Should You Keep Drinking Coffee?

It is nice, at last, to find something that is it is really enjoyable and that is good for us. Recent News Blogs have carried articles on the harmful effects of red meat,[1] milk,[2] and alcohol.[3] But what about coffee? A recent article, based on over 500,000 people in ten European countries confirmed the already extensive literature showing that coffee is beneficial for health.[4] In fact, overall death rates were reduced by over 10%. There was a massive (over 50%) reduction in diseases of the digestive system, confirming the well-known beneficial effect of coffee on the liver. The trend was also favourable for heart disease and stroke. Many biochemical markers also moved in a favourable direction, including glycated haemoglobin, and C-reactive protein. The only bit of bad news pertained to ovarian cancer, where a 30% increased risk of death was noted. Reverse causality is always a possibility in non-experimental studies, even if, like this one, they are prospective. However, this is unlikely since the hazard ratios were unaltered if patients who died within eight years of recruitment were excluded.

— Richard Lilford, CLAHRC WM Director


  1. Lilford RJ. An Issue of BMJ with Multiple Studies on Diet. NIHR CLAHRC West Midlands News Blog. 4 August 2017.
  2. Lilford RJ. Two Provocative Papers on Diet and Health. NIHR CLAHRC West Midlands News Blog. 12 December 2014.
  3. Lilford RJ. Alcohol and its Effects. NIHR CLAHRC West Midlands News Blog. 18 August 2017.
  4. Gunter MJ, Murphy N, Cross AJ, et al. Coffee Drinking and Mortality in 10 European Countries. Ann Intern Med. 2017; 167: 236-47.

An Extremely Fascinating Debate in JAMA

You really should read this debate – Steven Goodman, a statistician for whom I have the utmost regard, wrote a brilliant paper in which he and colleagues show the importance of ‘design thinking’ in observational research.[1] The essence of their argument is that in designing and interpreting observational studies one should think about how the corresponding RCT would look. This way one can spot survivorship bias, which arises when the intervention group has been depleted of the most susceptible cases. This way of thinking encourages a comparison between new users of an intervention with new users of the comparator. Of course, it is not always possible to identify ‘new users’, but at least thinking in such a ‘design way’ can alert the reader to the danger of false inference.
One of the examples mentioned concerns hormone replacement therapy (HRT) where the largest RCT (Women’s Health Initiative trial) gave a very different result to the largest observational study (Nurses’ Health Study). The latter suggests a protective effect for HRT, while the former suggest the opposite. It looks as though this might not have been a very good example because, as Bhupathiraju and colleagues point out, there is a much simpler and more convincing explanation for the difference in the observed effects of HRT across the two studies.[2] The hormone replacement was given to much younger women in the observational study than in the trial. Subsequent meta-analysis of subgroups across all RCTs confirms that HRT is only protective in younger woman (who do not have established coronary artery disease). Thus, HRT is probably effective if started sufficiently early after the menopause.

This does not mean, of course, that Goodman and colleagues are wrong in principle; they may simply have selected a bad example. This is an extremely interesting exchange conducted politely between scholars and is interesting from both of the methodological and the substantive points of view.

— Richard Lilford, CLAHRC WM Director


  1. Goodman SN, Schneeweiss S, Baiocchi M. Using design thinking to differentiate useful from misleading evidence in observational research. JAMA. 2017; 317(7): 705-7.
  2. Bhupathiraju SN, Stampfer MJ, Manson JE. Posing Causal Questions When Analyzing Observational Data. JAMA. 2017; 318(2): 201.

A Secondary Sanitary Revolution? What About the First One?

Water and sanitation is being taken increasingly seriously in Low- and Middle-Income Countries (LMICs). This is a good thing because, despite improved treatment of diarrhoea and vaccination against rotavirus,[1] gastrointestinal diseases are one of the two biggest causes of death in children under the age of five.[2] Yet recent evaluations of water and sanitation interventions show patchy results [3] and are overall disappointing.[4] [5] Very few studies have been done in urban areas, but infant death rates in slums are unconscionably high.[6]

Why are water and sanitation interventions so disappointing in the LMICs of today when the Sanitary Revolution around the turn of the 19th century was so successful? Well it turns out that the Sanitary Revolution was a bit of a myth – Thomas McKeown, Professor of Social Medicine at the University of Birmingham, caused quite a stir by pointing this out in the 1970s.[7] The ‘historical epidemiology’ of this time period is intensely interesting. While sequential chlorination of water in North American cities in the early years of the 1900s was associated with corresponding dramatic drops in the incidence of typhoid fever,[8] establishment of water and sanitation in the Netherlands [9] and Estonia [10] produced no benefit whatsoever on gastrointestinal deaths. Only one-third of the reduction in gastrointestinal-related deaths observed in around the turn of the 18th century Germany could be attributed to water and sanitation improvements.[9]
So why do water and sanitation interventions produce such variable, and often disappointing, benefits? In rural India this can often be attributed to low use of facilities, but little or no health benefit has been observed, even when uptake has been high. A number of (non-exclusive) theories can be adduced:

1) The inadequate dose theory. This holds that the type of intervention deployed in LMICs has generally been inadequate. For example, pit latrines (classed as ‘improved sewerage’ by the UN) do not clean up the environment adequately.[11] Similarly, water pipes may be installed, but the water may be contaminated en route.[12] St Petersburg is a notorious example.

