You are here

Patterns of Prescribing (Part 1): Making the Switch

Understanding data can help to make prescribing safer, better, and more cost-effective.

For patients and the general public, a better understanding of the options for treatment can help them make changes that will benefit the quality, safety and cost of care. Prescribing forms a part of this, along with other non-medicine treatments and changes. Patients are often mystyfied and sometimes resistant to change in the drug treatment that they are currently on. If can be demonstrated that they will receive the same, or improved, quality of treatment that will achieve a cost-saving for the health service, they will be more receptive of change.

For prescribers, awareness of the different options is key. It is possible to examine the prescriptions patterns of individual practices so that clinicians can identify how they compare to other practices, and address how they might make positive change to their prescribing choices.

In this post, we examine a couple of precribing areas where cost savings can be achieved. Other prescribing practices are important too, not just cost-cutting. In the next post, we'll examine how prescribing patterns affect patient safety and quality of care, including what the data tells us about how antibiotics are prescribed.

Can any money be saved?

Northern Ireland's annual bill for prescriptions written by GP practices is around £400m. This is less now than it was in 2010, when £417m was spent, in the context of overall rising demand for care. 

Despite an increasing number of prescriptions of all kinds being dispensed, the cost of the whole GP prescribing bill (non-inflation adjusted) has decreased overall since 2010. There are a number of factors influencing this broad measure aside the number of packages dispensed, mainly which drugs were prescribed, and how much they cost over the counter.

The Public Accounts Committee published a report on Primary Care Prescribing in 2015, finding that Northern Ireland prescribes more items per patient, and had higher levels of prescribing of more expensive drugs, when compared to England, Wales and Scotland. The NI Audit Office had also examined spending and found scope for more cost-effective prescribing. The PAC recommended some changes for the health service to implement - that a more proactive approach to examining prescribing patterns, identifying savings, and to benchmark GP practices in order to achieve greater accountability (not just limited to prescribing).

The health service takes steps to find efficiencies in primary care prescribing, led by the Pharmacy Efficiency Review Team. This includes making particular recommendations for certain medical treatments in the NI Formulary. The Health and Social Care Board has estimated that by the end of the current financial year (2016/17), savings totalling £194m will have been realised since 2010 through changes in prescribing.

HSCB estimate of efficiency savings from generic switching

Source: HSCB

Where can we find the data?

Of course, in order to examine what is being prescribed, when, by whom and at what cost, we need to have access to detailed data.

For English practices, the OpenPrescribing.net project (by the EBM Data Lab at the Nuffield Department of Primary Care Health Services, Oxford) examines a number of areas where prescribers can make efficiency savings and improvements in health outcomes through their choice of prescriptions. 

Northern Ireland practices are not included in the OpenPrescribing charts and dashboards, however, HSCNI Business Services Organisation (BSO) publishes GP prescribing data on a monthly basis for every individual practice in Northern Ireland on the OpenDataNI portal. The data is openly licensed, meaning that anyone can access and use it as they wish.

Taken together, the data totals over 19 million rows and is over 2GB in size. At the time of this post, October 2016 is the most recently published dataset, so data is examined up to then.

Below we look at a couple of specific examples where cost savings can be achieved from switching the drug prescribed rather than withdrawing its use altogether.

In general, in examining prescription data keep in mind:

  • There may be appropriate clinical reasons for preferring one option over another, even when evidence for cost-saving is clear. The choice of what drug best suits a patient is, ultimately, the discretion of the clinician in consultation with their patient.
  • There are more influences on the overall cost of prescribing than what is dispensed and in what quantity. Medicine costs change regularly and prescribing patterns can be influenced by market effects. Population effects, new or altered medical treatments and disease prevelance will also have effects.
  • Differences in patient cohorts (or therapeutic groups) may result in different prescribing requirements between one practice and another, and a demographic shift may have an effect within a practice over long periods of time. There are some methods for statistical adjustment that allow for more meaningful comparisons. These have not been applied to the analysis below as it is a simple measure of the proportion of one drug against a very similar alternative.
  • If you really are interested in this, and aim to work with the source data yourself, it is worth reading the FAQs and Explanatory Notes.

Cerazette® and Desogestrel

One of the most significant ways in which money can be saved from medical prescriptions is switching from expensive branded products to much cheaper non-proprietary (generic) alternatives. This is especially true when the 'alternative' is exactly the same medicine and compound. This is the case with desogestrel, a contraceptive pill that exsits in its generic form, and as a proprietary brand called Cerazette.

For most patients and doctors, this simple switch is a viable option. In a small number of cases, patients may find the alternative drug unpaletable, and be less likely to take it when prescribed.

OpenPrescribing includes Cerazette vs. desogestrel as part of its selected performance measures. There is another branded form of desogestrel prescribed in Northern Ireland named Cerelle, but as it costs only 4p per tablet (Cerazette is around 10p and non-branded desogestrel 3p), switching from branded to generic in that case would not be a significant cost-saver. Also, it is worth keeping in mind the recent news that a pharmaceutical manufacturer was found to have inflated the price of a product by 12,000%. That happened despite the drug (hydrocortisone tablets) being generic. With brands, cost controls can actually be agreed with the manufacturer. This is the reason why Cerelle is recommended as the first choice for oral progestogen-only contraceptives.

