Predictive model assessment for count data

The COVID-19 vaccination campaign in Switzerland and its impact on disease spread

We analyse infectious disease case surveillance data to estimate COVID-19 spread and gain an understanding of the impact of introducing vaccines to counter the disease in Switzerland. The data used in this work is extensive and detailed and includes information on weekly number of cases and vaccination rates by age and region. Our approach takes into account waning immunity. The statistical analysis allows us to determine the effects of choosing alternative vaccination strategies. Our results indicate greater uptake of vaccine would have led to fewer cases with a particularly large effect on undervaccinated regions. An alternative distribution scheme not targeting specific age groups also leads to fewer cases overall but could lead to more cases among the elderly (a potentially vulnerable population) during the early stage of prophylaxis rollout.

Predicting subnational incidence of COVID-19 cases and deaths in EU countries

BACKGROUND

Recurring COVID-19 waves highlight the need for tools able to quantify transmission risk, and identify geographical areas at risk of outbreaks. Local outbreak risk depends on complex immunity patterns resulting from previous infections, vaccination, waning and immune escape, alongside other factors (population density, social contact patterns). Immunity patterns are spatially and demographically heterogeneous, and are challenging to capture in country-level forecast models.

METHODS

We used a spatiotemporal regression model to forecast subnational case and death counts and applied it to three EU countries as test cases: France, Czechia, and Italy. Cases in local regions arise from importations or local transmission. Our model produces age-stratified forecasts given age-stratified data, and links reported case counts to routinely collected covariates (e.g. test number, vaccine coverage). We assessed the predictive performance of our model up to four weeks ahead using proper scoring rules and compared it to the European COVID-19 Forecast Hub ensemble model. Using simulations, we evaluated the impact of variations in transmission on the forecasts. We developed an open-source RShiny App to visualise the forecasts and scenarios.

RESULTS

At a national level, the median relative difference between our median weekly case forecasts and the data up to four weeks ahead was 25% (IQR: 12-50%) over the prediction period. The accuracy decreased as the forecast horizon increased (on average 24% increase in the median ranked probability score per added week), while the accuracy of death forecasts was more stable. Beyond two weeks, the model generated a narrow range of likely transmission dynamics. The median national case forecasts showed similar accuracy to forecasts from the European COVID-19 Forecast Hub ensemble model, but the prediction interval was narrower in our model. Generating forecasts under alternative transmission scenarios was therefore key to capturing the range of possible short-term transmission dynamics.

DISCUSSION

Our model captures changes in local COVID-19 outbreak dynamics, and enables quantification of short-term transmission risk at a subnational level. The outputs of the model improve our ability to identify areas where outbreaks are most likely, and are available to a wide range of public health professionals through the Shiny App we developed.

Modelling spatiotemporal patterns of visceral leishmaniasis incidence in two endemic states in India using environment, bioclimatic and demographic data, 2013-2022

BACKGROUND

As of 2021, the National Kala-azar Elimination Programme (NKAEP) in India has achieved visceral leishmaniasis (VL) elimination (<1 case / 10,000 population/year per block) in 625 of the 633 endemic blocks (subdistricts) in four states. The programme needs to sustain this achievement and target interventions in the remaining blocks to achieve the WHO 2030 target of VL elimination as a public health problem. An effective tool to analyse programme data and predict/ forecast the spatial and temporal trends of VL incidence, elimination threshold, and risk of resurgence will be of use to the programme management at this juncture.

METHODOLOGY/PRINCIPAL FINDINGS

We employed spatiotemporal models incorporating environment, climatic and demographic factors as covariates to describe monthly VL cases for 8-years (2013-2020) in 491 and 27 endemic and non-endemic blocks of Bihar and Jharkhand states. We fitted 37 models of spatial, temporal, and spatiotemporal interaction random effects with covariates to monthly VL cases for 6-years (2013-2018, training data) using Bayesian inference via Integrated Nested Laplace Approximation (INLA) approach. The best-fitting model was selected based on deviance information criterion (DIC) and Watanabe-Akaike Information Criterion (WAIC) and was validated with monthly cases for 2019-2020 (test data). The model could describe observed spatial and temporal patterns of VL incidence in the two states having widely differing incidence trajectories, with >93% and 99% coverage probability (proportion of observations falling inside 95% Bayesian credible interval for the predicted number of VL cases per month) during the training and testing periods. PIT (probability integral transform) histograms confirmed consistency between prediction and observation for the test period. Forecasting for 2021-2023 showed that the annual VL incidence is likely to exceed elimination threshold in 16-18 blocks in 4 districts of Jharkhand and 33-38 blocks in 10 districts of Bihar. The risk of VL in non-endemic neighbouring blocks of both Bihar and Jharkhand are less than 0.5 during the training and test periods, and for 2021-2023, the probability that the risk greater than 1 is negligible (P<0.1). Fitted model showed that VL occurrence was positively associated with mean temperature, minimum temperature, enhanced vegetation index, precipitation, and isothermality, and negatively with maximum temperature, land surface temperature, soil moisture and population density.

CONCLUSIONS/SIGNIFICANCE

The spatiotemporal model incorporating environmental, bioclimatic, and demographic factors demonstrated that the KAMIS database of the national programmme can be used for block level predictions of long-term spatial and temporal trends in VL incidence and risk of outbreak / resurgence in endemic and non-endemic settings. The database integrated with the modelling framework and a dashboard facility can facilitate such analysis and predictions. This could aid the programme to monitor progress of VL elimination at least one-year ahead, assess risk of resurgence or outbreak in post-elimination settings, and implement timely and targeted interventions or preventive measures so that the NKAEP meet the target of achieving elimination by 2030.

A zero-inflated endemic-epidemic model with an application to measles time series in Germany

Count data with an excess of zeros are often encountered when modeling infectious disease occurrence. The degree of zero inflation can vary over time due to nonepidemic periods as well as by age group or region. A well-established approach to analyze multivariate incidence time series is the endemic-epidemic modeling framework, also known as the HHH approach. However, it assumes Poisson or negative binomial distributions and is thus not tailored to surveillance data with excess zeros. Here, we propose a multivariate zero-inflated endemic-epidemic model with random effects that extends HHH. Parameters of both the zero-inflation probability and the HHH part of this mixture model can be estimated jointly and efficiently via (penalized) maximum likelihood inference using analytical derivatives. We found proper convergence and good coverage of confidence intervals in simulation studies. An application to measles counts in the 16 German states, 2005-2018, showed that zero inflation is more pronounced in the Eastern states characterized by a higher vaccination coverage. Probabilistic forecasts of measles cases improved when accounting for zero inflation. We anticipate zero-inflated HHH models to be a useful extension also for other applications and provide an implementation in an R package.

