Assessment of both GI and CV risks vs. the benefits of low-dose aspirin for individual patients can be difficult in clinical practice.
Assessment of both GI and CV risks vs. the benefits of low-dose aspirin for individual patients can be difficult in clinical practice.
To develop a tool to estimate CV and GI risks to facilitate the clinical decision-making process.
We constructed risk-ratio estimations and determined the incidence of CV events and upper GI complications according to the presence of different risk factors. For upper GI complications we assumed a baseline incidence of 1 case/1000-persons-year, a twofold increased risk with low-dose aspirin, and estimated a 60% GI risk reduction with proton pump inhibitors (PPI) co-therapy and a 60% risk reduction with H. pylori eradication in patients with a history of peptic ulcer.
The calculator can be found at http://www.asariskcalculator.com. In patients with low CV risk the number of GI complications induced by low-dose aspirin may be greater than the number of CV events prevented. In patients with high CV risk, low-dose aspirin is recommended, but the number of GI complications induced may still overcome the CV events saved. The use of PPI reduces the number of complication events induced by low-dose aspirin, but the number of CV events saved may still be offset by the number of GI complications induced in patients at very high GI risk.
There are many clinical situations where the number of potential upper GI complications induced by low-dose aspirin may exceed the number of potentially prevented CV events. A risk calculator should guide physicians in choosing appropriate therapy and maximise the aspirin benefit.
Cardiovascular disease is the leading cause of mortality in the word. The cardiovascular (CV) benefits of low-dose aspirin (LDA) in primary and secondary prevention of cardiovascular events may be offset by its association with major upper gastrointestinal (GI) complications.[2-4] Although LDA (dose of ≤325 daily) has been associated also with GI perforation, this is a rare event and the vast majority of complications are bleeding events from the upper GI tract (oesophagus, stomach and duodenum) manifested as the presence of melena, haematochezia and⁄or haematemesis confirmed by hospital staff.
The benefits of LDA in terms of reduction in the number of CV events are clear in the secondary prevention of CV events.[3, 5-7] In contrast, for primary prevention of CV events, the net benefits of LDA have been questioned as the risk of developing GI complications could outweigh the net CV benefits.[7-10] Nevertheless, in clinical practice, LDA, alone or combined with other anti-thrombotic drugs, is increasingly prescribed for both primary and secondary prevention. Also, the effect of aging population with several comorbidities is likely to increase its use. Clopidogrel, may reduce further the combined risk of ischaemic stroke, myocardial infarction, or vascular death by 8% more than LDA alone in patients at high risk for atherothrombotic events. The use of oral anticoagulation combined with LDA in patients with coronary artery disease reduces the risk of CV events by 56%, although the episodes of major bleeding are increased 1.9-fold.
The risk of upper GI bleeding is not uniform in patients requiring LDA therapy and depends on the presence of risk factors. Age, gender, uncomplicated ulcer history, complicated ulcer history, history of dyspepsia, H. pylori infection and co-therapy with clopidogrel, anticoagulants, corticosteroids or nonsteroidal anti-inflammatory drugs (NSAIDs) are associated with an increased risk of upper GI bleeding in patients receiving low-dose aspirin. Furthermore, 5–12% of patients who develop an ulcer bleeding event will eventually die due to uncontrolled bleeding or more probably to comorbidities aggravated during the bleeding event.[15, 16]
Because the evidence based on well-designed and controlled studies is not well developed in this field, expert consensus of two of the major Digestive and Cardiology USA societies have been published. The consensus recommends the use of a proton pump inhibitors (PPI) in patients receiving LDA who have GI risk factors. However, risk factors for GI bleeding in LDA users are not well defined and may be unknown by those who prescribe it. Furthermore, the assessment of both GI and CV risks and the benefits of LDA for any individual patient may be difficult to determine in clinical practice. To overcome some of these problems, we aimed to develop a tool to estimate both CV and GI risks for individual patients to facilitate the therapeutic decision-making process in routine clinical practice.
