Comparing models for early warning systems of neglected tropical diseases

Background

Early warning systems (EWS) are management tools to predict the occurrence of epidemics of infectious diseases. While climate-based EWS have been developed for malaria, no standard protocol to evaluate and compare EWS has been proposed. Additionally, there are several neglected tropical diseases whose transmission is sensitive to environmental conditions, for which no EWS have been proposed, though they represent a large burden for the affected populations.

Methodology/Principal Findings

In the present paper, an overview of the available linear and non-linear tools to predict seasonal time series of diseases is presented. Also, a general methodology to compare and evaluate models for prediction is presented and illustrated using American cutaneous leishmaniasis, a neglected tropical disease, as an example. The comparison of the different models using the predictive R ² for forecasts of “out-of-fit” data (data that has not been used to fit the models) shows that for the several linear and non-linear models tested, the best results were obtained for seasonal autoregressive (SAR) models that incorporate climatic covariates. An additional bootstrapping experiment shows that the relationship of the disease time series with the climatic covariates is strong and consistent for the SAR modeling approach. While the autoregressive part of the model is not significant, the exogenous forcing due to climate is always statistically significant. Prediction accuracy can vary from 50% to over 80% for disease burden at time scales of one year or shorter.

Conclusions/Significance

This study illustrates a protocol for the development of EWS that includes three main steps: (i) the fitting of different models using several methodologies, (ii) the comparison of models based on the predictability of “out-of-fit” data, and (iii) the assessment of the robustness of the relationship between the disease and the variables in the model selected as best with an objective criterion.     

Blunted Response to Combination Antiretroviral Therapy in HIV Elite Controllers: An International HIV Controller Collaboration

Objective

HIV “elite controllers” (ECs) spontaneously control viral load, but some eventually require combination antiretroviral treatment (cART), due to a loss of viral control or a decline in CD4 T-cell counts. Here we studied the CD4 T-cell count dynamics after cART initiation among 34 ECs followed in U.S. and European cohorts, by comparison with chronically viremic patients (VIRs).

Methods

ECs were defined as patients with at least ≥5 viral load (VL) measurements below 400 copies/mL during at least a 5-year period despite never receiving ART and were selected from the French ANRS CO18 cohort, the U.S. SCOPE cohort, the International HIV Controllers study and the European CASCADE collaboration. VIRs were selected from the ANRS COPANA cohort of recently-diagnosed (<1 year) ART-naïve HIV-1-infected adults. CD4 T-cell count dynamics after cART initiation in both groups were modelled with piecewise mixed linear models.

Results

After cART initiation, CD4 T-cell counts showed a biphasic rise in VIRs with: an initial rapid increase during the first 3 months (+0.63/month), followed by +0.19/month. This first rapid phase was not observed in ECs, in whom the CD4Tc count increased steadily, at a rate similar to that of the second phase observed in VIRs. After cART initiation at a CD4 T-cell count of 300/mm³, the estimated mean CD4 T-cell gain during the first 12 months was 139/mm³ in VIRs and 80/mm³ in ECs (p = 0.048).

Conclusions

cART increases CD4 T-cell counts in elite controllers, albeit less markedly than in other patients.     

Persistence of Pathogens with Short Infectious Periods in Seasonal Tick Populations: The Relative Importance of Three Transmission Routes

Background

The flaviviruses causing tick-borne encephalitis (TBE) persist at low but consistent levels in tick populations, despite short infectious periods in their mammalian hosts and transmission periods constrained by distinctly seasonal tick life cycles. In addition to systemic and vertical transmission, cofeeding transmission has been proposed as an important route for the persistence of TBE-causing viruses. Because cofeeding transmission requires ticks to feed simultaneously, the timing of tick activity may be critical to pathogen persistence. Existing models of tick-borne diseases do not incorporate all transmission routes and tick seasonality. Our aim is to evaluate the influence of seasonality on the relative importance of different transmission routes by using a comprehensive mathematical model.

