|Year : 2021 | Volume
| Issue : 1 | Page : 8-12
A diffused community tuberculosis outbreak that could be detected earlier using surveillance data, Japan, 2012–2014
Masaki Ota1, Kazuhiro Uchimura2, Susumu Hirao1
1 Department of Technical Assistance to National Tuberculosis Programmes, Tokyo, Japan
2 Department of Epidemiology and Clinical Research, Research Institute of Tuberculosis, Tokyo, Japan
|Date of Submission||24-Nov-2020|
|Date of Acceptance||30-Nov-2020|
|Date of Web Publication||28-Feb-2021|
3-1-24, Matsuyama, Kiyose City, Tokyo 2048533
Source of Support: None, Conflict of Interest: None
Background: Early detection of an outbreak is a role of disease surveillance systems; however, tuberculosis (TB) surveillance systems were underutilized to detect the outbreaks. In mid-2015, a local health office of central Japan noticed the number of TB cases of city in 2012–2014 were well above the expected numbers. This study was conducted to determine whether and when a community TB outbreak could be detected and characterize the cases using the national surveillance data. Method: The surveillance data of the A city and surrounding areas were retrospectively reviewed and analyzed for 2006–2018. Results: The TB notification rates of the A city from 2012 to 2014 were 28.0 (95% confidence interval [CI]: 20.3–38.4), 26.0 (95%CI: 18.6–36.0), 28.2 (95%CI: 20.3–38.4) per 100,000 population, respectively, higher than that of the entire prefecture (13.6, 13.0, 13.3, respectively). Similarly, in the neighboring B city, the rates of 2012 and 2014 were 51.0 (95%CI: 27.2–87.2) and 51.2 (95%CI: 27.3–87.5), respectively, higher than that of its parent prefecture (13.4 and 12.7, respectively). By the end of July 2012 (A city) or August 2012 (B city), the accumulated numbers of TB cases exceeded the previous annual TB cases. The average TB notification rates of A and B cities for 2012–2014 were higher than the surrounding areas. Conclusion: A community TB outbreak without well-defined setting could be detected by monitoring TB surveillance data.
Keywords: Disease outbreak, early warning, epidemiology, tuberculosis
|How to cite this article:|
Ota M, Uchimura K, Hirao S. A diffused community tuberculosis outbreak that could be detected earlier using surveillance data, Japan, 2012–2014. Int J Mycobacteriol 2021;10:8-12
|How to cite this URL:|
Ota M, Uchimura K, Hirao S. A diffused community tuberculosis outbreak that could be detected earlier using surveillance data, Japan, 2012–2014. Int J Mycobacteriol [serial online] 2021 [cited 2021 Jun 14];10:8-12. Available from: https://www.ijmyco.org/text.asp?2021/10/1/8/310507
| Introduction|| |
Public health surveillance is the continuous, systematic collection, analysis, and interpretation of health-related data needed for the planning, implementation, and evaluation of public health practice, and one of the roles is to serve as an early warning system for impending public health emergencies. Public health surveillance systems in various countries detected the outbreaks of Guillain–Barre syndrome, enterohaemorrhagic Escherichia coli O157, cholera, leptospirosis, and typhoid fever among others.,,,, In tuberculosis (TB) control, the United States Centers for the Disease Control and Prevention monitors TB surveillance data and detected resurgence of TB in late 1985. Furthermore, in the United States, outbreaks were retrospectively predicted using the national TB surveillance data and genotyping., However, there are not many detailed reports that reviewed how early a TB outbreak could be detected using the surveillance data and that described the TB outbreak in other countries, including Japan. In Japan, TB notification rate has declined in the past six decades from 698 per 100,000 population in 1951 to 17.7 per 100,000 population in 2013. However, about 8000 smear-positive TB cases are still reported every year, and these infectious TB cases pose a public health threat to the community. TB outbreaks have also been reported, involving hospitals, workplaces, schools, and sometimes homeless people.,,,
In mid-2015, a local health office of P prefecture, central Japan, had noticed that the number of TB cases reported from A city within its jurisdiction from 2012 to 2014 seemed to exceed the expected numbers. It was also found that most TB cases were patrons of a pachiko, parlor located in the neighboring B city of Q prefecture. Pachinko parlors are a Japanese style of casinos, which often have 100–1000, sometimes thousands of pinball and slot machines to accommodate gamble lovers. In early 2016, one of the authors (MO) was consulted about the possible community TB outbreak. Reviewing the national TB surveillance data, this study aims to determine whether and when the TB outbreak could be detected, to characterize the TB cases of A and B cities of the outbreak, and to identify the specific persons at risk.
| Method|| |
The A city of P prefecture is located inland in the central Japan with a population of 140,000 people, 26% of which are aged 65 years or older. B city of Q prefecture is located to North-west of a city. It has a population of about 25,000, 27% of which are aged 65 years or older. Both cities are about 2 h car rides from Tokyo, the capital area.
