Incidence for rotavirus and intussusception was estimated from standardized discharge data for all Hong Kong government hospitals (1997 to 1999). Intussusception incidence in infants (78 to 100 of 100 000) was relatively high. The distinct winter seasonality of rotavirus was not evident for intussusception. During the first 5 years of life an estimated 1 child of 28 is admitted with rotavirus infection (4% of all medical admissions).
With the licensing and planned routine use of the oral tetravalent rotavirus vaccine in the United States, it was anticipated that rotavirus diarrhea would soon be preventable in children. However, during the first large scale rotavirus vaccination program in the United States, an association with intussusception was identified, and the vaccine was withdrawn. 1,2 These events have raised serious questions about the future development of live oral rotavirus vaccines. To provide data for use in evaluating the risks and benefits of future rotavirus vaccines, a WHO meeting recommended that studies were needed to assess the disease burden of rotavirus diarrhea and the epidemiology of intussusception. 3 The 7 million people of Hong Kong have access to a comprehensive publicly funded government hospital system [Hospital Authority (HA)], and since 1997 standardized and computerized discharge data have been available. We used this new data source to estimate the incidence of rotavirus diarrhea and intussusception among hospitalized children in Hong Kong during a 2-year period from July 1, 1997 to June 30, 1999.
In 1997 the Clinical Management System (CMS) was introduced to collect uniform discharge and other information on all patients admitted to HA hospitals throughout Hong Kong. Information collected by using the CMS include patient identifiers, date of birth, sex, a maximum of 15 diagnoses and 15 procedures [classified by International Classification of Diseases (ICD), 9th revision, Clinical Modification code] and admission and discharge dates. Two databases from the CMS were analyzed: (1) all hospitalized pediatric medical patients admitted from July 1, 1997 to June 30, 1999; and (2) all patients of any age admitted to any ward (medical or surgical) with any diagnosis coded as ICD 560.0 (intussusception) or any procedure coded as ICD 46.80 (reduction of intussusception) during the same period.
The database of pediatric medical admissions was used to assess admissions for rotavirus and other causes of diarrhea. ICD 9th revision Clinical Modification codes were used to define diarrhea-associated hospitalizations: bacterial diarrheas (001 to 005, excluding 003.2, and 008.0 to 008.5); parasitic diarrheas (006 to 007, excluding 006.2 to 006.6); rotavirus diarrhea (008.61); other viral diarrheas (008.6 to 008.8); diarrhea of undetermined etiology, including that presumed to be infectious (009.0 to 009.3); and other noninfectious diarrheas (558.9 and 787.91). The ICD codes entered into the CMS are selected by the responsible medical officer at the time of the patient’s discharge. The CMS allows the doctor to type in the ICD code directly if known or to use a keyword search. The ICD codes entered may thus be influenced by a number of factors including the availability of laboratory results at the time of discharge.
The second database of all patients with intussusception or reduction of intussusception was analyzed to determine the number of children younger than the age of 5 years with intussusception. Those admissions with equivalent dates of birth were checked to see whether these represented readmissions of the same patient. Annual birth data were obtained from the Census Department of the Hong Kong Government.
During the 2-year study period 106 919 children younger than the age of 5 years were admitted to medical pediatric wards. A primary diagnosis indicating a diarrhea-associated admission was noted in 11.5% (12 257 of 106 919). A primary or secondary diagnosis indicating a diarrhea-associated admission was noted in 13.8%. The percentage of all discharges with a primary or secondary diarrhea-associated code ranged from 10.2 to 19.2% among the 12 HA hospitals. Four deaths were recorded in these diarrhea-associated hospitalizations, but in none was diarrhea coded as the primary diagnosis.
