Article, Pediatrics

Relationship between the number of pediatric patients with rotavirus and telephone triage for associated symptoms

a b s t r a c t

Background: Earlier syndromic surveillance may be effective in preventing the spread of infectious disease. How- ever, there has been no research on syndromic surveillance for rotavirus. The study aimed to assess the relation- ship between the incidence of rotavirus infections and the number of telephone triages for associated symptoms in pediatric patients under 4 years old in Osaka prefecture, Japan.

Methods: This was a retrospective observational study for which the study period was the 3 years between Jan- uary 2015 and December 2017. We analyzed data on children under 4 years old who were triaged by telephone Triage nurses using software. The primary endpoint was the number of rotavirus patients under 4 years triaged old per week. Using a linear regression model, we calculated the R square value of the regression model to assess the relationship between the number of patients with rotavirus and the number of telephone triages made for associated symptoms. Covariates in the linear regression model were the week number indicating seasonality and the weekly number of telephone triages related to rotavirus symptoms such as stomachache and vomiting. Results: During the study period, there were 102,336 patients with rotavirus, and the number of people triaged by telephone was 123,720. The highest correlation coefficient was 0.921 in the regression model with the number of telephone triages for “stomachache + nausea/vomiting” and “stomachache + diarrhea + nausea/vomiting”. Conclusion: The number of telephone triage symptoms was positively related to the incidence of PEdiatric pa- tients with rotavirus in a large metropolitan area of Japan.

(C) 2020 The Author(s). This is an open access article under the CC BY-NC-ND license

(http://creativecommons.org/licenses/by-nc-nd/4.0/).

  1. Introduction

Rotavirus infection is the most common gastrointestinal infectious disease among children aged 4 years or less, causing 215,000 deaths around the world every year [1]. Recently, surveillance for this disease has been conducted in African countries based on World Health Organi- zation recommendations [2-5]. However, it takes much time from occur- rence to publication of the results of traditional surveillance conducted by public health departments [6]. Therefore, earlier syndromic surveil- lance using existing data may be effective in preventing the spread of

* Corresponding author at: Department of Traumatology and Acute Critical Medicine, Osaka University Graduate School of Medicine, 2-15, Yamada-oka, Suita 565-0871, Japan.

E-mail address: [email protected] (Y. Katayama).

infectious disease. There are many studies on syndromic surveillance for gastrointestinal disease and waterborne disease such as Cryptospo- ridium using absenteeism records [7] and pharmacy drug sales [8-10]. However, there is no research on syndromic surveillance for rotavirus. Vaccination against rotavirus has spread in many developed countries in recent years. If syndromic surveillance for rotavirus can be conducted, it may be possible to prevent the spread of rotavirus infection in children through such surveillance.

In Osaka prefecture, Japan, telephone triage service has been pro- vided to the local population since 2012. The triage nurse uses software to determine the urgency of the client for each symptom and provides necessary services such as ambulance dispatch and guidance from med- ical institutions based on the result. Therefore, the number of telephone triages by symptom can be calculated in real time with this software. If

https://doi.org/10.1016/j.ajem.2020.03.039

0735-6757/(C) 2020 The Author(s). This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/).

there is a relationship between the number of telephone triage symp- toms such as stomachache, nausea/vomiting and diarrhea that are re- lated to rotavirus infection and the number of patients with rotavirus infection, it may be possible to predict the spread of rotavirus infections earlier. The purpose of this study was to assess the relationship between the incidence of rotavirus infections and the number of telephone tri- ages for associated symptoms in pediatric patients under 4 years old in Osaka prefecture, Japan.

  1. Materials and methods

    Study design, population and study setting

This was a retrospective observational study for which the study pe- riod was the 3 years between January 2015 and December 2017. Osaka prefecture is the largest urban area in western Japan, with an area of 1905.14 km2 and a population of 342,645 children who are 4 years old or younger among 8.8 million people [11]. In this study, we analyzed the data on children under 4 years old who were triaged by telephone triage nurses using software in Osaka prefecture. This study was ap- proved by the Ethics Committee of Osaka University Graduate School of Medicine (approval no. 16070). As the telephone triage data was anonymized, the necessity of obtaining informed consent from the sub- jects was waved.

