Limits...
Time series analysis of malaria in Afghanistan: using ARIMA models to predict future trends in incidence

View Article: PubMed Central - PubMed

ABSTRACT

Background: Malaria remains endemic in Afghanistan. National control and prevention strategies would be greatly enhanced through a better ability to forecast future trends in disease incidence. It is, therefore, of interest to develop a predictive tool for malaria patterns based on the current passive and affordable surveillance system in this resource-limited region.

Methods: This study employs data from Ministry of Public Health monthly reports from January 2005 to September 2015. Malaria incidence in Afghanistan was forecasted using autoregressive integrated moving average (ARIMA) models in order to build a predictive tool for malaria surveillance. Environmental and climate data were incorporated to assess whether they improve predictive power of models.

Results: Two models were identified, each appropriate for different time horizons. For near-term forecasts, malaria incidence can be predicted based on the number of cases in the four previous months and 12 months prior (Model 1); for longer-term prediction, malaria incidence can be predicted using the rates 1 and 12 months prior (Model 2). Next, climate and environmental variables were incorporated to assess whether the predictive power of proposed models could be improved. Enhanced vegetation index was found to have increased the predictive accuracy of longer-term forecasts.

Conclusion: Results indicate ARIMA models can be applied to forecast malaria patterns in Afghanistan, complementing current surveillance systems. The models provide a means to better understand malaria dynamics in a resource-limited context with minimal data input, yielding forecasts that can be used for public health planning at the national level.

Electronic supplementary material: The online version of this article (doi:10.1186/s12936-016-1602-1) contains supplementary material, which is available to authorized users.

No MeSH data available.


Related in: MedlinePlus

Malaria cases per month, from January 2005 up to September 2015, reported from health facilities throughout Afghanistan. a Adjusted for monthly cases per 10,000 outpatient clients, as reported from health facilities. b Unadjusted monthly malaria cases. c Total number of outpatient cases, reflecting trends health services utilization and reporting. Although the unadjusted data do not exhibit any trend beyond seasonality, because fewer centers were reporting at the beginning of the period (around 1000 centers compared to well over 2000 in 2015 [42]) and health services utilization increased substantially and proportionally for all parts of the country, adjustment was necessary to account for under-reporting. Subsequent analyses were performed using the adjusted rates
© Copyright Policy - OpenAccess
Related In: Results  -  Collection

License 1 - License 2
getmorefigures.php?uid=PMC5120433&req=5

Fig2: Malaria cases per month, from January 2005 up to September 2015, reported from health facilities throughout Afghanistan. a Adjusted for monthly cases per 10,000 outpatient clients, as reported from health facilities. b Unadjusted monthly malaria cases. c Total number of outpatient cases, reflecting trends health services utilization and reporting. Although the unadjusted data do not exhibit any trend beyond seasonality, because fewer centers were reporting at the beginning of the period (around 1000 centers compared to well over 2000 in 2015 [42]) and health services utilization increased substantially and proportionally for all parts of the country, adjustment was necessary to account for under-reporting. Subsequent analyses were performed using the adjusted rates

Mentions: No public census has been conducted in Afghanistan since 1979 [35], and other sources of demographic data [e.g. WHO, International Monetary Fund (IMF), Central Statistics Office (CSO)] cannot be corroborated with each other. In addition, utilization of health services was not homogenous throughout the study period (Fig. 2c), as the number facilities has risen from under 1000 to over 2000 centres since 2004. Hence, data on the total monthly new outpatient department visits were used as denominator in order to control for demographic and reporting trends. To verify that this did not lead to a bias in the trends over time due to recent changes in outpatient health service utilization occurring primarily in regions of either low or high malaria incidence, the overall of trend of malaria obtained after adjustment was compared with the weighted average of individual trends of provinces adjusted for their level of health service utilization.


Time series analysis of malaria in Afghanistan: using ARIMA models to predict future trends in incidence
Malaria cases per month, from January 2005 up to September 2015, reported from health facilities throughout Afghanistan. a Adjusted for monthly cases per 10,000 outpatient clients, as reported from health facilities. b Unadjusted monthly malaria cases. c Total number of outpatient cases, reflecting trends health services utilization and reporting. Although the unadjusted data do not exhibit any trend beyond seasonality, because fewer centers were reporting at the beginning of the period (around 1000 centers compared to well over 2000 in 2015 [42]) and health services utilization increased substantially and proportionally for all parts of the country, adjustment was necessary to account for under-reporting. Subsequent analyses were performed using the adjusted rates
© Copyright Policy - OpenAccess
Related In: Results  -  Collection

License 1 - License 2
Show All Figures
getmorefigures.php?uid=PMC5120433&req=5

Fig2: Malaria cases per month, from January 2005 up to September 2015, reported from health facilities throughout Afghanistan. a Adjusted for monthly cases per 10,000 outpatient clients, as reported from health facilities. b Unadjusted monthly malaria cases. c Total number of outpatient cases, reflecting trends health services utilization and reporting. Although the unadjusted data do not exhibit any trend beyond seasonality, because fewer centers were reporting at the beginning of the period (around 1000 centers compared to well over 2000 in 2015 [42]) and health services utilization increased substantially and proportionally for all parts of the country, adjustment was necessary to account for under-reporting. Subsequent analyses were performed using the adjusted rates
Mentions: No public census has been conducted in Afghanistan since 1979 [35], and other sources of demographic data [e.g. WHO, International Monetary Fund (IMF), Central Statistics Office (CSO)] cannot be corroborated with each other. In addition, utilization of health services was not homogenous throughout the study period (Fig. 2c), as the number facilities has risen from under 1000 to over 2000 centres since 2004. Hence, data on the total monthly new outpatient department visits were used as denominator in order to control for demographic and reporting trends. To verify that this did not lead to a bias in the trends over time due to recent changes in outpatient health service utilization occurring primarily in regions of either low or high malaria incidence, the overall of trend of malaria obtained after adjustment was compared with the weighted average of individual trends of provinces adjusted for their level of health service utilization.

View Article: PubMed Central - PubMed

ABSTRACT

Background: Malaria remains endemic in Afghanistan. National control and prevention strategies would be greatly enhanced through a better ability to forecast future trends in disease incidence. It is, therefore, of interest to develop a predictive tool for malaria patterns based on the current passive and affordable surveillance system in this resource-limited region.

Methods: This study employs data from Ministry of Public Health monthly reports from January 2005 to September 2015. Malaria incidence in Afghanistan was forecasted using autoregressive integrated moving average (ARIMA) models in order to build a predictive tool for malaria surveillance. Environmental and climate data were incorporated to assess whether they improve predictive power of models.

Results: Two models were identified, each appropriate for different time horizons. For near-term forecasts, malaria incidence can be predicted based on the number of cases in the four previous months and 12 months prior (Model 1); for longer-term prediction, malaria incidence can be predicted using the rates 1 and 12 months prior (Model 2). Next, climate and environmental variables were incorporated to assess whether the predictive power of proposed models could be improved. Enhanced vegetation index was found to have increased the predictive accuracy of longer-term forecasts.

Conclusion: Results indicate ARIMA models can be applied to forecast malaria patterns in Afghanistan, complementing current surveillance systems. The models provide a means to better understand malaria dynamics in a resource-limited context with minimal data input, yielding forecasts that can be used for public health planning at the national level.

Electronic supplementary material: The online version of this article (doi:10.1186/s12936-016-1602-1) contains supplementary material, which is available to authorized users.

No MeSH data available.


Related in: MedlinePlus