First, this report shows that an import restriction requiring certificates that animals are not carrying Salmonella is an effective public health measure for protecting the general population from RAS. Sweden noted a dramatic increase in the number of reported cases of RAS in 1996. No changes in the reporting system were made that could explain the increase in reported cases, making it likely that synchronization of import rules of animals and reptiles within the EU in March 1996 caused the sudden increase in the number of RAS cases in Sweden that year.

Secondly, our report indicates that actively informing the public about RAS can cause a decrease in cases. Immediately after the harmonization of rules, the availability of reptiles in pet stores increased, and many parents gave their children a pet turtle, lizard, or snake. The exact number of animals sold is unknown. No information about the risk of contracting salmonellosis was given to the families who bought the animals. In the United States, the Association of Reptilian and Amphibian Veterinarians (ARAV) produced a client education handout with basic facts on how to avoid transmission of Salmonella bacteria from reptiles to humans (13). The Centers for Disease Control and Prevention recommends that children under the age of 5 should avoid contact with reptiles and that households with children <1 year should not own reptiles.

Information about the risk of contracting salmonellosis from pet reptiles was communicated to the public from late 1997 and onwards, and the information given to the public in Sweden has closely followed ARAV guidelines. After the information activities began, the number of reported RAS cases decreased significantly in subsequent years. The decrease might have been even greater since it is likely that more patients with diarrhea would have informed their doctors of reptile contact once they knew of the association between reptiles and salmonellosis.

Third, our study demonstrates that RAS poses a threat to human health that cannot be ignored. The true number of such cases is hard to estimate. We have little hard data on the proportion of all Salmonella cases in the community being diagnosed, although according to conservative estimates, ≈90% of cases remain undiagnosed (11). The Swedish system, with its dual notifications from clinicians and laboratories, captures >99% of cases actually being diagnosed, and >95% of all diagnosed cases are clinically reported (A. Jansson, unpub. data). We assume that at least 50% of RAS patients have been reported as such, once a Salmonella infection has been diagnosed, for the following reasons: 1). only a few clinical notifications lack information on likely route of transmission, 2) domestically acquired salmonellosis is comparatively rare, and 3) most physicians are aware of this route of transmission. We estimated the true annual number of RAS in Sweden after 1996 to be ≈1,000 cases per year, or >10 cases per 100,000 inhabitants. The same rate applied in the United States would correspond to 30,000 to 40,000 cases per year. This number is substantially less than the 93,000 cases per year estimated by the Centers for Disease Control and Prevention (14), which hypothesized that 7% of all U.S. cases of salmonellosis were RAS.

On the basis of these assumptions, we estimated the true annual number of RAS in Sweden after 1996 would be ≈1,000 cases per year, or >10 cases per 100,000 inhabitants. This estimate should correspond to 30,000–40,000 cases in the United States, which is less than the estimated 93,000 RAS cases per year in the United States, accounting for 7% of all cases of salmonellosis (14).

The incidence of RAS in a country is naturally dependent on the magnitude of reptile imports. We therefore tried to obtain some estimates of the numbers of imported reptiles in the years under study. This attempt was unsuccessful because the responsible national authority (the Swedish Board of Agriculture) has no figures on the number of imported animals, only records on the importer. Furthermore, according to direct private import rules, anyone can legally bring <3 animals into the country without any registration. Although we have no indication that the numbers of imported reptiles decreased in the later years (rather, we think the opposite is true), we cannot rule out that some of the decrease in RAS cases was due to decreased imports.

Fourth, we obtained information on the epidemiology of RAS that is likely to hold true for other Western countries as well. Children are the most affected age group, with an incidence of 1.3/100,000 inhabitants. Those patients who had acquired salmonellosis from a turtle were younger than patients who acquired it from a snake or lizard, a reflection of the age groups for which turtles are bought as pets. Young boys 5 to 9 years of age who had contracted the disease from a turtle were shown to be at particular risk in this study.

In the United States during the 1970s, small pet turtles were a major source of RAS infections and accounted for 14% of all cases of salmonellosis in children <10 years of age. This fact was the reason for the ban on commercial distribution of pet turtles with a carapace length <4 inches. This ban has prevented an estimated 100,000 RAS cases among children in the United States yearly (14,15). Nevertheless, the incidence of RAS in the United States has been increasing because of an increase in pet reptile owners; experts estimate that the number of pet reptiles owners doubled from 1991 to 2001 (3).

In most cases included in this study, patients experienced gastrointestinal infection with symptoms severe enough to seek medical care; some even required hospitalization. Many of the serotypes isolated from the Swedish RAS patients were serotypes also commonly found in foodborne Salmonella infection. S. serotype Enteritidis and S. Typhimurium, the 2 most common serotypes found in humans in Europe and the United States (16,17), accounted for 33% of all RAS cases, while S. Subspecies III, known to be reptile-associated, caused 9% of cases. Some serotypes such as S. Abaetetuba, S. Bonn, S. Lome, S. Matadi, S. Nima, S. Shubra, S. Spartel, S. Tamberma, S. Veijle, and S. Windemere were only found in RAS cases and could therefore be assumed to be more reptile-associated than other serotypes. Previously known reptile-associated serotypes, including S. Poona, S. Stanley, S. Pomona, and S. Java, were also isolated from Swedish RAS patients (15). Some serotypes may be more species-specific; 24 of the 51 isolated serotypes were isolated from only 1 kind of reptile (turtle, snake, or lizard). In most of these cases, the numbers were too low to allow for any firm conclusions. However, S. Litchfield, S. Saintpaul, and S. Stanley seem to be associated with turtles and S. Adelaide with snakes. The wide variety of Salmonella serotypes in persons with RAS demonstrates that reptiles are well adapted to Salmonella and could harbor and transmit many different serotypes, which confirms that a risk is always involved in handling these kinds of animals.

Having the same close contact with pet reptiles and turtles as with cats and dogs, or even just keeping them in the private home, increases the risk for transmission of Salmonella bacteria from animals to humans. Pet stores, veterinarians, and other appropriate sources should provide better information on how to avoid these risks, such as the ARAV guidelines, to all prospective buyers of turtles and other reptiles.