The balanced discrete triplet Lindley model and its INAR(1) extension: properties and COVID-19 applications

This paper proposes a new flexible discrete triplet Lindley model that is constructed from the balanced discretization principle of the extended Lindley distribution. This model has several appealing statistical properties in terms of providing exact and closed form moment expressions and handling all forms of dispersion. Due to these, this paper explores further the usage of the discrete triplet Lindley as an innovation distribution in the simple integer-valued autoregressive process (INAR(1)). This subsequently allows for the modeling of count time series observations. In this context, a novel INAR(1) process is developed under mixed Binomial and the Pegram thinning operators. The model parameters of the INAR(1) process are estimated using the conditional maximum likelihood and Yule-Walker approaches. Some Monte Carlo simulation experiments are executed to assess the consistency of the estimators under the two estimation approaches. Interestingly, the proposed INAR(1) process is applied to analyze the COVID-19 cases and death series of different countries where it yields reliable parameter estimates and suitable forecasts via the modified Sieve bootstrap technique. On the other side, the new INAR(1) with discrete triplet Lindley innovations competes comfortably with other established INAR(1)s in the literature.

Forecasting emergency department arrivals using INGARCH models

BACKGROUND

Forecasting patient arrivals to hospital emergency departments is critical to dealing with surges and to efficient planning, management and functioning of hospital emerency departments.

OBJECTIVE

We explore whether past mean values and past observations are useful to forecast daily patient arrivals in an Emergency Department.

MATERIAL AND METHODS

We examine whether an integer-valued generalized autoregressive conditional heteroscedastic (INGARCH) model can yield a better conditional distribution fit and forecast of patient arrivals by using past arrival information and taking into account the dynamics of the volatility of arrivals.

RESULTS

We document that INGARCH models improve both in-sample and out-of-sample forecasts, particularly in the lower and upper quantiles of the distribution of arrivals.

CONCLUSION

Our results suggest that INGARCH modelling is a useful model for short-term and tactical emergency department planning, e.g., to assign rotas or locate staff for unexpected surges in patient arrivals.

Trending against the grain: Bird population responses to expanding energy portfolios in the US Northern Great Plains

Future global energy demand may be met through increased extraction of fossil fuels and production of renewable energy such as biofuels. Renewable energy from biofuels is often proposed as an environmentally friendly alternative to fossil fuels; however, impacts of renewable energy sources on wildlife populations have rarely been evaluated in working landscapes. We used North American Breeding Bird Survey data (1998 to 2021) to assess whether the joint effects of oil and gas and biofuel crop production explained grassland bird population declines. We modeled location-specific effects of land use on grassland bird habitat use for four grassland bird species (bobolink [Dolichonyx oryzivorus], grasshopper sparrow [Ammodramus savannarum], Savannah sparrow [Passerculus sandwichensis], and western meadowlark [Sturnella neglecta]) in North Dakota, a state experiencing rapid growth in both energy sectors. Our analysis showed that grassland birds responded more negatively to biofuel feedstocks (i.e., corn and soybeans) on the landscape compared with oil and gas development. Furthermore, we found that the effect of feedstocks was not generalizable to other forms of agricultural land use. When combined, these land use changes resulted in distributional shifts of grassland birds, with use by birds being lower in regions dominated by biofuel production, which appears partially responsible for observed abundance trends at the state level. Our results indicate that expansion of oil and gas development has negatively affected habitat use by some grassland birds, but this impact was more localized when compared to biofuel crops. Conservation practitioners may need to adapt their conservation strategies to account for widespread and rapid land use change driven by United States energy policies.

Using surveillance data for early warning modelling of highly pathogenic avian influenza in Europe reveals a seasonal shift in transmission, 2016-2022

Avian influenza in wild birds and poultry flocks constitutes a problem for animal welfare, food security and public health. In recent years there have been increasing numbers of outbreaks in Europe, with many poultry flocks culled after being infected with highly pathogenic avian influenza (HPAI). Continuous monitoring is crucial to enable timely implementation of control to prevent HPAI spread from wild birds to poultry and between poultry flocks within a country. We here utilize readily available public surveillance data and time-series models to predict HPAI detections within European countries and show a seasonal shift that happened during 2021-2022. The output is models capable of monitoring the weekly risk of HPAI outbreaks, to support decision making.

Evaluation of predictive capability of Bayesian spatio-temporal models for Covid-19 spread

BACKGROUND

Bayesian models have been applied throughout the Covid-19 pandemic especially to model time series of case counts or deaths. Fewer examples exist of spatio-temporal modeling, even though the spatial spread of disease is a crucial factor in public health monitoring. The predictive capabilities of infectious disease models is also important.

METHODS

In this study, the ability of Bayesian hierarchical models to recover different parts of the variation in disease counts is the focus. It is clear that different measures provide different views of behavior when models are fitted prospectively. Over a series of time horizons one step predictions have been generated and compared for different models (for case counts and death counts). These Bayesian SIR models were fitted using MCMC at 28 time horizons to mimic prospective prediction. A range of goodness of prediction measures were analyzed across the different time horizons.

RESULTS

A particularly important result is that the peak intensity of case load is often under-estimated, while random spikes in case load can be mimicked using time dependent random effects. It is also clear that during the early wave of the pandemic simpler model forms are favored, but subsequently lagged spatial dependence models for cases are favored, even if the sophisticated models perform better overall.

DISCUSSION

The models fitted mimic the situation where at a given time the history of the process is known but the future must be predicted based on the current evolution which has been observed. Using an overall 'best' model for prediction based on retrospective fitting of the complete pandemic waves is an assumption. However it is also clear that this case count model is well favored over other forms. During the first wave a simpler time series model predicts case counts better for counties than a spatially dependent one. The picture is more varied for morality.

CONCLUSIONS

From a predictive point of view it is clear that spatio-temporal models applied to county level Covid-19 data within the US vary in how well they fit over time and also how well they predict future events. At different times, SIR case count models and also mortality models with cumulative counts perform better in terms of prediction. A fundamental result is that predictive capability of models varies over time and using the same model could lead to poor predictive performance. In addition it is clear that models addressing the spatial context for case counts (i.e. with lagged neighborhood terms) and cumulative case counts for mortality data are clearly better at modeling spatio-temporal data which is commonly available for the Covid-19 pandemic in different areas of the globe.