Based on currently available publications, we constructed risk-ratio estimations and determined incidences of CV and upper GI complications (bleeding) according to the presence of different risk factors. We first performed an evidence-based review of the literature using MEDLINE and EMBASE databases from 1 January 1988 to May 2011 to identify studies reporting effects of LDA in the prevention of CV events. We also searched for studies reporting upper GI complications associated with LDA treatment for the prevention of cardiovascular diseases.
The literature search was based on combinations of the following terms: ‘aspirin’, and ‘cardiovascular’ and ‘myocardial infarction’ or ‘stroke’ and ‘gastrointestinal bleeding’ or ‘gastrointestinal complications’. The limits were human, publication dates and randomised controlled clinical trial. The languages were limited to English and the five major Western European languages (French, German, Italian, Spanish and Portuguese). In addition, the Cochrane Library was searched for any recent systematic review or meta-analysis of the topics of interest. A manual check of the reference lists of all accepted studies and meta-analyses was performed to supplement the above searches and ensure optimal and complete literature retrieval.
The search provided 2438 potential relevant citations, of which 249 were selected for review after the reading of the abstracts. After review of the full text only 68 studies (8 primary, 60 related studies) were considered relevant to the purpose of the study (Figure 1). All these studies were assigned a level of evidence using the schema of evidence assignment developed by the Centre for Evidence-Based Medicine in Oxford, United Kingdom. Data were collected on paper extraction forms by two investigators (AL, RC) and independently verified by a third investigator (MP) and catalogued in a database. We eventually focused on eight studies that provided risk estimations and event incidences, and identified publications that reported data according to the presence of well-known cardiovascular and GI risk factors. We then used the best estimates of those studies for the beneficial and harmful effects of LDA in CV prevention to model the impact of LDA use in different common clinical scenarios. The protocol was approved by the Institution Research and Ethical Committee of our institution.
For CV risks we used the Framingham 10-year risk estimates; calculations were based on the method reported by Wilson et al. for the prediction of CV heart disease (angina pectoris, recognised and unrecognised myocardial infarction, coronary insufficiency and coronary heart disease death) (http://www.framinghamheartstudy.org/risk/coronary.html#tab3). It was assumed that the risk remains constant during the 10-year period to obtain a 1-year CV risk.
The recommendation for LDA therapy depends on the accurate assessment of CV risk as part of the decision-making process. Six primary prevention trials and a meta-analysis with a mean follow-up of approximately 6 years provided an overall 18% proportional reduction in major coronary events (rate ratio: 0.82; 95% CI: 0.75–0.90) and a 23% reduction (0.77; 95% CI: 0.67–0.89) in nonfatal myocardial infarction. The effects of LDA were more pronounced in patients with higher CV risk.[4, 5, 7, 8, 13, 19, 20] We have assumed a 15–20% CV risk reduction with aspirin; a 20% CV risk reduction in the case of predicted 10-year risk of coronary heart disease of ≥10%, and 15% CV risk reduction in the case of moderate or low predicted coronary heart disease risk (5–10% and 2.5–5%, respectively).[6, 7] Clopidogrel combined with LDA reduced the combined risk of ischaemic stroke, myocardial infarction or vascular death by 8% more than LDA alone in patients at high risk for atherothrombotic events (0.92; 95% CI: 0.86–0.99). Risk reduction with the addition of clopidogrel to aspirin was evident in secondary prevention.[21, 22]
The effect of LDA for primary prevention of CV events in adults with diabetes is currently unclear, due to the fact that available trials report mixed results. LDA use for primary prevention is recommended for adults with diabetes and no previous history of vascular disease who are at increased CVD risk (10 year risk of CVD events greater than 10%) and who are not at increased risk for bleeding.