Methodology/Principal Findings

We developed a stage-structured population model that includes tick seasonality and evaluated the relative importance of the transmission routes for pathogens with short infectious periods, in particular Powassan virus (POWV) and the related “deer tick virus,” emergent encephalitis-causing flaviviruses in North America. We used the next generation matrix method to calculate the basic reproductive ratio and performed elasticity analyses. We confirmed that cofeeding transmission is critically important for such pathogens to persist in seasonal tick populations over the reasonable range of parameter values. At higher but still plausible rates of vertical transmission, our model suggests that vertical transmission can strongly enhance pathogen prevalence when it operates in combination with cofeeding transmission.

Conclusions/Significance

Our results demonstrate that the consistent prevalence of POWV observed in tick populations could be maintained by a combination of low vertical, intermediate cofeeding and high systemic transmission rates. When vertical transmission is weak, nymphal ticks support integral parts of the transmission cycle that are critical for maintaining the pathogen. We also extended the model to pathogens that cause chronic infections in hosts and found that cofeeding transmission could contribute to elevating prevalence even in these systems. Therefore, the common assumption that cofeeding transmission is not relevant in models of chronic host infection, such as Lyme disease, could lead to underestimating pathogen prevalence.     

Comparison of Two Hybrid Models for Forecasting the Incidence of Hemorrhagic Fever with Renal Syndrome in Jiangsu Province,China

Background

Cases of hemorrhagic fever with renal syndrome (HFRS) are widely distributed in eastern Asia, especially in China, Russia, and Korea. It is proved to be a difficult task to eliminate HFRS completely because of the diverse animal reservoirs and effects of global warming. Reliable forecasting is useful for the prevention and control of HFRS.

Methods

Two hybrid models, one composed of nonlinear autoregressive neural network (NARNN) and autoregressive integrated moving average (ARIMA) the other composed of generalized regression neural network (GRNN) and ARIMA were constructed to predict the incidence of HFRS in the future one year. Performances of the two hybrid models were compared with ARIMA model.

Results

The ARIMA, ARIMA-NARNN ARIMA-GRNN model fitted and predicted the seasonal fluctuation well. Among the three models, the mean square error (MSE), mean absolute error (MAE) and mean absolute percentage error (MAPE) of ARIMA-NARNN hybrid model was the lowest both in modeling stage and forecasting stage. As for the ARIMA-GRNN hybrid model, the MSE, MAE and MAPE of modeling performance and the MSE and MAE of forecasting performance were less than the ARIMA model, but the MAPE of forecasting performance did not improve.

Conclusion

Developing and applying the ARIMA-NARNN hybrid model is an effective method to make us better understand the epidemic characteristics of HFRS and could be helpful to the prevention and control of HFRS.     

Application of a Hybrid Model for Predicting the Incidence of Tuberculosis in Hubei,China

Background

A prediction model for tuberculosis incidence is needed in China which may be used as a decision-supportive tool for planning health interventions and allocating health resources.

Methods

The autoregressive integrated moving average (ARIMA) model was first constructed with the data of tuberculosis report rate in Hubei Province from Jan 2004 to Dec 2011.The data from Jan 2012 to Jun 2012 were used to validate the model. Then the generalized regression neural network (GRNN)-ARIMA combination model was established based on the constructed ARIMA model. Finally, the fitting and prediction accuracy of the two models was evaluated.

Results

A total of 465,960 cases were reported between Jan 2004 and Dec 2011 in Hubei Province. The report rate of tuberculosis was highest in 2005 (119.932 per 100,000 population) and lowest in 2010 (84.724 per 100,000 population). The time series of tuberculosis report rate show a gradual secular decline and a striking seasonal variation. The ARIMA (2, 1, 0) × (0, 1, 1)₁₂ model was selected from several plausible ARIMA models. The residual mean square error of the GRNN-ARIMA model and ARIMA model were 0.4467 and 0.6521 in training part, and 0.0958 and 0.1133 in validation part, respectively. The mean absolute error and mean absolute percentage error of the hybrid model were also less than the ARIMA model.

Discussion and Conclusions

The gradual decline in tuberculosis report rate may be attributed to the effect of intensive measures on tuberculosis. The striking seasonal variation may have resulted from several factors. We suppose that a delay in the surveillance system may also have contributed to the variation. According to the fitting and prediction accuracy, the hybrid model outperforms the traditional ARIMA model, which may facilitate the allocation of health resources in China.     