Automatic Forecast of Intensive Care Unit Admissions: The Experience During the COVID-19 Pandemic in Italy

The experience of the COVID-19 pandemic showed the importance of timely monitoring of admissions to the ICU admissions. The ability to promptly forecast the epidemic impact on the occupancy of beds in the ICU is a key issue for adequate management of the health care system.Despite this, most of the literature on predictive COVID-19 models in Italy has focused on predicting the number of infections, leaving trends in ordinary hospitalizations and ICU occupancies in the background.This work aims to present an ETS approach (Exponential Smoothing Time Series) time series forecasting tool for admissions to the ICU admissions based on ETS models. The results of the forecasting model are presented for the regions most affected by the epidemic, such as Veneto, Lombardy, Emilia-Romagna, and Piedmont.The mean absolute percentage errors (MAPE) between observed and predicted admissions to the ICU admissions remain lower than 11% for all considered geographical areas.In this epidemiological context, the proposed ETS forecasting model could be suitable to monitor, in a timely manner, the impact of COVID-19 disease on the health care system, not only during the early stages of the pandemic but also during the vaccination campaign, to quickly adapt possible preventive interventions.

Opioid use following pelvic reconstructive surgery: a predictive calculator

INTRODUCTION AND HYPOTHESIS

Our objective was to evaluate the amount of opioids used by patients undergoing surgery for pelvic floor disorders and identify risk factors for opioid consumption greater than the median.

METHODS

This was a prospective cohort study of 18- to 89-year-old women undergoing major urogynecological surgery between 1 November2020 and 15 October 2021. Subjects completed one preoperative questionnaire ("questionnaire 1") that surveyed factors expected to influence postoperative pain and opioid use. At approximately 1 and 2 weeks following surgery, patients completed two additional questionnaires ("questionnaire 2" and "questionnaire 3") about their pain scores and opioid use. Risk factors for opioid use greater than the median were assessed. Finally, a calculator was created to predict the amount of opioid used at 1 week following surgery.

RESULTS

One hundred and ninety patients were included. The median amount of milligram morphine equivalents prescribed was 100 (IQR 100-120), whereas the median amount used by questionnaire 2 was 15 (IQR 0-50) and by questionnaire 3 was 20 (IQR 0-75). On multivariate logistic regression, longer operative time (aOR 1.64 per hour of operative time, 95% CI 1.07-2.58) was associated with using greater than the median opioid consumption at the time of questionnaire 2; whereas for questionnaire 3, a diagnosis of fibromyalgia (aOR=16.9, 95% CI 2.24-362.9) was associated. A preliminary calculator was created using the information collected through questionnaires and chart review.

CONCLUSIONS

Patients undergoing surgery for pelvic floor disorders use far fewer opioids than they are prescribed.

Evaluation of Bayesian spatiotemporal infectious disease models for prospective surveillance analysis

BACKGROUND

COVID-19 brought enormous challenges to public health surveillance and underscored the importance of developing and maintaining robust systems for accurate surveillance. As public health data collection efforts expand, there is a critical need for infectious disease modeling researchers to continue to develop prospective surveillance metrics and statistical models to accommodate the modeling of large disease counts and variability. This paper evaluated different likelihoods for the disease count model and various spatiotemporal mean models for prospective surveillance.

METHODS

We evaluated Bayesian spatiotemporal models, which are the foundation for model-based infectious disease surveillance metrics. Bayesian spatiotemporal mean models based on the Poisson and the negative binomial likelihoods were evaluated with the different lengths of past data usage. We compared their goodness of fit and short-term prediction performance with both simulated epidemic data and real data from the COVID-19 pandemic.

RESULTS

The simulation results show that the negative binomial likelihood-based models show better goodness of fit results than Poisson likelihood-based models as deemed by smaller deviance information criteria (DIC) values. However, Poisson models yield smaller mean square error (MSE) and mean absolute one-step prediction error (MAOSPE) results when we use a shorter length of the past data such as 7 and 3 time periods. Real COVID-19 data analysis of New Jersey and South Carolina shows similar results for the goodness of fit and short-term prediction results. Negative binomial-based mean models showed better performance when we used the past data of 52 time periods. Poisson-based mean models showed comparable goodness of fit performance and smaller MSE and MAOSPE results when we used the past data of 7 and 3 time periods.

CONCLUSION

We evaluate these models and provide future infectious disease outbreak modeling guidelines for Bayesian spatiotemporal analysis. Our choice of the likelihood and spatiotemporal mean models was influenced by both historical data length and variability. With a longer length of past data usage and more over-dispersed data, the negative binomial likelihood shows a better model fit than the Poisson likelihood. However, as we use a shorter length of the past data for our surveillance analysis, the difference between the Poisson and the negative binomial models becomes smaller. In this case, the Poisson likelihood shows robust posterior mean estimate and short-term prediction results.

Spatio-temporal dynamic of the COVID-19 epidemic and the impact of imported cases in Rwanda

INTRODUCTION

Africa was threatened by the coronavirus disease 2019 (COVID-19) due to the limited health care infrastructure. Rwanda has consistently used non-pharmaceutical strategies, such as lockdown, curfew, and enforcement of prevention measures to control the spread of COVID-19. Despite the mitigation measures taken, the country has faced a series of outbreaks in 2020 and 2021. In this paper, we investigate the nature of epidemic phenomena in Rwanda and the impact of imported cases on the spread of COVID-19 using endemic-epidemic spatio-temporal models. Our study provides a framework for understanding the dynamics of the epidemic in Rwanda and monitoring its phenomena to inform public health decision-makers for timely and targeted interventions.

RESULTS

The findings provide insights into the effects of lockdown and imported infections in Rwanda's COVID-19 outbreaks. The findings showed that imported infections are dominated by locally transmitted cases. The high incidence was predominant in urban areas and at the borders of Rwanda with its neighboring countries. The inter-district spread of COVID-19 was very limited due to mitigation measures taken in Rwanda.

CONCLUSION

The study recommends using evidence-based decisions in the management of epidemics and integrating statistical models in the analytics component of the health information system.

Development of a personalized fall rate prediction model in community-dwelling older adults: a negative binomial regression modelling approach

BACKGROUND

Around a third of adults aged 65 and older fall every year, resulting in unintentional injuries in 30% of the cases. Fractures are a frequent consequence of falls, primarily caused in individuals with decreased bone strength who are unable to cushion their falls. Accordingly, an individual's number of experienced falls has a direct influence on fracture risk. The aim of this study was the development of a statistical model to predict future fall rates using personalized risk predictors.

