Tag Archives: Director’s Choice – From the Journals

More on the Hygiene Hypothesis and Exposure to Coliform Organisms from the Birth Canal

The putative advantages of a deep draught of coliform organisms during a baby’s journey into the world has been discussed in a previous News Blog, with respect to prevention of allergy.[1] It now seems that it is not just allergy, but also cancer – more specifically the acute leukaemia of childhood – that is influenced by the process of birth.[2] And again, bypassing the birth canal by means of Caesarean section increases risk. The mechanism seems to conform with the three hits hypothesis, described in a past News Blog.[3] Here the hits might be:

  1. Genetic predisposition.
  2. Failure to ‘benefit’ from exposure to coliforms during birth.
  3. Subsequent severe infection.

Regarding the third ‘hit’ above, it is known that acute lymphoblastic leukaemia of children occurs in semi-epidemic fashion, suggesting that an acute infection is the trigger.

Some decades ago I carried out a decision-analysis that argued that when the risk of intra-partum C-section exceeded a threshold of around 35%, then a planned C-section was the best option for mother and baby.[4] For the mother because intra-partum C-section is more risky than planned C-section; and for the baby because situations where intra-partum C-section is common usually imply that the baby is also at increased risk – for example if the baby is coming by the breach. However, I can now see that my decision-analysis was incomplete – maybe I should have factored in the ‘unknown unknowns’.

— Richard Lilford, CLAHRC WM Director


  1. Lilford RJ. Exposure of the Baby to a Rich Mixture of Coliform Organisms from the Birth Canal. NIHR CLAHRC West Midlands News Blog. 22 April 2016.
  2. Greaves M. A causal mechanisms for childhood acute lymphoblastic leukaemia. Nat Rev Cancer. 2018.
  3. Lilford RJ. Three Hits Hypothesis. NIHR CLAHRC West Midlands News Blog. 7 April 2017.
  4. Lilford RJ, van Coeverden de Groot HA, Moore PJ, Bingham P. The relative risks of caesarean section (intrapartum and elective) and vaginal delivery: a detailed analysis to exclude the effects of medical disorders and other acute pre-existing physiological disturbances. Br J Obstet Gynaecol. 1990; 97(10): 883-92.

Reducing Risk of Postpartum Haemorrhage in LMICs

­Worldwide around 127,000 women die each year due to postpartum haemorrhage (the loss of 500ml of blood within the first 24 hours after giving birth), making it the most common cause of maternal death.[1] It is especially prevalent in low- or middle-income countries (LMICs). The standard treatment is administration of oxytocin – however, the drug needs to be kept at 20-25°C, which can be difficult in some countries where refrigeration isn’t reliable in either the hospital or whilst being transported. Thus, there is a need for an alternative option. Widmer and colleagues conducted a double-blind RCT of nearly 30,000 women across ten countries comparing postpartum administration of oxytocin with carbetocin, a heat-stable oxytocin analogue that can be stored at room temperature.[2] Both groups of women showed similar frequencies of blood loss – 14.4% of those given oxytocin lost >500ml of blood, compared to 14.5% given carbetocin (relative risk 1.01, 95% CI 0.95-1.06); while, respectively, 1.45% and 1.51% lost >1000ml of blood (relative risk 1.04, 95% CI 0.87-1.25). There were also no significant differences in necessary interventions or adverse events.

Carbetocin could even perform better than oxytocin in LMICs as the oxytocin was stored in optimum conditions, which may not accurately reflect real-life settings. It will be interesting to track the implementation to uptake of this new finding – any takers?

— Peter Chilton, Research Fellow


  1. World Health Organization. Making Pregnancy Safer. Issue 4. Geneva, CHE: World Health Organization; 2007.
  2. Widmer M, Piaggio G, Nguyen TMH, et al. Heat-Stable Carbetocin versus Oxytocin to Prevent Hemorrhage after Vaginal Birth. New Engl J Med. 2018.

