Can a machine learning approach help us predict what specific treatments work best for individuals with depression?

by Marcus Munafò @MarcusMunafo

This blog originally appeared on the Mental Elf site on 11th Febraury 2016.

9990016123_b58fa297eb_h

Understanding who responds well to treatment for depression is important both scientifically (to help develop better treatments) and clinically (to more efficiently prescribe effective treatments to individuals). Many attempts to predict treatment outcomes have focused on mechanistic pathways (e.g., genetic and brain imaging measures). However, these may not be particularly useful clinically, where such measures are typically not available to clinicians making treatment decisions. A better alternative might be to use routinely- or readily-collected behavioural and self-report data, such as demographic variables and symptom scores.

Chekroud and colleagues (2015) report the results of a machine learning approach to predicting treatment outcome in depression, using clinical (rather than mechanistic) predictors. Since there are potentially a very large number of predictors, examining all possible predictors in an unbiased manner (sometimes called “data mining”) is most likely to produce a powerful prediction algorithm.

Machine learning approaches are well suited to this approach, because they can identify patterns of information in data, rather than focusing on individual predictors. They can therefore identify the combination of variables that most strongly predict the outcome. However, prediction algorithms generated in this way need to be independently validated. By definition, they will predict the outcome in the data set used to generate the algorithm (the discovery sample). The real test is whether they also predict similar outcomes in independent data sets (the replication sample). This avoids circularity, and increases the likelihood the algorithm will be clinically useful.

Clinicians currently have no empirically validated mechanisms to assess whether a patient with depression will respond to a specific antidepressant.

Clinicians currently have no empirically validated mechanisms to assess whether a patient with depression will respond to a specific antidepressant.

Methods

The authors used data from a large, multicenter clinical trial of major depressive disorder (the STAR*D trial – Trivedi et al, 2006) as their discovery sample, and a separate clinical trial (the CO-MED trial, Rush et al, 2011) as their replication sample. Data were available on 1,949 participants in the STAR*D trial, and 425 participants in the CO-MED trial. The CO-MED trial consisted of three treatment groups, with participants randomised to receive either:

  1. Escitalopram-placebo
  2. Bupropion-escitalopram
  3. Venlafaxine-mirtazapine

The authors built a predictive model using all readily-available sources of information that overlapped for participants in both trials. This included:

  • A range of sociodemographic measures
  • DSM-IV diagnostic items
  • Symptom severity checklists
  • Eating disorder diagnoses
  • Whether the participants had taken specific antidepressant drugs
  • History of major depression
  • The first 100 items of the psychiatric diagnostic symptoms questionnaire.

In total, 164 variables were used.

For the training process, the machine learning approach divided the original sample (using the STAR*D data) into ten subsets, using nine of those in the training process to make predictions about the remaining subset. This process was repeated ten times, and the results averaged across these repeats. The final model built using the STAR*D data was then used to predict outcomes in the each of the CO-MED trial treatment groups separately.

The model was developed to detect people for whom citalopram (given to everyone in the first 12 weeks of the STAR*D trial) is beneficial, rather than predicting non-responders. It was constrained to require only 25 predictive features (i.e., clinical measures), to balance model performance (which should be greater with an increasing number of predictors) with clinical usability (since an algorithm requiring a very large number of predictors may be difficult to implement in practice).

Only 11-30% of patients with depression reach remission with initial treatment, even after 8-12 months.

Only 11-30% of patients with depression reach remission with initial treatment, even after 8-12 months.

Results

The top three predictors of non-remission were:

  1. Baseline depression severity
  2. Feeling restless during the past 7 days
  3. Reduced energy level during the past 7 days

The top three predictors of remission were:

  1. Currently being employed
  2. Total years of education
  3. Loss of insight into one’s depressive condition

Overall, the model predicted outcome in the STAR*D data with:

  • An accuracy of 64.6% – it identified 62.8% of participants who eventually reached remission (i.e., sensitivity), and 66.2% of non-remitters (i.e., specificity)
  • This is equivalent to a positive predictive value (PPV) of 64.0% and a negative predictive value (NPV) of 65.3%
  • The performance of the model was considerably better than chance (P = 9.8 × 10-33)

In the CO-MED data, the model:

  • Pedicted outcome in the escitalopram-placebo group:
    • Accuracy 59.6%, 95% CI 51.3% to 67.5%,
    • P = 0.043,
    • PPV 65.0%,
    • NPV 56.0%.
  • Escitalopram-bupropion group
    • Accuracy 59.7%, 95% CI 50.9% to 68.1%,
    • P = 0.023,
    • PPV 59.7%,
    • NPV 59.7%.

However, there was no statistical evidence that it performed better than chance in the venlafaxine-mirtazapine group:

  • Accuracy 51.4%, 95% CI 42.8% to 60.0%,
  • P = 0.53,
  • PPV 53.9%,
  • NPV 50.0%.
Could predictive models that mine existing trial data help us prospectively identify people with depression who are likely to respond to a specific antidepressant?

Could predictive models that mine existing trial data help us prospectively identify people with depression who are likely to respond to a specific antidepressant?

Conclusions

The authors conclude that their model performs comparably to the best biomarker currently available (an EEG-based index) but is less expensive and easier to implement.

The outcome (clinical remission, based on a final score of 5 or less on the 16-item self-report Quick Inventory of Depressive Symptomatology, after at least 12 weeks) is associated with better function and better prognosis than response without remission.

Strengths and limitations

There are some strengths to this study:

  1. First, it attempts to build a prediction algorithm using data that are already collected routinely in clinical practice, or could be easily incorporated into routine practice.
  2. Second, the prediction algorithm shows some evidence of generalisability to an independent sample.
  3. Third, the algorithm also shows some degree of specificity, by performing best in the escitalopram-treated groups in the CO-MED data.