METHODS

In the prospective cohort GERICO, several fall risk factor variables were collected in community-dwelling older adults at two time-points four years apart (T1 and T2). Participants were asked how many falls they experienced during 12 months prior to the examinations. Rate ratios for the number of reported falls at T2 were computed for age, sex, reported fall number at T1, physical performance tests, physical activity level, comorbidity and medication number with negative binomial regression models.

RESULTS

The analysis included 604 participants (male: 122, female: 482) with a median age of 67.90 years at T1. The mean number of falls per person was 1.04 and 0.70 at T1 and T2. The number of reported falls at T1 as a factor variable was the strongest risk factor with an unadjusted rate ratio [RR] of 2.60 for 3 falls (95% confidence interval [CI] 1.54 to 4.37), RR of 2.63 (95% CI 1.06 to 6.54) for 4 falls, and RR of 10.19 (95% CI 6.25 to 16.60) for 5 and more falls, when compared to 0 falls. The cross-validated prediction error was comparable for the global model including all candidate variables and the univariable model including prior fall numbers at T1 as the only predictor.

CONCLUSION

In the GERICO cohort, the prior fall number as single predictor information for a personalized fall rate is as good as when including further available fall risk factors. Specifically, individuals who have experienced three and more falls are expected to fall multiple times again.

TRIAL REGISTRATION

ISRCTN11865958, 13/07/2016, retrospectively registered.

Extracting Self-Reported COVID-19 Symptom Tweets and Twitter Movement Mobility Origin/Destination Matrices to Inform Disease Models

The emergence of the novel coronavirus (COVID-19) generated a need to quickly and accurately assemble up-to-date information related to its spread. In this research article, we propose two methods in which Twitter is useful when modelling the spread of COVID-19: (1) machine learning algorithms trained in English, Spanish, German, Portuguese and Italian are used to identify symptomatic individuals derived from Twitter. Using the geo-location attached to each tweet, we map users to a geographic location to produce a time-series of potential symptomatic individuals. We calibrate an extended SEIRD epidemiological model with combinations of low-latency data feeds, including the symptomatic tweets, with death data and infer the parameters of the model. We then evaluate the usefulness of the data feeds when making predictions of daily deaths in 50 US States, 16 Latin American countries, 2 European countries and 7 NHS (National Health Service) regions in the UK. We show that using symptomatic tweets can result in a 6% and 17% increase in mean squared error accuracy, on average, when predicting COVID-19 deaths in US States and the rest of the world, respectively, compared to using solely death data. (2) Origin/destination (O/D) matrices, for movements between seven NHS regions, are constructed by determining when a user has tweeted twice in a 24 h period in two different locations. We show that increasing and decreasing a social connectivity parameter within an SIR model affects the rate of spread of a disease.

Spatio-temporal model to investigate COVID-19 spread accounting for the mobility amongst municipalities

The rapid spread of COVID-19 worldwide led to the implementation of various non-pharmaceutical interventions to limit transmission and hence reduce the number of infections. Using telecom-operator-based mobility data and a spatio-temporal dynamic model, the impact of mobility on the evolution of the pandemic at the level of the 581 Belgian municipalities is investigated. By decomposing incidence, particularly into within- and between-municipality components, we noted that the global epidemic component is relatively more important in larger municipalities (e.g., cities), while the local component is more relevant in smaller (rural) municipalities. Investigation of the effect of mobility on the pandemic spread showed that reduction of mobility has a significant impact in reducing the number of new infections.

Bayesian nowcasting with leading indicators applied to COVID-19 fatalities in Sweden

The real-time analysis of infectious disease surveillance data is essential in obtaining situational awareness about the current dynamics of a major public health event such as the COVID-19 pandemic. This analysis of e.g., time-series of reported cases or fatalities is complicated by reporting delays that lead to under-reporting of the complete number of events for the most recent time points. This can lead to misconceptions by the interpreter, for instance the media or the public, as was the case with the time-series of reported fatalities during the COVID-19 pandemic in Sweden. Nowcasting methods provide real-time estimates of the complete number of events using the incomplete time-series of currently reported events and information about the reporting delays from the past. In this paper we propose a novel Bayesian nowcasting approach applied to COVID-19-related fatalities in Sweden. We incorporate additional information in the form of time-series of number of reported cases and ICU admissions as leading signals. We demonstrate with a retrospective evaluation that the inclusion of ICU admissions as a leading signal improved the nowcasting performance of case fatalities for COVID-19 in Sweden compared to existing methods.

A One Health real-time surveillance system for nowcasting Campylobacter gastrointestinal illness outbreaks, Norway, week 30 2010 to week 11 2022

BackgroundCampylobacter is a leading cause of food and waterborne illness. Monitoring and modelling Campylobacter at chicken broiler farms, combined with weather pattern surveillance, can aid nowcasting of human gastrointestinal (GI) illness outbreaks. Near real-time sharing of data and model results with health authorities can help increase potential outbreak responsiveness.AimsTo leverage data on weather and Campylobacter on broiler farms to build a risk model for possible human Campylobacter outbreaks and to communicate risk assessments with health authorities.MethodsWe developed a spatio-temporal random effects model for weekly GI illness consultations in Norwegian municipalities with Campylobacter monitoring and weather data from week 30 2010 to 11 2022 to give 1-week nowcasts of GI illness outbreaks. The approach combined a municipality random effects baseline model for seasonally-adjusted GI illness with a second model for peak deviations from that baseline. Model results are communicated to national and local stakeholders through an interactive website: Sykdomspulsen One Health.ResultsLagged temperature and precipitation covariates, as well as 2-week-lagged positive Campylobacter sampling in broilers, were associated with higher levels of GI consultations. Significant inter-municipality variability in outbreak nowcasts were observed.ConclusionsCampylobacter surveillance in broilers can be useful in GI illness outbreak nowcasting. Surveillance of Campylobacter along potential pathways from the environment to illness such as via water system monitoring may improve nowcasting. A One Health system that communicates near real-time surveillance data and nowcast changes in risk to health professionals facilitates the prevention of Campylobacter outbreaks and reduces impact on human health.

Refining epidemiological forecasts with simple scoring rules

Estimates from infectious disease models have constituted a significant part of the scientific evidence used to inform the response to the COVID-19 pandemic in the UK. These estimates can vary strikingly in their bias and variability. Epidemiological forecasts should be consistent with the observations that eventually materialize. We use simple scoring rules to refine the forecasts of a novel statistical model for multisource COVID-19 surveillance data by tuning its smoothness hyperparameter. This article is part of the theme issue 'Technical challenges of modelling real-life epidemics and examples of overcoming these'.