For GI risk estimations we used data reported by Hernandez-Diaz and Garcia-Rodriguez as a baseline for the construction of tables and algorithms. This study characterised LDA users in terms of major GI risk factors and provided incidence rates as well as excess risk of upper GI complications linked to LDA based on the General Practice Research Database (UK) and systematic reviews of the literature. Based on their results, we assumed an incidence rate for overall baseline upper GI bleeding of 1 per 1000 person-years. We then constructed absolute incidence rates within each risk subgroup based on pooled estimates reported from meta-analysis of others and our own studies. The risk of upper GI bleeding associated with LDA has been reported among a range of RRs between 1.5 and 3.0 in many studies. Here, we have followed the assumption provided by Hernandez-Diaz and Garcia-Rodriguez that the pooled relative risk of upper GI bleeding was 2.0 for LDA, considered as doses ≤325 mg/day. Major risk factors for the development of upper GI bleeding are: age, male gender, history of peptic ulcer and concomitant use of NSAIDs, anticoagulants, or clopidogrel. Table 1 summarises the relative risk and 95% CI estimates and the main bibliographic sources supporting the reported estimates.
|Risk factor||RR point estimators (95% CI)|
|Past UGI pain/Dyspepsia||2.0 (1.8–2.2)|
|Peptic ulcer history||5.9 (5.2–6.7)|
|In NSAID user|
|No previous history/Dyspepsia||5.0 (4.5–5.5)|
|Peptic ulcer history||2.5 (2.1–3.1)|
|LDA (≤325 mg)[9, 10, 24]||2.2 (2.1–2.4)|
|LDA + Clopidogrel use||3.7 (2.4–5.8)|
|Warfarin + LDA||6.5 (4.2–6.9)|
|Proton pump inhibitor[26, 38]||0.4 (0.2–0.6)|
Age was considered a risk factor with variables estimated depending on the age range. The incidence increases exponentially after the age of 50 up to a relative risk of 9.2 for patients aged 80 or older. The risk and incidence rates among men and women are different, with men having a twofold increase compared with women in all age groups. Together with older age, ulcer history is the primary risk factor, with a higher risk among patients with a history of bleeding ulcer. History of dyspepsia was also considered a risk factor within this group because several studies found this variable to be associated with an increased risk of upper GI bleeding; it is probably a subclinical marker of peptic ulcer in investigated individuals, especially in areas with substantial H. pylori infection rates. NSAID use is another important risk factor among patients who use aspirin; NSAIDs cause a fourfold increase in the risk of upper GI bleeding. In patients with a history of peptic ulcer disease, the absolute risk is higher but the relative risk lower, as described widely elsewhere.
There is wide consensus that antiplatelet and/or anticoagulant agents increase the risk of upper GI bleeding. We have assumed, as did others,[6, 24, 25] that LDA doubles the risk of upper GI bleeding in each of the risk groups described above. Hernandez and Garcia Rodriguez validated most of these assumptions in control groups of NSAID outcome trials. Other drugs that are widely used in CV treatments are clopidogrel and warfarin. Based on recent data, clopidogrel use has been reported to have a risk profile similar to aspirin for upper GI bleeding, whereas warfarin in general produces a higher risk. Combinations of these compounds further increase the risk, which may be additive (e.g. clopidogrel plus aspirin) or greater (aspirin plus warfarin).
Co-therapy with PPI reduces the risk of upper GI complications among LDA users. Risk reduction in case–control and cohort studies ranges in most cases from 40 to 80%. A recent meta-analysis of the three available randomised controlled studies found similar results with over 60% risk reduction. Other antisecretory or anti-ulcer drugs include H2- receptor antagonists and misoprostol, but the available evidence in the prevention of GI complications with low-dose aspirin is reduced (misoprostol) and somewhat inconclusive.
Helicobacter pylori infection is a risk factor for upper GI bleeding in low-dose aspirin users. The effect of H. pylori eradication on the reduction of peptic ulcer or peptic ulcer bleeding recurrence is well established but the evidence of risk reduction for upper GI bleeding in low-dose aspirin users is less consistent. Two small studies investigated the effect of H. pylori eradication on upper GI bleeding in high-risk patients with previous ulcer bleeding. One found that the effect was similar to that seen with omeprazole in patients with H. pylori infection, whereas another study found that the re-bleeding rate was vastly reduced (>80%) only in patients treated with a combination of H. pylori eradication and PPI therapy.
Coronary heart disease risks were estimated using the survival model described by Wilson et al. in the Framingham Heart Study. The exact survival model was recreated for the calculator. The predictors described by Wilson et al. were age, diabetes, smoking, hypertension (categorised according to blood pressure readings by JNC-V, Joint National Committee definitions), and cholesterol levels [according to NCEP, National Cholesterol Education Program, based on total cholesterol and low density lipoprotein (LDL) cholesterol]. Note that these predictors are different for men and women. All criteria and limitations of the model apply.
