Background
Statins have noteworthy benefits in stroke prevention as well as in cardiovascular diseases due to their well-established pleiotropic action [
1]. In fact, beyond their well-known lipid-lowering effect, their role in the stabilization of endothelial function, in immunomodulation and antithrombotic processes, has been hypothesized and subsequently demonstrated [
2‐
4]. After the SPARCL trial [
5] many other studies have indicated that statins are of benefit to the majority of ischemic stroke patients as they reduce the risk of recurrence and improve outcome [
6‐
13]. It has also been observed that daily intake of a statin after an acute ischemic stroke could improve functional outcome in patients with LDL levels ≤ 100 mg/dL [
2]. The last Italian guidelines for stroke [
14], in agreement with recommendations from other international guidelines [
15,
16], state that statins are indicated in patients with ischemic stroke even in the absence of hypercholesterolemia. Moreover, the most recent ACC-AHA guidelines for management of high blood cholesterol levels [
17], based on RCTs results, recommend a high-intensity statin therapy for
every patient with “clinical atherosclerotic CVD” (ASCVD). ASCVD patients include acute coronary syndromes, a history of myocardial infarction, stable or unstable angina, coronary or other arterial revascularization, stroke, TIA or peripheral atherosclerotic arterial disease. In the presence of at least one of these clinical events, patients should receive statin therapy regardless LDL cholesterol levels [
18].
Despite this scientific evidence and guideline recommendations, an unacceptably high proportion of stroke patients are neither on lipid-lowering therapy nor managed aggressively enough to achieve recommended target cholesterol levels [
19,
20]. Causes for non-adherence to current guidelines are multifactorial, and depend both on physicians and patients.
Understanding the gap between a physician’s knowledge and his actual actions may be essential for the development of strategies aiming to improve patient management: predictors of adherence and causes for non-adherence should be identified and evaluated carefully.
Most studies have examined indirectly physicians’ adherence to guidelines, through self-administered questionnaires and interviews. To our knowledge, there have been only a few attempts to describe the topic using data from real clinical practice (i.e., the prescribed therapies as they appear on the discharge letter), and to characterize the clinical factors interfering with statin prescription: the GWTG Stroke, the Swedish Stroke Register, the Paul Coverdell National Stroke Registry [
21‐
23]. These registries describe prescription trends but are not able to identify the barriers to prescription.
The aim of this study is to identify the clinical factors influencing statin prescription by physicians in acute ischemic stroke patients at discharge from Lombardia Stroke Units [
24]. We analyzed data collected from a web-based registry: the Lombardia Stroke Registry (LSR), describing the activity of our region’s Stroke Units.
Discussion
In the literature there is a growing body of papers concerning the factors that can influence physicians’ attitude to the prescription of new drugs. Statins are, in fact, quite old drugs, whose range of clinical indications has broadened after the recent RCTs.
After the publication of the SPARCL trial results in 2006, a progressive increase in the number of patients who receive statin therapy at discharge in the United States has been reported. In the GWTG-Stroke study, the percentage of statin prescription increased from 75.7% in 2005 to 83.5% in 2007, however it was not considered high enough [
21]. In our study, the percentage of patients treated with statins is only 42%, with a slight increase during the study period. Few other settings show worse results than ours. For example, in a recent meta-analysis conducted in China, only 24% of ischemic stroke patients were treated with statins [
29]. It has been hypothesized that epidemiological features of Chinese people – lower mean blood cholesterol levels, more frequent hemorrhagic complications than Western people – could play a role in physicians’ decision-making process.
Prescription decisions are affected by many factors related to physicians’ attitudes and patient characteristics. Concerning physician-related factors, some authors have identified as good predictors of adherence to guidelines the habitual reading of medical journals and maintenance of professional contacts, such as talking to colleagues and attending scientific meetings [
30]. It is also widely recognized that new drugs and guidelines are adopted earlier by specialists [
30] and that clinical trial investigators are usually more prone to follow guidelines because of their link to the world of research [
31]. Another study describes the so-called ‘therapeutic traditions’ which seem to influence the prescription behavior of individual physicians [
32]. All these factors are difficult to quantify. As for patient-related factors, those able to influence their attitude to comply with therapy (after discharge) are usually identified in the literature as patients’ demographic, social and cognitive features. On the contrary, clinical factors detected during hospitalization are not usually considered as being among those that influence physicians’ prescription.
