PEDIATRICS Vol. 104 No. 3
September 1999,
pp. 489-494
Use of State Hospital Discharge Data to Assess the Morbidity From
Rotavirus Diarrhea and to Monitor the Impact of a Rotavirus
Immunization Program: A Pilot Study in Connecticut
Umesh D. Parashar, MBBS, MPH*,
Marc A. Chung, MPH
,
Robert C. Holman, MS§,
Robert W. Ryder, MD,
James L. Hadler, MD, MPH
, and
Roger I. Glass, MD, PhD*
From the * Viral Gastroenteritis Section, Division of Viral and
Rickettsial Diseases, and the § Office of the Director, Division of
Viral and Rickettsial Diseases, Centers for Disease Control and
Prevention, Atlanta, Georgia; the Department of Epidemiology and
Public Health, Yale University, New Haven, Connecticut; and the
Department of Public Health, Hartford, Connecticut.
 |
ABSTRACT |
Objectives. Now that rotavirus
vaccines have been licensed and recommended for routine immunization of
US infants, there is an urgent need for data to assess the morbidity
from rotavirus diarrhea and to monitor the impact of a rotavirus
immunization program. In a pilot study, we have assessed the usefulness
of state hospital discharge data on diarrhea in children to
provide this information by examining data from Connecticut.
Design. Retrospective analysis of discharge records
from acute care, nongovernmental hospitals in Connecticut.
Patients. Children 1 month through 4 years of age with a
diarrhea-associated diagnosis listed on the discharge record.
Setting. Connecticut, 1987 through 1996.
Results. During the 10-year study period, a total of
11 324 diarrhea-associated hospitalizations (49.4 hospitalizations per
10 000 children) were reported. Diarrhea-associated hospitalizations peaked during February through April, especially among children 4 to 35 months of age. The seasonality and age distribution of diarrhea-associated hospitalizations of presumed noninfectious and
viral etiologies resembled those of rotavirus-associated
hospitalizations. During 1993 to 1996, rotavirus was coded for 10.4%
of diarrhea-associated hospitalizations increasing from 8.6% in 1993 to 14.7% in 1996. The unadjusted median cost of a diarrhea-associated
hospitalization during 1987 to 1996 and 1993 to 1996 was $1941 and
$2428, respectively.
Conclusions. Diarrhea causes substantial morbidity in
children from Connecticut. The winter seasonal peak of
diarrhea-associated hospitalizations in children 4 to 35 months of age
coinciding with the peak of rotavirus-specific hospitalizations
suggests that rotavirus is an important contributor to the overall
morbidity. Although our findings suggest incomplete coding of rotavirus
cases, state hospital discharge data should provide sensitive and
timely information to monitor the impact of a rotavirus
immunization program in Connecticut.
Key words:
diarrhea,
rotavirus,
hospitalizations,
children,
cost.
On August 31, 1998, the first rotavirus vaccine was
licensed for immunization of US infants by the Food and Drug
Administration. This vaccine, Rotashield (Wyeth Lederle Vaccines and
Pediatrics), is a live, orally administered preparation that has
been shown to be safe and efficacious in preventing severe rotavirus
diarrhea.1-4 Both the Advisory Committee on Immunization
Practices (unpublished data) and the American Academy of
Pediatrics5 have proposed recommendations for routine use
of 3 doses of this vaccine at 2, 4, and 6 months of age. In view of the
licensure and likely use of rotavirus vaccines in the near future,
there is an urgent need to establish a system to assess the disease
burden of rotavirus diarrhea and to monitor the impact of a rotavirus
immunization program.
To date, many studies of the disease burden of rotavirus diarrhea in
the United States have been based on National Hospital Discharge Survey
(NHDS) data.6-9 These data are easily accessible,
relatively inexpensive, and representative of national
patterns.10,11 Furthermore, the seasonal winter peak of
diarrhea-associated hospitalizations in children 4 to 35 months of age
and its unique geographic pattern of spread clearly reflect rotavirus
activity in the United States. Moreover, hospitalizations are the most
costly and severe events associated with rotavirus diarrhea in US
children, and rotavirus vaccines have greatest efficacy against severe
disease.1-4 Therefore, monitoring the trends of
diarrhea-associated hospitalizations in children using NHDS data should
provide a sensitive indicator of the impact of a US rotavirus
immunization program. However, the NHDS data have some limitations.
