Studies that look at risk factors can be pretty variable in terms of what they tell you, the impact they have and how accurate they are.

Some findings are pretty logical, clear and indicate something that should be done.

• Smoking is a risk factor for [insert many diseases here], so to reduce the risk of [whatever disease], stop smoking.

Others make sense but don’t necessarily lend themselves to an effective intervention.

• Being male is a risk factor for cardiovascular disease… not much I can do about that.

Sometimes, you have to remember that a risk factor for one thing doesn’t provide a clear answer when a broader context is considered.

• Moderate consumption of red wine can reduce the risk of various conditions, but alcohol consumption can also increase the risk of other conditions.

Sometimes, how the study is designed and performed can really affect the results.

• If I did a large study of the general population in Guelph, I could presumably show that going to a hospital greatly increases your risk of death. Does that mean you shouldn’t go to the hospital? No, because I could presumably also show that if you have chest pain and go to a hospital, you’re more likely to live. Knowing the study population and what question is really being asked are critical.

Sometimes, something that’s found to be a risk factor isn’t really the risk factor, but it’s associated with something else that is.

Sometimes, something can be "statistically significant," but of limited consequence.

• If doing something increases the risk significantly, but only by 0.0001%, does that mean anything?

Why do I write this? Because these are some of the things that we have to think about when assessing risk factor studies. While one Toronto radio station loves to give 10 second snippets on some new risk factor medical study, you can’t determine much about the study itself from a sound-bite (or internet post). You need to think about the details regarding how the study was done. Nevertheless, risk factor studies can provide useful information, but consider the results carefully, whether they are relevant, whether they indicate changes need to be made or whether they indicate that we need to look at the issue further.

Anyway, this rambling post was prompted by a couple of papers looking at risk factors the leptospirosis in dogs (Hennebelle et al J Am Vet Med Assoc 2013 and Raghavan et al Prev Vet Med 2012)

The first study compared dogs from northern California that had or didn’t have leptospirosis. They found a few things:

• There were differences in geographic distribution of the lepto cases and controls. That makes sense since we know lepto varies regionally, but living in different areas might also be associated with different behaviours and contacts (e.g. wildlife contacts).
• There was a temporal cluster, with more cases occurring between May 2003 and May 2004, compared to the rest of the 2001-2010 study period. That makes sense too since we see variation in cases within and between years.
• These results don’t change anything, but are an indication of what work needs to be done next. Looking at why things vary geographically and temporally might be important for figuring out how to reduce the risk of disease. It also indicates regions where more efforts to educate pet owners (and veterinarians) are indicated, and where vaccination is more important.

The second study looked at dogs from Kansas and Nebraska, with and without leptospirosis. They also found a few risk factors.

• Lepto was more common in houses lacking complete plumbing facilities. Presumably, this is a proxy for something else. Poor plumbing doesn’t likely result in lepto in dogs. Rather, it presumably means that a dog living in a house with poor plumbing has some other factor that increases its risk. For example, incomplete plumbing may be more likely in lower socioeconomic (i.e. lower income) households, which might then correspond to other more direct risk factors for the dog (e.g. poorer nutrition, less veterinary care). It could also be that houses lacking complete plumbing tend to be in a different area where there’s more exposure to wildlife reservoirs. A couple of other indicators of poverty status were also significant, highlighting the potential impact of owner poverty on pet health.
• Dogs that lived within 2500 m of a university or college, or a park, were also at increased risk. The park risk factor makes sense since they could be exposed to sites infected by wildlife reservoirs (e.g. raccoons). Living close to a university or college is tougher to figure out. Maybe it’s associated with economy, as students are typically at lower income levels. Maybe it’s because colleges and universities usually have lots of green space that might harbor wildlife.

So, these studies tell us some new information, reinforce some previous knowledge (or perceptions) and raise some new questions that we need to answer. By themselves, they won’t result in major changes in how we try to prevent lepto in dogs, but little steps is typically how science progresses.