Key themes of our work that cut across organisational and disciplinary groups:
The goal is to lay foundations for new preventive efforts by identifying and evaluating causal risk factors for selected major diseases, including vascular diseases, cancers, neurodegenerative diseases, diabetes, and other age-related conditions.
Achievements We have led the discovery of >350 genetic loci in a range of common diseases (eg, breast, prostate, and ovarian cancers; coronary disease; type 2 diabetes) and risk factors (eg, lipids; insulin resistance) that have been reported in >30 publications in Nature and NatureGenetics since 2008. These findings have opened new avenues of biology and mechanistic understanding and helped develop new risk prediction tools in widespread use. Using the principle of “Mendelian randomisation”, we have identified risk factors likely to have causal roles (eg, triglyceride pathways and IL-6 signalling in coronary disease [Lancet 2011;Lancet 2012]; natriuretic peptides in type 2 diabetes [PLoS Medicine 2011]), and others that appear non-causal (eg, C-reactive protein [BMJ 2012, Lancet 2011]). These studies have yielded implications for industry, eg: Novartis (triglycerides results), Roche (IL-6 results), and ISIS pharmaceuticals (C-reactive protein results). We have quantified the impact of risk factors amenable to therapeutic modification, such as subclinical hypothyroidism in coronary disease (JAMA 2010) and dysglycaemia in relation to vascular and non-vascular conditions (NEJM 2011). We are testing the latter hypothesis in an HTA-funded trial of metformin (see iv in this section). We have shown that different subtypes of breast cancer defined by biological markers show important differences in prognosis (PLoS Medicine 2010), suggesting implications for targeting adjuvant chemotherapy.
Future plans – We are integrating “multi-omics” (eg, genomics, epigenomics, metabolomics) in participants linked to multiple NHS e-health records in 100,000 well-phenotyped individuals. We are establishing an e-health study involving 2 million blood donors in England. We are extending studies of 90,000 participants in non-European populations in which certain risk factor levels are unusually high (eg, toxic metals in Bangladesh; consanguinity in Pakistan; hepatitis C infection in Africa) and in which cardiometabolic risk is also high. We are extending major genetic sequencing efforts in Africans (eg, the African Genome Variation Project) to >20,000 South Asians.
Behaviour and health
The goal is to elucidate determinants of health behaviours, estimate the impact of these behaviours on chronic disease outcomes, and develop and evaluate individual and population-level interventions that modify behaviours to promote health and prevent disease.
Achievements – To define the wider determinants of risk behaviours, we study datasets throughout the lifecourse, drawing on health psychology, neuroscience, and behavioural economics to understand the basis for behaviour cued by environments and social patterning, with implications for our intervention studies. Our cohort studies have suggested that a combination of four health behaviours in adults (not smoking, moderate intake of alcohol, physical activity, consumption of ≥5 servings of fruit and vegetables daily) is associated with a 14 year greater life expectancy (PLoS Medicine 2008). Our meta-analysis has shown that higher levels of physical activity in adults attenuate the influence of FTO, the strongest known genetic susceptibility factor for obesity (PLoS Medicine 2011). Our meta-analysis has shown that physical activity promotion in primary care to sedentary adults resulted in sustained increases in physical activity levels (BMJ 2012). However, our randomised trial in sedentary individuals at familial risk of diabetes showed that a theory-based behaviour intervention was no more effective than an advice leaflet for promotion of physical activity (Lancet 2008). Our trials have shown that tailored smoking cessation messages delivered by post increase the number of smokers who try to quit, and that for web based information, tailoring did not appear to increase the effectiveness of messages encouraging patients to quit.
Future plans – We are identifying which interventions at the population-level (eg, taxation, legislation, ‘nudging’, environmental changes) are most likely to influence behaviour. We are developing and evaluating tailored interventions for smoking cessation, physical activity and medication adherence focusing on very brief interventions delivered by health care practitioners and computer-based approaches (eg, self-monitoring, mobile phone texting, smartphone apps). We have established a portfolio of planned interventions (eg, the Baby Milk Study to evaluate a complex behavioural intervention to avoid excessive formula-milk intake during infancy) and natural experiments (eg, a cohort study to assess the impact of the introduction of the Cambridgeshire Guided Busway on travel and physical activity).
