International Journal of Health & Allied Sciences

ORIGINAL ARTICLE
Year
: 2013  |  Volume : 2  |  Issue : 2  |  Page : 95--98

Seasonality of tuberculosis in rural West Bengal: A time series analysis


Ranadip Chowdhury1, Abhijit Mukherjee1, Somnath Naska1, Mrinmoy Adhikary2, Saibendu Kumar Lahiri1,  
1 Department of Community Medicine, RG Kar Medical College and Hospital, Kolkata, India
2 Department of Community Medicine, Vardhaman Mahavir Medical College and Safdarjung Hospital, New Delhi, India

Correspondence Address:
Ranadip Chowdhury
Department of Community Medicine, 3rd Floor Academic Building, RG Kar Medical College and Hospital, 1, Kshudiram Bose Sarani, Kolkata - 700 004
India

Abstract

Background: There is a recent concern about the global climatic change that is expected to have broad health impacts. The health effects of extreme weather events include a spectrum of wide variety of impacts. According to a study in Northern India, tuberculosis (TB) diagnosis peaked between April and June, and it reached a nadir between October and December. However, no seasonality was reported from South India. Aims: This study is aimed to assess the seasonality of TB in rural West Bengal and to develop a univariate time series model. Settings and Design: Retrospective record-based study was carried out at Amdanga tuberculosis unit (TU), North 24-parganas, West Bengal. Materials and Methods: A total of 1507 new TB cases were registered in the TB register of the TU during January-2008 to December-2011 period were taken for this study. Statistical Analysis: Seasonal adjusted factor (SAF), autocorrelation function (ACF), partial autocorrelation function (PACF), and seasonal autoregressive integrated moving average (SARIMA) methods were applied by using the SPSS 16.0 version. Results: ACF and PACF at lag 12 shows significant pick suggesting seasonal component of the TB series. SAF showed peak seasonal variation from March to June and nadir from October to December in additive model. Univariate model by expert-modeler in the SPSS showed SARIMA ((0,0,0)(1,2,0) 12 could best predict the model with 54.3% variability. Conclusion: A seasonal pattern of TB was observed. This information would be usefulfor administration and managers to take extra care to arrange and provide extra facilities during the peak seasons.



How to cite this article:
Chowdhury R, Mukherjee A, Naska S, Adhikary M, Lahiri SK. Seasonality of tuberculosis in rural West Bengal: A time series analysis.Int J Health Allied Sci 2013;2:95-98


How to cite this URL:
Chowdhury R, Mukherjee A, Naska S, Adhikary M, Lahiri SK. Seasonality of tuberculosis in rural West Bengal: A time series analysis. Int J Health Allied Sci [serial online] 2013 [cited 2024 Mar 28 ];2:95-98
Available from: https://www.ijhas.in/text.asp?2013/2/2/95/115684


Full Text

 Introduction



There is a recent concern about the global climate change that is expected to have broad health impacts. [1] Such human health impacts are most likely to occur where extreme weather and a vulnerable population combine together. The health effects of extreme weather events include a spectrum of wide variety of impacts, such as physical injury, poor nutritional status because of the effect of climate on food, an increase in a variety of respiratory, and diarrheal diseases, increased vector-borne diseases, an increase in allergic disorders and increased pollution levels. Seasonal variation of tuberculosis (TB) has been reported from different parts of the world; [2],[3],[4] although, no definite and consistent pattern has been observed. One earlier report from India has assessed trends using quarterly reports from districts with stable TB control programs. [4] According to this study in Northern India, TB diagnosis peaked between April and June, and it reached a nadir between October and December. However, no seasonality was reported from South India. [4] A hospital based study from Delhi showed seasonal pattern in pulmonary TB and extra pulmonary (EPTB) cases. [5]

The time series analyses methodology has been increasingly used in the field of epidemiological research on infectious diseases, particularly in the assessment of health services. [6] In health science research, autoregressive integrated moving average (ARIMA) models [6] as well as seasonal autoregressive integrated moving average (SARIMA), [7] models are useful tools for analyzing time series data containing ordinary or seasonal trends to develop a predictive forecasting model. This study is aimed to see the seasonality of TB in rural West Bengal and to develop univariate time series models.

