• No se han encontrado resultados

Evaluar la efectividad de las actividades del evento “MINGA PARA MI RÍO”

Pregunta 1. ¿De donde proviene el agua para la cuidad de Quito?

3. Evaluar la efectividad de las actividades del evento “MINGA PARA MI RÍO”

main, due to the general applicability of patient risk scoring techniques and the generality of the developed toolkits. Moreover, the research is expected to contribute to the academic communities, including the operational research and health and social care modellers, in order to implement more effective decision support systems and risk indices.

1.8

Thesis Overview

The remainder of this thesis is organised as follows. Chapter 2 provides a background on emergency readmission and risk scoring modelling. Then, complexity levels in data quality, feature generation, modelling and validation are presented inChapter 3. More- over, the modelling approaches that are used throughout this research are explained in Chapter 4. Next, the three main phases of the analyses are defined in Chapter 6, including the considered benchmarking models. Chapter 5 contains a brief overview of the healthcare data,NHSadministrative data and the description of the extracted sam- ples. Thereafter, the first phase of the project, the healthcare pre-processing framework is presented inChapter 7, which is based on the HESand the SUS, but has a generic structure. Chapter 8 presents the Predictive Risk Modelling of Hospital Readmission (ERMER). Then, the Temporal-Comorbidity Adjusted Risk Emergency Readmission (T-CARER) is presented inChapter 9. Chapter 10, describes the open-source toolkits developed for applying theERMER and theT-CARERsolutions. Finally,Chapter 11

Chapter 2

Background

For many healthcare providers and purchasers identification of high-risk events has been a major concern. According toLewis et al.(2011), there are three major sources of risk to the healthcare system:

• Ageing population and frailty;

• The increasing number of people with long-term conditions; • The rising rate of emergency admissions to hospitals.

Firstly, a major concern in healthcare organisations throughout the world is about coping with an ageing population (Caley and Sidhu, 2011). Survey results (Chitnis et al.,2012,Leadbeater and Garber,2010) show that many people would prefer to die with appropriate care support at home rather than at a hospital, and yet the number of death in hospitals can reach to 65% if there is no appropriate policy in place. Also, the average cost of hospital care is higher than the social care for older and terminally ill patients. And, the costs of care in the final phases before death are very high in hospitals. Looking further ahead, it is projected that people aged over 85 to almost double by 2030, with an additional 600,000 of the ageing population to need significant care (FCS,2011). While a quarter of people aged over 65 will need to spend very little in care over their life, half can expect the cost of up to £20,000, and one in 10 can expect the cost of £100,000 (FCS,2011). According to a recent Nuffield Trust report (Georghiou et al., 2012), the average cost of social care increases with the age of the patient. However, the cost of social care stays cheaper than hospital care for age below 85. Based on gender, the intersection point of the hospital and the social care

Chapter 2 Background 14 costs for male patients are close to age 90 and the female patients are approximately at age 80.

Furthermore,Figure 2.1presents the risk segmentation for a typical population accord- ing to Kaiser pyramid. Although it demonstrates that moderate to high-risk patients represent a very small percentage of the population, their future utilisation is extremely high compared to the majority of the population (Lewis et al.,2011).

Figure 2.1: Risk segmentation and future utilisation

Sometimes a hospital admission can be avoided by residential setting substitution or social care. According to the analysis by Bardsley et al. (2012) on a wide population in England, the use of social care may prevent the need for hospital care. The End-of- Life (EoL) research help patients to get appropriate support services towards EoLby better management of resources and patients. The ambition of theNHSis to increase the number of people who die in their usual place of residence to 60%. This baseline in 2007 was 38%, and with the EoLpractices that were in place, it reached to 42% in 2012 (NHS,2012,2013d).

Secondly, the ageing population and changes in lifestyles mean an increase in the num- ber of people with long-term conditions or comorbidities. For instance, there have been significant rises in chronic kidney disease, diabetes and cancer between 2006 and 2011 (DH, 2012). Also, it has been predicted that people with comorbidity conditions to rise from 1.9 million in 2009 to 2.9 million in 2018 (Fund,2013).

Moreover, the time that is spent in poor general health, a limiting chronic health or disability, can be attributed to frailty in some cases. Frailty refers to the condition of being weak and delicate, and it mainly develops as a result of ageing. It is associated with the state of high vulnerability and decreased the ability to sustain homoeostasis,

2.1 Preventable Emergency Admissions 15 which is correlated with high risk of adverse outcome including falls, delirium, im- mobility and disability, incontinence and susceptibility to medications and their side effects (Eeles et al.,2012,BGS,2014,2015,Walston et al.,2006). In the UK, the life expectancy is 17.8 years on average for a 65 years old male, of which about 43.3% is in poor general health, and about 41.6% is with a limiting chronic health condition or disability. Similarly, on average a 65-year-old female has a life expectancy of 20.4 years, of which 43.1% will be likely in poor general health and 45.1% with a limiting chronic health condition or disability (NICE,2016,ONS,2012).