Outpatient surveillance in Japan

The infectious disease surveillance program of Japan was initiated in 1981 and was revised and updated to its present form following the re- vision of the Infectious Disease Control Law in 1999 [6,12-14]. The sys- tem is currently called National Epidemiological Surveillance for Infectious Diseases (NESID), which includes a mandatory reporting sys- tem for nationally notifiable diseases and sentinel surveillance systems for various kinds of infectious diseases [15].

Rotavirus falls under the sentinel surveillance arm of the program. Weekly numbers of patients with rotavirus are reported from 3000 med- ical institutions nationwide to local health centers. Sentinel sites were designated among the hospitals that met certain criteria according to their geographic distribution and population densities. These sentinels use the following criteria for reporting patients with rotavirus: 1) diar- rhea of 3 times or more, 2) vomiting of one time or more and 3) any one of the following test results a) separation and identification by fecal culture, b) detection of pathogen antigens by immunochromatography and c) detection of pathogen genes by PCR. Sentinel sites report the age and sex of the patients on a weekly basis, but the report does not include personal information such as names or addresses. This information is transferred from local health centers to the prefectural government where it is aggregated into a prefectural report. The report is then forwarded to the National Institute of Infectious Diseases in Tokyo, which is affiliated with the Ministry of Health, Labour and Welfare. Within Osaka prefecture, 197 pediatric medical institutions report influ- enza patients to 10 local health centers.

Telephone triage service in Osaka

The telephone triage service in Osaka prefecture is a public service similar to that in Tokyo [16] and can be used freely by anyone. In this service, a triage nurse uses software with a protocol developed for tele- phone triage in Japan and determines the urgency of the client. There are 97 different protocols of telephone triage for chief complaints in Japan, and the urgency of the client is determined for each chief com- plaint by selecting the symptom and sign related to that complaint. As with telephone triage service in the departments of veterans’ affairs in the United States [17], Canada and United Kingdom [18-20], telephone service in Osaka prefecture provides the client with necessary services such as ambulance dispatch and guidance from medical institutions based on the result of the urgency [21]. The software records the sex and age group of the client, the times when the telephone triage was started and ended, the chief complaint and selected symptoms and

signs, the urgency of the client and the presence or absence of ambu- lance dispatch.

Endpoint

The primary endpoint was the number of patients with rotavirus per week in Osaka prefecture. The number of patients with rotavirus per week was obtained from the data published on the web site of the Osaka Institute of Public Health [22].

2.2. Statistical analysis

Using a linear regression model, we calculated the R square value of the regression model to assess the relationship between the number of telephone triages and the number of patients with rotavirus in Osaka prefecture. The covariates in the linear regression model were the week number indicating seasonality and the weekly number of tele- phone triages related to rotavirus symptoms such as stomachache and vomiting. We defined the week number as a binary variable, with the week including January 1st as “week number = 1”. In this study, we se- lected three chief complains related to rotavirus: stomachache, diarrhea and nausea/vomiting among the telephone triage categories. The seven input patterns of these three variables into the linear regression model are “stomachache”, “diarrhea”, “nausea/vomiting”, “stomachache + di- arrhea”, “stomachache + nausea/vomiting”, “diarrhea + nausea/ vomiting” and “stomachache + diarrhea + nausea/vomiting”. We cal- culated the correlation coefficient and R square and adjusted R square values of each linear regression model. Next, according to the season (January-March, April-June, July-September, October-December), we calculated the Spearman’s correlation coefficient, R square value be- tween the predicted weekly number of rotavirus patients from the best linear regression model and the weekly number of rotavirus pa- tients for each season. Statistical significance was defined as P b 0.05, and statistical analysis was performed with SPSS version 23.0J (IBM Corp., Armonk, NY).