However, there are also some limitations:

  1. First, there is a clear reduction in how well the algorithm predicts treatment outcome in the discovery sample (STAR*D) compared with the replication sample (CO-MED). This illustrates the need for an independent replication sample in studies of this kind.
  2. Second, and more importantly, although the algorithm performed better in the escitalopram-treated groups in CO-MED, it’s not clear that there was any evidence that performance was different across the three arms – the 95% confidence intervals for the venlafaxine-mirtazapine group (42.8% to 60.0%) include the point estimates for the other two groups (escitalopram-placebo: 59.6%, escitalopram-bupropion: 59.7%). Therefore, although there is some evidence of specificity, it is indirect, and the algorithm may in fact predict treatment outcome in general, rather than in those who have received a specific treatment, at least in part.
  3. Third, models of this kind cannot tell us whether the variables that predict treatment outcome are causal. This may not matter if our focus is on clinical prediction, although if they are not causal then the prediction algorithm may not generalize well to other populations. For example, in both the discovery and replication sample participants had been recruited into clinical trials, and therefore may not be representative of the wider population of people with major depressive disorder. Causal anchors are likely to be more important if we are interested in mechanistic (rather than clinical) predictors.

Summary

Ultimately, being able to simultaneously identify individuals likely to respond well to drug A and not respond to drug B will be clinically valuable, and is the goal of stratified medicine. This study represents only the first step towards being able to identify likely responders and non-responders for a single drug (in this case, citalopram); in particular, although there was some evidence for specificity in this study, it was relatively weak.

Ultimately, with larger datasets that include multiple treatment options (including non-pharmacological interventions), it may be possible to match people to the treatment option they are most likely to respond successfully to. The focus on routinely- or readily-collected data means that it gives an insight into what clinical prediction algorithms for treatment response in psychiatry may look like in the future.

This innovative study may open the door to predict more personalised medicine for people with depression.

This innovative study (and others like it) may open the door to predict more personalised medicine for people with depression.

Links

Primary paper

Chekroud AM, Zotti RJ, Shezhad Z, Gueorguieva R, Johnson MK, Trivedi MH, Cannon TD, Krystal JH, Corlett PR. (2015) Cross-trial prediction of treatment outcome in depression: a machine learning approach. Lancet Psychiatry 2015. doi: S2215-0366(15)00471-X [Abstract]

Other references

Trivedi MH, Rush AJ, Wisniewski SR, Nierenberg AA, Warden D, Ritz L, Norquist G, Howland RH, Lebowitz B, McGrath PJ, Shores-Wilson K, Biggs MM, Balasubramani GK, Fava M; STAR*D Study Team. (2006) Evaluation of outcomes with citalopram for depression using measurement-based care in STAR*D: implications for clinical practice. Am J Psychiatry. 2006 Jan;163(1):28-40. [PubMed abstract] [Wikipedia page]

Rush AJ, Trivedi MH, Stewart JW, Nierenberg AA, Fava M, Kurian BT, Warden D, Morris DW, Luther JF, Husain MM, Cook IA, Shelton RC, Lesser IM, Kornstein SG, Wisniewski SR. (2011) Combining medications to enhance depression outcomes (CO-MED): acute and long-term outcomes of a single-blind randomized study.Am J Psychiatry. 2011 Jul;168(7):689-701. doi: 10.1176/appi.ajp.2011.10111645. Epub 2011 May 2. [PubMed abstract]

Photo credits

– See more at: http://www.nationalelfservice.net/mental-health/depression/can-a-machine-learning-approach-help-us-predict-what-specific-treatments-work-best-for-individuals-with-depression/#sthash.wvpEojY8.dpuf

Later menopause linked with lower risk of depression

by Meg Fluharty @MegEliz_

This blog originally appeared on the Mental Elf site on 17th February 2016.

Women have twice the risk of developing major depression compared to men. This difference is most noticeable during the reproductive period years (Soares et al, 2008) (e.g. premenstrual, during pregnancy and postpartum, and perimenopause) when women are subject to large fluctuations of ovarian hormones.

Additionally, oestrogens are believed to utilise neuroprotective and antidepressive actions within the the brain (Arevalo et al, 2015), and transitioning to the postmenopausal period is associated with a large drop in oestrogen production (Burger al al., 2007).

Therefore, the authors, Georgakis et al (2016), are using ‘age at menopause’ and ‘duration of reproductive age’ as two markers of lifelong oestrogen exposure to measure the association with risk of depression in postmenopausal women.

Research shows that the median age for final menstrual period is 52.5 years, and that 90% of women have their final period by the age of 56.

Research shows that the median age for final menstrual period is 52.5 years, and that 90% of women have their final period by the age of 56.

Methods

Search criteria

The authors conducted a search in MEDLINE using the following keywords: menopause, climacteric, reproductive period, depression, and mood disorders. The authors then searched reference lists of included studies to identify additional studies. There was no restriction on language, publication year or study design. Cross sectional and cohort studies were obviously going to be helpful, but randomised controlled trials were also considered for eligibility if they included depression measurements before intervention.

Definition of variables

  • Age of menopause was defined as 1 year following last menstruation (although studies examining age at final menstruation were also considered)
  • Duration of reproductive period was defined as age of menopause minus age of menarche
  • Diagnosis of depression was defined by clinical diagnosis or validated questionnaire

Excluded studies

Studies were excluded if they used questionnaires without defined cut-offs, or self-reported depression as a single question. Studies including only women with depression were excluded as were those which also had severe psychiatric disorders. Case series, case reports, in vitro and animal studies were excluded. Studies limited to perimenopausal (the period leading up to menopause) women, breast cancer survivors with medically induced menopause, or women with surgically induced menopausal transition were excluded.

Statistical analysis

The odds ratios (OR) and confidence intervals (CI) were pooled across the identified studies, and the analysis was conducted separately for the two exposure variables (age of menopause and duration of reproductive period). The variables were first analysed as continuous variables in 2 year increments, and age of menopause was analysed again as a categorical variables (≥40 vs <40).

Results

A total of 67,714 women were included across 10 cross sectional and 4 cohort studies.

  • 12 studies used self-report diagnosis of depression
  • 1 study used DSM-III-R diagnosis
  • 1 study used physician diagnosis.

Women without a diagnosis of depression were used as the control group.

Age at menopause

Pooling the effect estimates across 13 studies which treated age at menopause as a continuous variable (e.g. 2 year increments); increased age of menopause was associated with 2% decrease in risk of postmenopausal depression (OR, 0.98; 95% CI 0.96 to 0.99 heterogeneity I2=7.6%; P=.37). Sensitivity analyses for hormone therapy, premenopausal depression, or defining age at menopause as 1 year following last menstruation did alter the association.

In 4 studies with data on premature menopause (<40 years), twice the risk of depression was found compared to women with menopause onset over 40 years (OR, 0.49; 95% CI 0.29 to 0.81; heterogeneity I2=54.2%, p=.09).

