Epidemiology – Numbers Needed to Treat (NNT)


NNT = the number of patients that need to be treated in order for 1 extra patient to benefit

Alternatives to NNT include:

  • Numbers Needed to Screen (NNS = No. needed to be screened for 1 to benefit)
  • Numbers Needed to Harm (NNH = No. needed to be exposed to a risk factor for 1 to be harmed)


  • NNT = 1/ARR
  • Absolute Risk Reduction (ARR) is calculated by the difference between the rate of event in controls and the rate of event in cases = (a/a+c) – (b/b+d)
  • NNTs should always be reported with 95% Confidence Intervals for interpretation


  • The lower the NNT the better.
  • E.g. Drug FAB helps prevent strokes and has an NNT of 1.  By treating Bob with FAB this should prevent him having a stroke.  On the otherhand, drug BAD has an NNT of 50, so you would have to treat 50 Bobs in order to prevent one stroke.
  • If the treatment or exposure if harmful (i.e. result is a negative number) the omit the sign and measure is renamed as NNH


  • Useful to communicate benefit and harm – easy to understand (risk communication)
  • NNTs can be used either for summarising the results of trials
  • A clinically useful measure of the relative benefit of an active treatment over a control (better than RR or OR)
  • Takes into account the frequency of the outcome – thus reflects the ublic health impact of the intervention


  • Cannot be used for performing a meta-analysis. Pooled NNTs derived from meta-analyses can be seriously misleading because the baseline risk often varies appreciably between the trials
  • Do not compare NNTs for different therapies *unless* the baseline risks of the disease are similar…

For further info, check out http://www.thennt.com/the-nnt-explained/


Epidemiology – Odds Ratio (OR)


The Odds Ratio is a measure of association which compares the odds of disease of those exposed to the odds of disease those unexposed.


  • OR = (odds of disease in exposed) / (odds of disease in the non-exposed)


I often think food poisoning is a good scenario to consider when interpretting ORs:  Imagine a group of 20 friends went out to the pub – the next day a 7 were ill.  They suspect that it may have been something they ate, maybe the fish casserole… here are the numbers:

(not ill)
(ate fish)
5 3 8
(didn’t eat fish)
2 10 12
7 13 20
  • Odds of exposure in cases = a/c = 5/2 = 2.5
  • Odds of exposure in controls = b/d = 3/10 = 0.3
  • Odds Ratio = (a/c) / (b/d) = 2.5/0.3 = 8.33

Interpretation: What does this mean?

  • OR of 1 would suggests that there is no difference between the groups; i.e. there would be no association between the suggested exposure (fish) and the outcome (being ill)
  • OR of > 1 suggests that the odds of exposure are positively associated with the adverse outcome compared to the odds of not being exposed
  • OR of < 1 suggests that the odds of exposure are negatively associated with the adverse outcomes compared to the odds of not being exposed.  Potentially, there could be a protective effect

In the example above, we can conclude that those who ate the fish casserole (exposure) were 8.3 times more likely (OR = 8.3) to be ill (outcome), compared to those who did not eat the fish casserole.  Of course this is an entirely ficticious example, and I have nothing against fish


  • Appropriate to analyse associations between groups from case-control and prevalent (or cross-sectional) data.
  • For rare diseases (or diseases with long latency periods) the OR can be an approximate measure to the RR (relative risk)
  • Doesn’t require denominator (i.e. total number in population) unlike measuring risk
  • Good method to estimate the strength of an association between exposures and outcomes


  • Association does not infer causation! *epidemiology golden rule*

Exam Techniques

This week I attended: “Part A Revision Seminar: How to Answer Questions!”

This was a useful session which (thankfully) built up my confidence and also provided some useful practical tips, which I thought might be helpful to share.

(I would like to acknowledge and thank Kirsteen Macleod (ST5) for sharing her wisdom and collated advice – the following information have been taken from her presentation)

ARGH I’ve got to answer a question – how do I start?

  • Identify what aspect of the curriculum the question covers
  • Do a brain dump of all ideas/knowledge/critiques… e.g. spider diagram
  • Think laterally – why is this important to public health? Have wider determinents been considered?
  • Identify an appropriate structure (something logical, e.g. framework)
  • Write a structured answer with bullet points, neat handwriting and headings
  • Use examples to illustrate your answer

What (other than knowledge) is essential to PASS the exam?