Diagnosis and screening
The goal is to develop and evaluate approaches that can cost effectively enhance the diagnosis, early detection, and prediction of chronic disease outcomes. Achievements We have shown that there are marked differences in GP referral patterns across cancer types according to socio-demographic groups, suggesting the need to prioritise younger people and ethnic minorities for early diagnosis initiatives (Lancet Oncology 2012). Our trial of approaches to diagnosing suspicious moles reinforced the importance and effectiveness of a seven point checklist for GPs (BMJ 2012), for which Walter was awarded the RCGP cancer research paper of the year. We have established the cost-effectiveness of screening men for abdominal aortic aneurysm (BMJ 2009, NEJM 2010), and defined appropriate surveillance intervals (JAMA 2013). Our trial of offering antenatal screening for sickle cell disease and thalassaemia as part of consultation for pregnancy confirmation in primary care showed an increase in the proportion of women screened before 10 weeks’ gestation (BMJ 2010). We have integrated our gene-disease discoveries into widely used risk prediction algorithms for breast and ovarian cancer (BOADICEA) and cancer prognosis (PREDICT), and laid foundations for stratified approaches to breast cancer screening (NEJM 2008). We have quantified the benefits and harms of screening for type 2 diabetes (BMJ 2009, Lancet 2012) for which the ADDITION trial team (lead: Griffin) won the BMJ’s paper of the year and the RCGP diabetes research paper of the year. We have demonstrated the predictive utility of routine GP data and how incorporating it in a stepwise approach to screening for cardiovascular disease would be more cost-effective than the current UK strategy (BMJ 2010). Another of our modelling studies has suggested that screening middle-aged men with symptoms of gastroesophageal reflux disease using a Cytosponge, a device developed in Cambridge, would reduce mortality in a cost-effective manner (Gastroenterology 2012). We have influenced international cardiovascular guidelines through evaluation of the incremental value of assessing established and emerging risk factors for risk assessment, including inflammation biomarkers (NEJM 2012), lipids and lipoproteins (JAMA 2012, Lancet 2010), adiposity measures (Lancet 2011), and carotid ultrasound (Lancet 2012).
Future plans – We are studying the early diagnosis of lung, pancreatic and colorectal cancers. We are developing stepwise approaches to combined risk assessment in primary care for cognate conditions (eg, diabetes, cardiovascular disease, renal disease), in contrast with fragmented single disease approaches. We are evaluating the impact on attitudes, behaviours, risk factors, and, ultimately, mortality of different approaches to disease risk prediction (eg, lifestyle-enriched risk scores that promote behaviour change vs highly accurate biomarker-enriched scores).
Control of chronic disease
The goal is to develop and evaluate approaches that can cost effectively improve the control and management of chronic conditions in primary care.
Achievements – Our trial concluded that intervention to promote early intensive management of type 2 diabetes was not harmful and could yield moderate benefits (Lancet 2011). Our trial in people with poorly controlled hypertension showed that a strategy of patient self-management of hypertension in combination with telemonitoring of blood pressure was superior in decreasing blood pressure than usual care (Lancet 2010). Our meta-analysis has suggested that ambulatory monitoring of blood pressure before the start of lifelong drug treatment leads to more appropriate targeting of treatment than relying on clinic or home measurement of blood pressure (BMJ 2011). Our comparative cohort study of older patients with atrial fibrillation showed that current risk stratification schemes have only limited ability to predict stroke, suggesting classification of all such patients over 75 years as “high risk” to avoid systematic under-treatment (BMJ 2011). Our cross sectional study of primary care records showed that older people were disproportionately low users of statin medications, suggesting implications for clearer guidelines for people aged over 75 years (BMJ 2012). Our long-term cohort study indicated that the estimated median survival for incident dementia is 4.5 years (BMJ 2008), suggesting implications for prognosis and planning for patients, carers, and health services.
Future plans – We are conducting trials to evaluate “polypill” interventions in cardiovascular disease prevention and ways to improve outcome following stroke. We have commenced the feasibility phase for a trial of metformin for the prevention of cardiovascular disease in people at high risk of type 2 diabetes, with the full trial of 12,000 people expected to start in 2015.
The goal is to develop methods of measuring quality of care and burden of disease, and evaluating interventions designed to improve healthcare.
Achievements – Our work has had substantial impact on measurement of quality in the NHS, in particular developing measures of patient experience in primary care and interpreting measures of patient experience in cancer. Our studies in primary care have yielded new insights about the impact of pay-for-performance programmes in the NHS (NEJM 2009) on exception reporting in primary care (NEJM 2008), and how financial incentives have diverted attention from aspects of care that were not incentivised (BMJ 2011). We have also shown the impact of hospital pay for performance schemes on reducing in-hospital mortality (NEJM 2012). Other work has informed NHS policy on areas including targets for hospital readmission rates, the impact of initiatives designed to better integrate care, and the contribution of leadership training to improving hospital quality. Our two-decade cohort study of dementia in England has informed health service provision by producing reliable prevalence estimates, adding to our understanding of neuropathology (NEJM 2009), and showing that a cohort effect exists in dementia prevalence, with later-born populations having a lower risk of prevalent dementia than those born earlier in the past century (Lancet 2013).
Future plans – We will expand work on quality measurement in the NHS, showing how valid data on patient experience can be used to improve care, working in primary care, in acute hospitals, and with patients who have cancer. Our work will demonstrate how care can be better integrated and we will continue with long-term studies to improve the care of people with dementia. To inform the blood service in England, we will complete a trial of 50,000 blood donors to determine the optimum safe interval between blood donations, as well as conduct further trials to assess how best to screen for iron deficiency and to prevent anaemia through personally-tailored dietary advice.