This answer will be updated if and when I find more information. Please note that I'm not a bearing expert. If anyone has supplemental information, I'd welcome hearing about it in comments or in another answer!
The bearing dimensions are the first obvious parameter you should pay attention to, just to make sure you buy the correct ones for your wheel. The first video that @Greg posted discusses bearing sizes, which are denoted by the numbers in the bearing model. An interesting point: 69103 bearings are apparently equivalent to 6903 bearings, i.e. the bearing outer diameter (30mm), inner diameter (17mm), and depth (7mm) are identical. In fact, someone with Fulcrum wheels asked this on a forum and was told that they are identical. This may aid you in finding bearings from different manufacturers, although my impression is that SKF is generally a good manufacturer.
In summary, most cartridge bearings have seals on one or both sides. For winter riding, you want to make sure your bearings seals are good, preferably full contact. Determining what is a good full contact seal may be difficult, however. To some extent, you can trade sealing for friction, and this tradeoff clearly matters if you are building a bike specifically to challenge the hour record. For practical use, the watts saved aren't worth the risk of damage due to contamination or the additional drag. Even for a pro racer, poor sealing will lead to additional contamination, which could cause greater drag.
All bearings have seals to keep contaminants out of the bearing, and the grease inside it. Generally, going from least sealing to best, there are non-contact bearing shields, low-friction and non-contact rubber seals, and full-contact rubber seals. Bearing shields do not touch the races of the bearing (i.e. parts that touch the bearing bore and the axle), and they offer the least friction. The full-contact seals offer the greatest friction. That said, the friction gained versus a low-friction seal could be something on the order of magnitude of 1-3W per hub. This is a very small difference. I am basing this off a podcast by Friction Facts (podcast #8, I believe), and I will cite this when I can retrieve the source. Phil Wood's site explains that they have two bearing types: standard bearings with full contact seals, and another bearing for races and fair weather than has a partial contact seal. A page by SKF briefly describes the types of sealing available for bike bearings.
You can sometimes deduce the type of bearing seal using the letters after the bearing's size code. However, the codes may be unique to each bearing manufacturer. Adam Kerin of Zero Friction Cycling says that the suffix -ZZ almost always denotes a non-contact shield (see the paragraph at the very end of the linked PDF), but you don't want this type of bearing anyway as it doesn't offer sufficient protection. Post 16, by Josh Poertner of Silca, on this Slowtwitch.com thread appears to corroborate this also.
Bearings can also be sealed on one or both sides. A 6903-1RS bearing would have a rubber seal on one side. I believe the other side is open to the elements. In contrast, 6903-2RS bearings have rubber seals on both sides. I think that 1-RS bearings should have lower friction than bearings sealed on both sides. Enduro says that they offer some bearings with a full contact seal on one side and a medium-contact seal on the other sided. This is a possible compromise between friction and protection from both sides. As Kerin (link above) discusses, the exact type of seal can vary considerably from manufacturer to manufacturer, but he feels (based on experience) that -2RS shields from cycling manufacturers are likely to be low-contact seals, but among generic bearing manufacturers, -2RS might be anything.
From some sleuthing, it appears that Fulcrum's freehubs may have been equipped with -1RS seals. This YouTube video by Cycling Weekly offers a guide for servicing Campagnolo and Fulcrum freehubs. Starting at 6:12, the presenter confirms that the two cartridge bearings in the freehub (at least the one he serviced) are sealed on one side only. He says that he doesn't see the point in sealing on the other side, as it's not exposed to dirt ingress.
I am not positive about this. Say you get water inside the freehub. Depending on how you move the bike around, it might then intrude into the bearings from the unshielded side. I don't know how big an issue this could be, and I'd welcome other users' comments. In this thread on the BikeRadar forum, one commenter (Cyclobob) basically said what I did (water will eventually ingress through the unprotected side).
You might check the seal types on the bearings you ordered. The previous version of this answer said that I would be surprised if you had got -1RS bearings to begin with. I may have overstated things there! Nonetheless, if you have been getting -1RS bearings, then considering that you want to keep riding in the winter, it might be worth seeing if -2RS bearings improve durability. If you got bearings with low-friction seals, then you definitely want to find a full-contact seal.
