applied across two wires, where one of those wires can be connected to the PIC's circuit ground.

If the voltage to be measured is applied across two wires, but neither wire can be connected to the PIC's circuit ground (for example, if for some reason there is a large unknown voltage between the common-mode voltage of those two wires and the PIC's circuit ground), then true isolation is necessary. This will typically involve either an optical link, or a transformer.

A module like the AD202 integrates a transformer, plus all the electronics to couple a measurement of a DC voltage through that transformer. But, it's very expensive.

An opto-coupler is cheap, but it's hard to use one to accurately transmit an analog value. (There's tricks, like using two matched opto-couplers, with one in the feedback loop to cancel out the non-linearity; but they're still not that good.)

Or, you may prefer to connect a PIC to the same ground as the voltage to be measured, A/D convert the voltage with that PIC, and then transmit that digital value (for example, using the UART) over a digital opto-coupler, to another PIC connected to a different ground.

There's ICs that do a similar job, for example the AD7400, which integrates the A/D converter, a digital isolator (in this case a transformer, actually, not an opto), and the control logic on either side of the isolation. But that wouldn't be very easy to use from LDmicro.

What a useful question