That's not an easy question to answer. Parasitic infections are not normally distributed in a population but follow a negative binomial. In addition many fish species have complex behavioural and social factors that essentially mean that the population is not homogeneous (for example it is a rule that bigger fish live in deeper water).

Thus to answer the question you need to specify what is the population under study and what are the species of interest. Then you need to work out a sampling regime that balances the information obtained with the available funds and time available.  It may be that you end up looking for indicator parasites rather than a large suite of parasites, because some parasites will be easier (cheaper) to count than others that may (for example) require histology of PCR.

Are you focusing on parasites specifically, or do you want to assess the overall health?

Baseline information that would be helpful to interpret results would include species, weight/length, sex.

In terms of parasites - identification of parasites in tissue sections is possible, though it can be challenging to identify past class level. In most cases you need serial sections - this may still not provide a definitive identification to species, and as John suggested above, may require PCR. DNA extraction from formalin fixed tissues is possible, but more challenging than from fresh tissue. Also, collecting tissue samples in formalin may also cause ectoparasites to drop off - you could get around this by collecting wet skin scrapings/gill biopsies first (in any case, it is often easier to identify parasites in wet mounts as they move).

As Amina mentioned above, examination of tissue sections for lesions is also useful. However, many tissue changes in fishes are of questionable significance, such as eosinophilic droplets in renal tubules, and McKnight cells in the intestinal mucosa (particularly if you are undertaking ecotoxicological assessment). To be able to determine what is a true pathological change would require baseline information on 'normal' in your target species, so normal changes are not misinterpreted. Wolf et al (2015) published a particularly useful paper on this titled "Nonlesions, misdiagnoses, missed diagnoses, and other interpretive challenges in fish histopathology studies: a guide for investigators, authors, reviewers, and readers" in Toxicologic Pathology.

There was a lot of work done using fish health as a measure of pollution in the 1980s and 1990s. Dr Jan Thulin was heavily involved in that in Europe. Use Google Scholar to search on Thulin,  pollution and fish. You should get some useful papers either by him or quoting his work.