2) The tipping point theory. This theory is an elaboration of the above inadequate dose theory and postulates a non-linear relationship between the intensity, type of water and sanitation (facility), and coverage of interventions and health, with increasing and then decreasing returns to scale as shown in Figure 1. By this theory, many interventions (such as pit latrines) simply fail to reach the ‘tipping point’, especially in densely inhabited city areas.

085 DCB A Secondary Sanitary Revolution Fig 1

3) The deep contamination theory (Figure 2). By this theory contamination follows many routes and becomes embedded in local communities, with transmission routes that are frequently replenished, so that garbage dumps, flies, nappies, soil and the human gut all act as repositories of infection. Food may be contaminated along its supply chain, as well as in preparation. Floods sweep sewage out of drains and back into communities. Cleaning up such an environment moves the tipping point (shown in Figure 1) to the right (meaning it is harder to reach) and may also take time to effect – a point to which we return.


4) The multiple agent hypothesis. By this theory some germs are more easily eradicated than others. Typhoid is waterborne, but, unlike cholera, it cannot replicate in water. Ensuring uncontaminated water may be enough to eradicate this particular problem. However, hookworms are at the other end of the spectrum, since they can be carried asymptomatically and linger in soil. There is even some evidence that organisms gain virulence as they are passed rapidly from host to host.[13] So this theory predicts that some types of infection might decrease more rapidly than others in response to an intervention. Moreover, some real gains, with respect to some type of serious infection, might be obscured by little or no change in more common, but less serious infections.

5) The multiple causes theory. This theory relies on evidence that malnutrition and gastrointestinal infections are self-reinforcing. Certainly malnutrition is associated with an altered microbiome, which, in turn, reduces absorption of food, creating a vicious cycle.[14] The microbiome affects the immune system, which, in turn, affects resistance to infection.

6) The ‘double-handed’ hygiene hypothesis. Hygiene can compensate for dirty water and a contaminated environment, and some of the most consistently effective interventions in LMICs have been based on improved hygiene and near use decontamination.[4] [15] On the other hand, hygiene does not seem important in places where exposure to harmful pathogens is low and, in such circumstances, hygiene may be too fastidious, leading to allergic illnesses.

7) The insensitive outcome hypothesis. Measuring the health benefit of sanitation is not unproblematic – the standard question on diarrhoea enquires about loose stools over a three-day period, and the measurement error appears to be large.[16] An account of blood in stools, signifying dysentery (Shigella and amoeba) is more specific, but is much rarer, leading to imprecision (lack of statistical power). Anthropological measurements reflect long-term conditions, and many factors, including gastrointestinal infections and malnutrition (see above), and also age of weaning, birth weight, and mother’s weight (inter-generational effects). We are working on designing a better (equally sensitive, but more specific and less reactive) method to measure gastrointestinal health.[17]

There may well be an element of truth in all these hypotheses. If a fully functioning water and sewerage system was installed, lanes paved and drained, and garbage eliminated, then there would probably be an impressive and rapid improvement in gastrointestinal health, especially if malnutrition was also tackled. But the same water and sewerage system would probably have moderate and delayed benefits if not accompanied by the other measures mentioned. What nutrition and vaccination would achieve without water and sanitation is unknown, but as they are less expensive, the experiment should be tried in places where water and sanitation improvements are some time away. In-depth study of transmission routes will help explicate some of the other theories postulated and careful comparative studies will help identify the tipping point for the most cost-effective solutions. What is for sure is that science has a role to play in unravelling the process by which we may achieve a Second Sanitary Revolution.