To overcome this limitation, our work tried to identify those patient-specific clinical and care factors that are able to influence physicians’ decisions to prescribe statin after an acute ischemic stroke. In our study, patients were more likely to be treated with statins if they were younger than 80 years old and not disabled before stroke, they had a personal history of risk/disease of systemic atherosclerosis (dyslipidemia, diabetes, myocardial infarction, coronaropathy, peripheral arteropathy), they were treated with thrombolysis at the emergency department and discharged on antithrombotics for the secondary prevention of ischemic stroke.
It is interesting to note that, in our study, dyslipidemia represents the principal factor leading to the prescription of statin in ischemic stroke of all subtypes, including in LVD stroke.
Among demographics, older age, previous disability and lower social status, have been previously reported to negatively influence the adherence to guidelines in many settings [
33]. According to the Swedish Stroke Register [
22] the prescription of statins was strongly age-related: statin therapy, even if associated with reduced risk of death, seemed to be underused among elderly patients, with an inverse correlation with increasing age. In line with these findings, in our setting younger patients and those without prestroke disability were more likely to be treated with statins. A higher risk of hemorrhagic complications in older people could explain this trend, even if other authors have not found a different rate of side effects compared to younger patients [
34]. It has been suggested that older people [
35] and people with prior disability [
36] tend to discontinue treatment during follow up. The hypothetical reduced compliance could, at least partially, explain the low frequency of prescription in these subjects. Another important point is that most of the current evidence derives from RCTs based on patients with an average age of 65 years or younger and little comorbidity [
5]. Thus, the observed lower prescription rate in older and disabled patients may perhaps be a consequence of this signaling. Moreover, current literature on statins efficacy shows controversial results: a recent work examined the impact of statins on outcome in older ischemic stroke patients, finding benefits in functional outcome and 12-month survival [
34]; another paper analyzed the effect of statins in the elderly in the prevention of CV events, finding a lower rate of MI and stroke but worse survival rate [
37]. Regarding gender, it has been already reported that female patients are often undertreated [
33,
38]; our study confirms this trend. Probably, the well-known higher risk of cardio-cerebrovascular pathologies in men than in women plays a role in the prescription-making process, even if it cannot completely justify it.
Stroke severity, directly linked to disability and life expectancy, is another factor our model found as influencing prescription. Patients with lower NIHSS scores both on admission and at discharge are more likely to receive statins. These findings suggest that physicians tend to prescribe statins in patients with milder neurological damage, perhaps in order to obtain better compliance. Higher severity could thus represent a 'non-prescription' bias leading physicians to avoid secondary preventive therapies in patients who have had severe strokes and with poor prognosis. Then, as already pointed out, patients with significant disability and poor prognosis are excluded from the dedicated RCTs, whose results cannot be applied to them. Another possible explanation could be linked to the highest risk of hemorrhagic transformation characterizing patients with more severe neurological deficits [
39]. A post hoc analysis of the SPARCL trial results found that the use of statins in stroke prevention was associated with an increase in the number of patients having hemorrhagic stroke [
40]. Therefore, another factor in patients with higher stroke severity being less frequently treated with statins is the possibility of increased risk of hemorrhagic transformation in patients who are already susceptible.
We also found that patients who did not receive antithrombotics at discharge showed a low statin prescription rate. This could happen in a context of global non-adherence to guidelines. Alternatively, they could be untreated with antithrombotics because of a higher clinical severity (large infarction size, hemorrhagic transformation, etc.): probably physicians tend to avoid statin prescription in order to limit the possible higher risk of bleeding in the subgroup.
The presence of non-cardioembolic risk factors was among the main drivers of increased statin use in the GWTG-Stroke trial [
21]: our results were in line with it. It is interesting to note that patients with hypertension and diabetes are more likely to be treated with statins than patients who are not affected by these risk factors, even if more than a half of them do not usually receive the prescription (see Table
1).
