Evaluating Acute Viral Gastroenteritis Severity: Modified Vesikari and Clark Scoring Systems Open Access

Children presenting to an inpatient, ED, or OP setting in Nashville, Tennessee between December 1, 2012 and June 30, 2016 were enrolled as part the Centers for Disease Control and Prevention (CDC) New Vaccine Surveillance Network, an active, population-based AGE surveillance platform. Parents or legal guardians provided written informed consent; when applicable, patients provided assent. This study was approved by the institutional review boards at the CDC, the Tennessee Department of Health, and Vanderbilt University Medical Center.

Children were eligible to be enrolled if they presented to Monroe Carell Jr. Children’s Hospital at Vanderbilt within 10 days of onset of AGE symptoms and were residents of Davidson County. AGE illness was defined as at least 1 episode of vomiting and/or at least 3 episodes of loose stools within 24 hours. Children in the inpatient, ED, and OP settings were eligible for enrollment if they were between ≥15 days and <18 years old.⁴  Children were excluded if they met any of the following criteria: (1) transferred from another hospital more than 48 hours after admission, (2) had a noninfectious or another identifiable cause of AGE symptoms (eg, head trauma), (3) were immunocompromised, or (4) were previously enrolled for the same AGE illness.

Demographic and clinical data were collected through parent and guardian interviews by trained research staff, and data from medical records were abstracted to determine clinical outcome data.

Whole stool specimens were collected from AGE patients within 10 days of illness onset; patients whose stool was collected outside of the window period were excluded from these analyses. Laboratory testing was conducted by the Tennessee Department of Health by means of reverse transcription quantitative polymerase chain reaction for norovirus, sapovirus, and astrovirus RNA and by Vanderbilt University Medical Center using Rotaclone enzyme immunoassay (EIA) for rotavirus antigen, followed by either repeat EIA and/or reverse transcription quantitative polymerase chain reaction verification of antigen detection by the CDC.¹⁴  Patients with an inconclusive test result were excluded.

Severity of AGE symptoms at presentation was determined using the MVS and CS. Both scoring systems account for duration of diarrhea and vomiting, number of diarrheal and vomiting episodes in 24 hours, and fever. The MVS accounts for dehydration using the World Health Organization Integrated Management of Childhood Illness dehydration criteria¹⁵  (Supplemental Table 4), along with treatment (rehydration and hospitalization). This difference was made to eliminate the need for follow-up interviews performed in previous studies.^7,8  The CS accounts for fever duration and behavioral symptoms. We logically imputed temperature for cases for which a fever was reported but associated temperature was not. There were 411 of 2012 (20.4%) cases of subjective fever; these were imputed to 37.0°C to account for uncertainty in how the temperature was measured; these cases do not contribute to MVS or CS score. As the duration of behavioral symptoms was not collected in this study, it was excluded from the calculation of the CS. Therefore, we used a modified version of the original CS. The MVS is calculated on a 20-point scale and the CS on a 21-point scale (Supplemental Table 4). For both scoring systems, the range of points per symptom factor is 0 through 3, with the exception of dehydration (2–3) and treatment (1–2) for MVS.

Demographic variables were summarized descriptively as frequency (percentage) or mean and median (SD and interquartile range [IQR], respectively), as appropriate. Demographics were compared between virus-positive and test negative patients and across virus detection categories (ie, rotavirus, norovirus, astrovirus, sapovirus, viral codetection, and no virus). We further compared clinical symptoms of fever, vomiting, diarrhea, and days ill at presentation between virus-positive and test negative patients and across virus detection categories. We compared the mean MVS and CS between groups defined by demographic and pathogen status. Comparisons were performed using Pearson’s χ² test for categorical variables and linear regression with robust standard errors for continuous variables. Multiple linear regression was used to assess the association between each scoring system (eg, MVS and CS) and the following variables: age, sex, prematurity, daycare or school attendance, insurance status, medical setting, and virus group. Spearman’s correlation was used to measure the relationship between the MVS and CS. When comparing severity scores across different settings, hospitalization was removed from the treatment category of the MVS for patients presenting to the inpatient setting. This was done to allow for an unbiased comparison of symptom severity at presentation across settings, since hospitalization itself contributes to the MVS and patients in the inpatient setting are hospitalized by definition. Removing these 2 points from the score allows the MVS to reflect just the symptom severity for inpatients at the time of presentation before hospitalization, enabling a fair comparison with severity in other settings. Hospitalization was included in all other analysis using the MVS. A significance level of 0.05 (2-tailed, where appropriate) was used for all analyses. All statistical analyses were performed using R software (version 4.1.1).