Modeling to Predict Cases of Hantavirus Pulmonary Syndrome in Chile

Background

Hantavirus pulmonary syndrome (HPS) is a life threatening disease transmitted by the rodent Oligoryzomys longicaudatus in Chile. Hantavirus outbreaks are typically small and geographically confined. Several studies have estimated risk based on spatial and temporal distribution of cases in relation to climate and environmental variables, but few have considered climatological modeling of HPS incidence for monitoring and forecasting purposes.

Methodology

Monthly counts of confirmed HPS cases were obtained from the Chilean Ministry of Health for 2001–2012. There were an estimated 667 confirmed HPS cases. The data suggested a seasonal trend, which appeared to correlate with changes in climatological variables such as temperature, precipitation, and humidity. We considered several Auto Regressive Integrated Moving Average (ARIMA) time-series models and regression models with ARIMA errors with one or a combination of these climate variables as covariates. We adopted an information-theoretic approach to model ranking and selection. Data from 2001–2009 were used in fitting and data from January 2010 to December 2012 were used for one-step-ahead predictions.

Results

We focused on six models. In a baseline model, future HPS cases were forecasted from previous incidence; the other models included climate variables as covariates. The baseline model had a Corrected Akaike Information Criterion (AICc) of 444.98, and the top ranked model, which included precipitation, had an AICc of 437.62. Although the AICc of the top ranked model only provided a 1.65% improvement to the baseline AICc, the empirical support was 39 times stronger relative to the baseline model.

Conclusions

Instead of choosing a single model, we present a set of candidate models that can be used in modeling and forecasting confirmed HPS cases in Chile. The models can be improved by using data at the regional level and easily extended to other countries with seasonal incidence of HPS.     

Did Media Attention of the 2009 A(H1N1) Influenza Epidemic Increase Outpatient Antibiotic Use in France?: A Time-Series Analysis

Background

In France, the 2009 A(H1N1) influenza epidemic occurred between September 2009 and January 2010. Sparking widespread controversy, it was intensely reported in the media. Despite therapeutic inefficacy, antibiotic consumption and viral respiratory infections are positively correlated, particularly in France, where antibiotic overconsumption is well-known. We first determined the period when media coverage was high, and then compared, during this period, observed outpatient antibiotic consumption to estimated outpatient antibiotic consumption “without media attention”.

Materials and Methods

To evaluate media coverage, two online databases were consulted: Factiva and Europresse. To quantify outpatient antibiotic consumption, we used data on reimbursements of outpatient systemic antibiotics from the computerized databases of the two main National Health Insurance agencies. Influenza-like syndromes data came from the French GPs Sentinelles Network. Weekly time-series of antibiotic consumption were modeled by autoregressive moving-average models with exogenous inputs and interventions. Analyses were computed for the entire series and by age group (0–5, 6–15, 16–60, and >60 years).

Results

Media coverage was intense between April 2009 and January 2010. No effect on total outpatient antibiotic consumption was observed during the whole mediatic period. However, during the epidemic in France (September 2009-January 2010), we found an antibiotic underconsumption for the entire series, 0–5 and >60 years. Additionally, at the beginning of the pandemic, when cases were still outside France (June 2009-August 2009), we found an antibiotic overconsumption for patients >16 years.

Conclusion

The early period of A(H1N1) virus circulation compared with seasonal influenza or an overdeclaration of ILS cases might explain the antibiotic underconsumption observed during the period of active A(H1N1) virus transmission in France. At the pandemic onset, when uncertainty was high, the overconsumption observed for individuals >16 years might have been caused by alarmist media reporting. Additional analyses are needed to understand the determinants of antibiotic consumption during this period.     

Rotavirus Seasonality and Age Effects in a Birth Cohort Study of Southern India

Introduction

Understanding the temporal patterns in disease occurrence is valuable for formulating effective disease preventive programs. Cohort studies present a unique opportunity to explore complex interactions associated with emergence of seasonal patterns of infectious diseases.