  • Be vigilant with timing (you must attempt ALL questions)
  • Write legibly
  • Plan your answers
  • Structure sensibly
  • Work logically through all aspects of the question – be sure to ANSWER THE WHOLE QUESTION

*UPDATE* Health Knowledge have uploaded a page of useful information for Part A preparation here

Hot Topic: Advertising of Junk Food

Context for this post

As part of our Part A preparation, each week we write a briefing on a current ‘hot topic’ which has public health implications.  A useful document to guide choice of ‘hot topics’ has been the Faculty of Public Health  manifesto “12 Steps to Better Public Health”. Below is my briefing plus wider discussion on the advertising of junk food on TV.

FPH Manifesto Summary 

  • A recent Which? report criticised the 2006 Ofcom measures to ban junk food advertising between programmes where 20% of the audience were <16
  • The current measures are ineffectual and fail to cover programmes such as soaps which are still watched by large numbers of young people.
  • A complete ban before the 9pm watershed is needed to effectively reduce consumption of salt, saturated fats and sugars by children and adolescents, reducing the risk of cardiovascular disease later in life.

The problem

  • According to the Department of Health, £838million was spent on food and drink advertising in the UK in 2007.
  • Junk food marketing contradicts messages about healthy eating, undermining children’s ability to choose better food and parents’ efforts to feed them healthily.
  • To combat the rise of childhood obesity it is vitally important that children are persuaded to eat more healthily.
  • Food Standards Agency suggest junk food marketing directly influences children’s food preferences, and indirectly influences what family and friends consider to be a ‘normal’ diet. 

Current regulations

  • Prevent the advertising of “less healthy” products during children’s TV programming. 
    • 70 percent of the television that children watch is outside the hours of ‘children’s TV’ that these rules cover,
    • Research from Which? in 2007 reported that that 18 of the 20 most popular programs watched by children under 16 were not be covered.
  • Products must pass the FSA nutrient profile model prior to Ofcom giving permission to advertise.
    • Points are given for fat, salt and sugar.  If the score is too high, the product is not allowed to be advertised.

A solution being lobbied for is a 9pm watershed for junk food adverts, which has been suggested to have the following benefits:

  1. Protect children: This will eliminate over 80% of instances of kids watching junk food TV advertising.  The current rules only protect children from half this much advertising.
  2. Support parents: It will provide clarity on when junk food adverts will be shown, allowing parents to exercise responsibility over whether their kids see such adverts.  
  3. Improve children’s health: Estimates show that the health benefits from a 9pm watershed for junk food TV adverts will save the nation up to almost a billion pounds a year – at a cost to industry of  £130 million a year, and no cost to the government.

A series of surveys have demonstrated that the majority of parents are in favour of a protecting their children from junk food advertising:

  • BHF survey found that 68% of parents were in favour of pre-9pm junk food advertising restrictions, with only 7% against.
  • Which? found in 2006 that 79% of parents believe unhealthy foods should not be advertised during the times children are most likely to be watching television.

There are currently no legal restrictions on non-broadcast junk food marketing aimed at children. 

  • This category includes marketing through sponsorship, packaging, text messaging and the internet. 
  • This is a growing form of advertising aimed at children and its omission from statutory regulation is a loophole often exploited by food companies.

Therefore… there appears to be a lot of support for a 9pm watershed to be put into place, however what true effect this will have will be difficult to measure because there are so many sources of influence upon the perceptions and desires of children.  Although I feel supportive of this campaign, it’s underlying premise suggests that exposure to adverts is the most important factor… what other influences could be important?  Have these been considered?  E.g. previous exposure to different foods (e.g. junk vs. healthy), parenting styles, friends/peers at school, ease of accessing healthy food…etc.  In essense, my point is that this campaign will only be worthwhile if it is complimented by other relevant actions which encourage behaviour change as realistically we do not know which influences are the greatest (and influences are likely to be different for different individuals).  This policy is a step in the right direction, however when one considers that the sponsors for the 2012 Olympics include Coca-Cola, Cadburys and McDonalds (and no companies representing healthy consumption) one cannot help but be skeptical.

References:  http://www.sustainweb.org/childrensfoodcampaign/junk_food_marketing/

Summary of PHTwitJC #2: Prostate-Specific Antigen Concerntration at age 60

Participants and followers of #PHTwitJC voted on their preferred topic to discuss and critique, from a selection of case-control papers put forward.  It was intended for this paper to be discussed on 28th August, however due to low participation it was re-run.

On Sunday September 11th #PHTwitJC discussed a paper by Vickers, A et al ((2010) Prostate specific antigen concentration at age 60 and death or metastasis from prostate cancer: case-control study).  The transcript of discussions can be accessed here.