Reproductive period

Pooling the effect estimates across 5 studies that includes reproductive period as a continuous variable (e.g. 2 year incriminates); found similar associations to age at menopause: a 2% decrease in risk of postmenopausal depression for an increase in reproductive period of 2 years (OR, 0.98; 95% CI 0.94 to 1.01; heterogeneity I2=0.0%; P=.41).

This evidence suggests that women who have the menopause later in life, are less likely to experience depression in their postmenopausal years.

This evidence suggests that women who have the menopause later in life, are less likely to experience depression in their postmenopausal years.

Discussion

This meta-analysis displayed an inverse relationship between the age of menopause and subsequent risk of postmenopausal depression, which prevailed after controlling for hormone therapy and premenopausal depression. Additionally, a similar effect was found within an analysis of the duration of reproductive period. These findings indicate that shorter exposure to endogenous oestrogen is associated with oestrogen deficiency and consequently heightened risk of depression after menopause.

To put it another way, the longer the period between menarche (first menses) and menopause (defined as final menstrual period or 1 year after final menstrual period), the lower the risk that the woman will experience depression in her postmenopausal years.

If these findings are confirmed within culturally diverse studies, they can be used to identify at-risk women for postmenopausal depression whom may benefit from either psychological monitoring or oestrogen-based therapy (Georgakis et al 2016).

Strengths and limitations

This systematic review featured a well-conducted meta-analysis, including a total of 67,714 women across 14 studies; and took important confounders into consideration (age, obesity, hormone therapy, smoking, and marital status). The authors conducted sensitivity analyses where necessary and there was no evidence of publication bias in the ‘age at menopause’ studies.

However, there were some limitations to consider:

  • Limiting their literature search just to the MEDLINE database will have resulted in many trials been missed, which is clearly a big weakness for any systematic review.
  • 12 of the 14 included studies used a self-report diagnosis of depression, rather than a diagnosis reached by a validated diagnostic instrument.
  • There were differences in the definition of depression, and depression cut-offs across studies.
  • The association of pre-existing depression and hormone therapy use on later depression should be considered; however the authors did conduct sensitivity analyses where possible.

Many women report a huge lack of information about the menopause, fuelled by a continuing stigma relating to this ubiquitous part of female human existence. This study provides some important pointers to risk factors and later life mental illness, which could be used to help educate women about their risk of depression as they age. However, given the limitations of this current review, we should look for further confirmation of these findings before we consider this question well and truly answered.

Do you talk to your female friends about the menopause?

Do you talk to your female friends about the menopause?

Links

Primary paper

Georgakis MK, Thomopoulos TP, Diamantaras A, et al. (2015) Association of Age at Menopause and Duration of Reproductive Period With Depression After Menopause: A Systematic Review and Meta-analysis. JAMA Psychiatry. Published online January 06, 2016. doi:10.1001/jamapsychiatry.2015.2653. [Abstract]

Other references

Soares CN, Zitek B. (2008) Reproductive hormone sensitivity and risk for depression across the female life cycle: a continuum of vulnerability? J Psychiatry Neurosci. 2008;33(4):331-343.

Arevalo MA, Azcoitia I, Garcia-Segura LM. (2015) The neuroprotective actions of oestradiol and oestrogen receptors. Nat Rev Neurosci. 2015;16(1): 17-29. [PubMed abstract]

Burger HG, Hale GE, Robertson DM, Dennerstein L. (2007) A review of hormonal changes during the menopausal transition: focus on findings from the Melbourne Women’s Midlife Health Project. Hum Reprod Update. 2007;13(6):559-565. [PubMed abstract]

If you’re looking for a good overview of recent evidence-based research, please read the Evidently Cochrane blogs on the Menopause.

Photo credits

– See more at: http://www.nationalelfservice.net/mental-health/depression/later-menopause-linked-with-lower-risk-of-depression/#sthash.v4Zbt9Fx.dpuf

The European Tobacco Products Directive and the future of e-cigarettes in the UK

By Jasmine Khouja @Jasmine_Khouja

E-cigarettes have become a popular product among smokers and ex-smokers, and Action on Smoking and Health (ASH) estimates that there are 2.6 million current users of e-cigarettes in the UK. As an alternative to tobacco smoking, research commissioned by Public Health England estimates that e-cigarettes are likely to be roughly 95% less harmful. The evidence supporting these popular and effective quitting aids suggests that e-cigarettes could be a powerful tool for harm reduction amongst current smokers but there is still uncertainty over the safety of e-cigarettes. Limited research concerning the effects of long-term use and the current lack of strict regulation of the products has fuelled this uncertainty but new regulations have been introduced into the pre-existing European Tobacco Products Directive (TPD) to rectify this. The updated TPD will come into force on 20th May 2016 with a transitional period allowed by the TPD. UK e-cigarettes and refill containers which are not in compliance with the TPD will be allowed to be released for sale on the UK market until 20th November 2016, but from 20th May 2017 all products sold to consumers will need to be fully compliant with the TPD. The alternative to following the regulations set by the TPD will be for e-cigarettes to gain a medical licence from the Medicines and Healthcare products Regulatory Agency (MHRA) and be regulated as licenced medicinal products to be sold in the UK.

jaz blog

As I am about to commence a PhD investigating the reasons for e-cigarette use, I am interested in what the implications of the directive will be in the UK; will it encourage smokers to switch to e-cigarettes, consequently reducing harm to themselves and others, or will it result in a reduction of available products and cause an increase in relapses to smoking?

I have read the directive and listed some of the key changes that will happen and added my own thoughts on what may happen as a result.

