MPH Master pieces

More on alcohol tax

The USA is a great place to study the effects of legislation because there are 50 states, they all have different laws and they keep changing their laws. So plenty of data to study. Here's one on alcohol tax. The detailed statistical analysis is pretty mind boggling but the graphs speak clearly.

Better than expected

When I was a director of public health, our district had the lowest death rate from cardiovascular disease in the country and one of the lowest teenage pregnancy rates. Tribute, you might think, to the excellence of our public health endeavours. Except that we were also the most affluent district in England; much of our success was due to the favourable social milieu. But that still left a question - were we doing better than expected given our affluence?

Here's a paper that asks the question for infant and perinatal mortality. Deprivation, ethnicity and maternal age allow you to estimate pretty accurately what a PCT's infant and perinatal mortality will be. No district managed to beat its milieu, but a couple were markedly worse.

The authors offer no comment on why these two were poor, beyond saying that it wasn't a one-off. The report doesn't give raw data, and you can't even tell from the report whether the outcome is based on a single year's mortality data or more, which would seem to leave chance as a possibility.

Checking out the data source they cite (NCHOD) doesn't help because of recent mergers. The nearest match to one of the two is Wyre Forest, analysed as the district not the PCT. It had an infant mortality rate of 5.2 per 1000, based on 16 deaths over 3 years. This gives a confidence interval of 3.2 to 8.4. The rate for England was 4.9 per 1000. Hm!

Note that we are trying to use predictors to guess an outcome which is a number (of infant or perinatal deaths) so it's a Poisson model.

For a first-hand account of working in a performance outlier, see this entertaining article.

An impressive school programme

Might be worth checking this research out in person - the study area is Hawaii!

Solid research, strong and lasting results on substance use, violence and sexual behaviour e.g. 'carried a knife to use to hurt someone': 9% in the control group and 2.7% in the intervention group.

Note the details of the intervention - school-wide, family and community components, lasted from kindergarten through 12th grade. The authors also point out that the results really only start to bite after three years (message to research funders - stay with it for 5 years or more!).

Just one fly in the ointment - the authors (a) use one sided significance tests, (b) cite 90% confidence intervals instead of the usual 95% and (c) make no adjustment for multiple comparisons. Their results are impressive so why such dubious statistical practices, all of which will tend to overstate statistical significance?

Forty years later...

I love papers that report long term follow up. So here's one from the Perry Pre School project.  This was the definitive early intervention project which prompted the nation-wide Sure Start programme in England.

A couple of things caught the eye in this report - (1) the subjects really were very disadvantaged, with IQs in the range 70 to 85 and (2) the intervention really was very intensive, with 2.5 hours of 'interactive academic instruction daily' for two years, from teachers who all had a masters degree and had completed training in child development. 

Only 4 of the original study group of 128 children have been lost to follow up, though 7 have died: 2 in the intervention group and 5 in the control group. That's a lot of deaths when you remember that these people are still young - aged 40.

The other startling result is that 23% of the intervention group (and 29% of the control group) have used cocaine. I may have read that wrong since the authors don't deem it worthy of comment. This is not my field of experttise but I think recent figures from New York show cocaine use well below 5%, so it's not that 25% useage is normal for the USA.

This particular paper shows little effect of the intervention on health at age 40 but there is no doubt of the long lasting beneficial effect of the intervention in reducing crime and poverty in the intervention group.

Childhood illness in Bangladesh

 This study is worth mulling over. It's an attempt to reduce childhood mortality using a strategy of health worker training, health systems improvement and family and community activities. This is a strong model compared to the hit-and-run approaches we sometimes use.

The communities in which it was implemented had an infant mortality rate around 50 per 1000 (infant mortality in the UK is around 5 per 1000). The study excluded deaths in the first week of life; among children aged 7 days to 5 years half of the deaths were from injury (25%) or respiratory infection (22%). A further 8% were from malnutrition and 1% from measles. The IMCI intervention was aimed at diarrhoea, pneumonia, malaria, measles and malnutrition and so in principle targetted about 60% of the causes of death.

