This approach, named BaySTDetect, was applied to detect unusual trends for asthma and chronic obstructive pulmonary disease at Clinical Commissioning Group (CCG) level in England (211 in total) on monthly data between August 2010 and March 2011, across mortality, hospital admissions and general practice drug prescriptions.44, To illustrate the typical output obtained from this model, Figure 2 shows the area-specific time trends of the CCGs which were detected as unusual, plotted against the national trend. Published by Oxford University Press on behalf of the International Epidemiological Association. It is a particularly important issue in low-income settings because surveillance studies often need to rely on information from surveys, and the lack of financial resources may make comprehensive coverage of data sources (e.g. As can be seen, the disease-mapping approach using the standard threshold of 0.8 on the posterior probability scale (DM1) shows the worst performance. In later life with high chances of obesity, cardiovascular diseases, diabetes can lead to disability and premature death. Prevention and management of chronic obstructive pulmonary disease (COPD) and chronic Kidney disease (CKD); and better management of co-morbidities such as diabetes and tuberculosis are also considered under the programme. In this section, we first discuss how data availability is one of the key challenges in surveillance studies, before giving a generic overview of test-based approaches for NCD surveillance. But India has taken the unprecedented step of setting a tenth target to address household air pollution. Another is that the Monte Carlo sampling method allows the computation of whatever joint probability statements are required. One of the biggest challenges researchers face when analysing large and complex space-time datasets is their computational burden. We then focus on three commonly used models within the Bayesian Hierarchical Model (BHM) framework and, through a simulation study, we compare their performance. (a) Relative risks and 95% credible intervals of hospital admissions for asthma and COPD for the national (common) temporal trend and for Harrow CCG, classified as unusual. As there is an explicit relationship among areas globally and/or locally, through structured random effects, places belonging to a particularly small-area can influence results for other areas, hence alleviating the MAUP problem.77. The shared component model,37 originally developed for two diseases, includes a common component (likely to reflect common risk factors) and a disease-specific one, which can point towards specific risk factors otherwise masked in a single disease model. Whereas the importance of controlling for multiple testing is clear in classical significance testing, the analogous problem in predictive setting is less of a concern.58 One reason for this is that local predictions from hierarchical models are naturally smoothed towards the global mean, making these consequently less prone to false-positive findings than unsmoothed area-by-area interval estimates. In addition, it is crucial to control the proportions of observations that are falsely classified as unusual [false-discovery rate (FDR)] and common [false omission rate (FOR)], respectively; these should typically not exceed a value of 0.05, specified based on the standard P-value threshold. According to the World Health Organization (WHO), surveillance is the ‘ongoing systematic data collection, analysis and interpretation and dissemination of information in order for action to be taken’. For these 15 areas, we selected the signal to be increased by log(2) for time points 3 and 10, and decreased by log(2) for time points 6, 12 and 15 out of the total 15 time points. to exhibit a risk pattern not deviating from the expected one.43. STmix1 gave no false-positive results (FDR = 0) and a sensitivity of 0.773, whereas for STmix2 sensitivity increased to 0.969, but at the same time a much higher proportion of false-positives was detected (FDR = 0.220). NCD surveillance shares many objectives with infectious disease surveillance, including generating information to guide public health action and detecting the health impact of environmental exposures or of environmentally driven disease vectors; however, it also presents some different methodological challenges.7,8. Non-communicable Disease. These were developed originally in the temporal setting only13; here, a fixed length ‘scanning window’ is passed over the time-series data with the number of cases in the window being recorded. The UK Small Area Health Statistics Unit (SAHSU) is part of the MRC-PHE Centre for Environment and Health, which is supported by the Medical Research Council (MR/L01341X/1) and Public Health England (PHE). Non-communicable diseases (NCDs) have emerged as a major component of disease burden globally. Source: Environment and Health Atlas.35 (b) Area-specific posterior probability that an area is characterized by a relative risk of malignant melanoma above 1. There has been a shift from communicable to non-communicable diseases. Non-communicable diseases are also called non-contagious or non-infectious diseases. Introduction. A Survey on Cardiovascular Nursing Occupational Standard: Meeting the Needs of Employers. Concepts and Techniques in Modern