Using Polio to Treat Cancer

Up until the mid-1950s polio was epidemic across many countries with an estimated 500,000 people paralysed or killed each year. Two vaccines were developed (one by Jonas Salk launched in 1957, the other by Albert Sabin in 1962), which lead to a dramatic decline in cases, and eventually eradication in many high-income countries. Following this the World Health Organization, UNICEF and the Rotary Foundation began an eradication campaign in 1988, and as a result there were only 22 reported polio cases worldwide in 2017. However, recent research has suggested that polio may be able to help patients with grade IV malignant gliomas.[1] Such patients have low survival rates – less than 20 months following diagnosis, and less than 12 months for recurrent gliomas. Current therapies are ineffective, inconsistent in improving survival and have many toxic effects. The trial was conducted using PVSRIPO, a modified live attenuated poliovirus type 1 vaccine. This is able to recognise CD155, a poliovirus receptor that is widely expressed in tumour cells. In total 61 patients were given a dose of PVSRIPO and followed up. Although 19% of patients had moderate-severe (grade 3 or higher) adverse events attributed to PVSRIPO, the overall survival of patients reached a plateau of 21% (95% CI 11-33) at 24 months, which was sustained at 36 months. This was higher than the rate among historical controls.


  1. Desjardins A, Gromeier M, Herndon JE, et al. Recurrent Glioblastoma Treated with Recombinant Poliovirus. New Engl J Med. 2018.

Mediating Variables

I have long argued that service delivery research, especially when generic interventions are evaluated, should examine the entire causal chain from intervention uptake to patient outcome.[1] Such an approach, of course, includes observation of mediating variables – variables (often residing in people’s hearts and minds) that form part of the above causal chain. The advantages of such ‘medication analysis’ seems to be catching on – two papers in the journal ‘Implementation Science’ have covered this topic in the last year.[2][3] In one case the mediating variables explained most of the effect (Anselmi, et al.), while in the other (Lee, et al.) they explained none of it, suggesting that the intervention was working through a theoretical domain not previously considered. These papers used structural equations. However, I prefer Bayesian networks that our CLAHRC is pioneering.[4-6] This is for two reasons:

  1. They can capture information from outside of the index study.
  2. They can meld qualitative and quantitative information through elicitation of informative probability densities.

— Richard Lilford, CLAHRC WM Director


  1. 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.
  2. Anselmi L, Binyaruka P, Borghi J. Understanding causal pathways within health systems policy evaluation through mediation analysis: an application to payment for performance (P4P) in Tanzania. Implement Sci. 2017; 12: 10.
  3. Lee H, Hall A, Nathan N, et al. Mechanisms of implementing public health interventions: a pooled causal mediation analysis of randomised trials. Implement Sci. 2018; 13: 42.
  4. Hemming K, Chilton PJ, Lilford RJ, Avery A, Sheikh A. Bayesian cohort and cross-sectional analyses of the PINCER trial: a pharmacist-led intervention to reduce medication errors in primary care. PLoS One. 2012;7(6):e38306
  5. 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.
  6. Watson SI, Chen YF, Bion JF, Aldridge CP, Girling A, Lilford RJ; HiSLAC Collaboration. Protocol for the health economic evaluation of increasing the weekend specialist to patient ratio in hospitals in England. BMJ Open. 2018; 8(2): e015561.

More on Body-Worn Cameras for the Police

In our previous News Blog we talked about a study that looked at the effect body cameras had on the behaviour of police officers.[1] The study found there was no significant differences seen for those who did and did not wear a body camera, and there was no improvement in behaviours. However, as we pointed out the results may have been context-specific as the study was conducted in only one district. Now, News Blog reader Peter Lilford has directed us to a similar study based in the UK and the USA.[2] This study, by researchers at Cambridge University, looked at over 4,000 shifts across seven sites where police officers were randomised to either wear a camera or not. All officers taking part wore a camera at some point over the year-long study. They found that by the end of the study period complaints against police officers had fallen by 93% (p<0.001). However, there was no significant difference in the number of complaints received by those wearing or not wearing a camera. The authors believe that this could be due to ‘contiguous accountability’ from repeated surveillance that even affected departments not taking part in the study. Another factor that the authors believe to be important is that police officers issued a warning at the beginning of any encounter that there was a camera present, perhaps serving to influence the resulting behaviours.

— Peter Chilton, Research Fellow


  1. Lilford RJ. What are the Effects of Body-Worn Cameras on Police and Citizen Behaviour? NIHR CLAHRC West Midlands News Blog. 1 June 2018.
  2. Ariel B, Sutherland A, Henstock D, et al. Contiguous Accountability: Global Multisite Randomized Controlled Trial on the Effect of Police Body-Worn Cameras on Citizens’ Complaints Against the Police. Crim Just Behav. 2016.