Q1 – This paper uses a case-control study design – was this appropriate for this topic?

Participants agreed that a case-control design was appropraite as the outcome was suitably rare (death from prostate cancer).  Cases and controls were identified from a previous cohort study (otherwise known as a nested case-control study design); it was suggested that using a clearly defined closed cohort strengthened the design.

Q2 – Did recruitment for cases and controls affect selection bias?

It was unclear from the paper how participants were recruited in the original cohort study, therefore it is difficult to ascertain the probability of selection bias.  It was questioned whether a bias was created through the different inclusion criteria for cases and controls (i.e. cases were dead, but controls were all alive). This is likely to cause a sample selection bias within the control group by discounting those who were dead.  Participants discussed what the implications of this bias would be upon the effect size.

Q3 – What confounding factors could be present?  Have these been accounted for?

A number of potential confounding factors (such as socio-economic status (SES), ethnicity, family history and other health conditions) were not accounted for in this paper.

Q4 – Study concluded that “a single measure of PSA at age 60 is associated with a man’s lifetime risk of death from prostate cancer” – are these results valid?

Participants questioned whether the results of this study were clinically useful, for example

“Stats hold up but can you predict when someone needs treating ie at 60 or 80? Would treat at 60 prolong life?” (@TwitttwooSue)

Participants did not feel confident that a single measure can infer lifetime risk of death from prostate cancer, especially as the study does not consider other causes of death (sample selection bias).  It was noted that the authors used different thresholds for different parts of the analysis, which can be considered problematic.

Q5 – What are the PH implications of this study? How might these influence future practice/policy?

It was noted that other RCT evidence have demonstrated that prostate-specitic antigen can identify cancers, but it has not been shown to prolong life (all-casue mortality).  Although the authors stated that the findings should inform screening programmes, it was noted that this was not discussed in the paper.

A query was raised questioning why investment in research for evidence to support prostate cancer screening continues, despite evidence to date suggesting that prostate cancer is not suitable to a screening programme.  @carotomes wondered whether there was political weight attached to the research due to the lack of male screening programmes, however @TwittwooSue pointed out that the male-only screening programme for Abdominal Aortic Aneurysm (AAA) screening is due to be launched soon.

Overall it was felt that this paper provided insufficient evidence to support the use of PSA as an effective screening tool for prostate cancer.  #PHTwitJC participants agreed that this paper shouldn’t influence change in public health policy or practice.

CA Jan 2003 Past Paper: Antidepressant drugs and generic counselling

In this week’s study session we (5 registrars in the region who are preparing for the January exams) we discussed our critical appraisals of the following paper: Chilvers et al (2001) Antidepressent drugs and generic counselling for treatment of major depression in primary care: randomised trial with patient preference arms, BMJ

Q1 required candidates to write a structured abstract, which is pretty difficult to do after reading the whole paper, including the abstract… I shall attempt this next week!

Q2: Critically appraise the paper, paying particular attention to: the methods (including statistics), choice of outcomes, results, and conclusions drawn


  • Although GP practices were randomised, only patients who self-selected to join the randomisation group were randomised.  These patients tended to be less severely depressed and their characteristics may have differed from those refusing randomisation (such as levels of motivation, previous experience…etc)
  • GPs recruited participants which may have provided an element of recruitment bias.  Did GPs have preference for a particular treatment?  No details provided or discussion of ways to avoid recruitment bias.
  • Authors acknowledged recruitment to be difficult (and even adjusted their power calculation and study numbers as a result).  No baseline figures of number of patients approached/refused were provided… not assured this is a representative sample of the general population with depression.
  • Statistics: not all baseline characteristics were reported in table 1.  Chi-square and Fisher’s Exact test both appropriate for small sample size.


  • Paper title suggests an investigation of patients with diagnosed major depression, whereas throughout article reference is made to mild/moderate depression.  Furthermore depression is diagnosed by ‘research criteria’ for the study which limits it’s application to the real-world
  • Problematic use of the term “generic counselling” – what does that mean?  Full treatment detail is really important, especially if this paper hopes to inform commissioning of services.  Again – lack of information limits the real-world application
  • Also, although study details that GPs were provided with protocol for anti-depressents, no attempt to assess the efficacy of this, or monitor how GPs managed patients on anti-deps…etc GP consultations are in themselves interventions.  Many unknowns regarding treatment process that have important implications for considering the paper’s conclusions…
  • Paper assumed that if patient was well at 8 weeks and again at 12 months, they were in remission throughout that period.  Given the nature of depression, this is an unfounded assumption to make as most patients will experience a rise and fall of symptoms throughout their recovery.  Authors refers to MANY assumptions, which is suggestive that they did not have appropriately defined periods of follow up (nor, as mentioned above, did they assess frequency of GP consultations as part of treatment).