A TB case was defined as one who had bacteriologically positive TB in the sputum sample determined by smear microscopy, culture, or nucleic acid amplification, or was determined to have TB by a physician through a chest X-ray, histological or pathological tests, or clinically.
Epidemiological information was extracted from the national surveillance data for TB cases registered from 2006 through 2018, including sex, age, city of residence, and type of TB disease (either pulmonary or extrapulmonary). No patients' names were included in the national TB surveillance data.
TB cases notified from the A and B cities, and the surrounding areas were epidemiologically described in terms of time, place, and person, particularly in relation to sex and age groups. The notification rates of the A and B cities were compared with that of the P and Q prefectures by year and by sex and age group. When notification rates of the A and B cities of a specific year were significantly higher than their prefectures, we also determined which months of the year the rates exceeded expected notification rates at latest.
Statistical tests were carried out with the R (The R Foundation for Statistical Computing, Vienna, Austria). Ninety-five percent confidence intervals (CIs) were calculated assuming that notification rates of TB cases distributed in binomial distribution. Fisher's exact test was employed for the comparison of proportions. P < 0.05 was considered statistically significant.
We obtained a waiver of the ethical review for the study from the Institutional Review Board of the Research Institute of Tuberculosis (Decision letters #28-26 and #30-20 of RIT-IRB) because this study was retrospective, the secondary use of the data that had already been collected by the local health office, and it did not involve confidential information. However, the board instructed that the investigators take precautions to prevent the profiles of the TB patients from being identified because the study involved relatively small cities where the patients could be identified. Thus, the authors could not mention the exact names of the cities and the prefectures in the manuscript.
| Results|| |
From 2006 to 2018, a total of 456 TB cases reported in A city of P prefecture, whereas 75 reported in B city of Q prefecture. [Figure 1] shows TB notification rates of A and B cities compared with the entire P and Q prefectures, respectively. The TB notification rates of A city in 2012–2014 and in 2016–2017 were 28.0 (95% CI: 20.3–38.4), 26.0 (95% CI: 18.6–36.0), 28.2 (95% CI: 20.3–38.4), 23.5 (95% CI: 16.3–32.8), and 18.7 (95% CI: 12.3–27.1) per 100 000 population, respectively, being higher than that of the entire P prefecture (13.6, 13.0, 13.3, 12.2, and 11.2, respectively, per 100 000 population) with a statistical significance. Similarly, the notification rates of B city in 2012 and 2014 were 51.0 (95%CI: 27.2–87.2) and 51.2 (95%CI: 27.3–87.5) per 100 000 population, respectively, being higher than that of the entire P prefecture (13.4 and 12.7, respectively, per 100 000 population) with a statistical significance.
|Figure 1: Tuberculosis notification rates of city A of prefecture P (a) and city B of prefecture Q (b), 2006–2018. Pop.: Population, TB: Tuberculosis. Bars indicate 95% confidence intervals|
Click here to view
TB epidemic curves of A and B cities specifically for 2012 are shown in [Figure 2]. At the end of July 2012, the accumulated number of TB cases of A city from the beginning of the year had become 32 (notification rate up to the end of July: 21.8 per 100 000 population, 95% CI: 14.9–30.9) and the lower end of the 95% CI exceeded the TB notification rate of P prefecture (13.6 per 100 000 population) for 2012. Similarly, at the end of August 2012, the accumulated number of TB cases of B city from the beginning of the year had become 8 (notification rate up to the end of August: 31.4 per 100 000 population, 95% CI: 13.5–61.8) and the lower end of the 95% CI exceeded the TB notification rate of Q prefecture (13.2 per 100 000 population) for 2012. It is also worth noting that on March 2012, 11 TB cases were reported in A city, whereas normally only 0–6 cases were reported per month from 2006 to 2011 (detailed data not shown).
