Article Text

Extended report
Initiation and adherence to secondary prevention pharmacotherapy after myocardial infarction in patients with rheumatoid arthritis: a nationwide cohort study
  1. Jesper Lindhardsen1,
  2. Ole Ahlehoff1,
  3. Gunnar Hilmar Gislason1,
  4. Ole Rintek Madsen2,
  5. Jonas Bjerring Olesen1,
  6. Christian Torp-Pedersen1,
  7. Peter Riis Hansen1
  1. 1Department of Cardiology, Gentofte University Hospital, Copenhagen, Denmark
  2. 2Department of Internal Medicine – Rheumatology section, Gentofte University Hospital, Copenhagen, Denmark
  1. Correspondence to Jesper Lindhardsen, Copenhagen University Hospital Gentofte, Department of Cardiology, Hellerup 2900, Denmark; jeslin01{at}


Objectives To examine whether rheumatoid arthritis (RA) is associated with less optimal secondary prevention pharmacotherapy after first-time myocardial infarction (MI).

Methods The authors identified all patients with first-time MI in the Danish National Patient Register from 2002 to 2009 and gathered individual level information including pharmacy records from nationwide registers. Initiation of standard care post-MI secondary prevention drugs, that is, aspirin, β-blockers, clopidogrel, renin angiotensin system (RAS) blockers and statins, was determined after discharge. In addition, adherence to each drug was evaluated as the proportion of patients on treatment during follow-up and time to first treatment gap.

Results A total of 66 107 MI patients (37% women) were discharged alive; 877 were identified as RA patients (59% women). Thirty days after discharge, RA was associated with significantly lower initiation of aspirin (OR 0.80 (0.67–0.96)), β-blockers (0.77 (0.65–0.92)) and statins (0.69 (0.58–0.82)), while initiation of RAS blockers (0.80 (0.57–1.11)) and clopidogrel (0.88 (0.75–1.02)) was non-significantly reduced. These estimates were virtually unchanged at day 180 and the results were corroborated by Cox regression analyses. Adherence to statins was lower in RA patients relative to non-RA patients (HR for treatment gap of 90 days: 1.26 (1.07–1.48)), while no significant differences were found in adherence to the other drugs.

Conclusions In this nationwide study of unselected patients with first-time MI, a reduced initiation of secondary prevention pharmacotherapy was observed in RA patients. This undertreatment may contribute to the increased cardiovascular disease burden in RA and the underlying mechanisms warrant further study.

Statistics from


Patients with rheumatoid arthritis (RA) are at increased risk of cardiovascular disease (CVD) compared with the general population, which substantially contributes to their significantly higher mortality.1,,3 Increased prevalence of classical risk factors for CVD cannot fully explain this observation, and although other explanations have been put forward such as inflammation-driven accelerated atherothrombosis and adverse effects of pharmacological RA treatment, only few reports have focused on underuse of cardioprotective drugs as a contributor to the increased risk of CVD in RA.3,,5 Recent data have shown that the excess risk of CVD conferred by RA is under-recognised by primary care physicians and undertreatment of dyslipidemia and hypertension has also been reported in patients with RA.6,,9 In general, secondary prevention therapy after myocardial infarction (MI) is considerably more implemented than primary prevention, but nonetheless it was recently found that inhospital post-MI management including reperfusion therapy and optimal medical treatment was less provided for RA patients.10 However, information on out-of-hospital post-MI secondary prevention pharmacotherapy in RA patients compared with non-RA patients is currently lacking. Since undertreatment of RA patients after MI is likely to worsen prognosis and may indicate a more general undertreatment of CVD in RA patients, we compared the use of post-MI secondary prevention pharmacotherapy in RA and non-RA patients in a register-based nationwide Danish cohort study.