Time to Tweet

A number of studies have suggested various effects that the circadian rhythm has on our lives, from mood [1] to our immune system.[2] It is also likely to impact our day-to-day lives, as such a team at the University of Bristol looked into how it could affect our thoughts and emotions.[3] The authors analysed a sample of over 800m Tweets from UK users posted throughout the day over a four-year period by comparing the words used to linguistic lists that had been designed to infer various psychological states of a person. They found that two independent peak times were able to explain 85% of the variance in word usage seen. The first is a peak at 5-6am that is correlated with analytical thinking, motivational thinking, and personal concerns, and is negatively correlated with negative language and social concerns. The second is a peak at 3-4am, correlated with existential concerns, and negatively correlated with positive emotions. These time peaks are also associated with major changes in neural activity and hormonal levels.

The language we use changes drastically throughout the day, a reflection of changes in our main concerns, and cognitive and emotional processes.

— Peter Chilton, Research Fellow


  1. McClung CA. Circadian genes, rhythms and the biology of mood disorders. Pharmacol Therap. 2007; 114(2): 222-32.
  2. Lange T, Dimitrov S, Born J. Effects of sleep and circadian rhythm on the human immune system. Ann N Y Acad Sci. 2010; 1193:48-59.
  3. Dzogang F, Lightman S, Cristianini N. Diurnal variations of psychometric indicators in Twitter content. PLOS One. 2018.

Too Much Performance Measurement

The United States National Quality Measures Clearinghouse (NQMC) now lists over 2,500 performance measures for healthcare. Enough already!

Writing in the New England Journal of Medicine colleagues from Ann Arbor have audited the 271 measures in the Medicare quality payment program.[1] Expenditure on collecting quality measures costs the average US physician about $40,000 per year. However, only 32 of the above measures were found to be validated by a consensus process. There was no real evidence for most of the measures. The notion that health care can be monitored comprehensively like widgets on a production line Is a modern holy grail. In the last analysis we need well-informed, articulate, compassionate and, above all, committed clinicians. Let’s concentrate more on producing them and less on statistical process control.[2]

— Richard Lilford, CLAHRC WM Director


  1. MacLean CH, Kerr EA, Qaseem A. Time Out – Charting a Path for Improving Performance Measurement. New Engl J Med. 2018; 378: 1757-61.
  2. Lilford RJ. Comparing Statistical Process Control and Interrupted Time Series. NIHR CLAHRC West Midlands News Blog. 11 March 2016.

Which are the Most Cost-Effective Surgical Operations for Low- and Middle-Income Countries?

I recently came across a Lancet review on essential surgery: operations that should be available to even very poor communities.[1] The authors sensibly divide the list into those that should be available in first-level hospitals, and those that should be available in specialist (tertiary) hospitals.

Obstetrical and gynaecological procedure’s rank high in the first list, including caesarean section, surgery for ectopic pregnancy, and tubal ligation. There are a large number of operations related to trauma, including chest drain, complex fracture reduction, laparotomy, amputation, and debridement. Also included for the first-level hospitals are a range of general surgical procedures, such as repair of perforations, appendectomy, hernia repair, and colostomy.

Specialist procedures that should be available in tertiary hospitals include congenital abnormalities (especially cleft lip and club foot) and eye surgery (cataract extraction or eyelid repair for trachoma).

But how cost-effective are all these procedures? A recent study in PLOS One summarises some of the recent evidence.[2] The most cost-effective surgery is emergency caesarean section and voluntary male circumcision. This is followed in increasing order by: cataract surgery, cleft lip and palate surgery, hernia repair, breast cancer surgery, trauma surgery, colorectal surgery, and non-emergency orthopaedic conditions. I think vesicovaginal fistula should be high on the list.

One notes that most, but not all, of the highly cost-effective surgical conditions relate to acute presentations rather than chronic conditions. This underlines the critical importance of access to first-level hospital care. Such access is crucial, not only for surgical conditions, but also acute medical conditions such as meningitis, malaria and snake bites.