First, only 0.5% to 1% of all US hospitals are included in the
sampling frame of the NHDS10,11 making these estimates
subject to bias and less robust for analysis. Second, because there is
an ~2-year delay between the time the data are collected and when
they become available for analysis, the information may not allow a
timely assessment of the impact of rotavirus vaccines. Moreover,
because states may introduce rotavirus vaccines at different times,
each state may wish to assess the impact of immunization in their own
setting.
More than 30 states now collect complete information on hospital
discharges that are accessible in databases. These data could provide
sensitive, complete, inexpensive, and timely information for states to
assess the severe morbidity from rotavirus diarrhea and to monitor the
impact of the new rotavirus vaccines. Connecticut currently maintains
such an on-line database of all hospital discharges in the state; the
system is maintained by the Connecticut Health Information Management
and Exchange (CHIME) program.12 We examined Connecticut
hospital discharge data for the period 1987 through 1996 to assess
whether state hospital discharge data on diarrhea in children could be
used to assess the morbidity from rotavirus diarrhea and to monitor the
impact of rotavirus vaccines after their introduction.
| |
METHODS |
Hospital discharge data for 1987 through 1996 were obtained from
the CHIME data for the fiscal years 1987 through 1997.12 The CHIME program was initiated in 1980 with the collection of systematic data on inpatients from acute-care hospitals in Connecticut. Beginning in 1984, associated financial information also was collected. Data are submitted voluntarily to CHIME by each participating hospital.
Through data exchange or purchasing agreements, CHIME obtains discharge
data from bordering states (New York, Massachusetts, and Rhode Island)
for Connecticut residents treated at border-state hospitals, as well as
for residents of selected border-state towns treated at border-state
hospitals. Patient records are processed through a comprehensive error
identification and correction process and are available for analysis
within 6 to 12 months of their collection.
Children 1 month through 4 years of age with an ICD-9-CM code for
diarrhea listed as 1 of the 15 diagnoses on the discharge record were
included in the analysis.13 The ICD-9-CM codes used
included diarrhea of determined etiology (bacterial [001-005,
008.0-008.5, excluding 003.2], viral [008.6-008.8], and parasitic
[006-007, excluding 006.3-006.6]) and diarrhea of undetermined
etiology including those presumed to be infectious (009.0-009.3) and
noninfectious (558.9, 787.91). Rotavirus-associated hospitalizations
for 1993 through 1996 were defined by the specific ICD-9-CM code 008.61 introduced in October 1992. Neonates (<1 month of age) were excluded
from the analysis, because the ICD-9-CM classifies diarrheal illness in
this age group with distinct neonatal codes.
We examined trends in the children's all-cause diarrhea-associated and
rotavirus-specific hospitalizations by age, race/ethnicity (as provided
by CHIME), season, and etiology. Hospitalization rates (per 10 000
children) were calculated by using the census population of children
<5 years of age.14 For 1990 to 1996, rates for children
by age group (1-11 months and 1-4 years of age) were calculated by
using the Connecticut census population of children that were <1 year
and 1 to 4 years of age in the 1990 census, respectively. For 1987 to
1989, the population of children <1 year and 1 to 4 years of age was
estimated by assuming that infants comprised 20% of all children <5
years of age. RRs with 95% CIs were calculated using Poisson
regression analysis.15 We assessed the economic burden of
diarrhea by examining the median duration and cost of
diarrhea-associated hospitalizations. Comparisons of the duration and
cost of hospitalization by time period were performed by using the
Wilcoxon rank-sum test.16
| |
RESULTS |
A total of 11 324 diarrhea-associated hospitalizations were
reported in children living in Connecticut who were 1 month through 4 years of age during the 10-year study period (Table
1). Most (68.3%) diarrhea-associated
hospitalizations were of unspecified etiology; of the remaining
diarrhea-associated hospitalizations, viruses accounted for 25.3%,
followed by bacteria (5.4%) and parasites (<0.3%). The annual
incidence of diarrhea-associated hospitalizations was 49.4 per 10 000
(ie, the cumulative incidence of diarrhea-associated hospitalizations
over the first 5 years of life was 247 per 10 000 children, or 1 in 40 children was hospitalized with diarrhea by age 5; Table
2). The incidence of diarrhea-associated hospitalizations was greater in infants 1 to 11 months of age than in
children 1 to 4 years of age (RR: 3.0; 95% CI: 2.8-3.1) and was
greater among males than among females (RR: 1.1; 95% CI: 1.1-1.2),
although the RR for the latter comparison was small and may not be
clinically meaningful. Compared with non-Hispanic whites, the incidence
of diarrhea-associated hospitalizations was greater among non-Hispanic
blacks (RR: 1.4; 95% CI: 1.3-1.4), Hispanics (RR: 1.7; 95% CI:
1.6-1.8), and non-Hispanic others (RR: 2.5; 95% CI: 2.3-2.7). A
total of 13 deaths were reported among children hospitalized with
diarrhea for a case-fatality rate of 1 death per 871 hospitalized
children.