— Richard Lilford, CLAHRC WM Director


  1. GBD Diarrhoeal Diseases Collaborators. Estimates of global, regional, and national morbidity, mortality, and aetiologies of diarrhoeal diseases: a systematic analysis for the Global Burden of Disease Study 2015. Lancet Infect Dis. 2017; 17: 909-48.
  2. Global Burden of Disease Pediatrics Collaboration. Global and National Burden of Diseases and Injuries Among Children and Adolescents Between 1990 and 2013. JAMA Pediatr. 2016; 170(3): 267-87.
  3. Lilford RJ, Oyebode O, Satterthwaite D, et al. Improving the health and welfare of people who live in slums. Lancet. 2017; 389: 559-70.
  4. Wolf J, Prüss-Ustün A, Cumming O, et al. Assessing the impact of drinking water and sanitation on diarrhoeal disease in low- and middle-income settings: systematic review and meta-regression. Trop Med Int Health. 2014; 19(8): 928-42.
  5. Fuller JA, Westphal JA, Kenney B, Eisenberg JNS. The joint effects of water and sanitation on diarrhoeal disease: a multicountry analysis of the Demographic and Health Surveys. Trop Med Int Health. 2015; 20(3): 284-92.
  6. Feikin DR, Olack B, Bigogo GM, et al. The burden of common infectious disease syndromes at the clinic and household level from population-based surveillance in rural and urban Kenya. PLoS One. 2011; 6: e16085.
  7. McKeown T, Record RG, Turner RD. An interpretation of the decline of mortality in England and Wales during the twentieth century. Popul Stud. 1975. 29(3): 391-422.
  8. Cutler D & Miller G. The Role of Public Health Improvements in Health Advances: The 20th Century United States. NBER Working Paper 10511. Cambridge, MA: National Bureau of Economic Research; 2004.
  9. Van Poppel F & van der Heijden C. The effects of water supply on infant and childhood mortality: a review of historical evidence. Health Trans Rev. 1997; 7(2): 113-48.
  10. Jaadla H & Puur A. The impact of water supply and sanitation on infant mortality: Individual-level evidence from Tartu, Estonia, 1897-1900. Popul Stud. 2016; 70(2): 163-79.
  11. Nakagiri A, Niwagaba CB, Nyenje PM, Kulabako RN, Tumuhairwe JB, Kansiime F. Are pit latrines in urban areas of sub-Saharan Africa performing? A review of usage, filling, insects and odour nuisances. BMC Public Health. 2016; 16: 120.
  12. Eschol J, Mahapatra P, Keshapagu S. Is fecal contamination of drinking water after collection associated with household water handling and hygiene practices? A study of urban slum households in Hyderabad, India. J Water Health. 2009; 7(1): 145-54.
  13. Ewald PW. Waterborne transmission and the evolution of virulence among gastrointestinal bacteria. Epidemiol Infect. 1991; 106: 83-119.
  14. Rook G, Bäckhed F, Levin BR, McFall-Ngai MJ, McLean AR. Evolution, human-microbe interactions, and life history plasticity. Lancet. 2017; 390: 521-30.
  15. Freeman MC, Garn JV, Sclar GD, Boisson S. The impact of sanitation on infectious disease and nutritional status: A systematic review and meta-analysis. Int J Hyg Environ Health. 2017; 220(6): 928-49.
  16. Schmidt WP, Arnold BF, Boisson S, et al. Epidemiological methods in diarrhoea studies – an update. Int J Epidemiol. 2011; 40(6): 1678-92.
  17. Lilford RJ. Protocol to Test Hypothesis That Streptococcal Infections and Their Sequelae Have Risen in Incidence Over the Last 14 Years in England. NIHR CLAHRC West Midlands News Blog. 13 January 2017.

Measuring the Quality of Health Care in Low-Income Settings

Measuring the quality of health care in High-Income Countries (HIC) is deceptively difficult, as shown by work carried out by many research groups, including CLAHRC WM.[1-5] However, a large amount of information is collected routinely by health care facilities in HICs. This data includes outcome data, such as Standardised Mortality Rates (SMRs), death rates from ’causes amenable to health care’, readmission rates, morbidity rates (such as pressure damage), and patient satisfaction, along with process data, such as waiting times, prescribing errors, and antibiotic use. There is controversy over many of these endpoints, and some are much better barometers of safety than others. While incident reporting systems provide a very poor basis for epidemiological studies (that is not their purpose), case-note review provides arguably the best and most widely used method for formal study of care quality – at least in hospitals.[3] [6] [7] Measuring safety in primary care is inhibited by the less comprehensive case-notes found in primary care settings as compared to hospital case-notes. Nevertheless, increasing amounts of process information is now available from general practices, particularly in countries (such as the UK) that collect this information routinely in electronic systems. It is possible, for example, to measure rates of statin prescriptions for people with high cardiovascular risk, and anticoagulants for people with ventricular fibrillation, as our CLAHRC has shown.[8] [9] HICs also conduct frequent audits of specific aspects of care – essentially by asking clinicians to fill in detailed pro formas for patients in various categories. For instance, National Audits in the UK have been carried out into all patients experiencing a myocardial infarction.[10] Direct observation of care has been used most often to understand barriers and facilitators to good practice, rather than to measure quality / safety in a quantitative way. However, routine data collection systems provide a measure of patient satisfaction with care – in the UK people who were admitted to hospital are surveyed on a regular basis [11] and general practices are required to arrange for anonymous patient feedback every year.[12] Mystery shoppers (simulated patients) have also been used from time to time, albeit not as a comparative epidemiological tool.[13]