Geography, An integrated framework for the geographic surveillance of chronic disease, Spatial aggregation and the ecological fallacy. The two mixture models returned more comparable performances. areas and/or time points with trends different from the expected ones, adding a space-time interaction parameter into the latent process.38 The detection of anomalies may indicate the presence of an emerged localized risk factor, the impact of an intervention, or differences in the quality of data, such as misdiagnosis of a disease, and under- or over-reporting of cases. Resident Physician in Cardio-Thoracic and Vascular Surgery, Research Assistant Professor of Epidemiology, Copyright © 2020 International Epidemiological Association. Mixture models have been proposed as a formal approach to anomaly detection. Health promotion through social media is also being used to generate awareness about prevention and control of NCDs, such as use of mobile technology in applications called mDiabetes for diabetes control, mCessation to help for quit tobacco, and no more tension as a support for mental stress management. It was applied to male and female lung cancer38 and later extended to jointly model multiple diseases,30,39 with an application on oral cavity, oesophagus, larynx and lung cancers in males in the 544 districts of Germany from 1986 to 1990. Some work in this area includes Foreman et al.59 who, using annual vital statistics for 1974–2011 at the US state spatial resolution, forecasted mortality up to 2024; and Ugarte et al.67 used P-splines to forecast cancer mortality counts in Spanish regions for 2009–11 using data from 1975–2008. World Health Organization. mortality/cancer registries) over an entire country infeasible.9, In the past 15 years, a number of Health and Demographic Surveillance Systems (HDSS) have been established in low-income settings to provide a reliable source of health data, and are now linked together through the International Network for the continuous Demographic Evaluation of Populations and Their Health (INDEPTH10). A common choice for capturing temporal dependence is a random walk prior,29 but extensions to incorporate spatiotemporal interactions among neighbouring areas and time points have also been developed.30 This framework can also account for factors known to modify spatial and temporal trends that, in the context of NCDs, will include demographic variables (e.g. In what follows, we use BHM as a shorthand for Bayesian inference applied to a hierarchically specified model. The graph below shows the declines across respiratory, cancer and circulatory diseases. In addition to this, its detection mechanism does not consider specific patterns in the time trends. Through the exceedance probabilities, these maps give a perception of the uncertainty around the area-level relative risks estimates. An additional version of spatial scan statistic was proposed to account for correlation across spatial units, which was not considered before.17 Scan statistics have been extensively applied to numerous health care applications. National HIV Surveillance System (NHSS). This applies particularly in the small-area context, where the number of space-time units investigated can vary substantially depending on the chosen spatial and temporal resolution, from a few hundreds to hundreds of thousand units, particularly when several outcomes are jointly analysed (for instance, Foreman et al.59 considered jointly deaths/age/sex specific space-time trends in the USA). This idea was extended to a spatial version of the scan statistic,15 which was later further extended to the spatiotemporal setting.16 In this case, the scanning window is represented by a cylinder, where the diameter specifies the spatial dimension and the height the temporal dimension. Syndromic data, such as primary care data, drug prescriptions, nurse calls and home visits, which are indicative of a potential anomaly, may provide an additional level of information leading to a detection event before the data aberration occurs.68 Diggle et al.69,70 analysed NHS non-emergency telephone calls reporting symptoms of gastrointestinal diseases. Differences across the competing models were observed in terms of computation time, an important factor in assessing their performance. In the current context, an HM combines two elements: a process model that describes how disease risk varies over space and time, typically involving both extant covariate effects and a latent spatiotemporal stochastic process; and a data model that describes the statistical properties of the available health outcome data conditional on the realization of the underlying risk process. Comparison of two methods for cell count determination in the course of biocide susceptibility testing. Richardson S, Thomson A, Best N, Elliott P. Hansell AL, Ghosh RE, Fecht D, Fortunato L. Mahaki B, Mehrabi Y, Kavousi A, Schmid VJ. India’s National Monitoring Framework for Prevention and Control of NCDs has committed for a 50% relative reduction in household use of solid fuel and a 30% relative reduction in prevalence of current tobacco use by 2025. an area, period of time, or combination of space and time) when compared with the whole study region. Non-infectious are non-communicable diseases and caused by a variety of reasons. To know more about