Abstract
Introduction. Studies suggest no benefit from statins in heart failure (HF), but many individuals in primary care are prescribed statins before HF diagnosis. This study aimed to assess the effect of a statin prescription prior to HF diagnosis on survival in an incident HF population.
Material and methods. Cases of HF diagnosed between 1995 and 2004 in 315 UK primary care practices were identified from electronic case records and followed up for 2 years from diagnosis. Statin prescribing before and after HF diagnosis was assessed, and Cox regression was used to determine the contribution of statin treatment to survival.
Results. A total of 10,914 cases met the inclusion criteria of whom 20% (2185/10,914) were treated with a statin prior to HF diagnosis. Two per cent (191/8729) had an initial statin prescription following diagnosis. Cases prescribed a statin before heart failure diagnosis had a lower risk of death in the subsequent 2 years (hazard ratio 0.52; 95% confidence interval 0.39–0.68), after adjustment for confounders.
Discussion. Most people with heart failure prescribed statins in primary care commenced them prior to diagnosis and appeared to gain benefit, presumably through the effect of statins on cardiovascular co-morbidities. Primary care physicians should not discontinue prior statin treatment at the time of heart failure diagnosis.
Introduction
The use of statins in the primary and secondary prevention Citation[1–3] of coronary heart disease (CHD) is well established, but their role in the management of heart failure appears to be limited Citation[4], Citation[5]. The early statin trials excluded patients with heart failure because of concerns that lowering serum cholesterol could worsen prognosis, and recent trials in heart failure suggest that statin initiation is not effective despite earlier promising observational work Citation[4–6].
Controlled Rosuvastatin Multinational Trial in Heart Failure (CORONA) assessed the effectiveness and safety of rosuvastatin 10 mg versus placebo in addition to optimal heart failure treatment in prevalent symptomatic heart failure cases, with a history of myocardial infarction and not previously treated with a statin Citation[4]. The study recently reported no effect on all-cause mortality and hospitalization over the median follow-up period of 32 months (hazard ratio (HR) 0.95; P =0.31), though safety of the statin was demonstrated Citation[4]. The Gruppo Italiano per lo Studio della Sopravvivenza nell'Insufficienza cardiaca heart failure (GISSI-HF) trial enrolled patients aged 18 years or older with symptomatic chronic heart failure, randomly assigned them to rosuvastatin 10 mg daily (n=2285) or placebo (n=2289), and followed them up for a median of 3.9 years Citation[5]. About 20% of potentially eligible patients were excluded due to prior statin use. No effect was seen in all-cause mortality or combined mortality and hospital admissions (HR 1.0 and 1.1, respectively).
Key messages
Trials of statins in heart failure suggest no benefit is gained from them.
Most people prescribed a statin with heart failure in primary care commence treatment before heart failure is diagnosed, and this is associated with a survival benefit.
Primary care physicians should not discontinue prior statin treatment at the time of heart failure diagnosis.
CHD and hypertension are common treatable aetiological factors in heart failure in primary care with clear indications for a statin Citation[7]. However, the place of statins for individuals with heart failure and these co-morbidities has been questioned with the lack of observed benefits from the addition of a statin in well treated heart failure patients in the CORONA and GISSI-HF trials Citation[4], Citation[5]. This raises the question as to whether statins should be discontinued in patients developing heart failure.
This study therefore aimed to describe current use of statins in heart failure in UK primary care to assess the likely impact of these results on daily practice. Furthermore, the impact on mortality of those currently prescribed statins was evaluated in a generalizable UK population of people with a community diagnosis of heart failure.
Material and methods
Population
Incident cases of heart failure were drawn from primary care practices contributing data to The Health Improvement Network (THIN) database. At the time of the analysis, the database held the electronic patient records of patients from 315 UK general practices (2.2 million currently registered patients). The characteristics of the THIN database have been described elsewhere, but in brief it includes data from a group of practices using VISION clinical software, some of which also contribute to the General Practice Research Database (GPRD) Citation[8]. A recent validation of THIN data found expected associations between risk factors, prescriptions, and disease, with similar results from practices that co-contribute to GPRD or not Citation[9].
Abbreviations
Procedure
Practices were eligible for inclusion if they had been using the VISION general practice software system or its predecessor, VM, for a combined period of at least 3 years. This included a year lead-in, to allow the practices to become familiar with the software, and at least 2 years during which new cases of heart failure could be diagnosed and followed up. As not all primary care diagnoses are verified using objective investigations, incident heart failure cases were categorized according to the level of diagnostic certainty: diagnostic (Read) code alone (unconfirmed case), diagnostic code associated with either a hospital episode (in- or out-patient discharge or a hospital letter), or an abnormal echocardiogram result (confirmed case).