This picture is very different in Low- and Middle-Income Countries (LMIC) and, again, it is yet more difficult to assess quality of out of hospital care than of hospital care.[14] Even in hospitals routine mortality data may not be available, let alone process data. An exception is the network of paediatric centres established in Kenya by Prof Michael English.[15] Occasionally large scale bespoke studies are carried out in LMICs – for example, a recent study in which CLAHRC WM participated, measured 30 day post-operative mortality rates in over 60 hospitals across low-, middle- and high-income countries.[16]

The quality and outcomes of care in community settings in LMICs is a woefully understudied area. We are attempting to correct this ‘dearth’ of information in a study in nine slums spread across four African and Asian countries. One of the largest obstacles to such a study is the very fragmented nature of health care provision in community settings in LMICs – a finding confirmed by a recent Lancet commission.[17] There are no routine data collection systems, and even deaths are not registered routinely. Where to start?

In this blog post I lay out a framework for measurement of quality from largely isolated providers, many of whom are unregulated, in a system where there is no routine system of data and no archive of case-notes. In such a constrained situation I can think of three (non-exclusive) types of study:

  1. Direct observation of the facilities where care is provided without actually observing care or its effects. Such observation is limited to some of the basic building blocks of a health care system – what services are present (e.g. number of pharmacies per 1,000 population) and availability (how often the pharmacy is open; how often a doctor / nurse / medical officer is available for consultation in a clinic). Such a ‘mapping’ exercise does not capture all care provided – e.g. it will miss hospital care and municipal / hospital-based outreach care, such as vaccination provided by Community Health Workers. It will also miss any IT based care using apps or online consultations.
  2. Direct observation of the care process by external observers. Researchers can observe care from close up, for example during consultations. Such observations can cover the humanity of care (which could be scored) and/or technical quality (which again could be scored against explicit standards and/or in a holistic (implicit) basis).[6] [7] An explicit standard would have to be based mainly on ‘if-then’ rules – e.g. if a patient complained of weight loss, excessive thirst, or recurrent boils, did the clinicians test their urine for sugar; if the patient complained of persistent productive cough and night sweats was a test for TB arranged? Implicit standards suffer from low reliability (high inter-observer variation).[18] Moreover, community providers in LMICs are arguably likely to be resistant to what they might perceive as an intrusive or even threatening form of observation. Those who permitted such scrutiny are unlikely to constitute a random sample. More vicarious observations – say of the length of consultations – would have some value, but might still be seen as intrusive. Provided some providers would permit direct observation, their results may represent an ‘upper bound’ on performance.
  3. Quality as assessed through the eyes of the patient / members of the public. Given the limitations of independent observation, the lack of anamnestic records of clinical encounters in the form of case-notes, absence of routine data, and likely limitations on access by independent direct observers, most information may need to be collected from patients themselves, or as we discuss, people masquerading as patients (simulated patients / mystery shoppers). The following types of data collection methods can be considered:
    1. Questions directed at members of the public regarding preventive services. So, households could be asked about vaccinations, surveillance (say for malnutrition), and their knowledge of screening services offered on a routine basis. This is likely to provide a fairly accurate measure of the quality of preventive services (provided the sampling strategy was carefully designed to yield a representative sample). This method could also provide information on advice and care provided through IT resources. This is a situation where some anamnestic data collection would be possible (with the permission of the respondent) since it would be possible to scroll back through the electronic ‘record’.
    2. Opinion surveys / debriefing following consultations. This method offers a viable alternative to observation of consultations and would be less expensive (though still not inexpensive). Information on the kindness / humanity of services could be easily obtained and quantified, along with ease of access to ambulatory and emergency care.[19] Measuring clinical quality would again rely on observations against a gold standard,[20] but given the large number of possible clinical scenarios standardising quality assessment would be tricky. However, a coarse-grained assessment would be possible and, given the low quality levels reported anecdotally, failure to achieve a high degree of standardisation might not vitiate collection of important information. Such a method might provide insights into the relative merits and demerits of traditional vs. modern health care, private vs. public, etc., provided that these differences were large.
    3. Simulated patients offering standardised clinical scenarios. This is arguably the optimal method of technical quality assessment in settings where case-notes are perfunctory or not available. Again, consultations could be scored for humanity of care and clinical/ technical competence, and again explicit and/or implicit standards could be used. However, we do not believe it would be ethical to use this method without obtaining assent from providers. There are some examples of successful use of the methods in LMICs.[21] [22] However, if my premise is accepted that providers must assent to use of simulated patients, then it is necessary to first establish trust between providers and academic teams, and this takes time. Again, there is a high probability that only the better providers will provide assent, in which case observations would likely represent ‘upper bounds’ on quality.

In conclusion, I think that the basic tools of quality assessment, in the current situation where direct observation and/or simulated patients are not acceptable, is a combination of:

  1. Direct observation of facilities that exist, along with ease of access to them, and
  2. Debriefing of people who have recently used the health facilities, or who might have received preventive services that are not based in these facilities.