  1. CHANGE: New e-cigarette products must be notified to the MHRA six months before their release to the public. E-cigarette companies will be charged £150 to notify MHRA of a new product and £80 for a modification to an existing product, and will then be charged £60 annually thereafter. POSSIBLE OUTCOMES: The MHRA should have more control over the products on the market and be able to prevent unsafe products entering the market but it may take longer for new products to become available to buy. Additionally, some existing products will be unavailable from 20th May 2017 if they do not to comply with the regulations by 20th November 2016.
  1. CHANGE: Under the TPD, e-liquids will only be allowed where the nicotine concentration does not exceed 20 mg of nicotine per ml of liquid. E-liquids containing more than 20 mg of nicotine per ml of liquid will have to gain a medical licence authorised by the MHRA. POSSIBLE OUTCOMES: People may reduce their doses of nicotine and reduce their addiction if their preferred dosage is no longer available. Fewer high dosage products may be available as gaining a medical licence is an expensive process (estimated between £87,000 and £266,000 annually over ten years for a single device). When current products with high dosages such as 36 mg of nicotine per ml of liquid become unavailable, people may use lower dosages such as 20 mg of nicotine per ml of liquid as a substitute and inhale twice as much vapour to get the same nicotine hit. Nicotine is not the only constituent of vapour though; there are low concentrations of other toxicants, so inhaling more vapour means inhaling more toxicants. Alternatively, current higher dosage users may relapse to tobacco smoking if they feel the lower dosages do not effectively deliver the nicotine hit they need.
  1. CHANGE: Products regulated under the TPD must provide information to the MHRA on the safety and contents of e-cigarette products (including ingredients, toxicants and emissions). Health warnings, instructions for use, information on addictiveness and toxicity must also appear on the packaging and accompanying information leaflet. POSSIBLE OUTCOMES: This should allow e-cigarette users to make informed choices. The notification fees mentioned above will include the storage of this information but the companies may have to bear extra costs in testing their products for the amount of toxicants and emissions produced. These tests will have to comply with the standards set in the TPD and by the MHRA which may prove too costly for smaller e-cigarette companies, forcing them to withdraw products from the market. This could leave the market open to the tobacco industry who generally have greater financial resources available to them. The tobacco industry have to also sustain the tobacco market; a consequence of this may be the deliberate placement of ineffective e-cigarette products on the market to encourage current smokers continue to smoke tobacco and ex-smokers using e-cigarettes relapse.
  1. CHANGE: E-cigarette products will be child-safe, will not break or leak during the refill process, and containers will not exceed 10 ml (refill cartridges will not exceed 2 ml). POSSIBLE OUTCOMES: This should prevent accidents involving children consuming dangerous levels of nicotine. Most changes will be made to newer devices, which require e-liquid refills. If these modifications aren’t made by 20th November 2016 the products will be removed from the market by 20th May 2017.
  1. CHANGE: Under the TPD, cross-border advertising will be banned, which includes in newspapers, radio and TV, but not on billboards and posters. Products will not be allowed to make smoking cessation or health claims. Advertising of products with a medicinal license will be allowed under “over the counter” medicine rules. POSSIBLE OUTCOMES: This should minimise the amount of e-cigarette advertising seen by those who should not use e-cigarettes such as children and non-smokers. However, only e-cigarette companies who can afford a medical licence will be able to advertise on TV and this could mislead people into thinking that these products are more effective than other products.

A possible outcome for many of these changes is the loss of products from the market because of non-compliance with the regulations. Although increased reassurance that e-cigarettes on the market meet certain quality standards may encourage new users, the removal of any e-cigarette product from the market will provide an opportunity for e-cigarette users to relapse to smoking; without their favourite brand or flavour, it may be easier for them to resume smoking again than to find a replacement that suits their needs and taste. This in turn could lead to increased levels of smoking, and therefore harms to both individuals and society as a whole. Additionally, high nicotine dosage e-cigarette users may be encouraged to inhale more vapour and therefore unnecessary amounts of other constituents. However, recent preliminary research findings from ASH UK suggest there are few high dosage users meaning that this should not affect many.

The withdrawal of products is likely to be determined by the cost of making products compliant. Tobacco companies generally have greater financial resources than e-cigarette companies, with the top companies making billions in profit each year, meaning they can afford to make the necessary changes to meet the new regulations. The few e-cigarette companies that are owned by tobacco companies mainly produce ‘cigalikes’ which are the least effective design of e-cigarettes and there is a higher chance of relapsing to smoking when using them compared to later-generation devices. Given that the tobacco-owned e-cigarette companies will probably have greater resources available to them, they could end up with a monopoly on the e-cigarette industry. In fact, this may already be happening; the first medically licensed e-cigarette is a ‘cigalike’ owned by British American Tobacco. This means British American Tobacco could own the only TV-advertised e-cigarette (until another company gains a licence). Consequently, smokers looking to try e-cigarettes may choose less effective devices because they are more widely advertised.

These changes may reassure the general public that the devices will be safe but may lead to many ex-smokers relapsing because they are forced to use e-cigarettes and e-liquids that do not meet their needs, all the while lining the pockets of the tobacco industry by allowing them a monopoly on higher nicotine dosage products. Of course, the possible outcomes stated here are speculative; research will need to be undertaken to evaluate the ongoing impact of the new guidelines.

Links

  1. https://www.gov.uk/government/uploads/system/uploads/attachment_data/file/489981/TPD_Cons_Gov_Response.pdf).
  2. http://www.telegraph.co.uk/news/health/news/12079130/E-cigarettes-win-first-approval-as-a-medicine-opening-way-for-prescription-by-the-NHS.html
  3. https://www.gov.uk/government/consultations/regulatory-fees-for-e-cigarettes
  4. http://ash.org.uk/files/documents/ASH_1011.pdf
  5. https://nicotinepolicy.net/documents/reports/Impacts%20of%20medicines%20regulation%20-%2020-09-2013.pdf

Photo Credits

http://ecigarettereviewed.com/ – Lindsay Fox

New alcohol guidelines: what you need to know

by Olivia Maynard @OliviaMaynard17

This blog originally appeared on the Mental Elf site on 9th Febraury 2016.

Last month the UK Chief Medical Officers (CMOs) published new guidelines for alcohol consumption. These are the first new guidelines since 1995 and are based on the latest evidence on the effects of alcohol consumption on health.

The guidelines provide recommendations for weekly drinking limits, single drinking episodes and recommendations for pregnant women, drawing heavily on the Sheffield Alcohol Policy Model, which uses the most up to date evidence on both the short- and long-term risks of alcohol.

What are the new guidelines?

Guidelines for weekly drinking

For the new weekly drinking guidelines, the CMOs recommend that:

  • It’s safest for both men and women to not regularly drink more than 14 units of alcohol per week;
  • It’s best to spread these units over 3 days or more;
  • Having several drink-free days each week is a good way of cutting down the amount you drink;
  • The risk of developing a range of illnesses increases with any amount you drink on a regular basis.

There are two key changes here from the guidelines we’ve been used to:

First, there’s no difference in recommendations for men and women. This is because there is increasing evidence that although women are more at risk from the long-term health effects of alcohol, men are more at risk from the short-term effects of drinking (they’re more likely to expose themselves to risky situations while drinking).