As noted above, there were no exhaustive studies on GI risks, nor was there any survival data on this topic. We thus decided to use a different approach using the information available in the literature. A baseline incidence estimation of upper GI bleeding (1/1000 persons-year) was available as well as individual risk estimations for the factors of interest: age, gender, ulcer history, NSAID use, low-dose aspirin use or warfarin plus aspirin use and clopidogrel. A multiplicative model was calculated using these estimates. For those risks not present in a patient, the corresponding coefficient of that risk became one. If the risk was present, the reported estimation value was used. Thus, for a patient with no risk factors the estimated GI risk is the baseline risk of the general population; risk increases linearly with each risk factor by the coefficient estimated for each risk based on data reported in Table 1 and the Hernandez and Garcia Rodriguez report. Also the addition of NSAIDs increases further the risk of bleeding. It was assumed that there were no sex differences for the GI risk factors, but males had a twofold increase risk of GI bleeding compared with females.
An important aspect of the algorithms was the effect of treatments that reduce the risk of upper GI complications and that are the basis for current treatment recommendations. We assumed conservative estimates for the effect of PPI (60% risk reduction) and considered that H. pylori eradication effects should only be applied to those with ulcer history, since there is no consensus yet that H. pylori eradication should be pursued in all patients with H. pylori who take LDA. It was also assumed that when the risk is high and based on a history of peptic ulcers, H. pylori eradication should not be the only effective measure to reduce aspirin-associated GI risk. Based on current evidence,[25, 28, 29] these patients need additional PPI therapy, and therefore the GI risk reduction effect attributed to H. pylori eradication in these patients on LDA was added on top of the risk reduction effect obtained with the PPI. Based on a similar effect to PPI therapy found in one study, our best estimation was a 60% upper GI bleeding risk reduction with H. pylori eradication in patients with ulcer history taking low-dose aspirin (Table 1).
We carried out a sensitivity analysis using a range of effects within the 95% intervals of reported estimators. The analysis was only performed for GI events, since the impact of LDA on them has been less tested and accepted compared with the effect of LDA on the CV risk. Therefore, we considered different risks for the effect of aspirin on GI bleeding and the effects of PPI on that risk, since it is the factor that has a major impact in terms of prevention of GI bleeding associated with ASA use. Different clinical scenarios with GI highest/lowest risk and lowest/highest PPI effects were constructed.
Based on the best available risk estimators and algorithm models described, we constructed an automatic calculator available on-line (http://www.asariskcalculator.com) where data from individual patients can be entered and the CV and GI risks calculated. Required data include age, gender, total cholesterol level, LDL cholesterol, blood pressure, smoking, presence/absence of diabetes, CV treatment (aspirin, clopidogrel, anticoagulants) and GI history (including absence of GI history, dyspepsia, uncomplicated ulcer and complicated ulcer).
To evaluate the discrimination of the risks provided by the calculator we have tested retrospectively the risks of two different cohorts of patient; one cohort was of patients who developed peptic ulcer GI bleeding while taking long-term LDA; the other one was age and sex-matched controls also taking long-term LDA but who had not GI bleeding complications. These cohorts of patients were collected from a multi-centre national registry.
The calculator provides both the basal CV and GI risks for the individual patient. Also, the number of CV events that LDA treatment could save in terms of 1000 patient-year and also the estimated excess number of upper GI events induced by LDA in the same terms. This information should aid the consultant doctor in weighing the benefits and risks of aspirin use for individual patients.
Although the weight and severity of CV and GI events may be considered different, if the estimation of the number of upper GI complications induced by LDA exceeds the number of CV events saved, the calculator recommends the addition of a PPI to improve the benefit balance. The calculator then will provide to the consultant new figures on the number of upper GI complications reduced by co-therapy with PPI and the number of CV events prevented with aspirin. This should be crucial in the decision to use LDA for primary prevention, but will also provide figures for estimation of GI risk in secondary prevention and evaluation of the opportunity to modify additional risk factors. In case of ulcer history, reducing the risk of upper GI complications and the beneficial effect of additional therapy will require knowledge of H. pylori infection status. In this case the calculator will provide the additional number of upper GI complications prevented with H. pylori eradication on top of those already prevented by PPI treatment. Overall, this calculator should provide the consultant all necessary elements to determine a patient's CV and GI risks, and to weigh the benefits and risks of aspirin with or without additional therapies, having in mind that the severity of the events and prognosis implications of them may be different and eventually affect the final clinical decision.