From December 1, 2012 to June 30, 2016, a total of 6309 children were eligible and 4216 (67%) were enrolled, with 3256 (77%) participants providing a stool specimen (Fig 1). The median age of the cohort was 1.9 years, 52% were male, 55% were white, 58.5% were non-Hispanic, 9.3% were premature, and 40.5% attended daycare or school. Children presenting to the inpatient, emergency department, and outpatient settings experienced symptoms for 3.5 (2.0), 2.6 (1.8), and 3.2 (1.9) days on average (SD), respectively, before enrollment. Days from enrollment to specimen collection was, on average (SD), 1.0 (1.8) day for the inpatient setting, 1.9 (2.2) days for the emergency department setting, and 1.9 (2.4) days for the outpatient setting.

Overall, 1387 (43%) of the stool specimen tested were positive for at least 1 of the 4 viruses tested. The most common single-detected virus tested was norovirus (678; 49%), followed by sapovirus (266; 19%), rotavirus (228; 16%), and astrovirus (98; 7%); a total 117 (8%) specimens tested positive for more than 1 viruses (Figs 2A and 2B).

Age, sex, premature status, daycare or school attendance, insurance status, breastfeeding, and medical setting distribution were comparable between the group of children that tested positive for at least 1 of the 4 viruses compared with the test negative children (Table 1). The mean MVS was higher in the virus-positive group compared with those testing negative (8.14 vs 7.37, respectively; P < .001) and the mean CS was higher, though not statistically significant, in the virus-positive group compared with the negative group (7.17 vs 6.95, respectively; P = .051).

The distribution of MVS and CS by setting is displayed in Supplemental Fig 3. The MVS and CS scoring systems were highly correlated (r = 0.84). Comparisons of MVS and CS within levels of each patient characteristic are presented in Table 2. In our univariate analyses, factors associated with significantly higher mean for both MVS and CS included younger age, male sex, daycare or school attendance, hospitalization, and positive test for rotavirus (Table 2). In addition, private insurance was significantly associated with a higher mean MVS, although not significant when using CS. Similarly, prematurity was associated with significantly higher mean CS but was not significant when using MVS.

When further stratifying the comparison of scoring systems by 2 age groups: 15 days to 2 years and 3 to 17 years, the younger children had a higher average MVS and CS (8.1 and 7.4, respectively) compared with the older children (7.1 and 6.5, respectively), with P values < .001 for both scoring systems.

Results for the 2 multiple regression models are presented in Table 3. We found that younger age, male sex, hospitalization (as compared with the ED), and rotavirus detection were significantly associated with higher mean MVS and CS. In addition, prematurity and enrollment in the ED setting (compared with outpatient) were associated with higher mean CS.

Our large, prospective AGE surveillance study found that MVS and CS correlated well and that children who were rotavirus-positive had significantly higher MVS and CS scores when compared with children who tested virus-negative as well as children who tested positive for norovirus, sapovirus, and/or astrovirus. In addition, younger age, male sex, and hospitalization were associated with higher severity scores in children in our cohort. Our study highlights the identification of specific risk factors for more severe disease presentation using the MVS and CS and could help assess future vaccine effectiveness trials for AGE pathogens.

Our study found that virus-positive children had higher MVS scores when compared with virus-negative children. Defining severity of AGE illness is critical for directing future interventions aimed at decreasing morbidity and mortality because of specific AGE pathogens, especially when assessing vaccine efficacy and effectiveness. Our study significantly adds to the literature in that few studies have compared severity of virus-negative and virus-positive AGE in children. One study evaluating the CS did not identify important differences in severity between viral and bacterial causes of AGE¹³  and was limited by a small number of participants (n = 137). Other studies have used the MVS to compare severity of virus-negative and virus-positive patients in the inpatient and outpatient settings for all causes of AGE.^8,16–18  However, these studies were mostly limited to children under 5 years old and did not examine specific viral causes of AGE in relation to the MVS. Of note, the MVS and CS did not differ between the ED and OP setting; however, the scores in the inpatient setting were significantly higher even when excluding points for hospitalization from the MVS. Thus, our large prospective cohort was unique in that we used both severity scoring systems to identify differences between virus-negative and virus-positive patients ranging in age from infancy through middle adolescence and presenting to all major health care settings.