  1. Results

During the study period, there were 102,336 patients with rotavirus, and 123,720 in Osaka prefecture were triaged by telephone. Table 1 shows the characteristics of the children aged 4 years or less in Osaka prefecture between January 2015 and December 2017. The age group requesting the highest amount of telephone triage was the 0-year-old group (39,139; 31.6%). There were 67,606 (54.6%) males and 56,003 (45.3%) females. The most frequent telephone caller was the family of the patient (122,538; 99.0%). The number of telephone triages con- ducted during the daytime was 53,918 (43.6%) and that during the nighttime was 69,802 (56.4%). The number of telephone triages con- ducted in 2015 was 41,488 and that conducted in 2017 was 41,273. The number of telephone triages conducted during the study period did not change significantly. The most common chief complaint among the incidents of telephone triage was Vital sign abnormalities such as no response or breathing (119,355; 45.7%). Telephone triages were conducted for 2216 (0.8%) complaints of stomachache, 3271 (1.3%) complaints of diarrhea and 11,549 (4.4%) complaints of nausea/ vomiting.

Table 2 shows the correlation coefficient and R square and adjusted R square values for each regression model. The highest correlation coeffi- cient was 0.921 in the regression model with the number of telephone triages for “stomachache + nausea/vomiting” and “stomachache + diar- rhea + nausea/vomiting”. Although the highest R square was 0.848 in the regression model with the number of telephone triages for “stom- achache + diarrhea + nausea/vomiting”, the highest adjusted R square was 0.768 in the regression model with the number of telephone triages for “stomachache + nausea/vomiting”. Fig. 1 shows the weekly pre- dicted number of rotavirus patients for the linear regression model

Table 1

Demographic and clinical characteristics of children b5 years old judged by telephone tri- age between 2015 and 2017 in Osaka.

Characteristic Total

(n = 123,720)

Age, years, n (%)

0

39,139

(31.6)

1

36,178

(29.2)

2

20,350

(16.4)

3

15,801

(12.8)

4

12,252

(9.9)

Sex, n (%)

Male

67,606

(54.6)

Female

56,003

(45.3)

Unknown

111

(0.1)

Person who initiated telephone consultation, n (%)

Patient principal

703

(0.6)

Patient’s family

122,538

(99.0)

Other person

469

(0.4)

Unknown

10

(0.0)

Time of telephone consultation and triage, n (%)

Daytime (09:00 to 17:59)

53,918

(43.6)

Nighttime (18:00 to 8:59)

69,802

(56.4)

Year, n (%)

2015

41,488

(33.5)

2016

40,959

(33.1)

2017

41,273

(33.4)

Season, n (%)

January to March

28,361

(22.9)

April to June

33,114

(26.8)

July to September

31,536

(25.5)

October to December

30,709

(24.8)

Contents of telephone consultation and triage for chief complaint

Abnormal vital signs

119,355

(45.7)

Fever

30,856

(11.8)

Head injury

18,008

(6.9)

Nausea/vomiting

11,549

(4.4)

Rash/hives

8308

(3.2)

accidental ingestion of foreign substance

7570

(2.9)

Cough

5320

(2.0)

Toothache, tooth damage

4133

(1.6)

Diarrhea

3271

(1.3)

Injury to the face and extremities

3003

(1.2)

Stomachache

2216

(0.8)

Other

47,341

(18.1)

(Model 5) with the number of telephone triages for “stomachache + nau- sea/vomiting” and the actual weekly numbers of rotavirus patients.

Fig. 2 shows scatter plots of the weekly numbers of rotavirus pa- tients and the predicted numbers of rotavirus patients from the linear regression model for “nausea/vomiting” by season. The vertical axis shows the predicted number of rotavirus patients calculated from the regression model, and the horizontal axis shows the actual number of rotavirus patients per week. The season with the highest Spearman’s correlation coefficient (R = 0.923) was in October-December (P b 0.001), followed in order by April-June (R = 0.793, P b 0.001),

January-March (R = 0.712, P b 0.001) and July-September (R = 0.659, P b 0.001).

  1. Discussion

This study revealed a positive relationship between the telephone triage data and the number of pediatric patients with rotavirus under 4 years old in a large metropolitan community of Japan. We compared the R square value for each linear regression model, and the R square values were high in all seven models. Furthermore, a difference was found in the correlation coefficient of the linear regression model de- pending on season. In this study, the contribution rate of the linear re- gression model using the telephone triage data was high, and the forecast of rotavirus epidemics using these data may allow earlier an- nouncement than the warning based on the traditional surveillance and thus be useful in preventing the spread of rotavirus.

