Hospitalizations due to rotavirus gastroenteritis in Catalonia, Spain, 2003-2008

Rotavirus gastroenteritis (RVGE) is the most common cause of severe gastroenteritis among children under 5 years of age, causing considerable morbidity and mortality worldwide [1, 2]. In Europe, RVGE causes approximately 230 deaths, more than 87,000 hospitalizations and 700,000 outpatient visits annually, consuming substantial health resources [3]. Although costs for treatment of severe cases (i.e. those admitted to hospital) are relevant in Europe [4], rotavirus vaccine is still too expensive to encourage the implementation of a generalized immunization strategy in areas of low incidence [5].

Previous studies regarding the burden of rotavirus disease in Europe and in different autonomous regions within Spain have yielded varying estimates of the number of hospitalizations attributable to rotavirus [6–10]. Few studies in Spain have assessed the burden of nosocomial rotavirus infection, even though it is considered to be one of the principal hospital-acquired infections in young children [9, 11–13]. We describe here the characteristics of hospitalizations associated with rotavirus infection in the autonomous region of Catalonia during the years 2003-2008. In addition, we evaluated the cost-effectiveness and threshold vaccine cost for savings with rotavirus vaccination.

We analyzed hospital discharge data for Catalonia from the Minimum Basic Data Set for Acute-care Hospitals (MBDS) of the Catalan Health Services. This data set comprises data from all public acute care hospitals and almost 90% of private acute care hospitals.

Diagnoses are coded according to the International Classification of Diseases, 9th revision, Clinical Modification (ICD-9-CM). We analyzed data for all hospitalizations of children less than 5 years of age with a principal diagnosis of infectious gastroenteritis (IGE) to estimate the frequency of community-acquired RVGE that required hospitalization. We defined community-acquired RVGE as all hospitalizations with a primary ICD9 CM code for RVGE (008.61), as well as those with a secondary code for RVGE, when the primary code was an IGE-related code. To evaluate the burden of nosocomial RVGE we selected hospitalizations with primary diagnoses that were not related to IGE but with RVGE coded in any of the secondary diagnoses. We analyzed data regarding the length of hospital stay, age, gender, date of admission, and outcome at discharge. The six-year period evaluated started the 1^st of January 2003 and ended the 31^st of December 2008.

The following ICD-9-CM codes were selected for IGE: 003.0, 005.0-005.9, 008.0-008.8 and 009.0-009.3. We excluded codes that are most likely to be imported gastroenteritis: cholera (001.0-001.9), typhoid and paratyphoid fever (002.0-002.9), shigellosis (004.0-004.9) and amoebiasis (006.0-006.9), which rarely if ever cause hospitalization in Catalonia. The primary diagnosis was assumed to be the main reason for hospital admission. To estimate the incidence of hospitalizations due to RVGE and other types of IGE, we obtained population estimates for each study year published by the Statistic Institut of Catalonia (Idescat). To estimate the incidence of nosocomial hospitalizations due to RVGE, we obtained the annual number of hospitalizations of children under 5 years of age from the annual reports of MBDS. The average length of stay in hospital was calculated for the different types of IGE and for each of the study years. Statistical analyses and data processing were conducted using STATA v.11.0 and Excel Office 2007.

Total admission costs from the health care perspective were calculated by multiplying the estimated number of cases (community acquired RVGE) of each year by the average rotavirus admission costs. The latter were calculated using the cost of a proxy Diagnosis Related Group (DRG) for this disease (see below), which was obtained from the Spanish Ministry of Health (http://www.msps.es), and weighted using data from the Centers for Medicare and Medicaid Service (CMS-DRG version 22) as in López-de-Andres et al[9]. Since there is no specific DRG for rotavirus gastroenteritis, DRG 184 (esophagitis, gastroenteritis, and miscellaneous digestive disorders, age < 18) was used.

The lack of a specific surveillance system for RVGE makes it difficult to directly estimate the burden of this disease in Spain. The percentage of IGE hospital admissions due to rotavirus in Spain ranged from 31% to 45% in previous studies with similar methodology to ours [8–10]. We found that 22% of the IGE hospital admissions in children under 5 in Catalonia, from 2003 through 2008, were due to rotavirus infection.

The incidence of admissions for rotavirus during our study period, 104 per 100,000 children < 5 years, is also lower than the most recent estimate for the total of Spain of 135 cases per 100000, and lower than recent estimates for other European countries [3, 7, 9, 17–19]. Our estimate is also lower than that from a previous study conducted in Catalonia (study period 1999-2000) of 125 per 100000 [20]. The lower estimates in Catalonia compared to the rest of Spain could reflect actual geographic variability in incidence of rotavirus infection or disease, or could be due to differences in the sensitivity of the hospital discharge databases for detecting RVGE. Similarly, the apparent decreased incidence we found compared to the earlier estimate for Catalonia by Gil et al. might reflect a true decreased incidence over the years since 2005 or could be due to changes in detection sensitivity. However, the coverage of MBDS for public hospitals improved from 95% to 100% over the time between the two estimates [20].