Methods

We used data from 452 children participating in a birth cohort study to assess the seasonal patterns of rotavirus diarrhea by creating a weekly time series of rotavirus incidence and fitting a Poisson harmonic regression with biannual peaks. Age and cohort effects were adjusted for by including the weekly counts of number of children in the study and the median age of cohort in a given week. Weekly average temperature, humidity and an interaction term to reflect the joint effect of temperature and humidity were included to consider the effects of meteorological variables.

Results

In the overall rotavirus time series, two significant peaks within a single year were observed – one in winter and the other in summer. The effect of age was found to be the most significant contributor for rotavirus incidence, showing a strong negative association. Seasonality remained a significant factor, even after adjusting for meteorological parameters, and the age and cohort effects.

Conclusions

The methodology for assessing seasonality in cohort studies is not yet developed. This is the first attempt to explore seasonal patterns in a cohort study with a dynamic denominator and rapidly changing immune response on individual and group levels, and provides a highly promising approach for a better understanding of the seasonal patterns of infectious diseases, tracking emergence of pathogenic strains and evaluating the efficacy of intervention programs.     

A Hybrid Model for Predicting the Prevalence of Schistosomiasis in Humans of Qianjiang City,China

Backgrounds/Objective

Schistosomiasis is still a major public health problem in China, despite the fact that the government has implemented a series of strategies to prevent and control the spread of the parasitic disease. Advanced warning and reliable forecasting can help policymakers to adjust and implement strategies more effectively, which will lead to the control and elimination of schistosomiasis. Our aim is to explore the application of a hybrid forecasting model to track the trends of the prevalence of schistosomiasis in humans, which provides a methodological basis for predicting and detecting schistosomiasis infection in endemic areas.

Methods

A hybrid approach combining the autoregressive integrated moving average (ARIMA) model and the nonlinear autoregressive neural network (NARNN) model to forecast the prevalence of schistosomiasis in the future four years. Forecasting performance was compared between the hybrid ARIMA-NARNN model, and the single ARIMA or the single NARNN model.

Results

The modelling mean square error (MSE), mean absolute error (MAE) and mean absolute percentage error (MAPE) of the ARIMA-NARNN model was 0.1869×10⁻⁴, 0.0029, 0.0419 with a corresponding testing error of 0.9375×10⁻⁴, 0.0081, 0.9064, respectively. These error values generated with the hybrid model were all lower than those obtained from the single ARIMA or NARNN model. The forecasting values were 0.75%, 0.80%, 0.76% and 0.77% in the future four years, which demonstrated a no-downward trend.

Conclusion

The hybrid model has high quality prediction accuracy in the prevalence of schistosomiasis, which provides a methodological basis for future schistosomiasis monitoring and control strategies in the study area. It is worth attempting to utilize the hybrid detection scheme in other schistosomiasis-endemic areas including other infectious diseases.     

Ranking Malaria Risk Factors to Guide Malaria Control Efforts in African Highlands

Introduction

Malaria is re-emerging in most of the African highlands exposing the non immune population to deadly epidemics. A better understanding of the factors impacting transmission in the highlands is crucial to improve well targeted malaria control strategies.

Methods and Findings

A conceptual model of potential malaria risk factors in the highlands was built based on the available literature. Furthermore, the relative importance of these factors on malaria can be estimated through “classification and regression trees”, an unexploited statistical method in the malaria field. This CART method was used to analyse the malaria risk factors in the Burundi highlands. The results showed that Anopheles density was the best predictor for high malaria prevalence. Then lower rainfall, no vector control, higher minimum temperature and houses near breeding sites were associated by order of importance to higher Anopheles density.

Conclusions

In Burundi highlands monitoring Anopheles densities when rainfall is low may be able to predict epidemics. The conceptual model combined with the CART analysis is a decision support tool that could provide an important contribution toward the prevention and control of malaria by identifying major risk factors.     

Seasonal Variations in Notification of Active Tuberculosis Cases in China, 2005–2012

Background

Although seasonal variation in tuberculosis (TB) incidence has been described in many countries, it remains unknown in China.