The actual intervention started out rather modestly - a two day training course for village health workers. This was beefed up after early evaluations. Specially recruited workers were given a 15-day training plus 'essential commodities'.

But of course the government of Bangladesh was not standing still so plenty happened in the 'non-intervention' communities. The death rate dropped 8.6% in the intervention areas and 7.8% in the comparison areas.

The same problem affected the big trials of community intervention on cardiovascular disease in western countries in the 1990s: big drops in the intervention areas, but big drops in comparison areas too, so no significant difference between the two.

More teenage sex

Here is a paper that got a lot of press coverage. At first glance it appears to show that an intervention designed, among other things, to reduce teenage pregnancy had the opposite effect. With hindsight it seems obvious: bring together a bunch of rebellious teenagers and they'll end up having sex; what's more it'll be unprotected sex.

An accompanying editorial provides some context. Britain had the highest teenage pregnancy in Europe; a programme in America seemed to reduce teenage pregnancy by half; who wouldn't want to try it over here? Two problems with this.

Firstly what is the 'it' we want to try? What exactly do we replicate? Trial programmes are run by enthusiasts; results always degrade with wide spread (in our case nation-wide) implementation. You can't replicate enthusiasm.

Secondly the headline result from the USA provided only a partial picture from six of 12 pilot sites. Results from all 12 sites showed no benefit.

Even so, reversal of effect is unexpected, and we must congratulate the Department of Health policy research team for commissioning this evaluation.

Meanwhile over at the Times newspaper, an editorial ponders the difficulty of basing policy on evidence. It's not as easy as you think.

Death in Siberia and alcohol

Life expectancy in Russia has swung wildly up and down over the past 20 years. Here is a massive study, based in three Siberian cities, which shows how much of this is due to alcohol. The short answer - a lot.

The study used a simple case-comparison method, but on a massive scale. Given the effort involved there must be many more papers to come.

Dementia - concurrent validation of a test

 This paper reports a new 'screening' test for Alzheimers disease. More below on use of the word 'screening' in this context.

Good points:

a. validated against people with a firm clinical diagnosis, made by experts, of Alzheimers disease (i.e. concurrent validation, against a 'gold standard');

b. ROC comparison against the current screening test - the Mini Mental State Exam. The correct way to compare two screening tests is the ROC curve.

c. age specific norms published - i.e. what 'normal' people score on this test.

Although the test is described as a screening test I would classify it as a clinical algorithm, given its intended use in hosptial and other clinics. I certainly hope it won't be used in the general population without more validation, though unfortunately the authors have put the test on a website for anyone to try. 

There is some debate about whether tests like this are 'screening' or not.  Both population screening and clinical testing start with a prior probability of disease which changes to a higher (or lower)probability after the result is known. In a clinical setting the prior probability of disease is higher: the patient has come to the clinic because something is wrong.

On the other hand the ethics are certainly different. In population screening we invite ordinary people who are happily going about their business and tell them they may have some serious disease. With clinical tests, the doctor is sifting through the possibilities for someone who has asked what is wrong with them.

Salt!

Here is an excellent summary of the scientific position on salt. Skip the tables and read the text. Like flu, this is a 'must know' for anyone in public health.

Home births

A great theorem in mathematics proves rigorously that some things can never be proved rigorously. It seems that in public health too we will never know the answer to some important questions, no matter how much they are researched.

The question of whether home births are safe has been around for twenty years or more, but still the debate rumbles on. Here is the latest attempt to answer the question; and here is an editorial shredding the attempt (though some of the criticisms seem unfounded to me).

Strictly speaking the paper compares outcomes in a group delivered at home by independent (i.e. non NHS) midwives versus a group delivered in NHS hospitals. Note that most home deliveries are by NHS midwives so this study looks at a specific detailed question - the independent midwife - rather than the general problem of home birth. 

And of course the problem is that no group can be found to match the sort of woman who wants a home delivery by an independent midwife. To my eye the two groups look comparable on most measures (Table 1) but the expert editorialists think not; and the authors themselves comment on some important differences in their discussion section.

So in the end (a) we find differences but (b) we have not compared like with like, leaving us no wiser than before.