The impact of Indigenous-led programs on alcohol-related criminal incidents: a multiple baseline design evaluation

OBJECTIVES

To evaluate the impact of a multi-component, Aboriginal-led strategy to reduce alcohol-related criminal incidents for Aboriginal people in four rural/remote communities in New South Wales (NSW), Australia.

METHODS

A retrospective multiple baseline design (MBD), using interrupted time series analysis of routinely collected crime data.

RESULTS

A statistically significant reduction in alcohol-related criminal incidents was observed in one community for both victims of crime (parameter estimate -0.195; p≤0.01) and persons of interest (parameter estimate -0.282; p≤0.001). None of the analyses show level shifts, meaning there were no measurable changes immediately post the introduction of the Breaking the Cycle (BTC) programs.

CONCLUSIONS

It is not possible to conclude that the program was effective independently of any other community factors, because the statistically significant result was not observed across multiple communities. The statistically significant result in one community has clear practical benefits in that community: a sustained impact over two years would reduce Aboriginal victims of alcohol-related crime from an estimated 56 incidents per annum to 36, and reduce Aboriginal persons of interest in alcohol-related crime from an estimated 68 alcohol-related person of interest (POI) per annum to 40.

IMPLICATIONS FOR PUBLIC HEALTH

The statistically and practically meaningful result in Community 1 highlights the potential of multi-component, Aboriginal-led strategies to reduce alcohol-related criminal incidents. Earlier engagement with researchers, to identify best-evidence strategies to reduce alcohol harms and to facilitate the use of prospective evaluation designs, would help translate the positive outcome in one community across multiple communities.

Reaction to the COVID-19 pandemic in Seoul with biostatistics

This paper discusses our collaboration work with government officers in the health department of Seoul during the COVID-19 pandemic. First, we focus on short-term forecasting for the number of new confirmed cases and severe cases. Second, we focus on understanding how much of the current infections has been affected by external influx from neighborhood areas or internal transmission within the area. This understanding may be important because it is linked to the government policy determining non-pharmaceutical interventions. To obtain the decomposition of the effect, districts of Seoul should be considered simultaneously, and multivariate time series models are used. Third, we focus on predicting the number of new weekly confirmed cases for each district in Seoul. This detailed prediction may be important to the government policy on resource allocation. We consider an ensemble method to overcome poor prediction performance of simple models. This paper presents the methodological details and analysis results of the study.

Probabilistic forecasts of international bilateral migration flows

Accurate forecasts of migration trends are essential for effective migration policies. We develop a method for probabilistic forecasting of international migration flows between pairs of countries. Our model encodes the observation that the spatial distribution patterns of migration are stable over time. This model produces one-period-ahead probabilistic forecasts that are more accurate than a leading alternative and are well calibrated for international bilateral migration flows, as well as country-level migrant inflows, outflows, and net migration flows. The flow forecasts are sex and age specific and properly account for population change not attributable to migration.

We propose a method for forecasting global human migration flows. A Bayesian hierarchical model is used to make probabilistic projections of the 39,800 bilateral migration flows among the 200 most populous countries. We generate out-of-sample forecasts for all bilateral flows for the 2015 to 2020 period, using models fitted to bilateral migration flows for five 5-y periods from 1990 to 1995 through 2010 to 2015. We find that the model produces well-calibrated out-of-sample forecasts of bilateral flows, as well as total country-level inflows, outflows, and net flows. The mean absolute error decreased by 61% using our method, compared to a leading model of international migration. Out-of-sample analysis indicated that simple methods for forecasting migration flows offered accurate projections of bilateral migration flows in the near term. Our method matched or improved on the out-of-sample performance using these simple deterministic alternatives, while also accurately assessing uncertainty. We integrate the migration flow forecasting model into a fully probabilistic population projection model to generate bilateral migration flow forecasts by age and sex for all flows from 2020 to 2025 through 2040 to 2045.

Understanding the transmission dynamics of a large-scale measles outbreak in Southern Vietnam

OBJECTIVES

During 2018-2020, Southern Vietnam experienced a large measles outbreak of over 26,000 cases. We aimed to understand and quantify the measles spread in space-time dependence and the transmissibility during the outbreak.

METHODS

Measles surveillance reported cases between 1/2018 and 6/2020, vaccination coverage, and population data at provincial level were used. To illustrate the spatiotemporal pattern of disease spread, we employed the endemic-epidemic multivariate time series model decomposing measles risk additively into autoregressive, spatiotemporal, and endemic component. Likelihood-based estimation procedures were performed to determine the time-varying reproductive number R_e of measles.

RESULTS

Our analysis shows that measles incidence was associated with vaccination coverage heterogeneity and spatial interaction between provincial units. The risk of infections was dominated by between-province transmission (36.1% to 78.8%), followed by local endogenous transmission (4.1% to 61.5%) whereas the endemic behavior had a relatively small contribution (2.1% to 33.4%) across provinces. In the exponential phase of the epidemic, R_e was above the threshold with a maximum value of 2.34 (95%CI: 2.20-2.46).

CONCLUSION

Local vaccination coverage and human mobility are important factors contributing to the measles dynamics in Southern Vietnam and the high risk of inter-provincial transmission is of most concern. Strengthening disease surveillance is recommended, and further research is essential to understand the relative contribution of population immunity and control measures in measles epidemics.

A poisson-multinomial spatial model for simultaneous outbreaks with application to arboviral diseases

Dengue, Zika, and chikungunya are arboviral diseases (AVD) transmitted mainly by Aedes aegypti. Rio de Janeiro city, Brazil, has been endemic for dengue for over 30 years, and experienced the first joint epidemic of the three diseases between 2015-2016. They present similar symptoms and only a small proportion of cases are laboratory-confirmed. These facts lead to potential misdiagnosis and, consequently, uncertainty in the registration of the cases. We have available the number of cases of each disease for the n=160 neighborhoods of Rio de Janeiro. We propose a Poisson model for the total number of cases of Aedes-borne diseases and, conditioned on the total, we assume a multinomial model for the allocation of the number of cases of each of the diseases across the neighborhoods. This provides simultaneously the estimation of the associations of the relative risk of the total cases of AVD with environmental and socioeconomic variables; and the estimation of the probability of presence of each disease as a function of available covariates. Our findings suggest that a one standard deviation increase in the social development index decreases the relative risk of the total cases of AVD by 28%. Neighborhoods with smaller proportion of green area had greater odds of having chikungunya in comparison to dengue and Zika. A one standard deviation increase in population density decreases the odds of a neighborhood having Zika instead of dengue by 18% but increases the odds of chikungunya in comparison to dengue by 18% and by 43% in comparison to Zika.