Methods

A time series decomposition analysis (X-12-ARIMA) was performed to examine the seasonal variation in active TB cases nationwide from 2005 through 2012 in China. Seasonal amplitude was calculated for the evaluation of TB seasonal variation.

Results

A total of 7.78 million active TB cases were reported over a period of 8 years. A spring peak (April) was observed with seasonal amplitude of 46.3%, compared with the winter trough (February). Most cases in provinces with subtropical and tropical monsoon climate showed lower amplitudes than those in temperate continental, plateau and mountain climate regions. The magnitude of seasonality varied inversely with annual average temperature, r (95% CI) = -0.71 (-0.79, -0.61). The seasonal amplitudes were 56.7, 60.5, 40.6, 46.4 and 50.9% for patients aged ≤14, 15–24, 25–44, 45–64, and ≥65 years, respectively. Students demonstrated greater seasonal amplitude than peasants, migrant workers and workers (115.3% vs. 43.5, 41.6 and 48.1%). Patients with pulmonary TB had lower amplitude compared to patients with pleural and other extra-pulmonary TB (EPTB) (45.9% vs. 52.0 and 56.3%). Relapse cases with sputum smear positive TB (SS+ TB) had significantly higher seasonal amplitude compared to new cases with sputum smear positive TB (52.2% vs. 41.6%).

Conclusions

TB is a seasonal disease in China. The peak and trough of TB transmission actually are in winter and in autumn respectively after factors of delay are removed. Higher amplitudes of TB seasonality are more likely to happen in temperate continental, plateau and mountain climate regions and regions with lower annual average temperature, and young person, students, patients with EPTB and relapse cases with SS+ TB are more likely to be affected by TB seasonality.     

Suicide seasonality: complex demodulation as a novel approach in epidemiologic analysis

Background

Seasonality of suicides is well-known and nearly ubiquitous, but recent evidence showed inconsistent patterns of decreasing or increasing seasonality in different countries. Furthermore, strength of seasonality was hypothesized to be associated with suicide prevalence. This study aimed at pointing out methodological difficulties in examining changes in suicide seasonality.

Methododology/Principal Findings

The present study examines the hypothesis of decreasing seasonality with a superior method that allows continuous modeling of seasonality. Suicides in Austria (1970–2008, N = 67,741) were analyzed with complex demodulation, a local (point-in-time specific) version of harmonic analysis. This avoids the need to arbitrarily split the time series, as is common practice in the field of suicide seasonality research, and facilitates incorporating the association with suicide prevalence. Regression models were used to assess time trends and association of amplitude and absolute suicide numbers. Results showed that strength of seasonality was associated with absolute suicide numbers, and that strength of seasonality was stable during the study period when this association was taken into account.

Conclusion/Significance

Continuous modeling of suicide seasonality with complex demodulation avoids spurious findings that can result when time series are segmented and analyzed piecewise or when the association with suicide prevalence is disregarded.     

Malaria Incidence in Children in South-West Burkina Faso: Comparison of Active and Passive Case Detection Methods

Background

The aim of this study was to determine the incidence and seasonal pattern of malaria in children in South-West Burkina Faso, and to compare, in a randomized trial, characteristics of cases detected by active and passive surveillance. This study also enabled the planning of a malaria vaccine trial.

Methods

Households with young children, located within 5 kilometers of a health facility, were randomized to one of two malaria surveillance methods. In the first group, children were monitored actively. Each child was visited twice weekly; tympanic temperature was measured, and if the child had a fever or history of fever, a malaria rapid diagnostic test was performed and a blood smear collected. In the second group, children were monitored passively. The child’s parent or caregiver was asked to bring the child to the nearest clinic if he was unwell. Follow up lasted 13 months from September 2009.

Results

Incidence of malaria (Fever with parasitaemia ≥5,000/µL) was 1.18 episodes/child/year in the active cohort and 0.89 in the passive cohort (rate ratio 1.32, 95% CI 1.13–1.54). Malaria cases in the passive cohort were more likely to have high grade fever; but parasite densities were similar in the two groups. Incidence was highly seasonal; when a specific case definition was used, about 60% of cases occurred within the 4 months June-September.