Data sources

Individual level information was gathered from a number of nationwide registers in Denmark and combined by use of a unique and permanent personal identifier issued at birth or immigration. Data concerning vital and residency status as well as income were obtained from Statistic Denmark. Inpatient (from 1978) and outpatient (from 1995) hospital data were collected from the Danish National Patient Register, which contains the date and treatment facility along with one primary diagnoses and, if appropriate, one or more secondary diagnoses for each individual contact. The register diagnoses were coded according to International Classification of Disease with use of the eighth revision (ICD-8) from 1977 until 1994 and the tenth revision (ICD-10) from 1994 to 2009. Information on medication was obtained from the Danish Register of Medicinal Products Statistics, where all dispensed prescriptions in Denmark have been recorded with high accuracy since 1995 coded by the Anatomical Therapeutic Chemicals (ATC) classification along with drug strength and quantity.11

Study participants

The study cohort consisted of all Danish individuals who were discharged alive after MI (ICD-10: I21–I22) from 2002 to 2009. To ensure a consistent treatment guidelines in the study period, the earliest year of entry into the cohort was 2002, where secondary prevention pharmacotherapies with aspirin, clopidogrel, β-blockers and statins were established standard of care in the management of all patients after MI regardless of the characteristics of the individual patient in absence of contraindications to the respective drugs, as was also treatment with renin angiotensin system (RAS) blockers in patients with post-MI heart failure or diabetes.12,,14 As the sensitivities of diabetes and heart failure diagnoses in the registers are only moderate, we used anti-diabetic medication (ATC A10) and loop diuretics (ATC code C03), respectively, as proxies for these clinical diagnoses.15 Cohort participants were characterised with regard to use of cardiovascular medication based on dispensed prescriptions in the year prior to admission and previous use of disease-modifying antirheumatic drugs (DMARDs), respectively. Also, comorbidity was assessed by inclusion of the diagnoses used in the Ontario prediction rule for survival after MI as previously described.16 An index (1–5) of socio-economic status was calculated by dividing the average income of the previous 5 years up to admission into age-standardised quintiles. In order to be designated as a RA patient, two or more RA diagnoses prior to admission were required (ICD-8: 712.19, 712.39, 712.59; ICD-10: DM05, DM06). A detailed list of register codes used in this study is provided in online supplementary table S1.

Outcome measures

The primary outcomes were: (1) initiation of standard of care secondary prevention pharmacotherapy after MI with aspirin (ATC codes B01AC06 and N02BA01), β-blockers (C07), clopidogrel (B01AC04), RAS blockers (C09) and statins (C10AA) within 30 days after discharge and (2) patient adherence to these drugs when initiated within 180 days from discharge after MI (figure 1).

Figure 1

Study flowchart for one specific treatment.

Statistical analyses

For each secondary prevention drug and time period examined, we defined the total number of patients eligible for treatment initiation as the number of patients alive who did not already receive this specific agent prior to MI. Treatment initiation was defined as at least one redeemed prescription within the examined time period. Likewise, separate logistic regression models were fitted for each drug to assess the OR for treatment initiation associated with RA. Confounding by sex, age, socio-economic status and comorbidity was assessed by reporting ORS in models where these covariates were successively included. All statistical models were adjusted for the year of MI. Due to significant effect modification between RA and sex in the case of aspirin and to achieve close evaluation of the impact of sex and age, stratified analyses were performed.

Adherence to treatment was evaluated by estimating the number of individuals who had treatment available on a given day as calculated from the prescriptions filled and the number of pills dispensed, relative to the number of patients alive on that particular day (as previously described).15 In addition, adherence was evaluated in proportional hazard models fitted to estimate the RA-related risk of the first treatment gap of 90 days or more after drug initiation.15

CIs of 95% were reported and a two-tailed p value of 0.05 was considered significant. Base SAS V.9.2 (Cary, North Carolina, USA) and Stata/MP V.11.2 (College Station, Texas, USA) statistical packages were used for the analyses.


A total of 66 107 subjects were discharged alive after first-time MI during the study period of whom 867 (1.3%) were identified as RA patients. Baseline characteristics of study subjects are shown in table 1.