In cases of low CV risk (≤10% in 10 years) the use of LDA induces more GI complication events than saves CV events in almost all clinical scenarios. Concomitant PPI use with or without H. pylori eradication may overcome the increased risk associated with LDA, but will not be enough to overcome the overall (basal + LDA-induced events) risk of upper GI complications. In patients with a CV risk greater than 20%, LDA is recommended by guidelines. However, depending on the presence of GI risk factors, the number of upper GI complication events induced by LDA may overcome the number of CV events prevented. The addition of a PPI and H. pylori eradication re-balance the equilibrium in favour of a beneficial effect of LDA in most clinical scenarios. Nevertheless, in patients with complicated peptic ulcer, while the combined effect of both therapeutic actions (PPI + H. pylori eradication) may reduce and/or eliminate the GI events induced by LDA, the estimated overall risk of number of upper GI complication events over a 10-year period (basal risk + aspirin induced) is still greater than the number of CV events prevented. Table S1 (online supporting information only) illustrates and summarises the effects of LDA use on different CV and GI clinical scenarios for patients with different CV and GI risks.
We conducted a sensitivity analysis with a variety of clinical scenarios and drug effects considering that LDA and PPI therapy are the main factors affecting GI risk. We estimated fixed CV benefits with LDA and variable GI events induced by aspirin within a relative risk range from 1.5 to 3.0. The effects of PPI treatment in turn modified these estimations with a relative risk range from 0.2 to 0.6. Figure 2, and Figure S1 and S2 provide data on sensitivity analysis based on age, gender, the presence/absence of ulcer history and the impact of aspirin and PPI on GI events induced for different clinical situations. The figures also includes the range of extra number of GI events prevented with PPI treatment (see supporting information).
As described in 'Aim', we tested retrospectively the estimates of GI events provided by the calculator in two different cohorts of patient; one cohort was of patients who developed peptic ulcer GI bleeding while taking long-term LDA, the other one was of age and sex-matched controls also taking long-term LDA but who had not had a GI bleeding complications. The estimated number of upper GI complication events (Basal + ASA induced) predicted by the calculator was statistically and significantly higher in the cohort of patients who developed an upper GI peptic ulcer bleeding when taking LDA compared with their matched controls also taking LDA who had not presented the bleeding event up to the moment of the transversal observation (Figure 3).
The clinical decision-making process during patient management must take into account the benefits and risks of different diagnostic or therapeutic options. The plethora of information available and the complex interaction of many treatments on different body systems can be difficult to fully examine. The management of patients with increased CV risk presents a situation where the use of drugs that reduce CV risk may increase the risk of other serious complications, mainly involving the GI tract. Furthermore, physicians who are the usual prescribers of these medications for the primary disease (CV) are not specialist or may not be able to appropriately evaluate the associated GI risks. The potential need of other therapies, such as gastroprotective agents or the elimination of H. pylori infection further complicates the decision-making process. Although recent evidence-based expert recommendations have tried to address these issues, they are still general recommendations that often are difficult to translate into clinical practice. Other approaches have addressed the cost utility of LDA treatment with or without PPI at different risks for CV diseases and gastrointestinal bleeds. The conclusion of their study is that PPI is not cost-effective for men with average gastrointestinal bleed risk but may be cost-effective for selected men at increased risk for gastrointestinal bleed, which basically agrees with the recommendations the calculator will provide.