Endemic-epidemic models to understand COVID-19 spatio-temporal evolution

We propose an endemic-epidemic model: a negative binomial space-time autoregression, which can be employed to monitor the contagion dynamics of the COVID-19 pandemic, both in time and in space. The model is exemplified through an empirical analysis of the provinces of northern Italy, heavily affected by the pandemic and characterized by similar non-pharmaceutical policy interventions.

A Model for Highly Fluctuating Spatio-Temporal Infection Data, with Applications to the COVID Epidemic

Spatio-temporal models need to address specific features of spatio-temporal infection data, such as periods of stable infection levels (endemicity), followed by epidemic phases, as well as infection spread from neighbouring areas. In this paper, we consider a mixture-link model for infection counts that allows alternation between epidemic phases (possibly multiple) and stable endemicity, with higher AR1 coefficients in epidemic phases. This is a form of regime-switching, allowing for non-stationarity in infection levels. We adopt a generalised Poisson model appropriate to the infection count data and avoid transformations (e.g., differencing) to alternative metrics, which have been adopted in many studies. We allow for neighbourhood spillover in infection, which is also governed by adaptive regime-switching. Compared to existing models, the observational (in-sample) model is expected to better reflect the balance between epidemic and endemic tendencies, and short-term extrapolations are likely to be improved. Two case study applications involve COVID area-time data, one for 32 London boroughs (and 96 weeks) since the start of the COVID epidemic, the other for a shorter time span focusing on the epidemic phase in 144 areas of Southeast England associated with the Alpha variant. In both applications, the proposed methods produce a better in-sample fit and out-of-sample short term predictions. The spatial dynamic implications are highlighted in the case studies.

A spatio-temporal autoregressive model for monitoring and predicting COVID infection rates

The COVID-19 epidemic has raised major issues with regard to modelling and forecasting outcomes such as cases, deaths and hospitalisations. In particular, the forecasting of area-specific counts of infectious disease poses problems when counts are changing rapidly and there are infection hotspots, as in epidemic situations. Such forecasts are of central importance for prioritizing interventions or making severity designations for different areas. In this paper, we consider different specifications of autoregressive dependence in incidence counts as these may considerably impact on adaptivity in epidemic situations. In particular, we introduce parameters to allow temporal adaptivity in autoregressive dependence. A case study considers COVID-19 data for 144 English local authorities during the UK epidemic second wave in late 2020 and early 2021, which demonstrate geographical clustering in new cases-linked to the then emergent alpha variant. The model allows for both spatial and time variation in autoregressive effects. We assess sensitivity in short-term predictions and fit to specification (spatial vs space-time autoregression, linear vs log-linear, and form of space decay), and show improved one-step ahead and in-sample prediction using space-time autoregression including temporal adaptivity.

Beta-negative binomial nonlinear spatio-temporal random effects modeling of COVID-19 case counts in Japan

Coronavirus disease 2019 (COVID-19) caused by the SARS-CoV-2 virus has spread seriously throughout the world. Predicting the spread, or the number of cases, in the future can facilitate preparation for, and prevention of, a worst-case scenario. To achieve these purposes, statistical modeling using past data is one feasible approach. This paper describes spatio-temporal modeling of COVID-19 case counts in 47 prefectures of Japan using a nonlinear random effects model, where random effects are introduced to capture the heterogeneity of a number of model parameters associated with the prefectures. The negative binomial distribution is frequently used with the Paul-Held random effects model to account for overdispersion in count data; however, the negative binomial distribution is known to be incapable of accommodating extreme observations such as those found in the COVID-19 case count data. We therefore propose use of the beta-negative binomial distribution with the Paul-Held model. This distribution is a generalization of the negative binomial distribution that has attracted much attention in recent years because it can model extreme observations with analytical tractability. The proposed beta-negative binomial model was applied to multivariate count time series data of COVID-19 cases in the 47 prefectures of Japan. Evaluation by one-step-ahead prediction showed that the proposed model can accommodate extreme observations without sacrificing predictive performance.

Impact of intensified control on visceral leishmaniasis in a highly-endemic district of Bihar, India: An interrupted time series analysis

Visceral leishmaniasis (VL) is declining in India and the World Health Organization’s (WHO) 2020 ‘elimination as a public health problem’ target has nearly been achieved. Intensified combined interventions might help reach elimination, but their impact has not been assessed. WHO’s Neglected Tropical Diseases 2021–2030 roadmap provides an opportunity to revisit VL control strategies. We estimated the combined effect of a district-wide pilot of intensified interventions in the highly-endemic Vaishali district, where cases fell from 3,598 in 2012–2014 to 762 in 2015–2017. The intensified control approach comprised indoor residual spraying with improved supervision; VL-specific training for accredited social health activists to reduce onset-to-diagnosis time; and increased Information Education & Communication activities in the community. We compared the rate of incidence decrease in Vaishali to other districts in Bihar state via an interrupted time series analysis with a spatiotemporal model informed by previous VL epidemiological estimates. Changes in Vaishali’s rank among Bihar’s endemic districts in terms of monthly incidence showed a change pre-pilot (3rd highest out of 33 reporting districts) vs. during the pilot (9th) (p<1e-10). The rate of decline in Vaishali’s incidence saw no change in rank at 11th highest, both pre-pilot & during the pilot. Counterfactual model simulations suggest an estimated median of 352 cases (IQR 234–477) were averted by the Vaishali pilot between January 2015 and December 2017, which was robust to modest changes in the onset-to-diagnosis distribution. Strengthening control strategies may have precipitated a substantial change in VL incidence in Vaishali and suggests this approach should be piloted in other highly-endemic districts.