In the GWTG setting, patients with known dyslipidemia or who were on a cholesterol reducer at the time of hospital admission seem to be more likely to leave the hospital on a statin [
21]. In our study too, comparing ORs, dyslipidemia shows the strongest correlation with statin prescription, followed by myocardial infarction and PAD; the Classification Tree confirmed dyslipidemia as the leading factor in prescriptive decisions. This means that dyslipidemia and cardiovascular pathologies still appear crucial in determining the need for statin therapy. Does this mean that stroke is not
per se recognized as a vascular disease?
However, the relationship between total cholesterol levels and stroke outcome has not, as yet, been well characterized. Despite much clinical data supporting a linear relation between cholesterol lowering and cardiovascular risk reduction [
41], in some papers hypercholesterolemia has been associated with better stroke outcome [
42]. Recent works have described a ‘U-shaped’ association between cholesterol levels and stroke mortality, which was higher for patients with < 160 mg/dl [
43]; the same cut-off was identified as a risk factor for a hemorrhagic fatal stroke [
44].
Current guidelines for stroke [
14‐
16] recommend statins for all patients with non-cardioembolic stroke. The best evidence from the literature derives from the SPARCL trial subanalysis: statins are mostly effective in reducing the risk of vascular events in patients with carotid atherosclerosis [
45]. The last ACC-AHA guidelines [
17] for lipid management identify “patients with clinical atherosclerotic cardiovascular disease” (including TIA and stroke) as the first subgroup of patients that should receive a high-intensity statin therapy, if tolerated.
In our setting, statin prescription rates for cardioembolism and ‘other causes’ of stroke are in line with current guidelines. However, it is particularly surprising that patients with large vessel disease, in the absence of other clinical indications for statins, have a prescription rate of only 44%, despite the current guidelines strongly recommending statin therapy for atherosclerotic disease. The Classification Tree applied to the subgroup of LVD patients leads us to conclude that dyslipidemia is the leading factor for statin prescription, even in the presence of well-defined atherosclerosis, instead of recognizing LVD per se as the clinical indication for statin reception.
Concerning small vessel disease, the low rate of prescription we detected could be explained by the complex relationship between statins and lacunar stroke, and by the fact that literature lacks precise indications. Small vessel disease determines both ischemic (lacunar) and hemorrhagic (microbleeds) stroke, and severe arterial hypertension is the main risk factor in both cases. According to these factors, statins could somehow enhance the risk of hemorrhage in patients with lacunar stroke because of the underlying small vessel pathology and unbalanced hypertension. Recent papers have analyzed the connections between statins use and the risk of developing microbleeds, with non-univocal results. Day et al. [
46] found that previous statin therapy was not associated with the prevalence or degree of microhemorrhages in patients with acute ischemic stroke or transient ischemic attack. Haussen et al. [
47] found instead that statins use was independently associated with microbleeds in patients with ICH. Some authors [
48] underline that patients with non-atherosclerotic causes of stroke should not be treated with statins and that statin therapy should be initiated only after a careful consideration of all the potential risks and benefits that could derive from them.
Besides the low rate of prescription in atherosclerotic disease, the Classification Tree does not highlight the TOAST criteria among the predictors of statin prescription: the therapy is prescribed mostly according to the risk profile.
Unexpectedly, thrombolytic treatment in the acute phase of stroke seems to enhance statin prescription. Results of studies about the association between statin therapy and thrombolysis are somehow controversial. Some studies analyze the effect of statin treatment before stroke and functional outcome after thrombolysis, with non-coherent findings: some have found a positive influence from statins [
49], while others do not [
50‐
52]. Engelter et al. [
52] found that prior statins use was not an independent predictor of functional outcome or of ICH after thrombolysis, but it may be considered as an indicator of baseline characteristics that are associated with a less favorable course. Cappellari et al. [
53,
54] suggested that statin treatment initiated within 24 hours of IV thrombolysis, but not statin treatment started before stroke and continued in the acute phase, may positively influence functional outcome. It is important to note that most of papers analyze the relationship between prior statin therapy and the clinical outcome after thrombolysis while our work describes the rate of statin prescription after thrombolysis. Our results could be explained with Cappellari’s findings. Further, to our knowledge, there is no evidence proving an increased risk of hemorrhage in patients treated with both rtPA and statins. We could also hypothesize that patients treated with thrombolysis are more likely to be managed following guidelines, and this could perhaps explain why they are more likely to receive the recommended statin regimen at discharge.