View this table:
[in this window]
[in a new window]
|
TABLE 1
Diarrhea-Associated Hospitalizations in Connecticut Children 1 to 59 Months of Age, by Etiology, 1987-1996
|
|
View this table:
[in this window]
[in a new window]
|
TABLE 2
Characteristics of Diarrhea-Associated Hospitalizations in Connecticut
Children 1 to 59 Months of Age, 1987-1996
|
|
For the 4-year period for which rotavirus-specific information was
available (1993-1996), this pathogen was listed on a total of 400 (10.4%) records (Table 1). Discharges coded as rotavirus increased
after the introduction of the rotavirus-specific ICD-9-CM code,
accounting for 8.6%, 10.6%, 8.3%, and 14.7% of all
diarrhea-associated hospitalizations in 1993, 1994, 1995, and 1996, respectively. The annual incidence of rotavirus-associated hospitalizations was 4.4 per 10 000 (Table 2). The incidence of
rotavirus-associated hospitalizations was greater in infants 1 to 11 months of age than in children 1 to 4 years of age (RR: 2.0; 95% CI:
1.6-2.4) but was similar among males and females (RR: 1.1; 95% CI:
1.0-1.4). Compared with non-Hispanic whites, the incidence of
rotavirus-associated hospitalizations was similar among non-Hispanic
blacks (RR: 1.0; 95% CI: 0.7-1.4) and non-Hispanic others (RR: 1.7;
95% CI: 1.0-2.8) but was greater among Hispanics (RR: 2.1; 95% CI:
1.6-2.6). Only 1 death was reported in a child with rotavirus
diarrhea.
The annual incidence of diarrhea-associated hospitalizations in
children 1 month through 4 years of age declined by 39% from 63.0 per
10 000 in 1987 to 38.5 per 10 000 in 1996 (RR: 0.6; 95% CI:
0.6-0.7); over this period, the rate for infants (1-11 months of age)
declined by 53% from 146.0 per 10 000 to 68.8 per 10 000 (RR: 0.6;
95% CI: 0.6-0.7), whereas the rate for children 12 to 59 months of
age declined by 35% from 46.3 per 10 000 to 30 per 10 000 (RR: 0.7;
95% CI: 0.7-0.8; Fig 1). Examination of
diarrhea-associated hospitalizations by month of admission showed that
winter peaks associated with rotavirus in previous studies occurred
each year and were most prominent among children 4 to 35 months of age
(Fig 2).
View larger version (12K):
[in this window]
[in a new window]
|
Fig. 1.
Diarrhea-associated hospitalization rates by year and age group among
children from Connecticut who were 1 through 59 months of age in 1987 through 1996.
|
|
View larger version (15K):
[in this window]
[in a new window]
|
Fig. 2.