Table 1

Baseline characteristics of postmyocardial infarction patients with rheumatoid arthritis and reference population

Initiation of treatment

The crude proportions of subjects who initiated treatment with secondary prevention pharmacotherapies only slightly increased after day 30 and were consistently lower in RA patients regardless of the drug and time point analysed up to 1 year after discharge (table 2). The chance of initiating these evidence-based drugs 30 days after discharge in RA patients compared with non-RA patients was significantly reduced in primary models including sex and age for aspirin, β-blockers, clopidogrel and statins, while it was non-significantly lowered for RAS blocker initiation (table 3). Indeed, initiation of treatment was approximately 30% (statins), 25% (β-blockers and aspirin) and 15% (clopidogrel) less likely in RA patients compared with non-RA patients. These results did not change significantly when socio-economic status was added to the statistical models. However, the ORs were slightly increased, that is, differences in treatment initiation between RA and non-RA patients were slightly reduced when comorbidity covariates were included, and here the difference in the clopidogrel estimates was non-significant (table 3). The sex- and age-stratified analyses generally showed a marked reduction in the odds for treatment initiation with increasing age regardless of RA status (online supplementary table S2). Male non-RA patients consistently had higher odds than non-RA females, but with the exception of statin treatment this gender-specific pattern was less pronounced in RA patients. Consequently, treatment initiation was less likely in RA males than in RA females, although significant effect modification was only present in the case of aspirin treatment.

Table 2

Frequencies of treatment initiation (crude data) with secondary prevention drugs after myocardial infarction in patients with and without rheumatoid arthritis (RA)

Table 3

OR for treatment initiation of secondary prevention drugs within 30 days after myocardial infarction associated with rheumatoid arthritis and effects of successive inclusion of covariates

Adherence to treatment

Adherence to secondary prevention drugs after initiation of treatment was generally high as more than 80% of patients still used aspirin, β-blockers, RAS blockers and statins after 1 year, with aspirin being used most persistently (figure 2). Use of clopidogrel declined more rapidly in agreement with the prevailing treatment guidelines.12,,14 The graphical evaluation and proportional hazard analyses showed that adherence to statins was somewhat reduced in RA patients relative to non-RA patients, while there were no significant differences with the other secondary prevention drugs (aspirin: HR 0.97 (0.79 to 1.18), β-blockers: 1.05 (0.88 to 1.27), clopidogrel: 0.90 (0.79 to 1.03), RAS blockers: 1.17 (0.91 to 1.51) and statins: 1.26 (1.07 to 1.48)).

Figure 2

Adherence to postmyocardial infarction secondary prevention drugs after treatment initiation in patients with rheumatoid arthritis and reference population. RAS, renin angiotensin system.

Additional analyses

To evaluate whether a delay in treatment initiation between 30 and 180 days after discharge contributed to the observed differences in apparent drug initiation, the fully adjusted ORs for receiving each drug by day 180 after discharge were also calculated and these were not changed compared with ORs at day 30 for aspirin, β-blockers, clopidogrel and statins, while use of RAS blockers was significantly lower at this time point (figure 3). In order to complement the above analyses and to allow for a more close consideration of changes in the risk-set after discharge, initiation of secondary prevention pharmacotherapy within 180 days of follow-up was evaluated in proportional hazard models for each drug and this analysis yielded results consistent with the primary findings (aspirin: HR 0.82 (0.74 to 0.91), β-blockers: 0.77 (0.70 to 0.85), clopidogrel: 0.91 (0.83 to 0.99), RAS blockers: 0.89 (0.76 to 1.03) and statins: 0.71 (0.64 to 0.78)).

Figure 3

Adjusted OR for treatment initiation with secondary prevention drugs after myocardial infarction on days 30 and 180 after discharge associated with rheumatoid arthritis (RA). Reference=non-RA patients. RAS, renin angiotensin system.

Since percutaneous coronary intervention (PCI)is a strong predictor for initiation of standard of care secondary prevention pharmacotherapy that also may be used differently in RA patients with MI compared with the general MI population, we explored the impact on PCI within 7 days of MI on drug initiation.10 ,17,,19 Although PCI generally increased the likelihood of initiation of secondary prevention pharmacotherapies, the results were only minimally changed by the inclusion of PCI in the statistical model, by stratification according to PCI and by inclusion of coronary artery bypass surgery in addition to PCI in these analyses, respectively (data not shown).

Also, the influence of non-steroidal anti-inflammatory drugs (NSAIDs) on aspirin prescription was evaluated in models stratified for NSAID use at baseline as determined by two methods; (1) NSAID dispensed within 30 days prior to MI and (2) use of any NSAID at the time of admission for MI.20 There was no effect of NSAID treatment on initiation of aspirin treatment in patients with RA or in the reference group regardless of the method used to determine NSAID exposure.