Several syndromic surveillance models for gastrointestinal infec- tious disease have been reported previously. Pivette et al. reported that a syndromic surveillance model using non-prescription drug sales data could detect epidemics on average 2.25 weeks earlier than surveil- lance of traditional sentinel data [8]. Bjelkmar et al. also reported that the number of telephone consultations was effective in the detection of early outbreaks in the study of Cryptosporidium in Skelleftea, Sweden in 2011 [23]. However, in a systematic review of syndromic sur- veillance, it was reported that syndromic surveillance using drug sales data and nurse advice line calls was timely but non-specific, whereas traditional surveillance using hospitalization data and Laboratory test results was not timely but more specific [24]. Our study showed a high correlation between the traditional surveillance data and the num- ber predicted from the linear regression model, indicating that it would be possible to predict an epidemic of rotavirus earlier than with tradi- tional surveillance. If a system such as nurse advice line or telephone tri- age can be constructed in an area where traditional surveillance is used, that data may be useful in the surveillance of rotavirus.

Next, among the Linear regression models, the R square values were higher in the models including the number of triages for “diarrhea” than in those not including this number. The number of telephone triages for “diarrhea” was the largest, and there were fewer telephone triage cases for “stomachache” and “nausea/vomiting” than for “diarrhea”. Because we included children who were 4 years old or younger in this study, they might be unable to appropriately tell their parents about chief complaints such as stomachache and nausea. Thus, the number of tele- phone triages related to objective symptoms such as diarrhea may have been higher than that for subjective symptoms such as stomachache and nausea. Indeed, in syndromic surveillance, a high correlation was reported when the number of explanatory variables was large, but no correlations were found when the number of explanatory variables was small [17]. In the present study, the high correlation coefficient of the linear regression model including the number of telephone triages for “diarrhea” may be related to the high number of telephone triages for this symptom. As all of the linear regression models including the number of telephone triages for “diarrhea” had high correlation coeffi- cients, we consider that any of these models could be used in the predic- tion of rotavirus epidemics.

Table 2

spearman correlation coefficient and R square and adjusted R square values for each linear regression model.

Model no. Variables Spearman correlation coefficient R2 Adjusted R2a

1 Number of telephone triages for stomachache

0.864

0.747

0.619

2 Number of telephone triages for diarrhea

0.844

0.712

0.567

3 Number of telephone triages for nausea/vomiting

0.917

0.840

0.759

4 Number of telephone triages for stomachache and diarrhea

0.867

0.751

0.622

5 Number of telephone triages for stomachache and nausea/vomiting

0.921

0.847

0.768

6 Number of telephone triages for diarrhea and nausea/vomiting

0.917

0.840

0.757

7 Number of telephone triages for stomachache, diarrhea and nausea/vomiting

0.921

0.848

0.766

a Adjusted R2 was adjusted by the number of variables in the linear regression model.

Fig. 1. The total number of pediatric patients with rotavirus and the predicted number of these patients from linear regression model 5 during the study period.

Finally, in the subgroup analysis divided into the four seasons, the linear regression model in October-December showed the highest cor- relation. However, the mechanism for this finding was not sufficiently revealed in this study. Rotavirus infection is a gastrointestinal infection that infects infants and children year round. The fact that other infec- tious gastrointestinal diseases such as those caused by Salmonella and Vibrio parahaemolyticus are prevalent in warmer seasons may have in- fluenced the results in this study. In any case, syndromic surveillance with the present linear regression model for rotavirus showed a high correlation coefficient throughout the year, which could help to prevent epidemics of rotavirus infection.

There were several limitations in this study. First, weekly numbers of patients with rotavirus are reported from sentinel medical institutions to local health centers based on the Infectious Disease Control Law, and not all rotavirus patients may have been reported. Second, we simultaneously assessed the correlation between the number of children triaged by tele- phone and the number of pediatric patients with rotavirus; this study was not an observational study that followed up the children for whom telephone triage was conducted or that calculated the prevalence of rotavi- rus. Third, the generalizability of our findings is limited due to areas where people can use systematic telephone triage system. Fourth, as this study was an observational study, unknown confounding factors may be present.

Fig. 2. The actual weekly number of rotavirus patients and the predicted number of rotavirus patients from linear regression model 3 by each season.