Unmet need for COVID-19 vaccination coverage in Kenya

COVID-19 has impacted the health and livelihoods of billions of people since it emerged in 2019. Vaccination for COVID-19 is a critical intervention that is being rolled out globally to end the pandemic. Understanding the spatial inequalities in vaccination coverage and access to vaccination centres is important for planning this intervention nationally. Here, COVID-19 vaccination data, representing the number of people given at least one dose of vaccine, a list of the approved vaccination sites, population data and ancillary GIS data were used to assess vaccination coverage, using Kenya as an example. Firstly, physical access was modelled using travel time to estimate the proportion of population within 1 hour of a vaccination site. Secondly, a Bayesian conditional autoregressive (CAR) model was used to estimate the COVID-19 vaccination coverage and the same framework used to forecast coverage rates for the first quarter of 2022. Nationally, the average travel time to a designated COVID-19 vaccination site (n = 622) was 75.5 min (Range: 62.9 - 94.5 min) and over 87% of the population >18 years reside within 1 hour to a vaccination site. The COVID-19 vaccination coverage in December 2021 was 16.70% (95% CI: 16.66 - 16.74) - 4.4 million people and was forecasted to be 30.75% (95% CI: 25.04 - 36.96) - 8.1 million people by the end of March 2022. Approximately 21 million adults were still unvaccinated in December 2021 and, in the absence of accelerated vaccine uptake, over 17.2 million adults may not be vaccinated by end March 2022 nationally. Our results highlight geographic inequalities at sub-national level and are important in targeting and improving vaccination coverage in hard-to-reach populations. Similar mapping efforts could help other countries identify and increase vaccination coverage for such populations.

The impact of local vaccine coverage and recent incidence on measles transmission in France between 2009 and 2018

BACKGROUND

Subnational heterogeneity in immunity to measles can create pockets of susceptibility and result in long-lasting outbreaks despite high levels of national vaccine coverage. The elimination status defined by the World Health Organization aims to identify countries where the virus is no longer circulating and can be verified after 36 months of interrupted transmission. However, since 2018, numerous countries have lost their elimination status soon after reaching it, showing that the indicators defining elimination may not be associated with lower risks of outbreaks.

METHODS

We quantified the impact of local vaccine coverage and recent levels of incidence on the dynamics of measles in each French department between 2009 and 2018, using mathematical models based on the "Endemic-Epidemic" regression framework. After fitting the models using daily case counts, we simulated the effect of variations in the vaccine coverage and recent incidence on future transmission.

RESULTS

High values of local vaccine coverage were associated with fewer imported cases and lower risks of local transmissions, but regions that had recently reported high levels of incidence were also at a lower risk of local transmission. This may be due to additional immunity accumulated during recent outbreaks. Therefore, the risk of local transmission was not lower in areas fulfilling the elimination criteria. A decrease of 3% in the 3-year average vaccine uptake led to a fivefold increase in the average annual number of cases in simulated outbreaks.

CONCLUSIONS

Local vaccine uptake was a reliable indicator of the intensity of transmission in France, even if it only describes yearly coverage in a given age group, and ignores population movements. Therefore, spatiotemporal variations in vaccine coverage, caused by disruptions in routine immunisation programmes, or lower trust in vaccines, can lead to large increases in both local and cross-regional transmission. The incidence indicator used to define the elimination status was not associated with a lower number of local transmissions in France, and may not illustrate the risks of imminent outbreaks. More detailed models of local immunity levels or subnational seroprevalence studies may yield better estimates of local risk of measles outbreaks.

ARZIMM: A Novel Analytic Platform for the Inference of Microbial Interactions and Community Stability from Longitudinal Microbiome Study

Dynamic changes of microbiome communities may play important roles in human health and diseases. The recent rise in longitudinal microbiome studies calls for statistical methods that can model the temporal dynamic patterns and simultaneously quantify the microbial interactions and community stability. Here, we propose a novel autoregressive zero-inflated mixed-effects model (ARZIMM) to capture the sparse microbial interactions and estimate the community stability. ARZIMM employs a zero-inflated Poisson autoregressive model to model the excessive zero abundances and the non-zero abundances separately, a random effect to investigate the underlining dynamic pattern shared within the group, and a Lasso-type penalty to capture and estimate the sparse microbial interactions. Based on the estimated microbial interaction matrix, we further derive the estimate of community stability, and identify the core dynamic patterns through network inference. Through extensive simulation studies and real data analyses we evaluate ARZIMM in comparison with the other methods.

Identifying extreme COVID-19 mortality risks in English small areas: a disease cluster approach

The COVID-19 pandemic is having a huge impact worldwide and has highlighted the extent of health inequalities between countries but also in small areas within a country. Identifying areas with high mortality is important both of public health mitigation in COVID-19 outbreaks, and of longer term efforts to tackle social inequalities in health. In this paper we consider different statistical models and an extension of a recent method to analyze COVID-19 related mortality in English small areas during the first wave of the epidemic in the first half of 2020. We seek to identify hotspots, and where they are most geographically concentrated, taking account of observed area factors as well as spatial correlation and clustering in regression residuals, while also allowing for spatial discontinuities. Results show an excess of COVID-19 mortality cases in small areas surrounding London and in other small areas in North-East and and North-West of England. Models alleviating spatial confounding show ethnic isolation, air quality and area morbidity covariates having a significant and broadly similar impact on COVID-19 mortality, whereas nursing home location seems to be slightly less important.

Identifying extreme COVID-19 mortality risks in English small areas: a disease cluster approach

The COVID-19 pandemic is having a huge impact worldwide and has highlighted the extent of health inequalities between countries but also in small areas within a country. Identifying areas with high mortality is important both of public health mitigation in COVID-19 outbreaks, and of longer term efforts to tackle social inequalities in health. In this paper we consider different statistical models and an extension of a recent method to analyze COVID-19 related mortality in English small areas during the first wave of the epidemic in the first half of 2020. We seek to identify hotspots, and where they are most geographically concentrated, taking account of observed area factors as well as spatial correlation and clustering in regression residuals, while also allowing for spatial discontinuities. Results show an excess of COVID-19 mortality cases in small areas surrounding London and in other small areas in North-East and and North-West of England. Models alleviating spatial confounding show ethnic isolation, air quality and area morbidity covariates having a significant and broadly similar impact on COVID-19 mortality, whereas nursing home location seems to be slightly less important.

