In South America several larger urban areas have successfully eliminated rabies through legislation, education, and mass vaccination of dogs (10). Cities in poorer countries such as Bolivia, however, lag behind in control efforts, in large part because resources are scarce, and programs are poorly focused. In these situations, control efforts and resources must be more focused. For this, knowledge of risk factors for canine rabies in urban settings is needed to assess the danger to public health.

We analyzed 26 years of laboratory data on rabies diagnosis in dogs in the city of Santa Cruz, Bolivia, where rabies is endemic. We interpreted the results in light of possible biases to determine risk factors and temporal trends for rabies in the general dog population.

The equation of the regression line was fitted only from 1978 because the low number of tests from 1972 through 1978 was uncharacteristic of the rest of the data. The number of monthly cases dropped significantly after 1978 (b = 0.04: t-test for slope= -4.53, 239 degrees of freedom [df], p<0.001) (Figure 2).

The cyclicity of rabies incidence suggested by Figure 1 is clearly shown by the 24-month centered moving average on cases per month in Figure 2. Since 1972, the mean number positive and mean percentage positive by month have shown no significant variation (F-test= 0.47, p=0.92).

The data strongly suggested a 5- to 6-year cyclicity of rabies incidence; this cycle was most clearly apparent with a centered moving average of number of cases diagnosed per month. The cyclicity was independent of any changes in control measures and might explain the recent downturn in rabies cases. Some previous studies have reported a cyclical nature of rabies incidence (14,16), but this feature was not noted in an urban study in Delhi (17). Cyclicity is usually explained by increasing numbers of young, susceptible, unrestrained dogs in a population with low vaccine coverage. These factors lead to a drop in herd immunity, allowing rapid spread of the disease (6,16). However, even if underreporting is taken into account, the canine deaths from rabies in an epidemic year are unlikely to have substantially affected the number of susceptibles. A reduction in susceptibles may also be due to dogs’ becoming immune after recovering from clinical or inapparent infections. Recovery of dogs from rabies is well documented (18–20); one study showed 20% of experimentally infected dogs recovered (21). There is also serologic evidence that almost 20% of unvaccinated dogs in Thailand have been exposed to rabies (22).

Gender does seem to be a risk factor for sample positivity: male dogs had a significantly higher percentage of samples diagnosed positive (OR 1.14). This increased risk may be explained by males’ fighting over females. Studies in Mexico and India (6,17) show higher numbers of male dogs being affected, though the authors concluded that gender was not a risk factor, perhaps because of low numbers studied.

Rabies was not evenly distributed in Santa Cruz. The percentage of positive samples increased significantly with distance from the city center and as socioeconomic status dropped. In addition, a higher number of positives was reported from beyond the fourth ring-road, despite a similar-sized dog population. Some of the increase in percentage positive (but not in number positive) as distance increased from the center may be due to reporting bias. Fewer samples were submitted from outside the fourth ring, and these samples probably included a higher proportion of dogs that showed specific signs of rabies. An association of increased risk for rabies and low socioeconomic status has also been shown in Mexico (6). Lower vaccination coverage and increased densities of unrestrained dogs have previously been reported to be associated with poorer urban areas (23), a characteristic also shown in a recent survey of the canine population in Santa Cruz (LIDIVET, unpub. data).

Age was a clear risk factor for sample positivity in our study. The median age of a dog that tested positive for rabies was up to 1 year, as found in Mexico (6). The age group most at risk for testing positive for rabies, however, was 1- to 2-year-old dogs (OR=1.73). Dogs 3 months to 1 year of age were at intermediate risk (OR=1.49); however, this risk for rabies in dogs up to 1 year old may be underestimated. Perhaps because of the die-off of dogs of that age from all causes and the relative ease of carrying a puppy to the laboratory, the proportion of submissions (68% of all samples) from dogs <1 year was high, even relative to the population (37% of all dogs). This disproportion led to the finding of more positive samples but also to less specific reporting, with proportionally more nonrabid dogs with vague symptoms; such dogs would not have been submitted had they been older. This discrepancy may have decreased the percentage positive and thus underestimated the comparative risk for rabies in dogs <1 year old. It is nonetheless plausible that the risk for contracting rabies in dogs <1 year is lower than for 1- to 2-year-old dogs. Although dogs <1 year are less likely to have been vaccinated, sexually immature dogs are also less likely to roam, interact, and fight with other dogs. Puppies <3 months old may also benefit from passive immunity from their mothers. The large population of puppies, however, and their increased contact with children and adults make them a particular public health risk. The decreasing odds of sample positivity after 2 years of age may be due to increased likelihood of vaccination and less fighting among older dogs. Older dogs are also more likely to be owned.