Looking at all the actual spend we can see how overall spending on all generic and branded desogestrel products (including Cerazette) has changed. Notably, between April and July of this year, overall spending on all desogestrel items fell from £74,820 to £53,885 - a 28% cost reduction over only 4 months.

The recent reduction in desogestrel spending is not because the monthly number of desogestrel items prescribed has fallen, however. The chart below shows that the number of desogestrel items prescribed by general practices and fulfilled by pharmacies has actually increased, not decreased.

So where has the saving come from, if the desogestrel prescriptions have increased? If you've been following, you might have guessed that Cerazette is an expensive branded form of desogestrel. In October 2016, Cerazette-branded desogestrel comprised less than 50% of all desogestrel items prescribed, but represented almost 80% of the money that was actually spent on desogestrel.

To put it simply, if a single prescription normally written for Cerazette is changed so that a cheaper desogestrel alternative replaces it (either Cerelle or generic), a cost-saving will result. Multiplied across all of the prescriptions written monthly in Northern Ireland, these savings can build up to a significant level.

Each practice will have a different prescribing pattern when it comes to their choice between Cerazette and desogestrel. There will be differences, both between practices and over time in the same practice, in how many of the branded presentations rather than the generic option have been dispensed. The chart below shows a timeline for all of the practices, measuring the percentage of desogestrel items prescribed that were Cerazette.

It shows quite a wide variance across the spread of all practices, and also that the trend has now shifted so that less Cerazette is now prescribed than previously, relative to all desogestrel prescriptions. It's notable that up to July 2014, a group of 10% of practices (the 10th percentile) were prescribing a very low proportion of around 20-30% Cerazette, while another 40% (the 60th percentile and above) were prescribing all of their desogestrel items by the expensive proprietary option.

In August 2016, the median practice in Northern Ireland was prescribing 40% of their desogestrel items as Cerazette, as compared to a median of around 94% in 2014. This has chopped about 22% off the total desogestrel bill between these periods, despite the overall increase in desogestrel items being dispensed.

There has been an uptick in the proportion of Cerazette prescribed since August 2016 of 10 percentage points at the median practice. We should be cautious not to read too much in to this - seasonal trends and discrepancies do occur with prescribing data - but it would be worth keeping an eye on this measure to ensure that the desirable trend continues.

Looking at the average trends across Local Commissioning Groups some interesting findings emerge. GP Practices in the Western LCG led the change in desogestrel prescribing at a faster rate, and before the other LGCs started to make the switch.

Trends observed across different areas can tell us a bit more about approaches to prescribing. We'll explore this more in the next section with a drug that forms a larger part of overall spending.

High-cost statins

Statins are a lipid-modifying drug used to treat high cholesterol, along with other non-chemical prescriptions, such as more exercise and healthy eating. In part, due to the crisis in obesity, they are a relatively large share of the prescribing budget, but a decreasing one. Despite an increasing number of statins being prescribed, the total statins bill has halved, and the per patient cost of prescribing statins reduced from £16 in 2010, to £5.32 in 2013. 

This is also evidence here of an influence on prescribing costs that occur away from the doctor's office. Some of the savings can be attributed to the patent expiration of the commonly-prescribed atorvastatin in May 2012, which led to a change in the NICE guidelines recommending its use in place of rousuvastatin (simvastatin had previously been the generic go-to option for statins).

Meanwhile, there is evidence that further savings could be achieved. The NI Audit Office estimated that switching to less expensive statins would have saved around £2.7 million in 2012 and £2.5 million in 2013.

In addition to branded statins, there are more expensive statins that aren't necessarily prescribed under a brand name, but cost more due to their format (e.g. as liquid). In order to determine which statins are the most expensive, the per dosage cost can be compared across this group of medicines. Again, this is possible to do using the prescribing dataset.

The table below demonstrates just a small number of statins that were prescribed by practices in 2016, and shows their cost differences per dosage (usually a single tablet). The full table is here.

Per dosage costs of selected statins prescribed by NI GP Practices, January - October 2016
Formulary Name Total Cost Total Quantity (e.g. of tablets) Cost per dosage
Rosuvastatin 20mg tablets £1,337,082.95 1,596,558 £0.84
Crestor 20mg tablets £57,101.52 68,186 £0.84
Simvastatin 20mg / Ezetimibe 10mg tablets £18,871.73 17,551 £1.08
Inegy 10mg/20mg tablets £6,337.09 5,872 £1.08
Atorvastatin 20mg tablets £473,284.02 12,770,166 £0.04
Atorvastatin 20mg chewable tablets sugar free £15,983.26 20,201 £0.79
Atorvastatin 30mg tablets £10,118.23 12,834 £0.79
Atorvastatin 60mg tablets £30,004.90 33,294 £0.90
Lipitor 20mg tablets £53,770.70 67,813 £0.79
Fluvastatin 20mg capsules £4,754.73 65,286 £0.07
Fluvastatin 80mg modified-release tablets £9,907.64 16,013 £0.62
Lescol 20mg capsules £701.80 1,422 £0.49
Pravastatin 20mg tablets £29,052.21 788,498 £0.04
Lipostat 20mg tablets £1,077.72 1,289 £0.84
Simvastatin 20mg tablets £162,702.13 6,293,120 £0.03
Simvastatin 20mg/5ml oral suspension sugar free £65,471.05 91,607 £0.71
Zocor 20mg tablets £5,348.86 5,594 £0.96