Monthly diarrhea-associated hospitalizations by age group among
children from Connecticut who were 1 through 59 months of age in 1987 through 1996.
|
|
We examined the monthly number of diarrhea-associated hospitalizations
during 1993 to 1996 by age group and by etiology to determine age- and
season-related trends (Fig 3). A winter seasonal peak in diarrhea-associated hospitalizations was observed among children 4 to 35 months of age; this peak was not observed in
children 1 to 3 months of age and was less prominent in children 36 to
59 months of age (Fig 3A). By etiology, the seasonal trends of
diarrhea-associated hospitalizations reflected those of
hospitalizations of presumed noninfectious and viral etiology (that
together accounted for 94% of all diarrhea-associated
hospitalizations), whereas the remaining small fraction (6%) of
hospitalizations specified as bacterial or parasitic etiology peaked
during the summer months (Fig 3B). Hospitalizations for rotavirus diarrhea showed a distinct peak in the late winter and early spring months (February through April).
View larger version (18K):
[in this window]
[in a new window]
|
Fig. 3.
Composite monthly diarrhea-associated hospitalizations among children
from Connecticut who were 1 through 59 months of age by age group (A)
and etiology (B) in 1993 through 1996. The proportion of monthly
hospitalizations is provided for each group separately.
|
|
The median duration and cost of a diarrhea-associated hospitalization
were 2 days and $1941, respectively (Table 2). The median duration of a
diarrhea-associated hospitalization decreased from 3 days in 1987 to
1990 to 2 days in 1993 to 1996 (P < .001), whereas the
median unadjusted cost increased by 65% from $1472 to $2428 over the
same period (P < .001). The median duration and cost of a
rotavirus-associated hospitalization was 2 days and $2511,
respectively. Overall, the most common payer type listed on the
discharge records was commercial/private (43%), followed by Medicare
(39%) and health maintenance organization (HMO)/preferred provider
organization (PPO; 13%). The percentage of diarrhea-associated hospitalizations with commercial/private payer decreased from 55% in
1987 to 1990 to 32% in 1993 to 1996, whereas those of HMO/preferred provider organization payer increased from 5% to 23% over the same
period.
| |
DISCUSSION |
The goal of this pilot study was to assess whether state hospital
discharge data could be used for surveillance of hospitalizations from
rotavirus diarrhea and to monitor the impact of the introduction of a
rotavirus vaccination program. For surveillance of rotavirus-associated hospitalizations, we were encouraged that 10.4% of all
diarrhea-associated hospitalizations in children from Connecticut
during 1993 to 1996 (an average of 100 cases per year) were coded as
rotavirus, an estimate that is consistent with NHDS data for states in
the Northeast.9 However, these rotavirus-specific data on
disease burden likely underestimate the true magnitude of the problem
for 2 reasons. First, because physicians may choose not to order a test
for rotavirus, because it adds cost but does not alter therapy, many
cases of rotavirus diarrhea may not be diagnosed specifically with a
proper assay or coded. This hypothesis is supported by the distinct
winter seasonality and predisposition for children 4 to 35 months of age (epidemiologic characteristics of rotavirus
diarrhea)6-9 of hospitalizations of presumed
noninfectious and nonrotavirus viral etiologies. Second, some
physicians or nosologists may not be aware of the new ICD-9-CM code for
rotavirus that was introduced in October 1992. This hypothesis is
supported by the increase in the percentage of diarrhea-associated
hospitalizations coded as rotavirus from 8.6% in 1993 to 14.7% in
1996. Based on previous studies17,18 that indicate that
rotavirus causes 30% to 45% of diarrhea-associated hospitalizations
in US children, we estimate that only one third to one half of all
rotavirus hospitalizations are being coded currently.
Although the underreporting of rotavirus limits an accurate assessment
of the morbidity from rotavirus diarrhea, Connecticut hospital
discharge data should provide timely and reasonably sensitive means by
which to monitor the impact of rotavirus vaccines. These data are
available for analysis within 1 year of their collection and provide
complete information on hospital discharges in Connecticut. The steady
increase in the coding of rotavirus in recent years suggests
improvements in the awareness of the new ICD-9-CM code for rotavirus
and indicates that underreporting should be less of a problem in the
future. Even in the absence of complete and accurate information on
rotavirus-associated hospitalizations, the impact of rotavirus vaccines
during the first year that they begin to be used widely should be
clearly visible in the reduction of winter peaks of diarrhea-associated
hospitalizations in children 4 months through 1 year of age. In each
subsequent year, this decline should extend progressively to cohorts of
older children.
