 |
 |
22 March 2010
A study of administrative data from one PCT suggested that practices which tended to use more lower-cost statins had slightly fewer patients with a total cholesterol of 5mmol/L or less in certain QOF measures. However, national data does not confirm this finding, and numerous factors limit the practical clinical significance and generalisation of these findings. There is high quality evidence that simvastatin 40mg daily is effective in reducing the risk of CV events, and it is still first choice statin for most patients, in accordance with NICE guidance.
Level of evidence:
Level 3 (other evidence) according to the SORT criteria.
Action
This study should not change practice: simvastatin 40mg daily is first choice statin for most patients, in accordance with NICE guidelines. This was a ‘snapshot’ study of one PCT, using historical administrative data, not related to clinical outcomes, whose findings were not replicated at national level.
What is the background to this?
NICE guidance on lipid modification and type 2 diabetes recommends simvastatin 40mg as first choice statin for primary and secondary prevention of CV disease. According to the most recent Better Care, Better Value indicator data available (quarter 1 2008/09), if PCTs with below 79% use of lower cost statins (achieved by the top quartile of trusts) increased this to 79% over £70m would be saved in a year.
A number of PCTs have set local prescribing targets to increase the use of lower cost statins over higher cost statins. This study, by a Somerset dispensing GP whose practice prescribing rate for lower cost statins was 49% at the time of writing, looked at administrative data to compare use of lower cost statins with proportion of patients with a total cholesterol of 5mmol/L or less, as measured by CHD8 and STROKE8 indicators of the QOF.
What does this study claim?
Using prescribing data for Somerset practices (number not stated) for a period before April 2007 (details not stated) and QOF data for the financial year 2006/7, the author found a small negative correlation between use of lower cost statins and the proportion of patients whose cholesterol levels were reported as 5mmol/L or less for CHD8 and STROKE8: Pearson’s correlation (r) –0.26, P=0.028. (Perfect correlation with a negative slope would be indicated by –1.00, no correlation would be indicated by 0.0).
What is the national picture?
A study of national data from 2005/6 found that there was no statistically significant link between use of low-cost statins and proportion of patients achieving QOF targets. More recent data (personal communication, NHS Information Centre for Health and Social Care) show that although the proportion of lower cost statins prescribed has increased, the proportion of patients achieving QOF targets has remain largely unchanged:
|
2006/7 |
2007/8 |
2008/9 |
|
| CHD8 |
81.9% |
82.5% |
82.1% |
| Stroke8 |
76.2% |
77.0% |
77.0% |
| DM17 |
83.1% |
83.2% |
82.6% |
| % low cost (q4 of full year) |
67.9% |
72.5% |
74.3% |
So what?
This study has numerous shortcomings that severely limit its usefulness in deciding on prescribing policy. First, it looked only at cholesterol measurements, not CV outcomes. Although as a general principle lower total cholesterol and lower LDL-cholesterol is desirable, the studies of high versus standard dose statins (IDEAL and TNT) have not shown convincingly that there is an overall advantage in using high doses in patients with stable heart disease. In addition, simvastatin 40mg and atorvastatin 10mg reduce lipid levels by the same amount (see the NICE full lipids guideline). It should also be noted that association does not necessarily mean causation: the apparent slight negative association, if true, might be caused in part by choice of statins, but could also be explained, at least in part, by other factors not considered in this analysis. Past and more recent national level data do not suggest an association.
The percentage of patients with a recorded total cholesterol of 5mmol/L or less a practice needs to achieve for full points in QOF is 70% in CHD8 and 60% in STROKE8 (and 70% in DM17). This study did not look at the relationship between use of lower cost statins and attainment of QOF indicators. Nevertheless, the scatter graph in the paper (figure 1) suggests that all bar one of the ten highest users (>80%) of lower cost statins would have achieved full points for the CHD8 and STROKE8 indicators.
The upper threshold in QOF measures reflects the clinical evidence base and is in keeping with NICE lipids guidance (Clinical Guideline [CG] 67). The author of this paper seems confused about NICE guidance, as he says “the QOF markers are less stringent than the targets set by NICE in its most recent guidance [CG 67], which recommends a maximum cholesterol target of 4mmol/L for secondary prevention.” This does not seem to be an accurate reflection of CG 67. Paragraph 1.4.25 of this says (in regard to non-ACS secondary prevention patients) “An ‘audit’ level of total cholesterol of 5mmol/litre should be used to assess progress in populations or groups of people with CVD, in recognition that more than a half of patients will not achieve a total cholesterol of less than 4mmol/litre or an LDL cholesterol of less than 2mmol/litre”. Page 207 of the NICE full guideline specifically says “the use of a target figure can be helpful in guiding increases of lipid lowering drugs as long as it is clear that this figure is intended to guide treatment rather than be a figure patients are expected to achieve”.
NICE does say that in non-ACS secondary prevention patients, health professionals should “consider increasing to simvastatin 80mg or a drug of similar efficacy and acquisition cost if a total cholesterol of less than 4mmol/litre or an LDL-cholesterol of less than 2mmol/litre is not attained”. However, the full guideline makes the thinking behind this clear by stating (page 190) “The results of modelling suggest that titration using a threshold target of 4mmol/L total cholesterol is cost-effective so long as titration stops at simvastatin 80mg. Most patients would not achieve a target of 4mmol/L total cholesterol and modelling suggests that it is not cost-effective to try to take more patients to target using higher cost statins such as atorvastatin”.
NICE also says that “Any decision to offer a higher intensity statin should take into account the patient’s informed preference, comorbidities, multiple drug therapy, and the benefits and risks of treatment”. We have obtained clarification from NICE that the statement above should be understood to mean that increasing the dose should be considered only for patients whose total cholesterol is greater than 4mmol/L and also whose LDL-cholesterol is greater than 2mmol/L. If either figure is below that level, then increasing the dose is not recommended. Moreover, these figures are intended to guide treatment, they are not targets patients are expected to attain.
In a previous blog we highlighted that a single cholesterol reading may over- or under estimate a person’s true average cholesterol by up to 14%. In practical terms, this means that for 95 people out of 100 with a single cholesterol measurement of 5.0mmol/L, their true average cholesterol level will be between 4.3mmol/L and 5.7mmol/L. (this is due to normal variations in a patient’s lipid levels as well as factors relating to the analysis itself).
Finally, the paper mentions exception reporting. There can be very good clinical reasons to exception report patients. That facility is there within QOF to avoid driving so-called ‘cookbook medicine’, but instead to allow the patient and prescriber to individualise treatment. For example, if a patient on simvastatin 40mg has a total cholesterol of 5.2mmol/L on repeat testing, having come down from 7.5mmol/L, it could be quite reasonable (and entirely in line with NICE guidance) for the patient to make an informed decision not to pursue further lipid lowering, and exception report the patient accordingly.
Design: statistical analysis of administrative data
Patients: data relate to practices in Somerset PCT in 2006/7
Intervention and comparison: correlation between proportion of lower costs statin use and proportion of patients with total cholesterol 5mmol/L or less, for STROKE8 and CHD8 in the Quality and Outcomes Framework
Outcomes and results: r -0.26 (P=0.028) for both measures
Sponsorship: None stated
More information can be found on the lipids section of NPC
Please comment on this blog in the NPC discussion rooms, or using our feedback form.
Make sure you are signed up to NPC Email updates — the free email alerting system that keeps you up to date with the NPC news and outputs relevant to you