  1. Conclusion

The number of telephone triage symptoms was positively related to the incidence of rotavirus infections in pediatric patients in a large met- ropolitan area of Japan. Early prediction of rotavirus epidemics using telephone triage data may help to prevent the spread of this common infection.

CRediT authorship contribution statement

Yusuke Katayama:Conceptualization, Methodology, Software, For- mal analysis, Investigation, Resources, Writing – original draft, Visualiza- tion, Project administration, Funding acquisition.Kosuke Kiyohara: Software, Formal analysis.Sho Komukai:Software, Formal analysis. Tetsuhisa Kitamura:Methodology, Investigation, Resources, Writing – review & editing, Visualization.Kenichiro Ishida:Methodology.Tomoya Hirose:Writing – review & editing.Tasuku Matsuyama:Methodology. Takeyuki Kiguchi:Writing – review & editing.Takeshi Shimazu:Super- vision, Project administration.

Declaration of competing interest

YK has received a grant from the JR-West Relief Foundation.

Acknowledgements

The authors thank the Osaka Metropolitan Fire Department. We also thank our colleagues from Osaka University Center of Medical Data Sci- ence and Advanced Clinical Epidemiology Investigator’s Research Pro- ject for providing their insight and expertise to improve our research.

Data sharing statement

No additional data.

Sources of support

This article was supported by the JR-West Relief Foundation.