We do not think that the above mentioned shortcomings of these methods is a reason to eschew assessment of service quality in community settings (such as slums) in LMICs – after all, one of the most powerful levers to improvement is quantitative evidence of current care quality.[23] [24] The perfect should not be the enemy of the good. Moreover, if the anecdotes I have heard regarding care quality (providers who hand out only three types of pill – red, yellow and blue; doctors and nurses who do not turn up for work; prescription of antibiotics for clearly non-infectious conditions) are even partly true, then these methods would be more than sufficient to document standards and compare them across types of provider and different settings.

— Richard Lilford, CLAHRC WM Director


  1. Brown C, Hofer T, Johal A, Thomson R, Nicholl J, Franklin BD, Lilford RJ. An epistemology of patient safety research: a framework for study design and interpretation. Part 1. Conceptualising and developing interventions. Qual Saf Health Care. 2008; 17(3): 158-62.
  2. Brown C, Hofer T, Johal A, Thomson R, Nicholl J, Franklin BD, Lilford RJ. An epistemology of patient safety research: a framework for study design and interpretation. Part 2. Study design. Qual Saf Health Care. 2008; 17(3): 163-9.
  3. Brown C, Hofer T, Johal A, Thomson R, Nicholl J, Franklin BD, Lilford RJ. An epistemology of patient safety research: a framework for study design and interpretation. Part 3. End points and measurement. Qual Saf Health Care. 2008; 17(3): 170-7.
  4. Brown C, Hofer T, Johal A, Thomson R, Nicholl J, Franklin BD, Lilford RJ. An epistemology of patient safety research: a framework for study design and interpretation. Part 4. One size does not fit all. Qual Saf Health Care. 2008; 17(3): 178-81.
  5. Brown C, Lilford R. Evaluating service delivery interventions to enhance patient safety. BMJ. 2008; 337: a2764.
  6. Benning A, Ghaleb M, Suokas A, Dixon-Woods M, Dawson J, Barber N, et al. Large scale organisational intervention to improve patient safety in four UK hospitals: mixed method evaluation. BMJ. 2011; 342: d195.
  7. Benning A, Dixon-Woods M, Nwulu U, Ghaleb M, Dawson J, Barber N, et al. Multiple component patient safety intervention in English hospitals: controlled evaluation of second phase. BMJ. 2011; 342: d199.
  8. Finnikin S, Ryan R, Marshall T. Cohort study investigating the relationship between cholesterol, cardiovascular risk score and the prescribing of statins in UK primary care: study protocol. BMJ Open. 2016; 6(11): e013120.
  9. Adderley N, Ryan R, Marshall T. The role of contraindications in prescribing anticoagulants to patients with atrial fibrillation: a cross-sectional analysis of primary care data in the UK. Br J Gen Pract. 2017. [ePub].
  10. Herrett E, Smeeth L, Walker L, Weston C, on behalf of the MINAP Academic Group. The Myocardial Ischaemia National Audit Project (MINAP). Heart. 2010; 96: 1264-7.
  11. Care Quality Commission. Adult inpatient survey 2016. Newcastle-upon-Tyne, UK: Care Quality Commission, 2017.
  12. Ipsos MORI. GP Patient Survey. National Report. July 2017 Publication. London: NHS England, 2017.
  13. Grant C, Nicholas R, Moore L, Sailsbury C. An observational study comparing quality of care in walk-in centres with general practice and NHS Direct using standardised patients. BMJ. 2002; 324: 1556.
  14. Nolte E & McKee M. Measuring and evaluating performance. In: Smith RD & Hanson K (eds). Health Systems in Low- and Middle-Income Countries: An economic and policy perspective. Oxford: Oxford University Press; 2011.
  15. Tuti T, Bitok M, Malla L, Paton C, Muinga N, Gathara D, et al. Improving documentation of clinical care within a clinical information network: an essential initial step in efforts to understand and improve care in Kenyan hospitals. BMJ Global Health. 2016; 1(1): e000028.
  16. Global Surg Collaborative. Mortality of emergency abdominal surgery in high-, middle- and low-income countries. Br J Surg. 2016; 103(8): 971-88.
  17. McPake B, Hanson K. Managing the public-private mix to achieve universal health coverage. Lancet. 2016; 388: 622-30.
  18. Lilford R, Edwards A, Girling A, Hofer T, Di Tanna GL, Petty J, Nicholl J. Inter-rater reliability of case-note audit: a systematic review. J Health Serv Res Policy. 2007; 12(3): 173-80.
  19. Schoen C, Osborn R, Huynh PT, Doty M, Davis K, Zapert K, Peugh J. Primary Care and Health System Performance: Adults’ Experiences in Five Countries. Health Aff. 2004.
  20. Kruk ME & Freedman LP. Assessing health system performance in developing countries: A review of the literature. Health Policy. 2008; 85: 263-76.
  21. Smith F. Private local pharmacies in low- and middle-income countries: a review of interventions to enhance their role in public health. Trop Med Int Health. 2009; 14(3): 362-72.
  22. Satyanarayana S, Kwan A, Daniels B, Subbaramn R, McDowell A, Bergkvist S, et al. Use of standardised patients to assess antibiotic dispensing for tuberculosis by pharmacies in urban India: a cross-sectional study. Lancet Infect Dis. 2016; 16(11): 1261-8.
  23. Kudzma E C. Florence Nightingale and healthcare reform. Nurs Sci Q. 2006; 19(1): 61-4.
  24. Donabedian A. The end results of health care: Ernest Codman’s contribution to quality assessment and beyond. Milbank Q. 1989; 67(2): 233-56.