  • Reports no systematic different between proportion of patients asigned to different trial arms – however no test of hetergeneity for the baseline charactertics (age, sex, ethnicity)
  • Patients choosing counselling reported to do better compared to those randomised to counselling – but 95% confidence interval reported as 0.0 – 9.2 ….. this is fairly wide considering a mean difference of 4.6, plus a lower limit of 0.0… is this considered as ‘crossing zero’ and therefore increase the likelihood that this result could be due to change?


  • Reported conclusion suggested that “12 months after starting treatment, generic counselling is as effective as anti-depresants”… however I disagree as firstly) the term ‘generic’ counselling is meaningless, secondly) not clear what severity of depression was being treated through this trial, and thirdly) the authors have omited to specify which type of anti-depressants were used… the lack of specific details make it difficult to agree with this conclusion
  • Not confident that confounding factors were well accounted for e.g. counselling has a gender bias, self-selection vs. willing to be randomised may have an effect on time to remission due to personal motivation…
  • Article further concludes that both treatments (again, lacking specific details of what treatments were?) are effective (again, lacking specific details of treating what type of depression? and defined using criteria which is not applicable to general practices).

(NB: Ideally… I should be preparing my critical appraisal under timed conditions, and will do in future.)

Q3 A review group is being pressed to fund the expansion of counselling services.  What would be your response; what additional information is needed?

As a group, we struggled to identify a framework to structure the answer to this question… any ideas?

In order to review a funding application it would be important to have access to the following sources of information:

  • Defintion of the condition being considered (is the expansion specifically for patients with depression? Counselling can be used to address a number of problems)
  • Epidemiology of the condition (including local incidence and prevalence, so trends and local needs can be considered) and local projections
  • Literature review / NICE guidance to help inform how best to commission services for the condition and ensure the evidence base is known and applied
  • Full details of the current services available (including numbers attending services for this condition, waiting lists if available (as this would evidence unmet need))
  • Funding application ought to include cost-effectiveness analysis whereby a return on investment can be demonstrated, with clear timelines as to when that ROI would be expected

My response would be to firstly identify information needed (as above) and collate this into a report which has a particular focus on the evidence-base, local need + projections, and cost-effectiveness of counselling to address this condition.  It would be important to involve stakeholders in this process of collating information, but also to present the report findings to the group reviewing mental health services to ensure that any altered commissioning arrangements are based on evidence of need and consider the bigger public health picture

9/11 – Reflections from a Public Health Perspective

It’s September.  This is the month when the critical appraisal papers *generally* are chosen for the January Part A FPH Exam, *typically* from the BMJ… (so I’m making a mental-note to review papers which were published at the end of the month).  But this isn’t just any old September…

This September marks 10 years since the ‘9/11’ attacks.  On Tuesday September 11th 2001, four coordinated suicide attacks by al-Qaeda were carried out on the United States.  Reflections on this 10 year anniversary have started to creep onto all the news channels, and I assume these reflections will be sustained until the anniversary day (a week today). But how can a public health perspective contribute to such reflections…?

Handily, The Lancet have today published a series of articles which examine the health consequences of the 9/11 events.

The editorial “9/11 – Ten Years On” observed that the US governmental responce focused on defense, security and emergency preparedness.  Not only did this encourage fear and anxiety amongst the general population, it stigmatised many muslim communities and individuals.  Furthermore, health was pushed onto the backseat of political agendas, and the author suggests that:

“.. 9/11 was a huge opportunity cost for the health of the American people.”

As well as US domestic health effects, the 9/11 events have had international consequences.  One postitive outcome has been an increased post 9/11 commitment to global health encompassed by the US national security strategies.  The US recognised that investing in global health (amongst other aspects of development) had positive outcomes in terms of stability and security for their country. 

The 9/11 events, responses and consequences have provided a lot of reflections and learning for public health; from the micro (such as individual health protection, occupational health, suicide attacks), to the macro (health policies, emergency preparedness and reponse).

Articles of interest include:

“The events of 9/11 not only represent an example of a local act with global consequences, but also an instance where poverty and perceived injustice can contribute to catastrophic global instability and insecurity. It is now abundantly clear that human-made crises will, if not resolved decisively through politics and diplomacy, create the conditions for human-made disasters.”