Second, there is an explicit statement that there is no ‘safe’ level of alcohol consumption. Over the past 21 years, the link between alcohol and cancer has become much clearer. For example, we now know that while the lifetime risk ofbreast cancer is 11% among female non-drinkers, the lifetime risk for a woman drinking within the new guidelines is 13%. A woman drinking over 35 units a week increases her risk of breast cancer to 21%.

In their report, the CMOs are also at pains to point out that the evidence supporting alcohol’s protective effects on ischaemic heart disease is now weaker than in 1995. Furthermore, any potential protective effect of alcohol is mainly observed among older women at very low levels of consumption. Previously some have used this to claim that drinking is better than abstinence – the new guidelines refute this.

The new guidance says it's safest for men and women to drink no more than 14 units each week.

The new guidance says it’s safest for men and women to drink no more than 14 units each week.

Guidelines for single drinking episodes

The new guidelines are the first to provide guidance on drinking on single occasions, recommending drinkers:

  • Limit the total amount consumed on any occasion;
  • Drink slowly, with food and alternating with water;
  • Avoid risky places and activities and ensure they have a safe method of getting home.

These new recommendations reflect the fact that many alcohol consumers may drink heavily on occasion and provide guidance to avoid the risk of injury and ischaemic heart disease which increase with heavy drinking.

Heavy drinking episodes are linked with a higher risk of injury.

Heavy drinking episodes are linked with a higher risk of injury.

Guidelines for drinking during pregnancy

The new guidelines suggest that:

  • The safest approach for pregnant women is not to drink alcohol at all, to keep risks to the baby to a minimum.
  • Drinking during pregnancy can lead to long-term harm to the baby, with the risk increasing with the more alcohol consumed;
  • The risk of harm to the baby is likely to be low if a woman has drunk only small amounts of alcohol before she knew she was pregnant or during pregnancy.

The CMOs report that while the evidence on the effects of low alcohol consumption during pregnancy remains ‘elusive’, taking a precautionary approach is most prudent when it comes to a baby’s long term health. However, given the elusive evidence, the guidance is also careful to note that mothers should not be too concerned if they have drunk early in their pregnancy, as this kind of stress may be even more harmful to the developing baby.

Pregnant women are advised not to drink alcohol at all.

Pregnant women are advised not to drink alcohol at all.

A note on risk

These recommendations are based on a level of alcohol consumption which confers a 1% lifetime risk of death from alcohol. Their purpose is therefore tominimise risk from alcohol, rather than eliminate it. Indeed, the guidelines explicitly state that there is no safe level of alcohol consumption. So what does a 1% lifetime risk mean and how does this compare to other health behaviours?

Lifetime mean risks

  • Being killed through BASE jumping (0.3%);
  • Being killed in a car accident (0.4%);
  • Being diagnosed with bowel cancer from eating three rashers of bacon every day (1%);
  • Dying from an alcohol related disease, if drinking within the new guidelines (1%);
  • Smokers dying from a smoking related disease (50%, although new estimates suggest that this may be as high as 67%).

Put in the context of smoking, the risk posed by drinking within the new guidelines seems tiny (although it’s still more risky than BASE jumping!) However, it’s important to note that alcohol consumption and smoking are quite different. Alcohol consumption is perceived as normal in our society and is much more prevalent than cigarette smoking. By contrast, the acceptability of smoking is reducing and unlike social alcohol consumers, smokers are constantly being told to quit smoking.

This 1% risk level is that which is deemed ‘acceptable’ by the CMO. However, everyone will have a different ‘acceptable’ level of risk, which depends in part on how much pleasure is obtained from drinking. While some will think that increasing their risk of death from alcohol to 5% is acceptable, others will not accept any risk and will use these guidelines to cut out alcohol completely.

Criticisms of the new guidelines

As expected, the ‘nanny state’ criticism has been bandied around in pubs, on message boards and on social media since the publication of these guidelines. Others claim that these new guidelines are simply scaremongering. However, it’s important to remember that these are recommendations, not rules.

The last word must go to CMO Professor Dame Sally Davies, who addressed this criticism by saying that:

What we are aiming to do with these guidelines is give the public the latest and most up- to-date scientific information so that they can make informed decisions about their own drinking and the level of risk they are prepared to take.

What do you think? Are these new guidelines useful? Will they help reduce alcohol related harm?

What do you think? Are these new guidelines useful? Will they help reduce alcohol related harm?

Links

Primary paper

Department of Health (2016) Health risks from alcohol: new guidelines. Open Consultation, 8 Jan 2016 (Consultation closes on 1 April 2016)

Department of Health (2016). Alcohol Guidelines Review – Report from the Guidelines development group to the UK Chief Medical Officers.

Other references

Centre for Public Health (2016). CMO Alcohol Guidelines Review – A summary of the evidence of the health and social impacts of alcohol consumption. Liverpool John Moores University.

Centre for Public Health (2016). CMO Alcohol Guidelines Review – Mapping systematic review level evidence. Liverpool John Moores University.

Department of Health (1995). Sensible drinking: Report of an inter-departmental working group.

Photo credits

 

The Rosalind Franklin Appathon Prize and Tech day: A celebration of pioneering women in STEMM past, present and future.

by Sarah Griffiths @SarahGriff90

Last week Angela Attwood and I attended the Rosalind Franklin Appathon Prize and Tech day. The appathon was set up by UCL Professor Rachel McKendry with funds from her Royal Society Rosalind Franklin Award. The Rosalind Franklin Award recognises outstanding women in science with an idea for how to raise the profile of other women in their field. Professor McKendry’s idea was to hold an appathon with two challenges. The first challenge was to invent an app that would empower women to become leaders in STEMM (science, technology engineering, mathematics and medicine). The second challenge was to recognise a woman who had pioneered a new app for research, societal good or enterprise. Marcus had nominated me for the second challenge for my work developing the iPad app About Face for teaching emotion recognition to children with autism. Although I was not short listed in the end, I was pleased to be invited along to the prize and tech day to celebrate the winners and their apps.