Conclusion

Passive case detection required at least a 30%–40% increase in the sample size for vaccine trials, compared to active detection, to achieve the same power. However we did not find any evidence that parasite densities were higher with passive than with active detection. The incidence of malaria is highly seasonal and meets the WHO criteria for Seasonal Malaria Chemoprevention (SMC). At least half of the malaria cases in these children could potentially be prevented if SMC was effectively deployed.     

Cutaneous Leishmaniasis and Sand Fly Fluctuations Are Associated with El Ni?o in Panamá

Background

Cutaneous Leishmaniasis (CL) is a neglected tropical vector-borne disease. Sand fly vectors (SF) and Leishmania spp parasites are sensitive to changes in weather conditions, rendering disease transmission susceptible to changes in local and global scale climatic patterns. Nevertheless, it is unclear how SF abundance is impacted by El Niño Southern Oscillation (ENSO) and how these changes might relate to changes in CL transmission.

Methodology and Findings

We studied association patterns between monthly time series, from January 2000 to December 2010, of: CL cases, rainfall and temperature from Panamá, and an ENSO index. We employed autoregressive models and cross wavelet coherence, to quantify the seasonal and interannual impact of local climate and ENSO on CL dynamics. We employed Poisson Rate Generalized Linear Mixed Models to study SF abundance patterns across ENSO phases, seasons and eco-epidemiological settings, employing records from 640 night-trap sampling collections spanning 2000–2011. We found that ENSO, rainfall and temperature were associated with CL cycles at interannual scales, while seasonal patterns were mainly associated with rainfall and temperature. Sand fly (SF) vector abundance, on average, decreased during the hot and cold ENSO phases, when compared with the normal ENSO phase, yet variability in vector abundance was largest during the cold ENSO phase. Our results showed a three month lagged association between SF vector abundance and CL cases.

Conclusion

Association patterns of CL with ENSO and local climatic factors in Panamá indicate that interannual CL cycles might be driven by ENSO, while the CL seasonality was mainly associated with temperature and rainfall variability. CL cases and SF abundance were associated in a fashion suggesting that sudden extraordinary changes in vector abundance might increase the potential for CL epidemic outbreaks, given that CL epidemics occur during the cold ENSO phase, a time when SF abundance shows its highest fluctuations.     

Latitudinal Variations in Seasonal Activity of Influenza and Respiratory Syncytial Virus (RSV): A Global Comparative Review

Background

There is limited information on influenza and respiratory syncytial virus (RSV) seasonal patterns in tropical areas, although there is renewed interest in understanding the seasonal drivers of respiratory viruses.

Methods

We review geographic variations in seasonality of laboratory-confirmed influenza and RSV epidemics in 137 global locations based on literature review and electronic sources. We assessed peak timing and epidemic duration and explored their association with geography and study settings. We fitted time series model to weekly national data available from the WHO influenza surveillance system (FluNet) to further characterize seasonal parameters.

Results

Influenza and RSV activity consistently peaked during winter months in temperate locales, while there was greater diversity in the tropics. Several temperate locations experienced semi-annual influenza activity with peaks occurring in winter and summer. Semi-annual activity was relatively common in tropical areas of Southeast Asia for both viruses. Biennial cycles of RSV activity were identified in Northern Europe. Both viruses exhibited weak latitudinal gradients in the timing of epidemics by hemisphere, with peak timing occurring later in the calendar year with increasing latitude (P<0.03). Time series model applied to influenza data from 85 countries confirmed the presence of latitudinal gradients in timing, duration, seasonal amplitude, and between-year variability of epidemics. Overall, 80% of tropical locations experienced distinct RSV seasons lasting 6 months or less, while the percentage was 50% for influenza.

Conclusion

Our review combining literature and electronic data sources suggests that a large fraction of tropical locations experience focused seasons of respiratory virus activity in individual years. Information on seasonal patterns remains limited in large undersampled regions, included Africa and Central America. Future studies should attempt to link the observed latitudinal gradients in seasonality of viral epidemics with climatic and population factors, and explore regional differences in disease transmission dynamics and attack rates.     