|Figure 2: Epidemic curves of tuberculosis cases reported in city A (Left) and city B (Right), 2012. Arrows indicate the months on which the notification rates derived from the accumulated tuberculosis cases from the beginning of the year exceeded expected rates of each prefecture. TB: Tuberculosis|
Click here to view
The geographic distribution of the average notification rates per 100,000 population from 2012 through 2014 by city is shown in [Figure 3]. The B city had the highest average notification rates, followed by the A city. The surrounding cities had lower average notification rates than the A and B cities.
|Figure 3: The geographic distribution of average tuberculosis notification rates by city surrounding A city, Japan, 2012–2014. A: City A, B: City B. Rates are expressed per 100,000 population. Thick lines indicate the boundaries between prefectures, whereas thin lines the boundaries of cities|
Click here to view
[Figure 4] shows age and sex distribution of TB cases of A city and P prefecture, and B city and Q prefecture from 2012 through 2014. In A city, the TB notification rates in males aged 35–64 years and in females aged 15–24, 35–54, and over 65 years were higher than that of the entire P prefecture with a statistical significance. Similarly, in B city, the TB notification rates in males aged 15–34, 45–74, and over 85 years were higher than that of the entire Q prefecture with a statistical significance.
|Figure 4: Age and sex distribution of tuberculosis cases of city A and prefecture P (Upper) and of city B and prefecture Q (lower), Japan, 2012–2014. *P < 0.05, Pref.: Prefecture|
Click here to view
| Discussion|| |
We retrospectively reviewed the national TB surveillance data for the central Japan from 2012 through 2014 and verified that a community TB outbreak in two municipalities could be detected earlier: At the end of July 2012 (A city) or August 2012 (B city) at latest, the community outbreak could have been detected based on the significantly higher TB notification rates in these cities. Specifically for A city, the number of TB cases reported in March 2012 suddenly almost doubled, and this could also have given the health authority an opportunity of an early warning of the community outbreak. Higher TB notification rates in A city than the parent prefecture continued through 2017, suggesting that the outbreak may have continued for several years after its start. The geographic analysis for the outbreak found that the patients were concentrated only in the A and B cities. The review also found that mainly young to middle-aged males and almost all females except children (A city) and young to elder males (B city) had more risk of developing TB disease.
In B city, almost all age groups of males were at high risk of developing TB, whereas females were not at high risk, possibly supporting the local health office's observation that TB transmission occurred in the pachinko parlor because the majority of the patrons are males and they might have picked up TB bacilli there. In A city, not only the males but also females were at high risk of TB disease. The situation of the males might be the same with that in B city; however, for the females in the A city, the reason is unclear.
Public surveillance systems have been shown to be effective in detecting various disease outbreaks. A Guillain–Barre syndrome outbreak was detected in the United States in 1978 after vaccination for swine influenza was widely conducted. An enterohemorrhagic E. coli O157 outbreak was detected in Western Japan in early 2004. A small cholera outbreak was detected among those coming back from Indonesia to Japan in 2005. A leptospirosis outbreak was detected in Orissa, India in 2002. A typhoid fever outbreak was detected after a flood in North Western Ethiopia in 2014. TB outbreaks were also detected by TB surveillance systems in the United States using both the national TB surveillance data and TB genotyping;, however, as we have shown, TB outbreaks could readily be detected by only monitoring the TB surveillance data on the monthly basis. Pachinko parlors were often suspected as a site of TB infection in Japan, because patrons of pachinko parlors usually stay the premises long and frequent; however, only a few TB outbreaks were proven to be involving a pachinko parlor with field and/or molecular epidemiological studies.,
Our study has a few limitations. We conducted an epidemiological review on the national TB surveillance data but did not conduct field investigation, particularly a case–control study to identify possible risk exposures, such as the use of the pachinko parlor by the patients. Molecular epidemiological investigation was not comprehensively conducted, either, because the outbreak occurred in two prefectures: They employed different methods (choosing loci for comparison) for molecular epidemiology for Mycobacterium tuberculosis, and the results were not shared and matched. However, the authors were told that most cases within each city shared identical strains.