— Richard Lilford, CLAHRC WM Director


  1. Mock CN, Donkor P, Gawande A, Jamison DT, Kruk ME, Debas HT, for the DCP3 Essential Surgery Author Group. Essential surgery: key messages from Disease Control Priorities, 3rd edition. Lancet. 2015; 385: 2209-19.
  2. Horton S, Gelband H, Jamison D, Levin C, Nugent R, Watkins D. Ranking 93 health interventions for low- and middle-income countries by cost-effectiveness. PLoS ONE. 2017; 12(8): e0182951.

Recording University Lectures

In this News Blog we have often discussed various interventions aimed at improving the education of students.[1][2] Here is another one. Increasingly universities are encouraging (or mandating) that staff record their lectures so that they can be put online. Not only is this thought to help students who are unable to attend, but it also allows students to repeat sections they may have missed/misheard, etc. However, the pre-recorded nature means that students are not as easily able to get clarifications or take part in discussions, and could miss out by not attending the ‘live’ lecture. Further, knowing that the lecture is available anytime, may mean that students delay watching indefinitely – ‘I’ll watch it tomorrow’, etc.

A before-and-after study by Edwards and Clinton looked at the impact making recorded lectures available had on second year undergraduate students.[3] They found that introducing lecture recording was associated with a significant decrease in lecture attendance, while attendance mediated a negative relationship between the availability of recorded lectures and students’ attainment. When lecture attendance was factored in, there was no significant relationship between watching recorded lectures and attainment. On balance, the study showed that the effects of introducing lecture recording were negative. Based on this, it could be worth reconsidering the move to relying on recorded lectures.

— Peter Chilton, Research Fellow


  1. Lilford RJ. Another Interesting Trial of an Educational Intervention – This Time Concerning Access. 21 July 2017. NIHR CLAHRC West Midlands News Blog.
  2. Lilford RJ. Evidence-Based Education. 24 February 2017. NIHR CLAHRC West Midlands News Blog.
  3. Edwards MR & Clinton ME. A study exploring the impact of lecture capture availability and lecture capture usage on student attendance and attainment. High Educ. 2018.

RCTs versus Observational Studies: This Time in the Advertisement Industry

There is a substantive body of medical methodological research in which the results of RCTs for a given treatment are compared to the results of observational studies for that same treatment. They show that, as compared to the RCTs, effect sizes in the observational studies are similar to those in RCTs, but that they are widely scattered around the gold standard (i.e. RCT) estimate.[1] [2]

A similar result was obtained in a study of RCTs versus observational studies in the economics literature back in the 1980s, except that in their study RCTs yield more conservative estimates than those in observational studies.[3] Now, a similar study has been carried out in the advertising industry using advertisements carried on Facebook as the basis for a field experiment.[4] The results of 15 RCTs of advertisements on Facebook were compared to the results of observational studies in which standard statistical methods were used to control for potential identifiable confounders. The findings of this methodological study corroborate those of earlier studies in economics. The observational studies, even after risk adjustment, showed a wide scatter of results around those of the corresponding RCTs and the observational studies tend to produce more strongly positive results. This contradicts the prevailing view in the advertisement industry that observational studies produce reliable estimates of the effectiveness of advertisements. Interestingly, if just one factor accounted for all of the difference between the observational studies and the RCTs, then this single factor would account for more explanatory power than all other variables taken together.

The selection bias in the case of advertising likely relates to a link between exposure to the environment featuring the advertisement and responding to the ad when observed. To put this another way, people who are exposed to an advertisement are already pre-disposed to respond to the advertisement. In health care we would say that exposure and response are on the same causal chain. Economists would say that they were ‘endogenous’.

— Richard Lilford, CLAHRC WM Director


  1. Benson K & Hartz AJ. A comparison of observational studies and randomized, controlled trials. N Engl J Med. 2000; 342(25): 1878-86.
  2. Anglemyer A, Horvath HT, Bero L. Healthcare outcomes assessed with observational study designs compared with those assessed in randomized trials. Cochrane Database Syst Rev. 2014; 4: MR000034.
  3. Banerjee AV, Duflo E, Kremer M. The Influence of Randomized Controlled Trials on Development Economics Research and on Development Policy. 2016
  4. Gordon BR, Zettelmeyer F, Bhargava N, Chapsky D. A Comparison of Approaches to Advertising Measurement: Evidence from Big Field Experiments at Facebook. 2018.