Costs per dosage can range from just 3p to over £1. Stronger dosages of the same drug tend to cost more. Less-commonly prescribed dosages of the same drug also tend to be more expensive (e.g. 30mg and 60mg versions of atorvastatin), as are the chewable or liquid presentations. And there might be good reasons why a particular drug or format suits a patient.

But there are also a lot of different options for clinicians when prescribing which could achieve the same positive results for patients while potentially achieving a significant cost reduction.

  • Atorvastatin and Lipitor is an example of where a direct switch could be made. The branded version, Lipitor, is 20 times more expensive per 20mg tablet than generic atorvastatin.
  • Rosuvastatin is different compound from atorvastatin, but it has the same purpose (to lower LDL-cholesterol in patients). It is probably as safe and effective as atorvastatin, yet it is around 21 times more expensive. As with almost any medicine, there is a range of opinion and competing evidence in the debate over which is best for patients, which is why it is important to understand that not every case should receive the same treatment, even when that is shown to be the most cost-effective.

Guidelines from the National Institute for Clinical Evidence (NICE) states that Given the considerably higher cost of using rosuvastatin, it would need to be considerably more effective than atorvastatin for there to be a possibility that its use could be cost effective for secondary prevention, and that for primary prevention: it would be prudent to recommend atorvastatin 20 mg as the preferred initial treatment. Similarly, guidance for Northern Ireland is that atorvastatin should be used as the first choice of lipid-modifying drugs, and that other statins (pravastatin, fluvastatin, rosuvastatin, simvastatin) may occasionally be used if there are tolerability issues or, under secondary care guidance, for complex lipid management.

While atorvastatin is the most regularly prescribed statin in line with this guidance, rosuvastatin is the most regularly prescribed of the high-cost statin options (note that rosuvastatin appears twice in the prescribing data, under its generic chemical name and branded as Crestor®: these are the same drug with the same cost). 

For January to October 2016, the bill for 5.7 million rosuvastatin tablets is around £3.86m (projected at £4.6m for all of 2016). Though six times the amount of atorvastatin tablets have been dispensed (35 million tablets), the bill for atorvastatin is around half the amount, at £1.45m. If it were possible to change a proportion of rosuvastatin prescriptions to atorvastatin, a not unremarkable saving could be realised. Conversely, the Crestor/Rousuvastation patent expiration should soon have a cost impact in the UK.

The full table of statins has highlighted the high-cost options. By selecting this particular group and comparing them to all statins, an analysis of the proportion of statins prescribed as the more costly option can be performed.

Trends show that not much has changed in the proportion of high-cost statins prescribed in Northern Ireland since 2013, but there has been a very gradual and slight reduction of a couple of percentage points across most percentile groupings of practices. A median of 10% to 11% does seem to be relatively high, given that OpenPrescribing identifies the English national median for high-cost statins at around 3%. This indicates that there is a good amount of room to drive this measure down across Northern Ireland and therefore reduce costs.

The high-cost statins are low in proportion relative to the number of all statins prescribed. However, they are significantly more proportionate in terms of their cost.

The chart below shows the difference in use of the high-cost statins between each of the LCGs.

This shows some mixed results. Since mid-2013, Southern, South-Eastern and Western practices have, on the whole, improved their measure so that high-cost statins make up a lower proportion. Belfast and Northern LCG practices, however, have on average increased the proportion of high-cost statins.

When Prescribing Analytics did a similar analysis of generic vs branded statin prescribing across England, they also found regional variations, highlighting that "wherever the percentage of branded items is high, it represents potential to make big savings by switching". The benefit of this analysis is that it looks at differences between larger groupings of practices (by their Clinical Commissioning Groups), therefore reducing the variations caused by particular individual practices which may have some outlying characteristics, for example, in their patient cohort. There is no apparent reason why practices in CCGs in Kent, the North East, Lancashire and Shropshire should have such a higher reliance on branded statins than those in other CCGs. It is therefore likely that practices within these areas could reasonably achieve a better measure without presenting risks to patients, and therefore saving money for the health service.

For Northern Ireland, if those LCGs with the higher measure were able reduce their measure of high-cost statins to the same level as the lower measure LCGs, an overall cost reduction would follow. If all LCGs were able to reduce their measure even further, then that saving would be magnified. 

 

The data used in the charts is from Business Services Organisation (BSO). The code for this analysis is available on Github.

The photo header is based on a work by Ralph Arvesen (CC-BY)