This study provides information on the disease burden and cost of
diarrhea-associated hospitalizations in children living in Connecticut,
data that should allow public health officials to assess the potential
benefits of rotavirus vaccines. Overall, 1 child in 40 was hospitalized
for diarrhea during the first 5 years of life at a median cost of $1941
per hospitalization and a most recent (1993-1996) cost of $2428 per
hospitalization. Diarrhea-associated hospitalization rates declined by
39% from 63.0 per 10 000 in 1987 to 38.5 per 10 000 in 1996. The
American Academy of Pediatrics began recommending the use of oral
rehydration therapy in 1985, and the decline in diarrhea-associated
hospitalization rates might be attributable, in part, to the prevention
of severe dehydration because of this intervention.19
However, this decline might reflect the increase in the number of
children receiving care in HMOs. It is recognized that
diarrhea-associated hospitalization rates are lower in children seeking
care in HMOs, possibly because of an improved access to medical care, a
shift in care from hospital to outpatient clinics and emergency rooms,
and disincentives to hospitalize children for a mild to moderate
diarrheal illness.20 This hypothesis is supported by the
observation that the percentage of diarrhea-associated hospitalizations
with HMO listed as the payer increased from 5% during 1987 to 1990 to
23% during 1993 to 1996.
The current study has some limitations related to the data collection
and coding procedures used. Although federal hospitals were excluded
from the CHIME data, this exclusion would have a negligible impact on
our conclusions, because the only federal hospitals in Connecticut are
veterans hospitals that serve a very small fraction of the population.
With regard to coding, our assumption that a laboratory diagnosis of
rotavirus is a prerequisite for coding this pathogen on the discharge
record may not be true. However, the marked winter seasonality of
rotavirus-associated hospitalizations and their predisposition for
children 4 to 35 months of age is consistent with previous
studies6-9 and provides some reassurance that our
assumptions are accurate.
More than 30 US states collect complete information on hospital
discharges. This study indicates that state hospital discharge data
should provide sensitive, inexpensive, and timely information to
monitor the impact of a rotavirus immunization program. The methods
described here can be used to examine data from other states, thereby
providing critical information for monitoring the impact of a national
rotavirus immunization program. At the same time, the incompleteness of
identification and reporting of rotavirus diarrhea and the lack of
information on laboratory-confirmed cases are limitations of state
hospital discharge data. More accurate data on the morbidity from
rotavirus diarrhea could be obtained by identifying a network of
sentinel hospitals that provide information on cases of rotavirus
confirmed by laboratory detection or by establishing such a
surveillance system within the state. However, this will require
additional resources. Several health insurance agencies and HMOs
routinely collect and monitor information on hospital discharges, and
these data could be used to augment rotavirus surveillance using NHDS
and state hospital discharge data. Combining population-based data from
a number of states conducting surveillance of hospital discharges
for diarrhea in children should allow a reasonable and timely
assessment of the impact of a rotavirus immunization program in the
United States.
| |
ACKNOWLEDGMENTS |
We thank Mary Lyon, Matthew J Clarke, and Abtin Shahriari for
technical assistance and Anne Mather for editorial assistance in the
preparation of the manuscript.
| |
FOOTNOTES |
Received for publication Dec 23, 1998; accepted Mar 25, 1999.
Reprint requests to (U.D.P.) Viral Gastroenteritis Unit,
Mailstop G-04, Centers for Disease Control and Prevention, 1600, Clifton Rd NE, Atlanta, GA 30333. E-mail: uap2{at}cdc.gov
| |
ABBREVIATIONS |
NHDS, National Hospital Discharge Survey;
CHIME, Connecticut Health Information Management and Exchange;
HMO, health
maintenance organization.
| |
REFERENCES |
-
Joensuu J,
Koskenniemi E,
Pang X-L,
Vesikari T
A randomized, double-blind placebo controlled trial of rhesus-human reassortant rotavirus vaccine for prevention of severe rotavirus gastroenteritis.
Lancet
1997;
350:1205-1209 [CrossRef][Medline]
-
Rennels MB,
Glass RI,
Dennehy PH,
Safety and efficacy of high-dose rhesus-human reassortant rotavirus vaccines
report of the national multicenter trial.