Diet and Socioeconomic Status

People looking to lose weight and/or get healthy try a wide variety of diets, from fad diets with highly specific restrictions on what can be eaten, to general healthy eating plans. One such nutritional recommendation is the Mediterranean diet, based on the “food patterns typical of Crete… Greece and southern Italy…”,[1] and entails consumption of high amounts of plant foods (fruit, vegetables, cereals, legumes, etc.) and olive oil, moderate amounts of dairy, fish and wine, and low amounts of poultry and red meat. A number of observational studies have shown associations between such a diet and lower incidences of cardiovascular disease (CVD) and associated mortality, cancer, neuro-degenerative disorders, and overall mortality. However, there is uncertainty whether such benefits differ across different socioeconomic groups.

Bonaccio et al. carried out a prospective analysis of nearly 19,000 Italians to see the effect of the Mediterranean diet on CVD.[2] While there was an overall reduction in CVD risk associated with adherence to the diet (HR=0.85, 95% CI 0.73-0.99), this was not seen across all socioeconomic groups – only in those who were educated to a postgraduate or higher level (HR=0.43, 0.25-0.72) and in those with a high (>€40,000) household income (HR=0.39, 0.23-0.66). Those with less education (HR=0.94, 0.78-1.14) and lower income (HR=1.01, 0.79-1.29) had no significant association. Why such a difference? Subgroup analysis of people with similar adherence to the diet showed that there were a number of differences in the diet of those with high compared to low education, and those with high compared to low income. These included consumption of organic vegetables (which would have higher antioxidants and lower levels of pesticides), monounsaturated fatty acids (found in avocado, nuts, olives, etc.), micronutrients, and whole-grain bread, as well as greater dietary diversity.

So perhaps it is more important to make sure the food you are eating is of high quality and varied, than just simple healthy eating. Of course, access to high quality food of high nutritional value is not easy for poor people.

— Peter Chilton, Research Fellow


  1. Willett WC, Sacks F, Trichopoulou A, Drescher G, Ferro-Luzzi A, Helsing E, Trichopoulos D. Mediterranean diet pyramid: a cultural model for healthy eating. Am J Clin Nutr. 1995; 61(6): 1402S–6S.
  2. Bonaccio M, Di Castelnuovo A, Pounis G, et al. High adherence to the Mediterranean diet is associated with cardiovascular protection in higher but not in lower socioeconomic groups: prospective findings from the Moli-sani study. Int J Epidemiol. 2017.

Using the Internet for More Than Just Cat Pictures

The Internet can be a highly useful tool – communicating with old or distant friends, finding out the latest news, purchasing the latest best-seller, looking at photos of cats, etc. People also go online when they, or someone they know, is ill, searching for information or posting on social media. Your Internet search provider tracks all of this, and this data can be used by researchers to track outbreaks and the spread of infectious diseases. A recent paper by Yang and colleagues [1] demonstrated such a feat with regards to dengue fever.

Dengue is quickly becoming one of the most endemic mosquito-borne disease worldwide, infecting around 390 million people each year in 128 countries,[2] and placing the local health services under immense pressure. The Aedes mosquito that transmits dengue thrives in slums / shanty towns.[3] One of the ways to reduce infection rates is to improve early case detection – identifying outbreaks early means that preventive measures, such as mosquito population control, providing mosquito screens or nets, etc., can be undertaken. However, there is no current surveillance system for dengue that is comprehensive, effective and reliable – governments tend to use reports from hospitals that are often delayed and/or inaccurate.

Yang, et al. combined dengue-related Internet searches with historical incidence data to track dengue activity in five areas, Mexico, Brazil, Thailand, Singapore and Taiwan. They were able to successfully estimate dengue activity one month prior to the publication of official local health records, with their method outperforming benchmark models across accuracy metrics in all areas, except Taiwan. The authors note that Taiwan had little previous dengue prevalence on which to base predictions, suggesting the methodology works best in areas where dengue is already endemic.