Patients with a diagnosis of heart failure were eligible if they were registered in their practice at least 1 year prior to the first diagnostic code for heart failure, between 1995 and 2004, and aged 45 years or over at the time of diagnosis. We expected that incident heart failure would be under-recorded in electronic patient records among cases who died within the first month as many of these are admitted and die in hospital with subsequent cause of death not routinely coded other than on the death certificate Citation[10]. Therefore, we excluded all cases that died within the first month of diagnosis from the analysis.
Statin exposure prior to HF diagnosis was defined as at least one prescription of a statin in the 12 months to 1 month period prior to HF diagnosis. Statin exposure after HF diagnosis was defined as at least one prescription of a statin in the period from 1 month prior to HF diagnosis to 2 years following diagnosis. Statins that were initiated in the month prior to HF diagnosis were categorized as post-HF because of the possibility that the first mention of HF in a patient's record was the date when the diagnosis was confirmed by hospital letter.
This study was approved by the Eastern Multi-Centre Research Ethics Committee 06/MRE05/20.
Analysis
Demographic data regarding those prescribed a statin or not were tabulated. A Cox proportional hazards regression was used to quantify the hazard associated with statin treatment on survival in the 2 years following a diagnosis of heart failure. Log-log graphs were used to assess proportional hazards in categorical and continuous covariates, and covariate-adjusted survival estimates were used to compare graphically survival in different groups. Stratification was used for year of diagnosis as the proportional hazards assumption was not met for this variable, and a robust standard error was calculated to allow for clustering of patients within practices.
Covariates included in the model were: risk factors for HF (smoking status, male sex, age), objective confirmation of HF diagnosis, treatment with angiotensin-converting enzyme inhibitors or angiotensin-II receptor antagonists (ACEI/ARB) and/or beta-blockade, and co-morbidities (diabetes, peripheral vascular disease (PVD), renal disease, coronary heart disease, cerebrovascular disease, valvular heart disease, chronic obstructive pulmonary disease (COPD), cancer, and dementia). We did not include past history of hypertension as a separate covariate in the model to avoid collinearity as HF cases without prior CHD may have been due to hypertension.
Exposure to statins was included in the model in two separate categories: exposure prior to HF diagnosis and exposure following diagnosis only, as the effect of statin treatment is known to increase with duration of treatment, and as the exposure of interest was statins when initiated prior to HF diagnosis. We also tested for an interaction between statin exposure prior to HF diagnosis and CHD to differentiate between cases prescribed a statin as secondary prevention of CHD and cases prescribed it as primary prevention. Hazard ratios with a probability of less than 0.01 were regarded as significant: 95% confidence intervals (CI) are also presented to provide a more conventional view of the precision of the HR estimates.
Results
The THIN database contained 2,972,319 practice patients from 315 general practices in the UK; 11,110 patients met the initial criteria for inclusion in the study. Approximately 2% (196/11,110) of recorded cases died in the first month, and were excluded. The remaining 10,914 cases were followed up from the date of diagnosis until they died, left their practice, or the end of the 2-year study period.
Base-line characteristics
The base-line characteristics of those prescribed or not prescribed statins are summarized in , along with information on drug treatment following diagnosis. Statin-treated cases were on average 8 years younger than non-statin-treated cases and less likely to be female. Deprivation (Townsend) scores were similar between the groups. People receiving a statin were more likely to have received this in the last 3 years of the study period. A greater proportion of cases receiving a statin had coded evidence confirming their HF diagnosis (abnormal echocardiogram, hospital letter, or in- or out-patient discharge) in their electronic record. People prescribed statins were more likely to have a history of hypertension, CHD, and diabetes, but less likely to have dementia. Statin prescription was associated with higher rates of prescribing of other heart failure treatments (ACEI/ARB and beta-blockers).
Table I. Patient characteristics.
Statin prescription
One-fifth (2185/10,914) of HF cases were prescribed a statin before the diagnosis of HF. An additional 2% (191/8729) began statin treatment in the year following HF diagnosis, and none who were previously prescribed a statin had it withdrawn following HF diagnosis. Although the overall statin treatment rate post diagnosis was 22%, there was also a large year-on-year increase in the use of statins, rising from 1% of new cases in 1995 to 52% in 2004 (). The most commonly prescribed statins were simvastatin and atorvastatin, accounting for 85% of the statins prescribed at the end of follow-up.