Modeling digital camera monitoring count data with intermittent zeros for short-term prediction

Digital camera monitoring has revolutionised survey designs in many fields, as an important source of information. The extended sampling coverage offered by this monitoring scheme makes it preferable compared to other traditional methods of survey. However, data obtained from digital camera monitoring are often highly variable, and characterized by sparse periods of zero counts, interspersed with missing observations due to outages. In practice, missing data of relatively shorter duration are mostly observed and are often imputed using interpolation techniques, ignoring long-term trends leading to inherent estimation biases. In this study, we investigated time series forecasting methods that adequately handle intermittency and produced plausible estimates for imputation and forecasting purposes. The study utilised a yearlong digital camera monitoring data set of hourly counts of powerboat launches at three boat ramps in Western Australia. Several time series forecasting methods were evaluated and the accuracies of their point estimates of forecasts for various lead times in hours of up to one week were assessed using cross-validation techniques. Intermittent demand forecasting techniques, including Croston's method and Syntetos-Boylan Approximation (SBA) models, and count data forecasting methods including autoregressive conditional Poisson (ACP) models, integer-valued moving average (INMA) models, and integer-valued autoregressive (INAR) models were evaluated. ACP and INAR models performed better than intermittent demand forecasting techniques for short forecast horizons and provided some evidence of their sufficiency in predicting the dynamics in recreational boating activities. This result established that, in as much as intermittency may be a key feature for a given dataset, it should not override the systemic characteristics of data in the application of forecasting techniques. Our results provide plausible estimates for short-term missing data and forecasts for monitoring events, with applications in supporting proper tracking of usage of facilities, guiding resource allocations and providing insightful perspectives for management decisions.

Epidemiological and spatio-temporal characteristics of COVID-19 in Rwanda

BACKGROUND

The coronavirus disease 2019 (COVID-19) has taken millions of lives and disrupted living standards at individual, societal, and worldwide levels, causing serious consequences globally. Understanding its epidemic curve and spatio-temporal dynamics is crucial for the development of effective public health plans and responses and the allocation of resources. Thus, we conducted this study to assess the epidemiological dynamics and spatio-temporal patterns of the COVID-19 pandemic in Rwanda.

METHODS

Using the surveillance package in R software version 4.0.2, we implemented endemic-epidemic multivariate time series models for infectious diseases to analyze COVID-19 data reported by Rwanda Biomedical Center under the Ministry of Health from March 15, 2020 to January 15, 2021.

RESULTS

The COVID-19 pandemic occurred in two waves in Rwanda and showed a heterogenous spatial distribution across districts. The Rwandan government responded effectively and efficiently through the implementation of various health measures and intervention policies to drastically reduce the transmission of the disease. Analysis of the three components of the model showed that the most affected districts displayed epidemic components within the area, whereas the effect of epidemic components from spatial neighbors were experienced by the districts that surround the most affected districts. The infection followed the disease endemic trend in other districts.

CONCLUSION

The epidemiological and spatio-temporal dynamics of COVID-19 in Rwanda show that the implementation of measures and interventions contributed significantly to the decrease in COVID-19 transmission within and between districts. This accentuates the critical call for continued intra- and inter- organization and community engagement nationwide to ensure effective and efficient response to the pandemic.

The impact of long-term non-pharmaceutical interventions on COVID-19 epidemic dynamics and control: the value and limitations of early models

Mathematical models of epidemics are important tools for predicting epidemic dynamics and evaluating interventions. Yet, because early models are built on limited information, it is unclear how long they will accurately capture epidemic dynamics. Using a stochastic SEIR model of COVID-19 fitted to reported deaths, we estimated transmission parameters at different time points during the first wave of the epidemic (March-June, 2020) in Santa Clara County, California. Although our estimated basic reproduction number ([Formula: see text]) remained stable from early April to late June (with an overall median of 3.76), our estimated effective reproduction number ([Formula: see text]) varied from 0.18 to 1.02 in April before stabilizing at 0.64 on 27 May. Between 22 April and 27 May, our model accurately predicted dynamics through June; however, the model did not predict rising summer cases after shelter-in-place orders were relaxed in June, which, in early July, was reflected in cases but not yet in deaths. While models are critical for informing intervention policy early in an epidemic, their performance will be limited as epidemic dynamics evolve. This paper is one of the first to evaluate the accuracy of an early epidemiological compartment model over time to understand the value and limitations of models during unfolding epidemics.

A comparison of multiple neighborhood matrix specifications for spatio-temporal model fitting: a case study on COVID-19 data

Establishing proper neighbor relations between a set of spatial units under analysis is essential when carrying out a spatial or spatio-temporal analysis. However, it is usual that researchers choose some of the most typical (and simple) neighborhood structures, such as the first-order contiguity matrix, without exploring other options. In this paper, we compare the performance of different neighborhood matrices in the context of modeling the weekly relative risk of COVID-19 over small areas located in or near Valencia, Spain. Specifically, we construct contiguity-based, distance-based, covariate-based (considering mobility flows and sociodemographic characteristics), and hybrid neighborhood matrices. We evaluate the goodness of fit, the overall predictive quality, the ability to detect high-risk spatio-temporal units, the capability to capture the spatio-temporal autocorrelation in the data, and the goodness of smoothing for a set of spatio-temporal models based on each of the neighborhood matrices. The results show that contiguity-based matrices, some of the distance-based matrices, and those based on sociodemographic characteristics perform better than the matrices based on k-nearest neighbors and those involving mobility flows. In addition, we test the linear combination of some of the constructed neighborhood matrices and the reweighting of these matrices after eliminating weak neighbor relations, without any model improvement.

Association between respiratory hospital admissions and air quality in Portugal: A count time series approach

Although regulatory improvements for air quality in the European Union have been made, air pollution is still a pressing problem and, its impact on health, both mortality and morbidity, is a topic of intense research nowadays. The main goal of this work is to assess the impact of the exposure to air pollutants on the number of daily hospital admissions due to respiratory causes in 58 spatial locations of Portugal mainland, during the period 2005-2017. To this end, INteger Generalised AutoRegressive Conditional Heteroskedastic (INGARCH)-based models are extensively used. This family of models has proven to be very useful in the analysis of serially dependent count data. Such models include information on the past history of the time series, as well as the effect of external covariates. In particular, daily hospitalisation counts, air quality and temperature data are endowed within INGARCH models of optimal orders, where the automatic inclusion of the most significant covariates is carried out through a new block-forward procedure. The INGARCH approach is adequate to model the outcome variable (respiratory hospital admissions) and the covariates, which advocates for the use of count time series approaches in this setting. Results show that the past history of the count process carries very relevant information and that temperature is the most determinant covariate, among the analysed, for daily hospital respiratory admissions. It is important to stress that, despite the small variability explained by air quality, all models include on average, approximately two air pollutants covariates besides temperature. Further analysis shows that the one-step-ahead forecasts distributions are well separated into two clusters: one cluster includes locations exclusively in the Lisbon area (exhibiting higher number of one-step-ahead hospital admissions forecasts), while the other contains the remaining locations. This results highlights that special attention must be given to air quality in Lisbon metropolitan area in order to decrease the number of hospital admissions.