Thirdly, the rise in the rate of emergency admission to hospitals is another contributing element. Discharging patients is a primary way of providing free beds in healthcare sectors. But if the estimated risks by healthcare administrators and decision support systems are not correct, it may lead to readmission of patients. Patient-flow mod- elling solutions, like Length-of-Stay (LoS), enable managers to better understand the operational and clinical functions (Adeyemi et al.,2013). TheLoSmodelling includes capturing the flow of patients from admission to discharge. The flow is through a num- ber of conceptual (virtual) phases that patients go through. The predictive models of

LoSuse the time spent in phases, in addition to the clinical data and the demographic data, to identify events.

In the following subsections, first, the preventable emergency admissions are defined and discussed. Then, the emergency readmission predictive modelling and the comor- bidity risk index modelling are summarised.

2.1

Preventable Emergency Admissions

According to a recent report by the Organisation for Economic Co-operation and Devel- opment (OECD), the healthcare spending have fallen in the half of the European Union (EU) countries in real terms1, including the UK between 2009 and 2012 after about forty years (OECD,2014). The spending in real terms per-capita was increasing by an average 4.9% per-year over the previous decade in theUK, until 1.3% decline between 2009 and 2012. In general, these declines were due to cuts in workforce and salaries, reductions in fees and pharmaceutical prices and increase in patient co-payments. Many countries are developing strategies to reduce down avoidable hospital care (OECD,

2014, Nolte and McKee, 2008). Over the last decade, the National Health Service (NHS) in England has been transformed through efficiency savings measures, such

2.1 Preventable Emergency Admissions 16 as the payment reform and quality improvement measures like marginal rate tariffs (Charlesworth et al., 2014). However, increasing demands for emergency admission still remain a major issue. A well-performing healthcare system must be able to pro- vide necessary policies (NICE,2016) for preventive care. In below, four major policies that are directly related to emergency readmissions are highlighted, and their impacts are discussed.

Firstly, there is sound evidence that the quality of care at the primary care level can reduce down potentially avoidable admissions. One approach is to use admissions of patients with the Ambulatory Care Sensitive Conditions (ACSCs)2 as a general indicator for optimality assessment of primary care, community services and outpatient care (Ansari et al.,2006,Billings et al.,1993,Purdy et al.,2009a,b). At present, twenty- seven ACSCs are specified in the NHS Outcomes Framework (Bardsley et al., 2013,

Blunt,2013b) as markers of improved health outcomes (Section 2.3.1).

Moreover, ACSCs are identified by experts and do not usually take into account the population and the quality of care. Consequently, they can be misleading and may reduce down the predictive model’s accuracy. Therefore, the rate of admission for care sensitive conditions may be adjusted by the characteristics of local population, such as age, deprivations, morbidity levels, area of residence, ethnicity, environmental factors, prevalence rates of diagnosed and undiagnosed based using the Quality and Outcomes Framework (QOF), QOF patient experience and QOF clinical quality of care (Calder´on-Larra˜naga et al., 2011, HSCIC,2014c, Purdy et al.,2009b,Sanderson and Dixon,2000, Tian et al.,2012).

Secondly, the Payment by Results (PbR) strategy is based on the Healthcare Resource Group (HRG) and is likely to be the heart of payment system for the coming decade. Most countries in Europe use a similar system known as case-based payment, which is based on the Diagnosis-Related Group (DRG) Fetter et al. (1980), Mistichelli (1984) and several other metrics (e.g. metrics based on assessments of demands and supplies) to calculate fixed annual budget. Since the introduction of thePbRin theNHS, about a decade ago, the general evaluations have been positive. But, there is no robust evidence on its long-term impact or its health system efficiency (Busse et al., 2011,

Charlesworth et al.,2014,Quentin et al.,2011).

Thirdly, the marginal rate was originally introduced to discourage unnecessary emer- gency admissions. But, according to the recent report by the DoH (DH, 2014), the introduction of the national 30% marginal rate tariff to limit the incentive for increased emergency admission did not meet the costs, and demand still continues to increase.

2

The Ambulatory Care Sensitive Conditions (ACSCs) is also known as the primary care sensitive condition (PCSCs)

2.2 Emergency Readmission Prediction Modelling 17

Documento similar