Table II. Prescribing of statins after heart failure diagnosis.
Mortality
There were 2308 observed deaths among the 10,914 incident cases of HF in the first 2 years following diagnosis (total 17,486 patient-years; death rate 131 per 1000 person-years at risk (pyrsk); 95% CI 127–137). The crude death rate among statin-treated cases was 72 per 1000 pyrsk (95% CI 64–81; 295/4089 per year) and 150 per 1000 pyrsk (95% CI 144–157; 2013/13,396 per year) among non-statin-treated cases.
The full survival model is shown in and survival curves in . Statin exposure, the main treatment of interest, was associated with a halving of the risk of death when initiated prior to HF diagnosis, after adjusting for other relevant treatments, co-morbidities, demographic, and other risk factors (HR 0.53; P <0.001). Results were similar whether or not the diagnosis of HF was confirmed by objective evidence coded in the patients’ record. The covariates are split into three groups: statin exposure, significant hazard ratios (HR), and non-significant HRs, and ordered by the relative size of the HR.
Figure 1. Survival in the 2 years following HF diagnosis: comparison of cases prescribed a statin prior to heart failure (HF) and cases not prescribed a statin prior to HF for both confirmed and unconfirmed heart failure.
Table III. Survival model.
Effects of co-prescription of heart failure treatment, co-morbidity, and diagnostic criteria
Non-statin-treated cases were more likely to have received a diuretic (96% versus 89%) but were less likely to have received angiotensin system or beta-blockade, both when looked at individually and in combination. In particular, the statin-treated cases were almost 1.5 times more likely to have received one of these two types of treatment (94% and 66%) and five times more likely to have received a beta-blocker and ACE/ARB in combination than those in the non-statin-treated group (48% and 12%). This treatment was associated with a highly statistically significant reduction in risk in the model (HRs 0.73 (95% CI 0.67–0.80) and 0.64 (0.57–0.73), respectively; P <0.001 in both cases) which was taken into account in the survival analysis ().
Four of the co-morbidities included in the model (diabetes, PVD, renal disease, CHD) were associated with a significantly increased risk of death (). Current smokers, males, and older adults were also at an increased, highly significant risk (HRs 1.36, 1.28, 1.03; P <0.001 in each case). The risk of death in cases with additional evidence confirming their HF diagnosis (abnormal echocardiogram, or a hospital letter, or in- or out-patient discharge associated with a coded diagnosis of HF) was similar to that of cases without such evidence (HR 1.00; P =0.985).
Discussion
This study included over 10,000 newly diagnosed cases of heart failure from a large number of primary care practices drawn from throughout the UK and followed for 2 years after diagnosis. The vast majority of people prescribed a statin received this before a diagnosis of heart failure had been made, and these individuals had a 47% lower risk of death in the 2 years following diagnosis (HR 0.53; 95% CI 0.40–0.70), after adjustment for confounders including major co-morbidities, heart failure treatment, probable HF aetiology, and year of diagnosis. This suggests that most statin prescription for people with heart failure in UK primary care is appropriate, and the observed mortality benefits in this group should encourage practitioners to continue to prescribe statins in heart failure where indicated for co-morbidities.
Cases prescribed a statin after diagnosis also had a lower risk (32% (HR 0.68; 95% CI 0.46–0.99)) albeit smaller and not significant at the P <0.01 level. The benefit of statin treatment prior to HF diagnosis did not appear to differ in cases with and without CHD (HR 0.83; 95% CI 0.62–1.12). There was a large increase in the prescription of statins for people with heart failure during the 10-year study period, from a maximum of 14% before 2000 to 52% in 2004 which presumably reflects increased use of these agents in primary and secondary prevention of CHD.
Almost all cases were treated with a diuretic (94%), two-thirds with an ACE inhibitor or ARB (66%), and almost one-quarter with a beta-blocker (26%). One-fifth of cases were treated with both an ACE or ARB, and a beta-blocker (20%). These seem low but are similar to the results of the Improvement programme in evaluation and management of heart failure (IMPROVEMENT) survey from 1999, roughly the mid-point of our study, as well as the findings from a recent study of heart failure using a different primary care database Citation[11], Citation[12]. In the latter study set 2002–2006 (i.e. the end of the current study period), 37% of heart failure patients were prescribed a beta-blocker, but a further 44% had a relative or absolute contraindication for such a prescription suggesting that the room for improvement may not be as great as might be expected Citation[12]. A final source of confounding might be the issue of drug dosing, and previous work has suggested that only a minority of primary care patients receive target doses of both ACEI and beta-blockers Citation[12]. The current analysis does not allow consideration of this.