Finally, the impact of the criteria used for RA was assessed by rerunning the fully adjusted primary analyses (initiation of treatment by day 30) with use of two different criteria for RA status. When the definition of RA status was made less conservative (only one RA diagnosis required), results were only minimally changed. When previous DMARD treatment was required for the RA status, however, OR increased slightly, that is, resulting in less differences in treatment initiation (aspirin: 0.82 (0.69 to 0.97), β-blockers: 0.88 (0.74 to 1.05), clopidogrel: 0.86 (0.74 to 0.99), RAS blockers 0.90 (0.69 to 1.19) and statins: 0.74 (0.62 to 0.88)).


In a contemporary cohort of more than 60 000 unselected patients alive after first-time MI, we found that RA patients were less likely to initiate evidence-based standard of care secondary prevention pharmacotherapy 30 days after discharge compared with non-RA patients. This apparent undertreatment was not improved after 180 days after MI, which indicated that absence of drug initiation rather than delay in treatment initiation may be the most important barrier for achieving optimal secondary prevention pharmacotherapy. In addition, we found that post-MI RA patients exhibited lower adherence to statins compared with non-RA patients.

These results expand the currently available evidence on use of secondary CVD prevention treatments in RA patients, which is limited to small studies of the acute management of MI. Van Doornum and colleagues recently reviewed hospital admissions to ascertain secondary prevention drug treatment in a small (n=90) cohort of Australian RA patients with MI and matched controls and found a substantially lower inhospital use of statins, aspirin and β-blockers (but not RAS blockers), while these differences were not significant in an earlier smaller (n=40) study from the UK.10 ,18 The use of statins for primary prevention in RA patients with no overt CVD was recently studied by Toms and colleagues, who found that more than half of high risk patients eligible for statin therapy were untreated.6 While this finding indeed is indicative of statin undertreatment in RA patients, no control group was evaluated and the results could therefore potentially be explained by a general underuse of statins. Similar methodological limitations apply to findings of suboptimal antihypertensive treatment of RA patients.7,,9

Undertreatment with secondary prevention drugs in RA patients after MI may contribute to a worsened prognosis post-MI compared with the general population.18 ,21 ,22 Several studies have shown that female sex, high age, low socio-economic status and/or educational level as well as the presence of chronic comorbidities are major risk factors for non-initiation of evidence-based medications following MI.17 ,23,,25 The present study also found these significant associations, but their inclusion into the models did not considerably change the observation that RA patients were less likely to receive guideline-recommended secondary prevention pharmacotherapy. Suboptimal treatment of RA patients may therefore represent a new aspect of the treatment-risk paradox, that is, where patients in the highest objective risk score categories receive less evidence-based treatment, frequently because of underestimation or ignorance of this risk.8 ,26 No firm conclusion can be made from our data on the impact of gender on post-MI undertreatment in patients with RA, although RA women showed a trend for less undertreatment than men, especially in the case of antiplatelet therapy, although this was only statistically significant for aspirin.

This study focused exclusively on post-MI secondary prevention pharmacotherapy because here the guideline-based decision-making is simply to prescribe aspirin, β-blocker, clopidogrel and a statin (with additional RAS blocker treatment in the presence of heart failure or diabetes) if no contraindications are present, as non-initiation of these therapies is known to reduce survival.12 ,13 ,17 ,27,,29 Consequently, it is reasonable to assume that undertreatment of RA patients is even more prevalent in primary prevention of CVD, where the treatment decision involves both assessment of risk factors and global CVD risk stratification, and this notion has been suggested by recent data.6 ,8 ,30 The present findings may therefore provide indirect support for the strategy of annual CVD risk assessment in RA patients recently recommended by the European League Against Rheumatism.3 ,25 Very limited data exist on adherence to secondary prevention pharmacotherapy and its consequences in RA patients, but statin discontinuation has been associated with increased risk of ischaemic events in both the general population and in subjects with RA.31,,33 Importantly, our findings suggested that adherence to statin treatment after MI is worse in RA patients relative to non-RA patients, which may contribute to a poor prognosis in these patients.