Our study also found that rotavirus was associated with both higher mean MVS and CS, even in the postrotavirus vaccine era. Given the global burden of rotavirus-associated illness, the MVS and CS were originally designed to assess rotavirus vaccine efficacy.^9,19–22  A case-control study of children less than 1 year of age who had a rotavirus infection from 1987 to 1995 identified male sex, prematurity, and public insurance as risk factors for severe rotavirus-associated AGE.²³  Another study found no difference between biological sex and severity of AGE, though it was limited to a smaller cohort size of only 650 participants.²⁴  Our study significantly strengthens the literature in showing that male sex is indeed a significant risk factor for more severe AGE presentation by assessing a large cohort with a broad age range and multiple contributing pathogens. Further research is needed to assess why these differences exist to form mitigation strategies for severe AGE. In the postrotavirus vaccine era, rotavirus is still associated with the highest severity and optimizing vaccine uptake is essential to help reduce the burden of rotavirus-associated disease.

Previous studies have consistently shown that younger age is associated with more severe AGE presentation. Our study agrees with previous findings and further supports research targeted toward lowering morbidity and mortality in younger children with AGE. Lastly, our study found that previous daycare or school attendance was associated with lower AGE severity at presentation. Previous studies have also shown similar findings,^25,26  with the hypothesis being that early exposure to pathogens grants protection from more severe presentation later in life. Future research should further focus on this association and its implication on public health policy.

We found a strong positive correlation between MVS and CS. Previous research studies in this area have focused on the original Vesikari and Clark scoring systems regarding severe rotavirus induced AGE, and they did not find a strong correlation between scores.^9–11  Apart from only focusing on severe rotavirus induced AGE, they also had a smaller cohort.^9–11  Our results show that the MVS and CS were concordant. Our study population differed from previous research in that we compared the systems using a large prospective cohort with multiple AGE pathogens who sought care at 3 different settings. This large and diverse cohort may have accounted for the differences seen in our study compared with previous literature. This is an important first step in establishing a scoring system that could be validated and used to aid clinicians in assessing AGE severity at presentation, including need for hospitalization. It also may aid in future AGE research (including postvaccine surveillance) by showing both systems are comparable.

Strengths of our study include a large sample size, multiple settings (emergency, outpatient, and inpatient), and enrollment over 4 consecutive seasons. However, this was a single-center study that may not be representative of all regions of the United States. Furthermore, children included in the study might not display the full spectrum of AGE symptom presentation. We also only tested for 4 common AGE viral pathogens (rotavirus, norovirus, sapovirus and astrovirus), and therefore, we are not able to assess severity of other AGE viral pathogens (eg, adenovirus, enterovirus). The study design is also potentially limited by survey biases such as sampling, recall, acquiescence, and/or extreme response. We attempted to mitigate this inherent limitation using a standardized interview technique with trained research personnel. Moreover, the definitions of dehydration and behavioral symptoms for the scales can be somewhat interview or clinician dependent as symptoms such as “lethargic, irritable, and skin pinch response” are not objective. We tried to minimize this subjectivity by structured in-depth training with our research assistants enrolling patients; however, these measurements were not replicated by a second person. Moreover, since different researchers collected data in each setting, there is a limitation of the inter-rater reliability of MVS and CS across settings. Efforts were made to promote consistency across settings through use of standardized symptom assessments across settings. However, inter-rater reliability was not formally evaluated between researchers. Furthermore, neither the modified CS system nor the adapted MVS system used for comparison across settings have been thoroughly validated, which could also impact reliability. This highlights an area for improvement in future work.

In summary, we found that the 2 major scoring systems, MVS and CS, were comparable when assessing AGE severity in a large cohort of children with AGE. Future studies should focus on comparing effectiveness between rotavirus vaccines in the postsurveillance period using these 2 scores. We also found that male sex and younger age were associated with higher MVS and CS. Furthermore, higher MVS and CS scores were demonstrated in hospitalized and rotavirus-positive children. These risk factors may aid clinicians in determining which children are at high-risk for severe AGE. Future studies should also focus on the development of a clinical decision-support tool that could identify children at high risk of severe disease requiring hospitalization.