In the literature, statin therapy is mostly analyzed when administered during in-hospital stays. In this setting, it seems to be associated with a lower rate of complications in stroke patients. The prophylactic administration of statins significantly reduces in-hospital mortality [
49] and is associated with a favorable outcome [
55,
56], with few incidences of adverse reaction [
57]. Montaner et al. [
58] reported that statin use between 3 and 12 hours after stroke was associated with neurological improvement at 3 days from ischemia, suggesting a rapid mechanism of action on coagulation or the fibrinolysis system. These findings should enhance physicians’ attitude to statin prescription. Considering our available data (i.e., statin prescription rate at discharge), however, we cannot hypothesize that statins could determine a lower occurrence of complications, but that patients with in-hospital complications are unlikely to be treated with statins, the only exception being hypertensive fits and hyperglycemia. This finding could be explained as previously done for age, disability and stroke severity: physicians avoid prescribing statins in critical patients, aside from situations that require statins for other indications than stroke. Besides, even if not confirmed by the logistic regression model, in our setting statins are not usually prescribed in patients with in-hospital bleeding, probably due to the higher risk of hemorrhage associated with statin use in the literature [
59].
Besides these reports from the literature, we have to note that the effects of statins in acute stroke have been extrapolated - often with uncertain results - especially from retrospective studies, while few randomized clinical trials are currently available [
60]. Consequently, current guidelines lack of some precise information, in particular concerning: their indication in non-atherothrombotic stroke [
48], their recommended dose [
60], the time to start therapy [
8,
58‐
62], their rate of side effects [
12,
40,
57], the absolute contra-indications [
48], the other concomitant therapies [
12,
49‐
53], the recommended approach with respect to old age (in RCTs concerning secondary prevention with statins the mean age is <65 years [
12]).
In this paper we have not explored the physician-related factors influencing statin prescription but, reviewing the literature in order to compare our results with other experiences, we found more than a few grey areas among recommendations. We can hypothesize that the low statin prescription rate by physicians may be partially due to these controversies.
Limits of the study
This study has some limitations. First, patient factors such as patient preferences and agreement to treatment are not addressed. As a matter of fact, these variables are not usually collected in a registry, and thus this information was not available in this study. Also, this study only provides the categorical data of vascular risk factors (e.g. diabetes mellitus, dyslipidemia, etc.) rather than the raw values of the biological parameters (e.g. blood glucose, total cholesterol, HDL, LDL etc.). The LSR Registry is a repository where mostly “qualitative” and process information is stored, focusing on the presence of a diagnosis rather than on its entity. Then, type and dosage of statin is not precisely reported in LSR, which may cause potential bias in statistical analysis. This is due to the Registry structure, modeled on the form of current clinical practice guidelines for stroke that in fact are not specific as well, recommending to use a statin without specifying which type.
Moreover, we performed the analyses without considering the type of providing treatment specialist, which might impact on the prescription decision-making. The reason why we did not consider such parameter is that the distribution of treatment specialty is very uneven, with only one center managed by internal medicine specialists, one whose management shifted during the study period and 40 managed by neurologists. In the internal medicine Stroke Unit, the percentage of statin prescriptions was higher (65.7%) than the overall average percentage (42%), in the one where the management shifted from Internal Medicine to Neurology it was slightly above the average (43.14%). As regards the centers managed by neurologists, the statins prescription rate has a high variability, with values ranging from 15.8% to 82.8%. From this data it is difficult to evaluate the role of the treating specialty as a predictor of prescription.
Although our Registry is provided with a follow-up section, in this study we have not included information on post-discharge statins use. The motivation underlying this choice is twofold. On the one hand, the follow-up is only partially completed for our patients. On the other hand, for those patients who died during follow-up, the registry doesn’t allow physicians to enter information about treatments in use before death. This would cause a bias, as the patients who died would result as not continuing the treatment even if they were.