 Materials and Methods



The present study was a retrospective record-based study, carried out at Amdanga tuberculosis unit (TU), North 24 Parganas West Bengal, India is a predominantly rural TU serving a population of approximately 0.47 million. The TU is located 12 km from the district headquarters at Barasat and 29 km from Kolkata, the capital of West Bengal. This TU covers two rural community developmental block of West Bengal namely Amdanga block and Habra-II block. A total 1507 new TB cases were registered in the TB register of the TU during January-2008 to December-2011 period were taken for this study.

Statistical analysis

Seasonality of the series was assessed by taking 12 window period moving average and then seasonal factors for the series was formulated. The stationarity of the series was made by means of seasonal differencing. The stationarity of the series was checked by autocorrelation function and partial autocorrelation function. Expert modeler of SPSS 16 software was used to fit the best suitable model for the series.

 Results



A total 1507 new TB cases were registered in the TB register of the TU during January-2008 to December-2011 period.

Among the total study population 51 (3.4%) were children and 1456 (96.6%) were adult according to Revised National Tuberculosis Control Programme (RNTCP) register [Table 1].{Table 1}

The series exhibited a number of peaks, aside from the small-scale fluctuations, the significant peak appeared to be separated by more than a few months [Figure 1].{Figure 1}

The autocorrelation function showed a significant peak at a lag of 12 (autocorrelation = 0.731; Box-Ljung statistics P = 0.000) suggested the presence of a seasonal component in the data [Figure 2].{Figure 2}

The significant peak at a lag of 12 in the partial autocorrelation function confirmed the presence of a seasonal component in the data [Figure 3].{Figure 3}

As the amplitude of both the seasonal and irregular variations does not change as the level of the trend rises or falls, here additive model was used [Figure 1]. In additive model, the seasonal adjustments are added to the seasonally adjusted series to obtain the observed values. This adjustment attempts to remove the seasonal effect from a series in order to look at other characteristics of interest that may be "masked" by the seasonal component in effect seasonal components that do not depend on the overall level of the series. Observations without seasonal variation have a seasonal component of 0. Above table showed that from the month of March to June the seasonal adjusted factor (SAF) of TB was more than 0 that is in these months the registered TB cases were more above the typical months. Among these months, March with SAF = 24.413 had the highest SAF, that is in this month, the registered TB cases were more than 24.4% above the typical months [Table 2].{Table 2}

Above figure showed that the predictive and actual values matched reasonably well and there was a consistency in the trend [Figure 4].{Figure 4}

ARIMA model types are listed using the standard notation of ARIMA (p, d, q) (P, D, Q), where P is the order of autoregression, d is the order of differencing (or integration), and q is the order of moving-average, and (P, D, Q) are their seasonal counterparts. [8] Although the time series modeler offers a number of different goodness-of-fit statistics, here stationary R-squared value was used. This statistic provides an estimate of the proportion of the total variation in the series that is explained by the model and is preferable to ordinary R-squared when there is a trend or seasonal pattern, as is the case here. Larger values of stationary R-squared (up to a maximum value of 1) indicate better fit. A value of 0.543 meant that the model could explain 54.3% of the observed variation in the series [Table 3].{Table 3}

The Ljung-Box statistic, also known as the modified Box-Pierce statistic, provides an indication of whether the model is correctly specified. A significance value less than 0.05 implies that there is structure in the observed series which is not accounted for by the model. The value of 0.50 shown here was not significant (P = 0.180), so the model was correctly specified. The expert modeler detected one point that was considered to be outliers [Table 3].