References

  1. Tate JE, Burton AH, Boschi-Pinto C, Parashar UD, World Health Organization- Coordinated Global Rotavirus Surveillance, Network. Global, regional, and national estimates of rotavirus mortality in children b5 years of age, 2000-2013. Clin Infect Dis 2016;62(Suppl. 2):S96-S105. https://doi.org/10.1093/cid/civ1013.
  2. Mwenda JM, Tate JE, Parashar UD, Mihigo R, Agocs M, Serhan F, et al. African rotavi- rus surveillance network: a brief overview. Pediatr Infect Dis J 2014;33(Suppl. 1): S6-8. https://doi.org/10.1097/INF.0000000000000174.
  3. Tsolenyanu E, Seheri M, Dagnra A, Djadou E, Togossou S, Nyada M, et al. Surveillance for rotavirus gastroenteritis in children less than 5 years of age in Togo. Pediatr Infect Dis J 2014;33(Suppl. 1):S14-8. https://doi.org/10.1097/INF.0000000000000046.
  4. Abebe A, Teka T, Kassa T, Seheri M, Beyene B, Teshome B, et al. Hospital-based sur- veillance for rotavirus gastroenteritis in children younger than 5 years of age in Ethiopia: 2007-2012. Pediatr Infect Dis J 2014;33(Suppl. 1):S28-33. https://doi. org/10.1097/INF.0000000000000048.
  5. Breiman RF, Cosmas L, Audi A, Mwiti W, Njuguna H, Bigogo GM, et al. Use of population-based surveillance to determine the incidence of rotavirus gastroenteri- tis in an urban slum and a rural setting in Kenya. Pediatr Infect Dis J 2014;33(Suppl. 1):S54-61. https://doi.org/10.1097/INF.0000000000000094.
  6. Okabe N, Yamashita K, Taniguchi K, Inouye S. Influenza surveillance system of Japan and acute encephalitis and encephalopathy in the influenza season. Pediatr Int 2000; 42:187-91. https://doi.org/10.1046/j.1442-200x.2000.01206.x.
  7. Marx MA, Rodriguez CV, Greenko J, Das D, Heffernan R, Karpati AM, et al. Diarrheal illness detected through syndromic surveillance after a massive power outage: New York City, August 2003. Am J Public Health 2006;96:547-53. https://doi.org/10. 2105/AJPH.2004.061358.
  8. Pivette M, Mueller JE, Crepey P, Bar-Hen A. Surveillance of gastrointestinal disease in France using drug sales data. Epidemics 2014;8:1-8. https://doi.org/10.1016/j. epidem.2014.05.001.
  9. Pivette M, Mueller JE, Crepey P, Bar-Hen A. Drug sales data analysis for outbreak de- tection of infectious diseases: a systematic literature review. BMC Infect Dis 2014; 14:604. https://doi.org/10.1186/s12879-014-0604-2.
  10. Kirian ML, Weintraub JM. Prediction of gastrointestinal disease with over-the- counter diarrheal remedy sales records in the San Francisco Bay Area. BMC Med In- form Decis Mak 2010;10:39. https://doi.org/10.1186/1472-6947-10-39.
  11. The Census of Japan in 2015. Available at: http://www.pref.osaka.lg.jp/attach/1891/ 00210094/27jinkoutoukihon.pdf. [Accessed November 11, 2019.]
  12. Shimada T, Sunagawa T, Taniguchi K, Yahata Y, Kamiya H, Yamamoto KU, et al. De- scription of hospitalized cases of influenza A(H1N1)pdm09 infection on the basis of the national hospitalized-case surveillance, 2009-2010. Japan Jpn J Infect Dis 2015; 68:151-8. https://doi.org/10.7883/yoken.JJID.2014.125.
  13. Nakamura Y, Sugawara T, Kawanohara H, Ohkusa Y, Kamei M, Oishi K. Evaluation of estimated number of influenza patients from national sentinel surveillance using the national database of electronic medical claims. Jpn J Infect Dis 2015;68:27-9. https:// doi.org/10.7883/yoken.JJID.2014.092.
  14. Murakami Y, Hashimoto S, Kawado M, Ohta A, Taniguchi K, Sunagawa T, et al. Esti- mated number of patients with influenza A(H1)pdm09, or other viral types, from 2010 to 2014 in Japan. PloS One 2016;11:e0146520. https://doi.org/10.1371/jour- nal.pone.0146520.
  15. The guidelines for National Epidemiological Surveillance of Infectious diseases: in- fluenza. Available at: https://www.mhlw.go.jp/stf/shingi/2r9852000002oeqs-att/ 2r9852000002oetv.pdf. [Accessed November 20, 2019.]
  16. Sakurai A, Morimura N, Takeda M, Miura K, Kiyotake N, Ishihara T, et al. A retrospec- tive quality assessment of the 7119 call triage system in Tokyo – telephone triage for non-ambulance cases. J Telemed Telecare 2014;20:233-8. https://doi.org/10.1177/ 1357633×14536347.
  17. Lucero-Obusan C, Winston CA, Schirmer PL, Oda G, Holodniy M. Enhanced influenza surveillance using telephone triage and electronic syndromic surveillance in the De- partment of Veterans Affairs, 2011-2015. Public Health Rep. 2017;132:16s-22s. doi: https://doi.org/10.1177/0033354917709779
  18. Cooper D, Chinemana F. NHS Direct derived data: an exciting new opportunity or an epidemiological headache? J Public Health (Oxf). 2004;26:158-160. doi:https://doi. org/10.1093/pubmed/fdh133
  19. Moore K. Real-time syndrome surveillance in Ontario, Canada: the potential use of emergency departments and Telehealth. Eur J Emerg Med 2004;11:3-11. https:// doi.org/10.3201/eid1505.081174.
  20. van Dijk A, Aramini J, Edge G, Moore KM. real-time surveillance for respiratory dis- ease outbreaks, Ontario, Canada. Emerg Infect Dis 2009;15:799-801. https://doi.org/ 10.3201/eid1505.081174.
  21. Telephone triage service in Osaka. Available at: https://www.city.osaka.lg.jp/shobo/ page/0000052526.html. [Accessed June 28, 2019.]
  22. The Information Center of Infectious Disease in Osaka Prefecture. Available at: http://www.iph.pref.osaka.jp/infection/2-old.html. [Accessed November 20, 2019.]
  23. Bjelkmar P, Hansen A, Schonning C, Bergstrom J, Lofdahl M, Lebbad M, et al. Early outbreak detection by linking health advice line calls to water distribution areas ret- rospectively demonstrated in a large waterborne outbreak of cryptosporidiosis in Sweden. BMC Public Health 2017;17:328. https://doi.org/10.1186/s12889-017- 4233-8.
  24. Berger M, Shiau R, Weintraub JM. Review of syndromic surveillance: implications for waterborne disease detection. J Epidemiol Community Health 2006;60:543-50. https://doi.org/10.1136/jech.2005.038539.