Acknowledgements
We would like to thank all the Lombardia Stroke Registry collaborators for their support and help in the realization of the LSR database. In alphabetical order:
Laura Adobbati (‘Auxologico Italiano’ IRCCS, Milan), Elio Agostoni (‘Manzoni’ Hospital, Lecco), Marco Arnaboldi (‘Sant’Anna’ Hospital, Como), Gianstefano Baietti (Cremona Hospital), Ubaldo Balducci (Chiari Hospital), Pierluigi Baron (‘Ca’ Granda’ Policlinico, Milan), Pietro Bassi (‘San Giuseppe’ Hospital, Milan), Paola Bazzi (Vimercate Hospital), Simone Beretta (Monza Hospital), Luciano Bet (‘San Donato’ Hospital, Milan), Luigi Bettoni (Cremona Hospital), Giacomo Bezzi (Sondrio Hospital), Giorgio Boncoraglio (‘Carlo Besta’ IRCCS, Milan), Giorgio Bono (‘Fondazione Macchi’ Hospital, Varese), Bianca Bordo (Desio Hospital), Giuseppina Borutti (Voghera Hospital), Nereo Bresolin (‘Ca’ Granda’ Policlinico, Milan), Fabio Brusaferri (Crema Hospital), Maria Vittoria Calloni (Legnano Hospital), Massimo Camerlingo (‘San Marco’ Policlinico, Zingonia), Bruno Censori (‘Riuniti’ di Bergamo Hospital), Carlo Cerini (Esine Hospital), Nicoletta Checcarelli (‘Valduce’ Hospital, Como), Francesco Chia (Chiari Hospital), Antonio Colombo (Desio Hospital), Giancarlo Comi (‘San Raffaele’ Hospital, Milan), Angelo Costa (‘Spedali Civili’ Hospital, Brescia), Vittorio Crespi (Vimercate Hospital), Maria Luisa Delodovici (‘Fondazione Macchi’ Hospital, Varese), Edoardo Donati (‘Poliambulanza’ Hospital, Brescia), Claudio Ferrante (Ponte San Pietro Hospital), Carlo Ferrarese (Monza Hospital), Fabio Frediani (‘San Carlo Borromeo’ Hospital, Milan), Paola Gambaro (‘Luigi Sacco’ Hospital, Milan), Antonella Gomitoni (Saronno Hospital), Giampiero Grampa (Saronno Hospital), Mario Guidotti (‘Valduce’ Hospital, Como), Patrizia Lattuada (‘San Carlo Borromeo’ Hospital, Milan), Ezio Lanza (Treviglio-Caravaggio Hospital), Andrea Magnoni (Merate Hospital), Simona Marcheselli (‘Humanitas’ Clinical Institute, Rozzano), Claudio Mariani (‘Luigi Sacco’ Hospital, Milan), Claudia Marsile (Vizzolo Predabissi Hospital), Alessandra Martignoni (‘San Matteo’ Policlinico IRCCS, Pavia), Marco Mattioli (Garbagnate Hospital), Giovanni Meola (‘San Donato’ Hospital, Milan), Graziella Molini (Vizzolo Predabissi Hospital), Irene Neromante (Busto Arsizio Hospital), Emanuele Poloni (‘San Carlo Borromeo’ Hospital, Milan), Eugenio Parati (‘Carlo Besta’ IRCCS, Milan), Giovanni Piccolo (Voghera Hospital), Marco Poloni (‘Riuniti’ Hospital, Bergamo), Daniele Porazzi (Busto Arsizio Hospital), Alessandro Prelle (Crema Hospital), Paolo Previdi (Mantova Hospital), Maurizio Riva (Lodi Hospital), Alessandro Romorini (Magenta Hospital), Ignazio Santilli (‘Niguarda Ca’ Granda’ Hospital, Milan), Francesco Sasanelli (Vizzolo Predabissi Hospital), Chiara Scaccabarozzi (‘Manzoni’ Hospital, Lecco), Maria Sessa (‘San Raffaele’ Hospital, Milan), Vincenzo Silani (‘Auxologico Italiano’ IRCCS, Milan), Giorgio Silvestrelli (Mantova Hospital), Carlo Sebastiano Tadeo (‘Città Studi’ Clinical Institute, Milan), Marinella Turla (Esine Hospital), Simone Vidale (‘Sant’Anna’ Hospital, Como), Maria Teresa Zaccone (‘Beato Matteo’ Clinical Institute, Vigevano), Davide Zarcone (Gallarate Hospital), Angelo Zilioli (Lodi Hospital).