On the day we were shown videos showcasing the shortlisted apps, some of which you can see here. The winner of Challenge 1 was Amazing STEMM Trailblazers an app for teaching children about influential women in STEMM through games. The runner up was STEMM Role Models, an app for finding female experts to speak at conferences. My personal favourite in this category was EyeSTEMM an app which suggests STEMM careers for young users based on photos they upload to show their interests. There were many exceptional apps shortlisted for Challenge 2, all of which would have been worthy winners. The winner was eSexual Health Clinic, developed by epidemiologist Pam Sonnenberg and her team at UCL. It provides Chlamydia test results and after care, including ordering antibiotic medication to the users local community pharmacy. This app demonstrates the potential of mobile technology to revolutionise health care and reduce strain on the NHS. The runner up was Findme  developed by psychologist Sue Fletcher-Watson from Edinburgh University. The app aims to teach children with autism social skills such as following eye gaze and gestures. It was created with consultation with individuals with autism, and the effectiveness of the app has been tested in a randomised control trial. Also shortlisted was Drink Less, which may be of particular interest to readers of this blog. This app was developed by Professor Susan Michie and her team of health psychology researchers at UCL. The app aims to help people keep track of, and to cut down on, their alcohol intake. The data from this app is being used to test how well particular techniques for cutting down work when delivered in an app, rather than by a clinician.

ggg

Professor Dame Athene Donald speaking about how to promote women to become leaders in science research.

As well announcing the winning apps, the afternoon also included a number of brilliant talks addressing the challenges and opportunities for women in STEMM industries today. Speakers included Jennifer Glynn, Rosalind Franklin’s sister, who spoke of male-only common rooms in universities which were the norm in the 1950s, reminding us of how far we have come in terms of gender equality in academia. Baroness Martha Lane Fox, founder of lastminute.com, spoke about gender imbalance in the technology industry (only 17% of tech jobs in the UK are held by women) and emphasised the importance of encouraging women to consider STEMM jobs in order to address skills shortages. Finally, Professor Dame Athene Donald gave an inspirational talk about her experiences as a leading female physicist and her ideas for how to promote women in science. This talk I found particularly interesting as a woman seeking a career in this field.

The whole day was very inspiring, both in terms of showing the potential of technology to make positive differences in research, health and education and in terms of showcasing the many women who are playing a leading role in this field. Although there are clearly still challenges being faced by women in STEMM, the message of the day was one of confidence that these can be overcome.

 

A behavioural insights bar: How wine glass size may influence wine consumption

by Olivia Maynard @OliviaMaynard17

Now that the festive season is almost upon us, I’ve been wading through the list of jobs I’ve been putting off for longer than I can remember, with the hope of starting afresh in 2016.

One of these jobs is wrapping up some of the studies I’ve been running this year, tidying up the data files and deciding what to do with the results. Obviously it’s best practice to write up all studies for publication in peer-reviewed journals, but sometimes this isn’t possible straight away (for example, when we’ve collected pilot data which will inform larger studies or research grants), although journals specifically for pilot and feasibility work do exist. However, it’s still important to share the findings, at the very least to prevent other research groups from running exactly the same pilot study (avoiding the file drawer effect).

The pilot study I’m trying to wrap up was conducted in September this year and is worth reporting, not only because the research is interesting, but also because the method of data collection was novel.

In December 2014 we were approached by the Behavioural Insights Team (BIT), who asked whether we’d be interested in running an experiment at their annual conference. Alongside a star-studded list of speakers, the BIT had planned to demonstrate to conference delegates the power of behavioural insights, by running a series of mini-experiments throughout the conference. We were asked to contribute, not only because I had previously worked in the BIT as part of a placement during my PhD, but also because of TARG’s track record in running behavioural experiments to influence alcohol consumption, both in the lab and in the ‘real-world’.

glassThe team asked us to run an experiment in the Skylon bar in the Royal Festival Hall – the venue of the conference drinks reception. After an initial assessment of the bar (yes, this is a tough job!) and discussing various possible experiments we could conduct, we finally decided to examine the impact of glass size on alcohol consumption. While considerable previous research has shown that plate size is an important driver in food consumption, and we have shown that glass shape (i.e., curved versus straight) influences alcohol consumption, there is very little research on the impact of glass size on alcohol consumption. Larger wine glasses are increasingly common and these may increase wine consumption and drinking speed by suggesting larger consumption norms to consumers, or by tricking consumers into thinking there’s a smaller amount in the glass than in a smaller glass which is equally full.

The primary aim of this pilot study was to determine the feasibility of implementing a glass size intervention study in a real-world drinking environment in order to inform future studies in this area.

Method

Prior to starting the study, as with every TARG study, we published the protocol online on the Open Science Framework. Depending on the side of the bar they were stood in, delegates attending the drinks reception were provided with either a small or a large wine glass, each of which was filled to the same volume. Every 15 minutes we counted the number of delegates on the two sides of the bar and every hour (for three hours) we counted the number of empty wine bottles on each side of the bar. We calculated the average volume of wine consumed per delegate each hour and then compared these between the two groups.

Results

From a feasibility point of view, the study worked as well as expected. Follow-up interviews with the manager of the bar indicated that bar staff enjoyed the process of participating in a study and were happy to participate again in future studies.

However, because we were conducting this in the real-world, rather than in our carefully controlled laboratory environment, we encountered a few logistical challenges. Here are the key points we learned from running this pilot study:

  1. In the real-world, there’s a necessary trade-off between collecting the data and not disrupting normal behaviour

bottles

Ideally we would have counted the number of empty bottles more frequently than every hour in order to get a more accurate measure of how much was consumed by the delegates. However prior to the start of the study, the bar manager suggested that this would interfere with their service and the bar staff reiterated this after the study had finished. As the bar staff were vital to the success of this pilot study, we didn’t think it was appropriate to push for more data collection than they felt comfortable with.

  1. Complete control of the environment isn’t possible in the real-world

controlkey

To prevent delegates from moving between the two sides of the bar we placed physical barriers between them, such as sofas, plants and lamps. However, inevitably, some delegates who wanted to ‘work the room’ at what was essentially a networking event did make their way past the barriers we set up. Other than instructing the waiters to replace the glass of those who had moved sides with the glass size appropriate for the side of the bar they were now in, there was very little we could do about this, short of frog-marching delegates back to their original side (which we thought wouldn’t go down very well on this occasion!)

