Sunday, 6 March 2011

Antidepressant prescribing in England: variation may not be as great as Guardian map suggests.

The Guardian published a story on March 5th, 2011, showing that the crude rate of anti-depressant prescribing in some areas of England was more than 3 times higher than in others.The highest rate of prescriptions was in Blackpool, and the lowest in Kensington and Chelsea.
The data was from publicly available sources and is linked to from the Guardian website. The journalists give this description of their method :
"How did we arrive at our figures? First, we gathered prescription data from the online database managed by the NHS Information Centre.

This quarterly information was compiled to get annual numbers covering 1 April 2009 to 31 March 2010 – the most recent full year with available data. In order to make the numbers comparable, we then linked the raw prescription numbers to the ONS mid-year population estimates.
This allowed us to calculate the prescriptions per 100,000 figure in the data below, which controls for the different sizes of PCTs, if not their different levels of wealth, employment and general illness."

The data is also not age standardised. In the comments section some have suggested that in areas with low numbers of prescriptions doctors may be prescribing several months prescriptions at one time. I have looked at the most recent data set available (July-August 2010). Data is available on the total cost of the prescriptions as well as the number of prescriptions and these are plotted below.

This shows a clear correlation between number of prescriptions and cost. However when one looks at individual PCTs the average cost of an antidepressant prescription in Kensington and Chelsea PCT is £7.01, whilst in Blackpool PCT it is £3.48. This does suggest that either more expensive antidepressants are being prescribed in Kensington and Chelsea, or more months prescriptions are being given at one time. Since the absolute number of prescriptions in Kensington and Chelsea is so much lower than in Blackpool, it may be that more antidepressants are being prescribed in each prescription. 

When looking at datasets it's good to make use of all that is available.

You can find my spreadsheet here.

EDIT 4.20pm 6/3/11 I've calculated a rate of prescribing of antidepressants per 1000 of population over 19. (This is not ideal as some teenagers may be prescribed antidepressants, but more accurate than using the total population including children). This has been plotted against average cost of antidepressant prescription. This shows that Blackpool and Kensington and Chelsea are outliers. 

There is not a strong relationship between cost of prescriptions and number of prescriptions. This may explain some of the variation between north and south of England, but the long-established relationships between deprivation and depression are likely to have greater explanatory power.

Here is a link to some papers on that relationship and a very interesting report (via @coxar) on the relationship between antidepressant prescribing, poisoning by antidepressants and deprivation.


  1. Good work. Recently we audited the uptake of CBT in my practice. There was a wide variation in the numbers referred for CBT by different partners. Interestingly, the referrals from a registrar topped the list by far, but the uptake amongst his referrals was nearly 100%. In contrast, that from the partners was only about 30-50%. Hard to draw any firm conclusion as the sample size was small.

  2. You are right, prescribing data is to crude a tool when used on its own to jump to the conclusions they have. Prescribing data is an extremely useful 'pointer'.

    Variable include choice of drugs, cost of those chosen, generic rate, access to other therapies such as CBT, socio-economic status of the area, age profile, rurality, access to prescribng support etc etc etc.

    Saying that, I wouldn't actually be that suprised if there was a difference between prescribing in the north compared to the south, but would be if it was as marked as suggested.

    The cost variation could be as result of prescribing intervals (ie 2 or 3 months worth in one go) and the use of newer more expensive therapies (pardoxically the newer they are the less they are proven!).

  3. Could it be that less people in Blackpool pay for prescriptions as they're on the NHS?
    The cost of a prescription it the same regardless of where you are in England... £7.20

  4. Thanks for the comments.

    To anonymous, when I talk about the cost of the prescription I mean to the NHS rather than to the end-user. The number of people paying for prescriptions in a given area is not taken into consideration when calculating this statistic.

  5. Had a look myself. NIC probably best to use, as av total has other assumptions (e.g. discount)? Also interesting to look at volumes in relation (and cost here) to population, if possible in terms of relationship to population at risk (PU, ASTRO-PU). Is it a primary or secondary service driven issue? If you know catchments of local secondary care, might be worth a look?

    As others have commented, could be down to duration, but most areas trying to push 1/12 only. As the diffs are large with NIC, it's not to do with who pays. Might even be partly driven by initial private sector consults - they'll prescribe the "latest" drugs no doubt?

    Interesting to see who locally picks up these issues once PCTs are abolished?

  6. Thanks Ian. What is the difference between NIC and total? They seemed pretty similar to me and I didn't know that one was better to choose than the other.

  7. Pauline Sweetman6 March 2011 at 14:38

    In my experience as a prescribing adviser, typically, in an area where people are on benefit and do not pay for their medicines it is common for prescriptions to be given for only one month's supply, to minimise waste. This may be the case in the north. Another factor that would bring the cost down would be the use of generic medicines.
    Although everywhere we are trying to increase generic prescribing and reduce the prescription length, where people are paying for their prescription it may be better to prescribe a longer supply to ensure compliance, especially if the patient is expected to be on a medicine for many months. Repeated prescription charges do not assist compliance. It is unlikely (though possible) that generic prescribing may be lower in Kensington & Chelsea, but this could be checked via PACT data (and as Ian has said, PU, ASTRO-PU figures would also be useful). A proportion of the wealthy are likely to be obtaining their medicines privately which would also affect the statistics.


I am reintroducing word verification to cut back on spam posts. I'm sorry if you find it frustrating,