Appropriate secondary preventative medication for heart failure was more likely to be prescribed to those already receiving a statin, and these individuals were younger and more likely to be male. This is similar to the findings by Ray and colleagues when considering a heart failure population prescribed a statin post diagnosis and, as with that study, our results suggest that the effect of statins is independent of co-prescription Citation[13].
The method of analysis allowed adjustment for important confounders including co-existing risk factors, co-treatment, and co-morbidities. However, this work utilized a large database from routine practice, and as with all observational studies cases were not randomly assigned to treatments so important biases between the treated and untreated groups may not have been controlled for. Ascertainment of cause of death can also be difficult and so the use of all-cause mortality in the current analysis is a strength Citation[14]. Furthermore, although we were able to identify prior diagnosed CHD in cases, it is not certain whether this was the main or only cause of their heart failure. The 47% reduction in risk among cases prescribed a statin before HF is larger than that observed in major statin trials (for example, 22% (95% CI 0–40) in West of Scotland Coronary Prevention Study (WOSCOPS) Citation[1] and 20% in Justification for the Use of Statins in Primary Prevention: An Intervention Trial Evaluating Rosuvastatin (JUPITER) Citation[15] trials). This may be an overestimate of the benefit of statins on mortality, as some high-risk statin-treated cases may have died before being diagnosed with HF, leaving a cohort of treated survivors who were at a lower risk of death following HF diagnosis than the untreated group.
There are many possible patterns of exposure to statins that can occur in clinical settings and that are not dealt with in this study. We only had information on statin exposure in the year prior to HF for most cases and not on total duration of treatment. Only half of the non-treated cases had a total cholesterol measurement taken, so we were unable to include this as a covariate or to separate any benefits from lipid-level reduction from other possible mechanisms. However, we observed no difference in the effect of statin exposure prior to HF diagnosis in cases with and without CHD, in line with the meta-analysis reported by Baigent Citation[3]. Although we were not able to quantify the duration of exposure in each case, cases prescribed statins before HF will have had a longer exposure to statins than those prescribed after HF diagnosis, and the difference in risk reduction observed may be due in part to this effect (HR 0.53 and 0.68, respectively).
Neither the CORONA nor GISSI-HF trials found benefit from the addition of statins in people with no firm indication for lipid-lowering treatment and who were receiving optimal HF management Citation[4], Citation[5]. Importantly, neither trial found any evidence of harm from statins. Furthermore, these and all previous observational studies of statins in heart failure to date have been set in secondary care, often following hospitalization or referral for clinical management. Such populations are likely to suffer from spectrum bias compared to primary care, including more severe disease and better application of evidence-based therapies (patients were very well treated at base-line), and will not reflect the full range of cases seen in the community. The current study found apparent, significant benefit from statins prescribed largely prior to diagnosis of heart failure in a large primary care population suggesting that heart failure should not be seen as a contraindication for statins prescribed for a co-morbidity.
These data support previous suggestions that the increased use of statins in primary prevention and secondary use in CHD, both in terms of increased dosage and number of patients prescribed a statin, might lead to benefit Citation[16–18]. More wide-spread and more effective use of statins to reduce cardiovascular disease risk in the general population, together with increased survival following Myocardial Infarction, is likely to alter the incidence, aetiology, and progression of heart failure seen in clinical settings and to mean that even higher proportions of individuals who eventually develop HF will be pretreated with a statin.
Our data strongly suggest that physicians should not discontinue statin therapy in patients with on-going indications when heart failure presents.
Acknowledgements
RJM and RR received a bursary from the database owners, EPIC, to perform the study following an open competition among members of the Society of Academic Primary Care (www.sapc.ac.uk). RJM was funded by a National Institute of Health Research Primary Care Post Doctoral Fellowship, JAAM by a National Institute of Health Research Career Scientist Fellowship, and FDRH part-funded as a National Institute of Health Research Senior Investigator. No funder had a role in the study design, analysis, or interpretation of data; in the writing of the report; or in the decision to submit the paper for publication. We would like to thank Martin Frischer and Heath Heatlie at the University of Keele who commented on the initial study protocol. Competing interests: Ryan RP: No competing interests. McManus RJ: RJM has received funding for research and travel from a number of companies which manufacturer drugs for heart failure. Mant JW: No competing interests.Macleod JAA: No competing interests. Hobbs FDR: FDRH has received intermittent research funding and speaking consultancies for a variety of pharmaceutical companies, including ones that market statins and therapies licensed for heart failure.