Our study was developed to provide recommendations at the individual patient level and to facilitate the clinical decision process for prescribers of aspirin for CV disease. By inputting basic clinical and laboratory data from any given patient, the calculator developed will provide estimations of both CV and GI event rates over 10 years time and the effects of LDA therapy on these rates. Estimations are based on the well-known Framingham risk score for the CV system and on the best available evidence concerning GI risk. Based on these data, the calculator will provide recommendations for the use of LDA or not, with or without gastroprotection via PPI and H. pylori eradication. Gastroprotection is recommended when the number of GI events induced by LDA over the baseline rates outnumbers the reduction in CV events. This may be seen as a simplistic therapeutic approach, as the weight in terms of severity (e.g. risk of death) or prognosis of a CV event can be seen as more relevant than a GI event. However, it is also clear that developing a GI bleeding event jeopardises the correct treatment of patients who need LDA and increases the risk of having a recurrent MI if stopped. Besides, the recommendations the calculator will provide at individual levels conform to the more general recommendations of major guidelines, including those for patients with diabetes and more recent recommendations given by expert consensus panels from both GI and CV societies to prevent upper GI tract complications during low-dose aspirin and antiplatelet therapy.[3, 25]
There are several advantages for using a system such as this CV/GI risk calculator. First, it can be used at any office level to obtain a rapid estimation of the CV risk and the need to use or not use low-dose aspirin. In addition to other calculators that already perform this, our calculator also offers estimations of the GI risks associated with LDA use and recommendations to reduce that risk. Second, the calculator will provide estimations of the rates of CV and GI events based on different patient factors and therapeutic options, which allows the healthcare provider to evaluate the benefits and risks to facilitate clinical decision-making. Third, by fitting with more broad recommendations of available guidelines in terms of LDA indication to prevent CV events and gastroprotection to prevent GI events, the calculator will provide evidence-based advice consistent with the most up to date clinical practice. Fourth, the calculator has proved to discriminate risk levels in this first approach study conducted with data of 904 patients taking LDA. Furthermore, since the calculator can be easily adapted to new evidences, recommendations can be updated to reflect new clinical data.
Over recent years there has been a heated debate over whether the pharmacokinetic drug interaction between clopidogrel and proton pump inhibitors reduces the antiplatelet effects of clopidogrel and leads to an increased risk of vascular events. The COGENT (Clopidogrel and the Optimization of Gastrointestinal Events) trial was a large randomised controlled clinical trial comparing a clopidogrel/omeprazole combination pill with clopidogrel alone. The conclusion was that there was no difference in the incidence of cardiovascular death and myocardial infarction between the two groups (HR for omeprazole 0.99, 95% CI: 0.68–1.44). By contrast, omeprazole significantly reduced gastrointestinal events (HR: 0.34, 95% CI: 0.18–0.63), suggesting substantial net benefit. Also, in a recent observational cohort study for 24 471 patients, it has been showed that this interaction is not clinically important.
In this article, the use of anti-H2 has not been described as a therapy for GI bleeding prevention. A meta-analysis showed that H2RA did not significantly reduce the risk of symptomatic ulcers among NSAID takers.[32, 33] According to this data, several scientific societies have recommended the use of PPI of drug of choice for GI bleeding prevention.[17, 25, 34]
This article has several limitations that must be understood by those who use it. We are using the 10-year CV risk estimation instead of the 2 year (Weibull model) that could provide a closer estimation of immediate risk. We preferred to apply the most frequently used table to estimate CV risk and the most accurate rates provided by this method. Although other scores are used in other parts of the world (e.g. Europe), the risk factors are basically the same. SCORE provides CV risk of death; we chose to use the risk of CV events, which could be more easily compared with GI events, as we currently lack reliable data on GI deaths in this context. As we have mentioned above, to compare number of CV events with GI events must be seen with caution, since the severity and prognosis can be seen as very different, but we have already suggested that both have serious clinical implications and both need to be eventually weighed by the prescribing physician, especially in primary prevention. Also, it must be understood that the CV risk can be modified with the use of other effective treatments, including statins, blood pressure control and smoking cessation, which are strongly recommended by multiple different guidelines. Although the risk reduction with these additional therapies does not occur immediately, their effects can be assumed to occur with rapidity sufficient to incorporate them in the initial decision-making process. An important