There were 1270 children with a primary diagnosis of rotavirus disease and 1607 with a primary or secondary diagnosis of rotavirus disease; 98% (1607 of 1639) of all rotavirus-associated admissions (primary or secondary coding) were younger than the age of 5 years. The percentage of those diarrhea-associated admissions coded as rotavirus varied considerably among the 12 hospitals from 0% for one hospital that never tested and reported rotavirus to 27.9% for another hospital where diarrhea admissions were routinely tested for rotavirus. Children with rotavirus-associated admissions had a median stay in hospital of 4 days (interquartile range, 2 to 6 days). Rotavirus was coded for one of the diarrhea deaths, with a primary diagnosis of congenital abnormalities and a hospital stay of >3 months, suggesting nosocomial rotavirus infection.
A total of 279 patients were coded with a discharge diagnosis of intussusception or a procedure code to reduce intussusception. Of these 222 (80%) were younger than 5 years of age. Thirty-two of these records were readmissions: 2 children had 4 admissions; 5 had 3 admissions; and 16 had 2 admissions. For 7 patients the apparent readmissions were the result of interhospital transfer. For 5 patients the readmissions were separated by >2 months, suggesting possible recurrent episodes of intussusception. Of these 190 individual children younger than the age of 5 years, 99 (52%) were in the age group 0 to 11 months (Fig. 1) and 58% were male.
Patients with rotavirus-associated diarrhea and intussusception did not differ substantially in their age distribution, but they did differ in seasonality, with intussusception showing no distinct seasonality and rotavirus showing a winter peak (Figs. 1 and 2). For the 190 children with intussusception, 44 (23%) had procedure codes indicating surgical intervention (bowel resection, laparotomy), 17 (9%) barium or other enema and 97 (51%) reduction of intussusception without mention of method (i.e. ICD 46.80). The median hospital stay for intussusception was 4 days (interquartile range, 3 to 7 days). The one death, coded only as intussusception, occurred on the day of admission, and the only procedure coded was an ultrasound examination.
Incidence rates per 100 000 for diarrhea coded as rotavirus (primary or secondary diagnosis) and intussusception were estimated for 1997 and 1998 using total Hong Kong births as denominators (Table 1). Based on the diarrhea admissions for July, 1998, to June, 1999 (7538 total diarrhea and 881 rotavirus), the cumulative risk of hospitalization for diarrhea by age 5 years was 1 of 8 (7538 per 63 160) for diarrhea and 1 of 72 (881 per 63 160) for rotavirus. If we assume that with active surveillance 27.9% of diarrhea admissions (hospital with the highest percentage) rather than 10.9% of admissions (mean percentage of all 12 HA hospitals) are caused by rotavirus, then the cumulative risk of hospitalization for rotavirus disease by age 5 years could be as high as 1 of 28 (2255 per 63 160). These rotavirus-associated admission rates do no account for possible readmissions during the same episode of illness.
In many countries incidence data are hard to obtain because denominator populations are not available to match numerators for diseases such as hospitalizations. Our data emphasize the importance in Hong Kong of diarrheal diseases, which accounted for 11.5% of all admissions under the age of 5 years. Although 10.9% of diarrhea-associated admissions (primary or secondary codes) were specifically coded as being caused by rotavirus, this figure is clearly an underestimate considering both the large difference in the percentage of diarrhea admissions classified as being caused by rotavirus in different hospitals (0 to 27.9%) and the higher prevalence of rotavirus (∼30%) identified previously in studies from individual hospitals that performed rotavirus screening for all children with diarrhea. 4, 5 The range in rotavirus coding among these Hong Kong hospitals reflects differences in the availability of rotavirus testing with 5 of the 12 hospitals noting that testing was performed only for selected diarrhea admissions. Despite this significant underreporting, our rates of hospitalization for rotavirus infection in Hong Kong (1.7 to 2.8 per 1000 children younger than 5 years and 4.4 to 6.1 of 1000 children younger than the age of 1 year) are comparable with those reported in other industrialized countries, e.g. Sweden (3.7 per 1000), 6 England and Wales (5 per 1000) and Denmark (4.8 per 1000) children younger than 5 years old per year. In the United States 1 of 78 children is hospitalized for rotavirus diarrhea. 7 The cumulative risk of hospitalization for rotavirus diarrhea by age of 5 years was estimated to be 1 of 71 to 1 of 117 for Hong Kong, but with active rotavirus surveillance it could be as high as 1 of 28, i.e. 4% of hospitalizations of children younger than 5 years old.