  1. Accurate enforcement of study conditions is more difficult in the real-world

pouring

If we had conducted this study in the laboratory, we would have randomised participants to receive one of two glass sizes and carefully poured the exact volume of wine into their glass. In this real-world study, however, we had to rely on the waiters to accurately pour the wine into the glasses. Although highly trained, the waiters may also have fallen foul of the visual illusion the different glasses present (an effect which has been shown in previous real-world experiments). Future studies could monitor waiter pouring behaviour before and during the study.

  1. Studies in real bars have some other unexpected challenges…

full glassess

The BIT had asked that we present the results at 9am the following morning, allowing a nine hour turnaround from the end of the study to data presentation. This time pressure was not helped by the large quantities of complimentary champagne being served at the event, which considerably slowed down data entry and analysis at midnight!

Despite this substantial challenge, the results of the study were presented the following morning.

These data suggested that there was no difference in volume of wine consumed between the groups drinking from larger glasses and those drinking from tablesmaller glasses. As this study wasn’t powered to detect a meaningful difference between the two groups, we weren’t really surprised by this finding. However, these pilot data, along with the lessons learned from conducting the study will be used to inform our future research studies and grant applications.

And there we have it – another pilot study out of the file drawer and another item crossed off my ‘to-do’ list.

I’d like to thank the entire Behavioural Insights Team, in particular Ariella Kristal and Gabrielle Stubbs, for making this study happen, Carlotta Albanese from the Skylon bar and David Troy and Jim Lumsden from TARG for helping with all the data collection (and data entry at midnight).

Can we use the inhalation of 7.5% CO2 as a model to probe cognition and behaviour in anxiety?

by Alex Kwong @tskwong

A lot of the work conducted in the Tobacco and Alcohol research group (TARG) mainly focuses around tobacco and alcohol research (funny that…). However, when we’re not getting people intoxicated in the name of science (yes we do that), we’re also carrying research ranging from body perception, to emotion recognition and anxiety research. The latter is something that I’ve focused on, and to cut a long story short, we make people anxious by making them breathe in air enriched with carbon-dioxide (CO2), about 7.1% more than what you would normally breathe. Once people are anxious, we assess them on a number of outcomes, some clinically relevant, some more practical and applied.

Needless to say, breathing in about 7% more CO2 for a period of up to 20 minutes should make you anxious for a number of reasons (to be explained later on). But can breathing in a gas that is enriched with CO2 act as a viable model for anxiety, capable of assessing cognition and behaviours that are susceptible to anxiety? In this post I’ll explore some of the previous research utilising this model, and look at some of the future directions of the model and how it could be used as a training tool to help improve performance under anxiety. By then, hopefully you’ll agree with me that the model is good at experimentally inducing anxiety, and you’ll sign up for all our studies.

Possibly the most influential research on the inhalation of CO2 has been by Bailey et al. (2005) and work from David Nutt’s former lab in Bristol. They found that breathing in CO2 enriched gas for a period of 20 minutes decreased positive mood (feelings of happiness and relaxation) and increased negative mood (worry and fear). Since then, a plethora of research has supported this, and also found that the model induces symptoms such as sweating, increased heart rate and blood pressure and hypoxia, all common in generalised anxiety disorder (GAD). Interestingly, other research has found that we can actually reduce these responses to the CO2 model by giving people anxiolytic drugs. As such, the model of 7.5% CO2 has been considered a validated model of human anxiety induction that is generalisable to anxiety disorders such as GAD.

But why does breathing in a gas that is enriched with CO2 cause these sort of feelings? One explanation is that breathing in CO2 causes chemoreceptors to mislead the body into thinking that it is starved of oxygen. This leads to fear like responses, as well as increased breathing rates and higher blood pressure and heart rate. If you’ve ever had the pleasure of taking part in one of these CO2 experiments, you’ll likely agree that these things happen. I’ll just stress at this point that effects of the gas are transient and usually disappear quickly after the inhalation. Some people even enjoy the experience, so I hope I’m still selling this to you.

CO2 set-up
A typical experimental set up with the CO2.

So if it makes you feel like you’re experiencing physiological anxiety, then it’s obviously a model of human anxiety right? Well what about the psychological components? People with GAD often have a hypervigilance to threat, even when there is nothing threatening around. Additionally, their attention to negative stimuli is increased, even in the presence of other emotional content. Anxious sufferers also interpret ambiguous information as potentially dangerous or threatening. Can the CO2 model can tap into some of these psychological components that are common in GAD?

To address this, one study found that the inhalation of 7.5% CO2 caused quicker eye-movements to be made towards threatening stimuli. Another study found that CO2 caused attention to reflect a hyper vigilance to threatening information. Otherresearch in preparation has found that people were worse at correctly identify emotional faces during CO2. Lastly, Cooper et al. (2013) found that CO2 caused people to interpret ambiguous information in CCTV footage as threatening. These findings support the 7.5% CO2 model affecting psychological processes similar to those in GAD.

Great! So the model seems to be similar to the experience of GAD, what next? Well, what’s also quite fascinating is that if we have a model for anxiety, we could predict how people will behave in situations like sport, musical performances, decision-making, medical and security services etc – behaviours that can induce feelings of anxiety or be affected by anxiety, even in those without a disorder. Understanding how people will behave in stressful situations might help improve performances in the future.

The CO2 model has been used to investigate this. Attwood et al. (2013) found that 7.5% CO2 impaired the ability to match pairs of faces, a finding which has tremendous implications for military and forensic settings (e.g., border crossings and proof of sale purchases like alcohol and tobacco). More recently, we also found that the inhalation of 7.5% CO2 impairs the ability to remember faces that have previously been seen. Importantly, ‘witnesses’ did not report lower confidence of their choices despite this impaired ability, which has implications for the judicial system (e.g., courtrooms and line-ups).

Upcoming research has suggested that CO2 impairs decision-making on a gambling task, by making people choose more exploratory decisions which in turn causes less money earned. Other research has suggested that the CO2 causes excessive force production which could affect military, surgical and sporting behaviours. The same research also suggested that people speed up when asked to tap in time with a metronome, which could detriment musical performances and any task that requires accurate bodily timing. Together, this research shows that the inhalation of 7.5% CO2 may be a useful tool for examine how anxiety may affect behaviours.