All the coefficient of SARIMA (0,0,0) (1,2,0) 12 model was significant [Table 4].{Table 4}

Both the above figures showed the stationarity of the transformed series [Figure 5] and [Figure 6].{Figure 5}{Figure 6}

 Discussion



In an assessment of seasonal trends, Thorpe et al. [4] reported that diagnosis of TB peaked between April and June, and reached nadir between October and December. Seasonal variation in TB incidence has also been reported from Europe and South Africa particularly in children. [9],[10] The present study demonstrated the seasonality of TB cases in a TU of rural West Bengal. The study corroborates with the observations by Thorpe et al. [4] showing the seasonal variation, which were peaked between March and June and dropped during October and December. Areas in the north India had the highest seasonal variation and low or no seasonality was noted in central and southern regions of India in that study. [4] Similar observations of seasonal variation has also been reported by others from different countries. [2],[3],[11],[12] The causes of seasonality of TB are still unclear. There could be many speculations about the high number of TB cases during summer; one of the reasons could be that the process of transmission of TB infection is intensified by increased time spent in overcrowded, poorly ventilated housing conditions, and by an increased frequency of coughing from other respiratory infections. It could also be due to a rise in temperature during summer. It is speculated that because of harvesting period of wheat crop in north India during this period from March and June, lots of dust particles come out when the wheat's seed is taken out and very fine particles are dispersed in the atmosphere producing air pollution. Other possible explanations of such variability given by different authors include low levels of vitamin D, which are known to affect macrophage function and cell-mediated immunity that might result in impaired cellular immunity leading to reactivation of dormant mycobacterial infection. [13] However, the exact cause of such variability is currently unexplained. Being a retrospective study, there are certain limitations of the study: Information on occupation, travel, support of family etc., was not available and this could not be taken into account. The study is an observational study and cause and effect could not be established. It is also to be noted that it is a TU-based study and more sick people are likely to attend the hospital as compared to a population-based study; therefore, it is difficult to predict the variation in the actual population.

 Acknowledgments



We want to acknowledge Mr. Tapas Das, LT Amdanga TU for his support in doing the study.

References

1Patz JA, Engelberg D, Last J. The effects of changing weather on public health. Annu Rev Public Health 2000;21:271-307.
2Luquero FJ, Sanchez-Padilla E, Simon-Soria F, Eiros JM, Golub JE. Trend and seasonality of tuberculosis in Spain, 1996-2004. Int J Tuberc Lung Dis 2008;12:221-4.
3Nagayama N, Ohmori M. Seasonality in various forms of tuberculosis. Int J Tuberc Lung Dis 2006;10:1117-22.
4Thorpe LE, Frieden TR, Laserson KF, Wells C, Khatri GR. Seasonality of tuberculosis in India: Is it real and what does it tell us? Lancet 2004;364:1613-4.
5Behera D, Sharma PP. A retrospective study of seasonal variation in the number of cases diagnosed at a tertiary care tuberculosis hospital. Indian J Chest Dis Allied Sci 2011;53:145-52.
6Allard R. Use of time-series analysis in infectious disease surveillance. Bull World Health Organ 1998;76:327-33.
7Naish S, Hu W, Nicholls N, Mackenzie JS, McMichael AJ, Dale P, et al. Weather variability, tides, and Barmah Forest virus disease in the Gladstone region, Australia. Environ Health Perspect 2006;114:678-83.
8Choudhury Z, Banu S, Islam MA. Forecasting dengue incidence in Dhaka, Bangladesh: A time-series analysis. Dengue Bulletin 2008;32:29-37.
9Douglas AS, Strachan DP, Maxwell JD. Seasonality of tuberculosis: The reverse of other respiratory diseases in the UK. Thorax 1996;51:944-6.
10Schaaf HS, Nel ED, Beyers N, Gie RP, Scott F, Donald PR. A decade of experience with Mycobacterium tuberculosis culture from children: A seasonal influence on incidence of childhood tuberculosis. Tuber Lung Dis 1996;77:43-6.
11Leung CC, Yew WW, Chan TY, Tam CM, Chan CY, Chan CK, et al. Seasonal pattern of tuberculosis in Hong Kong. Int J Epidemiol 2005;34:924-30.
12Atun RA, Samyshkin YA, Drobniewski F, Kuznetsov SI, Fedorin IM, Coker RJ. Seasonal variation and hospital utilization for tuberculosis in Russia: Hospitals as social care institutions. Eur J Public Health 2005;15:350-4.
13Janmeja AK, Mohapatra PR. Seasonality of tuberculosis. Lancet 2004;364:704-5.