In general, Danish RA patients are followed regularly by rheumatologists, which in apparent contrast to our findings should offer ample opportunity for initiation of post-MI secondary prevention therapies if these were not initiated during the MI hospitalisation. On the other hand, previous studies including one from Denmark have shown that in patients with MI such treatment is nearly always initiated by hospital physicians, and that patients who did not start treatment before or shortly after discharge had a low probability of starting treatment later.15 ,17 These considerations as well as the lack of improvement after 6 months observed in the current study may suggest that physician behaviour and opinion-based (rather than evidence-based) barriers not least during the index hospitalisation are among the key obstacles for optimal secondary prevention pharmacotherapy in RA patients. Interestingly, the chart reviews in the Van Doornum study also did not reveal any specific evidence-based reasons why the treatments were not prescribed in RA patients during the admission for MI.10 A possible cause for the reluctance of physicians to provide evidence-based secondary prevention pharmacotherapy may be that RA patients were perceived to be more complex than the average MI patient, for example, due to use of DMARDs. This may particularly have been the case for statins, where concerns of known adverse effects such as hepatotoxicity and myopathy could have caused physicians to avoid these drugs leading to both reduced initiation and adherence.34 ,35 Statin treatment post-MI was recommended regardless of lipid level during the study period but may have been considered to be inappropriate in cases with very low lipids levels. As RA patients with MI may have lower lipid levels than non-RA patients, this mechanism may have contributed to lower statin initiation.36 Also, aspirin may have been withheld due to concerns of increased bleeding risk as well as suggestions of a reduced protective effect of aspirin when coadministered with NSAIDs.37 Our sensitivity analyses, however, indicated that confounding by indication due to NSAID use was unlikely to explain the lower likelihood of aspirin initiation. These and other considerations, however, could lead to non-initiation or postponement of therapy which in turn can result in treatment denied. Conversely, after discharge the care giving general practitioner or rheumatologist may be reluctant to change CVD treatment initiated (or not) by cardiologists during the hospitalisation.

Strengths and limitations

There are important considerations in the interpretation of the results. A considerable strength of the study was the large number of unselected patients and the high accuracy of prescription claims due to national legislation, which enabled data collection on treatment initiation in a ‘real world’ cohort. Also, results were fairly robust to inclusion of numerous confounders and additional analyses, for example, pharmacotherapy 6 months after discharge and provision of PCI. Nonetheless, the results for RAS blocker initiation were more uncertain due to the relatively low number of RA patients in this group and they may also be less valid because filled prescriptions were used as a proxy for heart failure and diabetes instead of clinical variables including data on left ventricular ejection fraction and blood glucose levels, respectively. Furthermore, the identification of RA patients relied on register data rather than clinical diagnoses, but the results were not substantially changed when other RA definitions were used. Also, the study design precluded the possibility to evaluate potential contraindications for the individual drugs. While the information on prescriptions filled is known to be highly accurate,11 it was not possible to confirm whether patients actually ingested the medication. This limitation also applied to the adherence analyses, which were based on the assumption of full patient compliance. In addition, we cannot exclude that unmeasured confounders impacted the findings.


This nationwide study of unselected patients with first-time MI showed that RA patients were less likely to initiate secondary prevention pharmacotherapy compared with non-RA patients, with statins being the most underused drugs. In addition, adherence to statins, when they were initiated, was relatively lower in RA patients. As secondary post-MI prevention pharmacotherapy is generally more implemented than primary prevention drugs aimed at CVD, undertreatment of RA patients in the post-MI setting may suggest a more general problem with suboptimal cardiovascular pharmacotherapy in RA patients. The present findings may contribute to the increased CVD burden in RA and the underlying mechanisms warrant further study.


Supplementary materials

  • Supplementary Data

    This web only file has been produced by the BMJ Publishing Group from an electronic file supplied by the author(s) and has not been edited for content.

    Files in this Data Supplement:

    • Web Only Data - This web only file has been produced by the BMJ Publishing Group from an electronic file supplied by the author(s) and has not been edited for content.


  • Funding JL has received an unrestricted grant from The Danish Rheumatism Association. The funding source had no influence on the study design, interpretation of results, or decision to submit the article.

  • Competing interests None.

  • Ethics approval This study was approved by The Danish Data Protection Agency (ref. 2008-41-2685).

  • Provenance and peer review Not commissioned; externally peer reviewed.

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.