— Peter Chilton, Research Fellow


  1. Yang S, Kou SC, Lu F, Brownstein JS, Brooke N, Santillana M. Advances in using Internet searches to track dengue. PLoS Comput Biol. 2017; 13(7): e1005607.
  2. World Health Organization. Dengue and severe dengue. 2016.
  3. Ezeh A, Oyebode O, Satterthwaite D, et al. The history, geography, and sociology of slums and the health problems of people who live in slums. Lancet. 2017; 389: 547-58.

Alcohol and its Effects

News blog readers may be familiar with the famous ‘J curve’ relating alcohol consumption to health outcomes, including brain health.[1] The J curve shows a negative correlation between alcohol consumption and cognitive functioning at a low level of alcohol consumption (< 7 units/week), turning to a positive association in quantities exceeding about 28 units/week. One large glass of wine per day should be safe according to this finding. However, the data from which these findings are derived is cross-sectional. The BMJ has recently published a longitudinal study of alcohol and its effect on both cognition and brain structure (as measured by functional MRI).[2] The news is bad I am afraid. In the words of the editor, Fiona Godlee, ‘better’ research flattens the J curve.[3] The study seems to show a linear increase in risk with increasing intake of alcohol. The result was statistically significant for people drinking more than about two small glasses of wine per day. Why was a harmful effect at low dose detected in this longitudinal study but not the cross-sectional studies? So here is the thing – people with higher cognitive functioning tend to have higher alcohol consumption at baseline. In fact, the ‘cleverer’ the person, the more they tend to drink. The result is a difference in the findings of cross-sectional and longitudinal studies. While cross-sectional studies show no difference in cognition with moderate alcohol intake, the longitudinal studies show that cognition and brain structure decline at relatively low levels of alcohol consumption. To put this another way, moderate alcohol intake abolishes the cognitive advantage that moderate alcohol consumers have at baseline. Interestingly, not all parts of the brain are equally affected on MRI. Likewise the effect on cognition is not global; it affects lexical more than semantic fluency, for example. This is an extremely well-written, detailed and interesting study. The cohort of people who participated in the study were civil servants followed up for 30 years. The results are of immense public health importance. Human happiness, wealth and prosperity all relate to brain function. A person’s intellectual endowment is a precious gift and should not be lightly squandered.  I will take these findings too heart, both in my personal life and as a public health practitioner. It is really a question of long-term loss vs. short-term gain – alcohol is a pleasant social lubricant, much beloved of myself, and a small glass of wine has even been shown to improve creative problem-solving![4]

— Richard Lilford, CLAHRC WM Director


  1. Di Castelnuovo A, Costanzo  S, Bagnardi  V, Donati  MB, Iacoviello  L, de Gaetano    Alcohol dosing and total mortality in men and womenArch Intern Med. 2006; 166(22): 2437-45.
  2. Topiwala A, Allan C, Valkanova V, et al. Moderate alcohol consumption as risk factor for adverse brain outcomes and cognitive decline: longitudinal cohort study. BMJ. 2017; 357:j2353.
  3. Godlee F. Better research flattens the J shaped curve. 2017; 357: j2755.
  4. Benedek M, Panzierer L, Jauk E, Neubauer AC. Creativity on tap? Effects of alcohol intoxication on creative cognition. Consciousness Cognition. 2017. [ePub].

“We seek him here, we seek him there, Those Frenchies seek him everywhere.”

The notorious weekend mortality effect is every bit as elusive as the Scarlet Pimpernel. Recent studies have delved deeper into the possibility that the weekend effect is an artefact of admission of sicker patients at the weekend than on week days.[1] First, it has been shown that the mortality of all who present to the emergency department (i.e. admitted plus sent home) is the same over the weekend as over the rest of the week.[2] Second, patients who arrive by ambulance are generally much sicker than patients arriving by other means and the proportion who arrive by ambulance is higher over the weekend than over weekdays.[3] When controlling for method of arrival, most of the weekend effect disappears. Most, but not all. This paper provides further evidence that most estimates of the weekend effect are at least overestimates. Through Professor Julian Bion’s HiSLAC Study [4] we are evaluating the effect of weekend admission, not just on mortality, but also on the quality of care and the overall adverse event rate. We will use a Bayesian network to synthesise information across the causal chain and come up with a refined estimate of the effect of weekend admission, not only on mortality, but also on other adverse events.

— Richard Lilford, CLAHRC WM Director


  1. Bray BD, Steventon A. What have we learnt after 15 years of research into the ‘weekend effect’? BMJ Qual Saf. 2017; 26: 607-10.
  2. Aldridge C, Bion J, Boyal A, et al. Weekend specialist intensity and admission mortality in acute hospital trusts in England: a cross-sectional study. Lancet. 2016. 388: 178-86.
  3. Anselmi L, Meacock R, Kristensen SR, Doran T, Sutton M. Arrival by ambulance explains variation in mortality by time of admission: retrospective study of admissions to hospital following emergency department attendance in England. BMJ Qual Saf. 2017; 26: 613-21.
  4. Chen Y, Boyal A, Sutton E, et al. The magnitude and mechanisms of the weekend effect in hospital admissions: A protocol for a mixed methods review incorporating a systematic review and framework synthesis. Syst Rev. 2016; 5: 84.