The seasonality of RVGE has already been widely described by many studies in Europe and America, and our data also shows increased incidence during winter months. During the years we evaluated, there was a progressive increase of incidence through 2005, which is consistent with results of previous evaluations in Spain [9]. At least some of the increase could have been due to enhanced clinical awareness. The slight decrease in hospital admissions in 2007 and 2008 might be attributed to the introduction of the rotavirus vaccines in the Spanish market during 2006 and 2007. However, based on numbers of vaccine doses sold to pharmacies in Catalonia by pharmaceutical companies, we estimated vaccine coverage among children less than 1 year old at only 20% in 2008. The coverage among children up to 5 years of age would be much lower, thus limiting the impact of vaccination on hospitalization to date.

We found an average of 0.6 day longer mean hospital stay for hospitalizations due to RVGE compared with hospitalizations for other causes of IGE. Similar results were described in other studies using Vesikari's score to account for duration and severity of diarrhea [21, 22]. Thus we estimate that if there had been vaccination coverage similar to that for measles, mumps and rubella vaccine (approximately 96%)[23], and a 90% efficacy of a rotavirus vaccine [24, 25], over 7500 days of hospitalization attributable to rotavirus could have been prevented over the 6 year study period. Achieving high vaccine coverage would likely require inclusion of rotavirus vaccine in the national immunization program.

The average annual incidence of presumed nosocomial RVGE in our study period was 2.5 per 1000 hospitalizations in children < 5 years of age, which is lower than the incidence of 4.5 per 1000 admissions reported in another recent study for the whole of Spain [26]. However, in that study, the incidence for Catalonia of 2.0 cases per 1,000 children under 5 is quite consistent with our estimate and thus the incidence of nosocomial RVGE in Catalonia appears to be lower than in the rest of Spain. Our finding that children under 12 months old appear to be at higher risk of acquiring nosocomial RVGE than older children, is consistent with previous studies [26, 27].

In the economic evaluation of RVGE, the strategy "no vaccine" was substantially less expensive than the vaccination strategy at the current vaccine price of 187 €. At a vaccine cost of 1.93 € the two strategies had the same expected cost. A study from the United States estimated the median cost-effectiveness ratio of vaccination program per serious case prevented at 3024 US$ [28], substantially lower than our estimate for Catalonia. Thus, rotavirus vaccination in Catalonia does not appear to be as cost-effective as in the United States. The cost-effectiveness ratio estimated in our study is consistently higher due to the lower probability of hospitalization, higher vaccine price and lower costs of hospitalization.

There are limitations to using the MBDS to estimate RVGE incidence and severity. First, the dataset covers all public hospitals in our region but it may miss cases of RVGE that are not laboratory confirmed. Second, a recent study in Italy indicated that 27% of RV positive cases admitted to the hospital were not coded as RVGE in the hospital discharge database, and two studies in Spain found that 52% (Foster J. et al) and 47% (Gutierrez-Gimeno et al) of children hospitalized for gastroenteritis in Spain and prospectively tested for rotavirus infection were positive [21, 29, 30]. Similar coding and diagnostic error in our dataset would result in underestimation of the true number of RVGE cases. However, the MBDS system in Catalonia uses an internal and external validation system that monitors and updates the hospital discharge data before processing and storage, thus capturing diagnoses of cases that were discharged before the results of in-hospital microbiological tests. Third, there might be cases of nosocomial RVGE that are not detected if the symptoms start after discharge and do not provoke re-hospitalization. On the other hand, our definition for presumed nosocomial RVGE might lack specificity if the primary diagnostic code does not reflect the actual cause of admission and this would lead to labeling some cases of community-acquired RVGE as presumed nosocomial RVGE, resulting in an overestimate of nosocomial RVGE. Fourth, some hospitals might not run diagnostic tests for RVGE as frequently as others, yielding an underestimation of RVGE cases. Fifth, our economic evaluation only concerns community acquired RVGE and is calculated from the health system perspective, and therefore does not consider indirect costs associated with rotavirus admissions (such us travelling expenses and family members' days off from work). This would underestimate the total societal costs of each admission. However, such indirect costs are likely to be small compared to the direct costs of hospitalization [30]. Lastly, we have not taken into account the unknown rate of readmission for RVGE in our study population. Since the data from MBDS is anonymous, we could not distinguish if the same child was hospitalized more than once. Counting readmissions in our analysis could result in overestimation of the true incidence of both community acquired and nosocomial RVGE. However, since primary rotavirus infection generally protects against subsequent severe rotavirus illness, we believe that our estimates of the incidence of hospital-acquired and nosocomial RVGE are not substantially biased by readmissions.

In conclusion, rotavirus is an important cause of infectious gastroenteritis associated with hospitalization among children under 5 years old in Catalonia, and especially among children under 2 years of age. RVGE occurs mainly in winter months and causes longer hospital stays than gastroenteritis caused by other infectious agents. However, the incidence of community acquired and nosocomial RVGE in Catalonia appears to be lower than in other regions of Spain and other European countries. The cost of rotavirus vaccination would need to be substantially lower than the current cost in order to make rotavirus vaccination as cost-effective in Catalonia as in many other areas where universal vaccination has been implemented.