Patient Satisfaction with HIV/AIDS Care and Treatment in the Decentralization of Services Delivery in Vietnam

Objective

We evaluated the patient satisfaction with HIV/AIDS care and treatment and its determinants across levels of health service administration in Vietnam.

Methods

We interviewed 1016 patients at 7 hospitals and health centers in three epicenters, including Hanoi, Hai Phong, and Ho Chi Minh City. The Satisfaction with HIV/AIDS Treatment Interview Scale (SATIS) was developed, and 3 dimensions were constructed using factor analysis, namely “Quality and Convenience”; “Availability and Responsiveness”; and “Competence of health care workers”.

Results

In a band score of (0; 10), the mean scores of all domains were large; it was the highest in “Competence of health workers” (9.34±0.84), and the lowest in “Quality and Convenience” (9.03±1.04). The percentages of respondents completely satisfied with overall service quality and treatment outcomes were 42.4% and 18.8%, respectively. In multivariate analysis, factors related to higher satisfaction included female sex, older age, and living with spouses or partners. Meanwhile, lower satisfaction was found among patients who were attending provincial and district clinics; in the richest group; had higher CD4 count; and drug users.

Conclusion

This study highlights the importance of improving the quality of HIV/AIDS services at the provincial and district clinics. Potential strategies include capacity building for health workers, integrative service delivery, engagements of family members in treatment supports, and additional attention and comprehensive care for drug users with HIV/AIDS.     

Tuberculosis Incidence Correlates with Sunshine: An Ecological 28-Year Time Series Study

Background

Birmingham is the largest UK city after London, and central Birmingham has an annual tuberculosis incidence of 80 per 100,000. We examined seasonality and sunlight as drivers of tuberculosis incidence. Hours of sunshine are seasonal, sunshine exposure is necessary for the production of vitamin D by the body and vitamin D plays a role in the host response to tuberculosis.

Methods

We performed an ecological study that examined tuberculosis incidence in Birmingham from Dec 1981 to Nov 2009, using publicly-available data from statutory tuberculosis notifications, and related this to the seasons and hours of sunshine (UK Meteorological Office data) using unmeasured component models.

Results

There were 9,739 tuberculosis cases over the study period. There was strong evidence for seasonality, with notifications being 24.1% higher in summer than winter (p<0.001). Winter dips in sunshine correlated with peaks in tuberculosis incidence six months later (4.7% increase in incidence for each 100 hours decrease in sunshine, p<0.001).

Discussion and Conclusion

A potential mechanism for these associations includes decreased vitamin D levels with consequent impaired host defence arising from reduced sunshine exposure in winter. This is the longest time series of any published study and our use of statutory notifications means this data is essentially complete. We cannot, however, exclude the possibility that another factor closely correlated with the seasons, other than sunshine, is responsible. Furthermore, exposure to sunlight depends not only on total hours of sunshine but also on multiple individual factors. Our results should therefore be considered hypothesis-generating. Confirmation of a potential causal relationship between winter vitamin D deficiency and summer peaks in tuberculosis incidence would require a randomized-controlled trial of the effect of vitamin D supplementation on future tuberculosis incidence.     

Comparison of ARIMA and Random Forest time series models for prediction of avian influenza H5N1 outbreaks

Background

Time series models can play an important role in disease prediction. Incidence data can be used to predict the future occurrence of disease events. Developments in modeling approaches provide an opportunity to compare different time series models for predictive power.

Results

We applied ARIMA and Random Forest time series models to incidence data of outbreaks of highly pathogenic avian influenza (H5N1) in Egypt, available through the online EMPRES-I system. We found that the Random Forest model outperformed the ARIMA model in predictive ability. Furthermore, we found that the Random Forest model is effective for predicting outbreaks of H5N1 in Egypt.

Conclusions

Random Forest time series modeling provides enhanced predictive ability over existing time series models for the prediction of infectious disease outbreaks. This result, along with those showing the concordance between bird and human outbreaks (Rabinowitz et al. 2012), provides a new approach to predicting these dangerous outbreaks in bird populations based on existing, freely available data. Our analysis uncovers the time-series structure of outbreak severity for highly pathogenic avain influenza (H5N1) in Egypt.