Pediatrics
1996;
97:7-13 [Abstract/Free Full Text] -
Perez-Schael I,
Guntinas MJ,
Perez M,
Efficacy of the rhesus rotavirus-based quadrivalent vaccine in infants and young children in Venezuela.
N Engl J Med
1997;
337:1181-1187 [Abstract/Free Full Text]
-
Santosham M,
Moulton LH,
Reid R,
Efficacy and safety of high-dose rhesus-human reassortant rotavirus vaccine in Native American populations.
J Pediatr
1997;
131:632-638 [CrossRef][Medline]
-
Committee on Infectious Diseases
Prevention of rotavirus disease: guidelines for use of rotavirus vaccine.
Pediatrics
1998;
102:1483-1491 [Abstract/Free Full Text]
-
Glass RI, Lew JF, Gangarosa RE, LeBaron CW, Ho MS. Estimates of the morbidity and mortality from diarrheal diseases in American children. J Pediatr. 1991;118:27-33. Supplement
-
Jin S,
Kilgore PK,
Holman RC,
Clarke MJ,
Gangarosa EJ,
Glass RI
Trends in hospitalizations for diarrhea in United States children from 1979-1992: estimates of the morbidity associated with rotavirus.
Pediatr Infect Dis J
1996;
15:397-404 [CrossRef][Medline]
-
Glass RI, Kilgore PE, Holman RC, et al. The epidemiology of rotavirus diarrhea in the United States: surveillance and estimates of disease burden. J Infect Dis. 1996;174:5-11. Supplement
-
Parashar UD,
Holman RC,
Clarke MJ,
Bresee JS,
Glass RI
Hospitalizations associated with rotavirus diarrhea in the United States, 1993 through 1995: surveillance based on the new ICD-9-CM rotavirus specific diagnostic code.
J Infect Dis
1998;
177:13-17 [Medline]
-
National Center for Health Statistics. National Hospital Discharge Survey: Multi-year Data Tape Information, 1979-92 [data for the years 1993, 1994, and 1995 were also used]. Hyattsville, MD: National Center for Health Statistics; 1995
-
Graves EJ
National hospital discharge survey: annual summary, 1993.
Vital Health Stat
1995;
13:121
-
Connecticut Health Information Management and Exchange. Hospital Inpatient Data Tapes, Fiscal Years 1987-1997. Wallingford, CT: CHIME Inc; 1998
-
Public Health Service and Health Care Financing Administration. International Classification of Diseases, 9th Revision, Clinical Modification. 6th ed [CD-ROM]. Washington, DC: Public Health Service; 1997
-
US Bureau of Census. Intercensual Estimates of the Population of States by Age, Sex, and Race: 1970-1996. Washington, DC: US Bureau of Census; 1997
-
Kleinbaum DG, Kupper LL, Muller KE, Nizam A, eds. Applied Regression Analysis and Other Multivariable Methods. New York, NY: Duxbury Press; 1998
-
Lehman EL, ed. Nonparametrics: Statistical Methods Based on Ranks. San Francisco, CA: Holden-Day, Inc; 1975
-
Matson DO,
Estes MK
Impact of rotavirus infection at a large pediatric hospital.
J Infect Dis
1990;
162:598-604 [Medline]
-
Brandt CD,
Kim HW,
Rodriguez JO,
Pediatric viral gastroenteritis during eight years of study.
J Clin Microbiol
1983;
18:71-78 [Abstract/Free Full Text]
-
Mauer AM,
Dweck SH,
Finberg L,
American Academy of Pediatrics Committee on Nutrition. Use of oral fluid therapy and post-treatment feeding following enteritis in children in a developed country.
Pediatrics
1985;
75:358-361 [Abstract/Free Full Text]
-
Parashar UD,
Holman RC,
Bresee JS,
Epidemiology of diarrheal disease among children enrolled in four west coast health maintenance organizations.
Pediatr Infect Dis J
1998;
17:605-611 [CrossRef][Medline]
Pediatrics (ISSN 0031 4005). Copyright ©1999 by the American Academy of Pediatrics
Top
Abstract
Methods