Mask
The amount of Bane and Darth Vader impressions I got from participants was staggering – “It would be extremely anxious…, for you”

By now you should be getting the picture that a) the CO2 model is good for inducing anxiety and b) that I am incredibly biased in favouring this model. But I think there are good reasons to endorse this stance. Many previous studies that induce anxiety are time limited, meaning that ‘anxiety’ may only affect certain stages of the task. Other studies only produce one single ‘hit’ to cause anxiety (e.g., one phobic stimuli, one bodily stressor), which may not be characteristic of anxiety as a whole. However, one anxiety inducer that I think is quite neat is the threat of electric shock. Threatening people with electric shock is a great way to induce anxiety but in some experiments, the shock doesn’t actually come, so people quickly learn that there is no threat and thereby no longer remain anxious, which is a problematic for anxiety research.

The CO2 model is not without its flaws. Tasks can only be conducted within the 20 minute inhalation window. That said, there is no limit to how many times someone can be CO2’d. Practically, people may decide they no longer want to feel anxious during the inhalation and so drop out, but this is likely to be a problem in anxiety research generally. Perhaps most importantly, whilst we have conducted numerous CO2 experiments, we are still unsure exactly how the model works on all attentional and behavioural mechanisms. Future research is looking at how the CO2 model affects the brain, and our eye-movements. There is also research that has explored psychological interventions, such as mindfulness training, and whether this can reduce some of the symptoms brought on by the CO2 inhalation. It’ll also be really interesting to see whether the model can be utilised as a training tool for people who need to perform under anxious conditions. Research has shown that practising under conditions of anxiety can help improve performance at a later stage and so the next step would be to see if people can perform better in real life anxious situations, if they’ve practised on the CO2 model first.

In summary, the CO2 model seems to be a reliable way to induce anxiety that can impact on both cognition and behaviour. The model is validated by a wealth of research showing its similarity to GAD. Although the model is not perfect for inducing anxiety, it is one of the more promising tools we currently have, and subsequent research should continue to use the model as a viable probe for exploring cognition and behaviour under anxiety.

Researching abroad: Cannabis and decision making in the Big Apple

by Michelle Taylor @chelle_bluebird

Setting off for TARGs 2013 annual retreat to Cumberland Lodge in Great Windsor Park, I was looking forward to hearing a talk from an invited guest speaker. Gill had flown in from Columbia University to talk to the group about a recent drug administration study her lab group had completed. The research being conducted by their lab was very different to the epidemiological research that I am used to. Now don’t get me wrong, I thoroughly enjoy the research that I do, but these studies sounded new and exciting. After listening to the talk, the evening activities began with dinner and a quiz. Luckily, I ended up on the same quiz team as Gill, giving me the opportunity to ask more about her research. I decided to grab the bull by the horns and offer my help in one of her future studies, and so my trip to the Big Apple began…

central park 1Nine months later I was on my way to Heathrow for a two month stint collecting data on a cannabis administration study. I was both excited and apprehensive. I have never lived more than a 3 hour drive away from family, and have always been in a city where I have known people. I didn’t know whether I would get homesick, or whether I would make friends on my trip abroad. These feelings of apprehension soon disappeared in the first few hours of my first day at the New York Psychiatric Institute. Everyone I met was so friendly and welcoming, even the many morning commuters who stopped to help the lone Brit who was obviously puzzled by the subway map at 7.30am.

yankeesI was to spend the next six weeks collecting data for a study examining the neuro-behavioural mechanisms of decisions to smoke cannabis at the Substance Use Research Center in the New York Psychiatric Institute at Columbia University. This research centre is unique; it is one of the largest drug administration centre in the world and has licenses to administer a wide variety of drugs, including cannabis, cocaine and heroin. This means that much of the research conducted here is cutting edge. The aim of the study that I would be working on was to shed light on how and why drug abusers repeatedly make decisions to take drugs despite substantial negative consequences. The study used brain imaging (fMRI) to examine the neural and behavioural processes involved in decisions to self-administer cannabis, compared to decisions to eat food, in regular cannabis users. We also examined the influence of drug and food cues on the processes underlying these decisions. To do this, participants were recruited as inpatients and stayed with us in the lab for a week. Data collection for this study is still ongoing, but I will be sure to write another blog post with what we found when the results are available.

coney_2I found this research fascinating and it was a pleasure to be involved in the work carried out in this department. The experience was made even more enjoyable by the people I was working with. There were many office conversations about the British and American slang that was being used, many lunchtime trips to Chipotle (an American fast food restaurant that I am definitely missing since my return to the UK), and several Friday evening trips to the local Irish bar. One office memory that will always stick in my mind was meeting a very accomplished researcher in the field of my PhD, a researcher that was definitely someone I should be impressing. Upon entering this individuals office on an extreme
ly hot New York day, the fan was turned to the meeting area and the smell of cannabis filled the room as the flow of air reached me (I had been administered the drug to a participant earlier that afternoon). Probably not the best first impression I have ever made!

milkshakeI did, of course, take every opportunity to explore New York. I was lucky enough to get tickets to watch the New York Yankees beat the Boston Red Sox at the Yankee Stadium, which was also one of the last games played by baseball-legend Derek Jeter. I made several trips to the American Natural History Museum (my favourite type of museum, and this one cannot be done in a day), and while there saw a live spider show, a 3D film about Great White Sharks and a full T-Rex skeleton. The glorious weather allowed for several leisurely strolls around Central Park. And, of course, the American food definitely needs a mention. If anyone reading this takes a trip over the Atlantic, I would definitely recommend visiting Big Daddy’s Diner for what could be the best milkshake on the planet. And don’t be shy about trying a hotdog from one of the carts that can be found on nearly every street corner. The reason there are so many of them is that they’re delicious! I would also recommend a trip to the Russian Tea Rooms for caviar afternoon tea, an evening at the New York Metropolitan Opera (if that’s your cup of tea), and a trip to Coney Island.

t_rexAlthough it was daunting going abroad for that length of time to begin with, I don’t think I would be having those feelings again and I would definitely jump at any opportunity to work in a different environment in the future. I am very grateful that I am a PhD student in a large working group like TARG, as without this I probably would not have come across opportunities such as this one. This experience has taught me the importance of inter-disciplinary research, and the need for several fields contributing evidence to a much larger research question. Since this trip, I have been successful in a fellowship application allowing me 9 months in a different department at the University of Bristol, an application that I probably would not have made if it wasn’t for my experience at the Columbia University. I am an epidemiologist and do not have any plans to change that; however I do plan to conduct more interdisciplinary research in the future. I would like to that Gill (and everyone in her lab group) for welcoming me and making this trip possible. I look forward to hopefully working with you again in the future…