The authors alone are responsible for the content and writing of the paper.
References
- Shepherd J, Cobbe SM, Ford I, Isles CG, Lorimer AR, MacFarlane PW, et al. Prevention of coronary heart disease with pravastatin in men with hypercholesterolemia. West of Scotland Coronary Prevention Study Group. N Engl J Med. 1995; 333: 1301–7
- Heart Protection Study Collaborative Group. MRC/BHF Heart Protection Study of cholesterol lowering with simvastatin in 20,536 high-risk individuals: a randomised placebo-controlled trial. Lancet. 2002;360:7–22.
- Baigent C, Keech A, Kearney PM, Blackwell L, Buck G, Pollicino C, et al. Efficacy and safety of cholesterol-lowering treatment: prospective meta-analysis of data from 90,056 participants in 14 randomised trials of statins. Lancet. 2005; 366: 1267–78
- Kjekshus J, Apetrei E, Barrios V, Böhm M, Cleland JG, Cornel JH, et al. CORONA Group. Rosuvastatin in older patients with systolic heart failure. N Engl J Med. 2007; 357: 2301–4
- ; Gissi-HF InvestigatorsTavazzi, L, Maggioni, AP, Marchioli, R, Barlera, S, Franzosi, MG, Latini, R. Effect of rosuvastatin in patients with chronic heart failure (the GISSI-HF trial): a randomised, double-blind, placebo-controlled trial. Lancet. 2008;372:1231–9.
- van der Harst P, Voors AA, van Gilst WH, Bohm M, van Veldhuisen DJ. Statins in the treatment of chronic heart failure: a systematic review. PLoS Med. 2006; 3: e333
- Velagaleti RS, Vasan RS. Heart failure in the twenty-first century: is it a coronary artery disease or hypertension problem?. Cardiol Clin. 2007; 25: 487–95
- Bourke A, Dattani H, Robinson M. Feasibility study and methodology to create a quality-evaluated database of primary care data. Inform Prim Care. 2004; 12: 171–7
- Lewis JD, Schinnar R, Bilker WB, Wang X, Strom BL. Validation studies of the health improvement network (THIN) database for pharmacoepidemiology research. Pharmacoepidemiol Drug Saf. 2007; 16: 393–401
- Hippisley-Cox J, Coupland C, Vinogradova Y, Robson J, May M, Brindle P. Derivation and validation of QRISK, a new cardiovascular disease risk score for the United Kingdom: prospective open cohort study. BMJ. 2007; 335: 136
- Cleland JG, Cohen-Solal A, Aguilar JC, Dietz R, Eastaugh J, Follath F, et al. Management of heart failure in primary care (the IMPROVEMENT of Heart Failure Programme): an international survey. Lancet. 2002; 360: 1631–9
- Calvert MJ, Shankar A, McManus RJ, Ryan R, Freemantle N. Evaluation of the management of heart failure in primary care. Fam Pract. 2009; 26: 145–53
- Ray JG, Gong Y, Sykora K, Tu JV. Statin use and survival outcomes in elderly patients with heart failure. Arch Intern Med. 2005; 165: 62–7
- Mant J, Wilson S, Parry J, Bridge P, Wilson R, Murdoch W, et al. Clinicians didn't reliably distinguish between different causes of cardiac death using case histories. J Clin Epidemiol. 2006; 59: 862–7
- Ridker PM, Danielson E, Fonseca FA, Genest J, Gotto AM, Jr, Kastelein JJ, et al. JUPITER Study Group. Rosuvastatin to prevent vascular events in men and women with elevated C-reactive protein. N Engl J Med. 2008; 359: 2195–207
- Brady AJB, Norrie J, Ford I. Statin prescribing: is the reality meeting the expectations of primary care?. Br J Cardiol. 2005; 12: 397–400
- de Lusignan S, Dzregah B, Hague N, Chan N. Cholesterol management in patients with IHD: an audit-based appraisal of progress towards clinical targets in primary care. Br J Cardiol. 2003; 10: 223–8
- Hobbs FD, Southworth H. Achievement of English National Service Framework lipid-lowering goals: pooled data from recent comparative treatment trials of statins at starting doses. Int J Clin Pract. 2005; 59: 1171–7