This recorded conversation in the newsgroup rec.bicycles.tech (credit Sheldon Brown's site) alludes to some bearings with double contact seals or labyrinth seals. I have not yet found a cartridge bearing designation that specifically says they have these types of seal. It may be that there is not enough room in the common cartridge bearings to place seals like these and to maintain sufficient ball size to bear the loads. Or it may just not be common practice yet.
Grease in the bearings
Bearings are filled with grease. They may vary in the amount of fill. Less grease may produce less friction up to a point, but this will probably protect the bearings less. Phil Wood's site says that their fair weather racing bearings are 30% filled, and their regular bearings are 100% filled. On Slowtwitch.com, Josh Poertner posted that typical industrial bearings are 20-25% filled because they spin very fast and will eject grease if filled more. In contrast, bike bearings need a higher fill proportion, which adds miniscule drag but which prevents contaminant intrusion. He did not specify how much, but maybe more is better. If you over-fill the bearing, presumably it would eject the grease and you'd need to clean it up.
I'm not seeing fill rates specified on most of the bearing sites I've Googled, so I'm not clear how consumers might obtain this information. Nor is it clear what an optimal fill rate would be.
Note that it's possible to re-grease cartridge bearings. This involves picking one of the rubber seals out, cleaning out the grease, and injecting new grease. Do be careful not to damage the seal, but I believe careful use of a pick tool, or perhaps a pair of tweezers, would be sufficient. This Global Cycling Network YouTube vid demonstrates how to do this. They extracted the bearings from the hub before servicing. This vid by November Bicycles makes it clear that you may not need to remove the bearings, but it may depend on where they're situated. You could try re-greasing your bearings, perhaps halfway through winter.
One industry development that might be worth paying attention to is encasing the bearings in an oil-soaked polymer. SKF appears to have pioneered this design for use in food services, but it may be starting to move to bikes. As far as I can tell, the bearings are encased in a polymer that takes up all the free space inside the cartridge (recall: we would otherwise fill part of this space with grease). SKF argues that the polymer insert plus the external seals mean that contaminants physically can't intrude into the cartridge bearing. Normally, if you pressure wash the bike, the water might force the lip of the seal inside the bearing, but they think that there simply isn't any room for that to happen with this design. These bearings are cheaper than ceramic bearings, but are still fairly expensive. This development may be worth watching.
Manufacturing tolerances for the bearing could be an issue. For example, your 61903 / 6903 bearing has an outer diameter of 30mm. Really, this is 30.0 +/- x millimeters. If the bearing is too big for the hub's bore, this could be an issue. Most likely, some manufacturers have worse tolerances than others. However, I don't think there's a way for consumers to know this information. About the only thing you have to go on is general reputation. For example, SKF and Phil Wood have (as far as I know) good reputations.
Also, the tolerances of the hubs could be an issue. If the bearing seats aren't quite parallel, or if they're misaligned, this could cause your bearings to wear out faster. Again, I don't see a way to tell as a consumer.
Theoretically, a bike store could measure bearing sizes with an accurate pair of calipers as they come in and keep logs, but how many stores have time to do that? Relatedly, I think that Hambini has a video where he compares the variance in bottom bracket shell diameter among small samples of some common bike manufacturers, but again, it is hard to get a big enough sample as a consumer.
Bearing roundness (ABEC scale) - don't worry about this
You may see some bearings that are specified as ABEC-5, for example these units sold by Jenson USA (link only for example purposes). The ABEC scale is an ordinal rating of how round the balls are. The grades go 1, 3, 5, 7, and 9. My impression is that this is only one of a number of important characteristics for bearings in general. I think bike bearings may usually be grade 3 or 5, and that higher grades may offer no advantage for bikes. Our bearings don't rotate as fast as industrial ones). This online retailer and this article by Bike Radar seems to align with that point.
Bearing loads - this is a function of the hub design, and you can't change this post hoc
Last, the feedback I've got from mechanics and wheelbuilders is that small bearings will wear out faster. Adam Kerin (same link as before) appears to corroborate this. A bearing needs adequate size to bear the load. This isn't an issue with your hub, but I've had an ultralight front hub with 688 bearings. Kerin specifically calls them out as inadequate, and indeed, my hub bearings wore out very fast. You can't change the size of the bearings in the hub, and 6903 is a pretty common size for a rear hub anyway. However, where they're placed in the hub might make a difference, and also how many there are.