Patient Involvement in Patient Safety: Null Result from a High Quality Study

Most patient safety evaluations are simple before and after / time series improvement studies. So it is always refreshing to find a study with contemporaneous controls. Lawton and her colleagues report a nice cluster randomized trial covering 33 hospital wards in five hospitals.[1] They evaluate a well-known patient safety intervention based on the idea of giving patients a more active role in monitoring safety on their ward.

The trial produced a null result, but some of the measures of safety were in the right direction and there was a correlation between the enthusiasm/fidelity with which the intervention was implemented and measures of safety.

Safety is hard to measure (as the authors state), and improvement often builds on a number of small incremental changes. So, it would be very nice to see this intervention replicated, possibly with measures to generate greater commitment from ward staff.
Here is the problem with patient safety research; on the one hand the subject of patient safety is full of hubristic claims made on the basis of insufficient (weak) evidence. On the other hand, high quality studies, such as the one reported here, often fail to find an effect. In many cases, as in the study reported here, there are reasons to suspect a type 2 error (false negative result). Beware also the rising tide – the phenomenon that arises where a trial occurs in the context of a strong secular trend – this trend ‘swallows up’ the headroom for a marginal intervention effect.[2] What is to be done? First, do not declare defeat too early. Second, be prepared to either carry out larger studies or replication studies that can be combined in a meta-analysis. Third, make multiple measurements across a causal chain [3] and synthesise this disparate data using Bayesian networks.[4] Fourth, further to the Bayesian approach, do not dichotomise results on the standard frequentist statistical convention into null and positive. It is stupid to classify a p-value of 0.06 as null if other evidence supports an effect, or to classify a p-value of 0.04 as positive if other data point the opposite way. Knowledge of complex areas, such as service interventions to improve safety, should take account of patterns in the data and information external to the index study. Bayesian networks provide a framework for such an analysis.[4] [5]

— Richard Lilford, CLAHRC WM Director


  1. Lawton R, O’Hara JK, Sheard L, et al. Can patient involvement improve patient safety? A cluster randomised control trial of the Patient Reporting and Action for a Safe Environment (PRASE) intervention. BMJ Qual Saf. 2017; 26: 622-31.
  2. Chen YF, Hemming K, Stevens AJ, Lilford RJ. Secular trends and evaluation of complex interventions: the rising tide phenomenon. BMJ Qual Saf. 2016; 25: 303-10.
  3. Lilford RJ, Chilton PJ, Hemming K, Girling AJ, Taylor CA, Barach P. Evaluating policy and service interventions: framework to guide selection and interpretation of study end points. BMJ. 2010; 341: c4413.
  4. Watson SI & Lilford RJ. Essay 1: Integrating multiple sources of evidence: a Bayesian perspective. In: Challenges, solutions and future directions in the evaluation of service innovations in health care and public health. Southampton (UK): NIHR Journals Library, 2016.
  5. Lilford RJ, Girling AJ, Sheikh, et al. Protocol for evaluation of the cost-effectiveness of ePrescribing systems and candidate prototype for other related health information technologies. BMC Health Serv Res. 2014; 14: 314.

Patient and Public Involvement in Data Collection

Further to last fortnight’s News blog article [1] I have found a further study in which patients participated in data collection.[2] This paper, by and large, corroborates the procedural requirements for public and patient involvement in data collection that I had specified. For example, it was necessary for lay observers to undergo DBS checks; the ethics approval form had to include lay observers; and training had to be arranged for the lay observers. Recruitment of lay observers proved more difficult than anticipated. The lay observers had a positive experience and brought a different perspective to the research according to feedback. The extent to which observer perspective is a good thing is, however, contestable. Generally I think the role of the observer is to collect data for analysis, and not colour it with a ‘perspective’. The professional researchers on the project felt that having lay researchers involved increased their workloads. The thorny issues of payment and selection do not seem to have been fully discussed in this paper. Also not discussed was the idea that, in qualitative research, respondents may be less inhibited to disclose information to a lay observer. Let the debate continue!

— Richard Lilford, CLAHRC WM Director


  1. Lilford RJ. Patient and Public Involvement: Direct Involvement of Patient Representatives in Data Collection. NIHR CLAHRC West Midlands News Blog. 4 August 2017.
  2. Garfield S, Jheeta S, Jacklin A, Bischler A, Norton C, Franklin BD. Patient and public involvement in data collection for health services research: a descriptive study. Res Involve Engage. 2015; 1: 8.