consideration is that patients may acquire additional risk factors over time, which would necessitate a reassessment of their overall risk profile. CV risk tools, including this one, are primarily constructed to provide estimations for primary prevention of CV events with LDA, which is still the subject of debate concerning benefits and risks. The current tool will facilitate this process. We have introduced variables with drugs (e.g. clopidogrel, warfarin) that although are more often used in secondary prevention, they can also be used in primary prevention. Furthermore, although the tool will not provide accurate CV risk estimations for those who need LDA for secondary prevention, these patients must be considered at high risk and the tool allows for estimation of GI risk. Other limitations include the fact that, contrary to extensive literature data and various tools for the evaluation of CV risk, estimations of GI risk are much less developed. Although there are many publications concerning risk factors for GI complications in patients using vs. not using NSAIDs, there is less data for LDA users. Indeed, the evidence is poorly organised and some reported factors need confirmation. To overcome this limitation we have used, as baseline, GI risk factors reported by Hernandez-Diaz and Garcia-Rodriguez as the first approach to provide rates of GI complications, since it was based on data from solid databases. Based on this study we made our own estimations, as explained in the methods section, using the best available evidence concerning the effects of drugs and risk factors. Of all factors affecting GI risk, H. pylori infection probably has the least evidence supporting the effect of treatment in this patient population. We preferred to be conservative in estimating its protective effect and limited the recommendation to patients with ulcer history. New evidence in this field will easily be incorporated in the model. In this calculator, we assumed a lack of interaction of PPI with clopidogrel in terms of CV risk, as the clinical data do not support this and within the PPIs, there are a variety that are recommended by regulatory authorities for use due to the absence of evidence of clear interaction.
It is also possible that due to the limited precision of some of the estimations, either the CV or the GI excess number of events provided could be different and therefore the balance between the number of CV events and the number of GI events can be different depending on which end we compare from. In this way, we have conducted sensitivity analysis for the effect of aspirin and PPI in the GI risk/GI prevention effect, but not for the CV events, where LDA provides a widely proved but more narrow benefits, which we believe will have a much lower impact on the outcomes than in the GI side. The sensitivity analysis shows that in all cases the main conclusions and recommendations can be sustained. The beneficial effect of LDA that goes beyond the cardiovascular has not been included in this calculator. It should be noted that this effect also extends to the prevention of mortality due to cancer. This effect has been quantified in an absolute 20-year reduction in risk of cancer death reaching 7.08%.
Finally, the calculator allows the exercise of introducing clinical situations with maximal CV risk and maximal GI risk, which may result in ‘artificial’ clinical scenarios providing illogical rates of events. We discourage this because those patients are rarely or never seen in clinical practice. It must be understood that future work with this tool is needed to validate its capacity to provide accurate prognostic value at both GI and CV levels and to provide evidence of current clinical practice and outcomes among other options. Also, further research into the pathogenesis H. pylori infection will clarify the relation between low-dose aspirin and upper gastrointestinal damage.
In conclusion, the use of algorithms to integrate the stratification of individual CV risk with that of upper GI risk is an important clinical situation to consider in patients with suspected coronary artery disease. A proper quantification of GI risk should identify patients who may benefit more from the avoidance of LDA therapy, or from the addition of appropriate therapeutic modifications to avoid complications and obtain the maximal benefit of LDA. For this reason, the use of the aspirin CV/GI risk calculator should guide physicians in choosing appropriate treatment for primary CV prevention.
Guarantor of the article: A. Lanas.
Author contributions: None. All authors approved the final version of the manuscript.
The authors are indebted to Patricia Carrera and Carlos Sostres from for their expertise and efforts in the analysis of the data. We also thank Felipe Arroyo from Azormultimedia for constructing the web of the calculator based on the data and algorithms provided. Declaration of personal interests: Angel Lanas has served as consultant to AstraZeneca, Pfizer, and Bayer, and has received research funding from AstraZeneca and Pfizer for investigator-initiated research.
Declaration of funding interests: Funds for the study were provided by the Government of the Autonomous Region of Aragón, Spain (B01). RC was funded by a grant for international research post-residency, Spanish Society of Cardiology. Mónica Polo-Tomas was an employee of the CIBERehd working for the Group of Digestive Diseases of the IIS Aragón.