NCDs and National Programme Guidelines-  Click here, /www.who.int/news-room/fact-sheets/detail/noncommunicable-diseases, mohfw.gov.in/Major-Programmes/non-communicable-diseases-injury-trauma/Non-Communicable-Disease-II/national-programme-prevention-and-control-cancer-diabetes-cardiovascular-diseases-and, You would need to login or signup to start a Discussion. A non-communicable disease (NCD) is a disease that is not transmissible directly from one person to another. This can be particularly challenging for rare diseases where the numbers of cases at small-area level are very low. © The Author(s) 2020. Physical inactivity, unhealthy diets (diets low in fruit, vegetables, and whole grains, but high in salt and fat), tobacco use (smoking, secondhand smoke, and smokeless tobacco), and the harmful use of alcohol are the main behavioural risk factors for NCDs. Marta Blangiardo, Areti Boulieri, Peter Diggle, Frédéric B Piel, Gavin Shaddick, Paul Elliott, Advances in spatiotemporal models for non-communicable disease surveillance, International Journal of Epidemiology, Volume 49, Issue Supplement_1, April 2020, Pages i26–i37, https://doi.org/10.1093/ije/dyz181. In response to the “WHO Global Action Plan for the Prevention and Control of NCDs 2013-2020”, India is the first country to adopt the National Action Plan with specific national targets and indicators aimed at reducing the number of global premature deaths from NCDs by 25% by 2025. In a simulation study, we found that mixture models designed for detection perform better than standard disease mapping models. Leroux BG, Lei X, Breslow N. Estimation of disease rates in small areas: a new mixed model for spatial dependence. Non-communicable diseases (NCDs) are medical conditions or diseases that are not caused by infectious agents. Topics Covered: Issues relating to development and management of Social Sector/Services relating to Health, Education, Human Resources, issues relating to poverty and hunger. NCDs kill approximately 41 million people (71% of global deaths) worldwide each year, including 14 million people who die too young between the ages of 30 and 70. Surveillance methods must be able to capture spatial and temporal patterns in both lifestyle/environmental exposures and health outcomes. We selected 15 areas to deviate from the overall time trend over the last five time points. Tackling the epidemic of chronic diseases – or non-communicable diseases (NCDs) – is at the heart of this agenda, and it’s a major challenge. Following the design initially proposed by Li et al.,43 and later used by Boulierei et al.,47 we used real asthma hospital episode statistics (HES) data to generate 50 simulated datasets. However, there may be challenges in future due to selective data availability following perceived concerns about data security and confidentiality, as demonstrated by the newly implemented NHS National Data Optout Programme. This will potentially lead to bias in population representativeness due to non-random missingness12 which will need to be addressed using advanced statistical methods, for instance through the integration of data from appropriate surveys/cohorts, as proposed in the context of residual confounding.73, An important issue with surveillance studies is that of the spatial resolution and the type of geographical areas considered; modifying these might lead to different results, as the spatial distribution of the outcome will depend on these choices. Both elements are specified up to the values of a set of unknown parameters, which can be estimated by Bayesian or non-Bayesian versions of likelihood-based inference, typically implemented using Markov chain Monte Carlo integration and Monte Carlo likelihood maximization methods, respectively. Wang Y, Pirani M, Hansell AL, Richardson S, Blangiardo M. Yiannakoulias N, Svenson LW, Schopflocher DP. We consider: (i) two different thresholds for DM: 0.8 as commonly used and previously described (DM1): a more conservative threshold of 0.9 (DM2), under the assumption that false-positives are more important to minimize than false-negatives; and (ii) two different rules for STmix as presented in the original paper: an area is modified if at least for one time point the space-time interaction has a probability greater than 0.8 to be above 1 (STmix1); an area is modified if for at least three time points the space-time interactions have an average probability greater that 0.8 to be above 1 (STmix2). In particular, two alternative models are considered: the first one assumes a global time trend for all areas (common trend), and the second estimates a time trend for each area independently (area-specific trend). Later we introduce the computational aspects of the BHM modelling framework for NCD surveillance; then we run a simulation study to evaluate advantages and drawbacks of the approaches presented in detecting areas deviating from the expected trend. Finally, we consider how to use and interpret the complex models, how model selection may vary depending on the intended user group and how best to communicate results to stakeholders and the general public. World Health Organization. The Environment and Health Atlas for England and Wales36 (typical output from the Atlas was presented in Figure 1) is an example of work in this direction, providing stakeholders and