In conclusion, a community TB outbreak without well-defined setting could be detected by monitoring TB surveillance data. recommend that local health offices, and preferably prefectures as well, should monitor their TB surveillance data at least on the monthly basis and initiate an outbreak investigation in case numbers of TB patients reported exceed expectation. If it is a community outbreak and it does not involve a well-defined setting, the local health office should try to conduct a case–control study to identify possible exposure source(s).
The authors would like to thank the public health staff members of the relevant prefectures for their essential contributions to the tuberculosis surveillance system, without which this investigation could not have been taking place.
Financial support and sponsorship
This study was partially supported by the Japan Agency for Medical Research and Development (Grants #: JP20fk0108127).
Conflicts of interest
There are no conflicts of interest.
| References|| |
Retailliau HF, Curtis AC, Storr G, Caesar G, Eddins DL, Hattwick MA. Illness after influenza vaccination reported through a nationwide surveillance system, 1976-1977. Am J Epidemiol 1980;111:270-8.
Ota M, Kitsutani P, Tada Y, Ohyama T. Small cholera outbreak among those coming back from Indonesia. Infectious Disease Surveillance Report. 2006;8:7-8. Available from: https://idsc.niid.go.jp/iasr/27/311/dj3113.html 20
. [Last accessed on 11 Mar 2020].
Jena AB, Mohanty KC, Devadasan N. An outbreak of leptospirosis in Orissa, India: The importance of surveillance. Trop Med Int Health 2004;9:1016-21.
Ota M, Toyama Y, Kon M, Yoza T, Belay BB. Strengthening the communicable disease surveillance and response system, Amhara Region, Ethiopia, 2012-2014. J Int Health 2017;32:1-8.
Centers for Disease Control (CDC). Tuberculosis–United States, first 39 weeks, 1985. MMWR Morb Mortal Wkly Rep 1985;34:625-7.
Althomsons SP, Kammerer JS, Shang N, Navin TR. Using routinely reported tuberculosis genotyping and surveillance data to predict tuberculosis outbreaks. PLoS One 2012;7:e48754.
Althomsons SP, Hill AN, Harrist AV, France AM, Powell KM, Posey JE, et al
. Statistical method to detect tuberculosis outbreaks among endemic clusters in a low-incidence setting. Emerging Infect Dis 2018;24:573-5.
Katsuda N, Hirosawa T, Reyer JA, Hamajima N. Roles of public health centers in tuberculosis control in Japan. Nagoya J Med Sci 2015;77:19-28.
Tuberculosis Surveillance Center, RIT. JATA. Tuberculosis annual report 2013--(4) Tuberculosis treatment and treatment outcomes. Kekkaku 2015;90:595-604.
Tasaka M, Koeda E, Takahashi C, Ota M. A tuberculosis outbreak in a psychiatric hospital: Kanagawa, Japan, 2012. Epidemiol Infect 2020;148:e7.
Fujikawa A, Fujii T, Mimura S, Takahashi R, Sakai M, Suzuki S, et al
. Tuberculosis contact investigation using interferon-gamma release assay with chest x-ray and computed tomography. PLoS One 2014;9:e85612.
Tasaka M, Shimamura T, Iwata M, Toyozawa T, Ota Ml. A tuberculosis contact investigation involving a large number of contacts tested with interferon-gamma release assay at a nursing school: Kanagawa, Japan, 2012. Western Pac Surveill Response J 2018;9:4-8.
Endo M, Ota M, Kayebeta A, Takahashi I, Nagata Y. A tuberculosis outbreak at an insecure, temporary housing facility, manga café, Tokyo, Japan, 2016-2017. Epidemiol Infect 2019;147:e222.
Komoto Y, Shoun A, Akiyama K, Sakamoto A, Sato T, Nishimura N, et al
. Development and validation of the Pachinko/Pachi-Slot Playing Ambivalence Scale. Asian J Gambl Issues Public Health 2017;7:3.
Toyota M. Evaluation of tuberculosis transmission routes in an outbreak among young adults for developing an effective method for contact investigations. Kekkaku 2012;87:757-63.
Seto J, Ahiko T, Wada T, Hase A, Yamada K. Effectiveness of comprehensive variable number of tandem repeat (VNTR) analysis in areas with a low incidence of tuberculosis. Kekkaku 2013;88:535-42.
[Figure 1], [Figure 2], [Figure 3], [Figure 4]