The estimated rate of hospitalization for intussusception excludes admissions to private hospitals, which accounted for 13% of patients (of all ages) with ICD-9 code 560 (intestinal obstruction without mention of hernia) (HA Statistical Report, 1998/1999). However, even without the private hospital data, our rates for intussusception for infants younger than 1 year of age are high compared with the United States ((18 to 56 per 100 000), 8, 9 and the United Kingdom (66 per 100 000). 10 However, intussusception hospitalization rates in the United States vary by race, being highest among infants of other races (112 to 217), than in blacks (32 to 50) and whites (27 to 35). 8
The age distribution of intussusception- and rotavirus-associated hospitalizations shown in Figure 1 suggests that the incidence of intussusception cases peaks from 6 to 9 months and continues throughout the first 5 years of age. This peak is similar to those noted in the United Kingdom (3 to 7 months), 10 and the United States (5 to 7 months). 8 In contrast the rotavirus admissions appeared to have a wider peak from ∼4 to 16 months. As noted in other studies, 8 intussusception admissions demonstrated a seasonality distinct from those of rotavirus admissions. Although hospitalizations for both intussusception and rotavirus peaked in the winter months of December and January, a second peak of intussusception admissions was observed during May through July, when the incidence of rotavirus admissions was low.
Our data confirm that rotavirus is an important cause of hospitalization of children younger than the age of 5 years in Hong Kong. However, routinely collected HA hospital discharge data are likely to underestimate the true disease burden of rotavirus diarrhea in this population, emphasizing the importance of active surveillance. Intussusception incidence, despite underestimation because of the absence of data from private hospitals, was relatively high compared with other industrialized countries. Studies to evaluate the risk benefits of future rotavirus vaccines should consider regional variations in disease incidence.
We thank the Hospital Authority Head Office for providing data from the Clinical Management System and for estimating the percentage of non-Hong Kong-born children admitted with diarrhea and Dr. Fung Hong for helpful comments and advice.
1. Murphy TV, Gargiullo PM, Massoudi MS, et al. Intussusception
among infants given an oral rotavirus
vaccine. N Engl J Med 2001; 344: 564–72.
2. Anonymous. Intussusception
among recipients of rotavirus
vaccine: United States, 1998–1999. MMWR 1999; 48: 577–81.
3. Melton L. Lifesaving vaccine caught in an ethical minefield. Lancet 2000; 356: 318.
4. Chan PK, Tam JS, Nelson EA, et al. Rotavirus
infection in Hong Kong: epidemiology and estimates of disease burden. Epidemiol Infect 1998; 120: 321–5.
5. Biswas R, Lyon DJ, Nelson EA, Lau D, Lewindon PJ. Aetiology of acute diarrhoea in hospitalized children in Hong Kong. Trop Med Int Health 1996; 1: 679–83.
6. Johansen K, Bennet R, Bondesson K, et al. Incidence and estimates of the disease burden of rotavirus
in Sweden. Acta Paediatr 1999; 88 (Suppl): 20–3.
7. Anonymous. Rotavirus
vaccine for the prevention of rotavirus
gastroenteritis among children: recommendations of the Advisory Committee on Immunization Practices (ACIP). MMWR 1999; 48: 1–20.
8. Parashar UD, Holman RC, Cummings KC, et al. Trends in intussusception
-associated hospitalizations and deaths among US infants. Pediatrics 2000; 106: 1413–21.
9. Chang HG, Smith PF, Ackelsberg J, Morse DL, Glass RI. Intussusception
, rotavirus diarrhea
, and rotavirus
vaccine use among children in New York state. Pediatrics 2001; 108: 54–60.
10. Gay N, Ramsay M, Waight P. Rotavirus
vaccination and intussusception
. Lancet 1999; 354: 956.