the general public with a collection of maps to inform on the spatial distribution of environmental factors and diseases. We defined TP as the number of true-positives, FP as false-positives, TN as true-negatives and FN as false-negatives, respectively. Electronic Health Record Standards For India Helpdesk, Yash Respiratory & Medical Disease Centre, Northcentral Liver, Digestive Disease & Endoscopy Centre, Yash Prasutigruh & Women Disease Hospital, Dr. Kamras Urinary & Surgical Diseases Centre, Renew R. P Endoscopic & Enitre Women Diseases Clin, Escorts - Mehta Heart Center & Mehta Institute For Chest & Heart Diseases & Intensive Care Unit, Mehta Institute For Chest & Heart Diseases & I.C.U, Metro Centre For Liver & Digestive Diseases, Surya Clinic & Centre For Life Style Diseases, Jain Hospital And Institute Of Digestive Diseases - Solapur, International Center For Cardio Thoracic & Vascular Diseases(Frontier Life Hosp). Tackling the risk factors will therefore not only save lives; it will also provide a huge boost for the economic development of the country. There is increasing recognition of the importance of surveillance for NCDs. This shows a rapid epidemiological transition with a shift in disease burden to NCDs. They contribute to raised blood pressure (hypertension); raised blood sugar (diabetes); raised and abnormal blood lipids (dyslipidaemia); and obesity. Find Out NCDs include Parkinson's disease, autoimmune diseases, strokes, most heart diseases, most cancers, diabetes, chronic kidney disease, osteoarthritis, osteoporosis, Alzheimer's disease, cataracts, and others.NCDs may be chronic or acute. These methods extend the scan statistics to estimate the time trend via a regression-based model specifying either a linear or a quadratic function. Some of the reasons for the non-infectious disease are genetics, nutritional deficiency, age and sex of the individual and so on. We believe that epidemiological surveillance will be at the centre of future methodological research to match the continuous increase in data availability, e.g. In addition to estimating parameters, the scientific goals of health surveillance include prediction of relevant properties of the unobserved risk surface as it evolves in real time. These are caused by biological factors such as high blood pressure, obesity, high blood sugar and high blood cholesterol. They used exceedance probabilities to define maps of potential outbreaks. For instance, Dai et al.72 linked tweets with the American Community Survey and the Behavioral Risk Factor Surveillance System to study asthma prevalence at the state level in the USA. Piel FB, Parkes BL, Daby H, Hansell AL, Elliott P. Kulldorff M, Heffernan R, Hartman J, Assunçao R, Mostashari F. Sherman RL, Henry KA, Tannenbaum SL, Feaster DJ, Kobetz E, Lee DJ. However, there is increased importance of early warning detection, so that unusual behaviour can be detected at the earliest possible time. In the context of disease mapping, this leads to a spatial, temporal and spatiotemporal risk distribution that researchers can map, in terms of point estimates but also of associated measures of uncertainty. Therefore, NCDs and its risk factors have great importance to young people as well. Each of the 50 simulations took on average 33.4 min for models DM1 and DM2, 39.2 min for models STmix 1 and STmix2 and 66.8 min for FlexDetect. These are chronic diseases of long duration, and generally slow progression and are the result of a combination of genetic, physiological, environmental and behaviours factors. They are also known as chronic diseases as they remain for longer period of time. However, attention should be paid to the choice of threshold, as this affects the results. The majority of premature NCD deaths are preventable. Part of this work was supported by an Early Career MRC-PHE Fellowship awarded to A.B. A non-communicable disease, or NCD, is a medical condition or disease which by definition is non-infectious and cannot be passed from person to person. Welfare schemes for vulnerable sections of the population by the Centre and States and the performance of these schemes. Posterior mean and 95% credible intervals for the competing models in the simulation study. Perhaps the most popular test-based methods used for NCD surveillance are the scan statistics. The third UN High-Level Meeting on Non-Communicable Diseases (NCDs) on Sept 27, 2018, will review national and global progress towards the prevention and control of NCDs, and provide an opportunity to renew, reinforce, and enhance commitments to reduce their burden. In particular, Abellan et al.42 developed a BHM model (termed STmix) where a mixture of two normal distributions characterized by different variances is specified for the space-time interaction. Bonferroni correction has been extensively used in epidemiology to correct for multiple testing, particularly in omics studies,48,49 but it is well known that this approach leads to conservative results. Non-communicable diseases remain unabated – older ages 65 and above driving the burden of disease A total of 460 236 deaths were recorded in South Africa in 2015, indicating a decline of 3,0% in deaths processed between 2015 and 2014 (474 659), this is according to the Mortality and causes of death , 2015 report released by Statistics South Africa today.