We thank Drs. Robson and Leung for noting the low incidence of HUS after infection with E. coli 0157:H7 during an outbreak in Missouri. Of the 243 persons who acquired E. coli 0157:H7 infection, 2 developed illness compatible with HUS, defined as microangiopathic hemolytic anemia, acute renal failure (creatinine level >2.0 mg/dL or >50% above baseline for children), and thrombocytopenia (platelet count <150 x 10⁹/L). A careful review of the 32 hospitalized patients' medical records did not show any additional persons who had a diagnosis compatible with HUS. It is not surprising that the incidence of the syndrome was low because the outbreak was community-wide and affected all age groups. The incidence of HUS after infection with E. coli 0157:H7 is significantly higher among infants, elderly persons, and debilitated patients [1], which were the populations affected in the three outbreaks mentioned by Drs. Robson and Leung [2-4]. In the nursing home outbreak they cite, 12 cases of HUS occurred among 55 ill residents (incidence, 22%), but no cases occurred among 18 ill staff members [2]. In the outbreak they cite in handicapped persons, 8 of 20 ill residents (incidence, 40%), but none of 31 ill employees, developed the syndrome [3].

Other factors may also be responsible for the low incidence. A high proportion of patients (65%) reported having nonbloody diarrhea; these patients may have been less likely to develop HUS than those with bloody diarrhea [1]. For example, in the nursing home outbreak cited, 12 of 41 patients (29%) with bloody diarrhea but none of 14 persons with nonbloody diarrhea developed the syndrome [2]. The small number of cases may also relate to the size of the inoculum, which, by analogy with other water-borne outbreaks, may have been low. In addition, strains of E. coli 0157:H7 may differ in their propensity to cause HUS. Those producing both Shiga-like toxins I and II (as in the Missouri outbreak) may be less likely to cause HUS than those producing only Shiga-like toxin II [1].

In Missouri, many patients never sought medical attention, so those with asymptomatic or mildly symptomatic HUS may not have been identified. Similarly, some hospitalized patients may also have had undetected, mild HUS because blood smears, renal function tests, and platelet counts were not done for all patients. Interestingly, several patients may have had "partial HUS" [1], with isolated elevations in creatinine level, decreased hematocrit values, or decreased platelet counts.

We agree with Drs. Robson and Leung that the possibility of development of HUS should be considered for all persons with E. coli 0157:H7 infection. Although routine measurement of laboratory indices would increase the rate of diagnosis of mild or partial HUS, additional study is needed to determine whether early diagnosis would modify prognosis. In addition, measurement of laboratory variables in patients with E. coli 0157:H7 infection would be of interest to determine if long-term renal function is affected in persons with mild or partial HUS, as well as in those with the full-blown syndrome [5].