INTRODUCTION

Due to effective antiretroviral therapy (ART), patients living with HIV infection (PLWH) have an increased life expectancy compared with previous years.¹ However, as life expectancy increases in PLWH, so does the risk of developing atherosclerotic cardiovascular disease (ASCVD).² It is estimated that PLWH have approximately two-fold greater risk of developing ASCVD compared with patients without HIV.³ The reasons for this are multifactorial and include a higher prevalence of traditional cardiovascular disease risk factors (e.g. smoking, diabetes and hypertension in some studies), adverse metabolic effects from ART (e.g. insulin resistance and dyslipidaemia), and HIV-related immune activation and inflammation.^{3, 4}3-Hydroxy-3-methyl-glutaryl-coenzyme A reductase inhibitors (‘statins’) have become a standard of care for both primary and secondary ASCVD prevention, including in PLWH.^5-7 In spite of current guideline recommendations specifically noting HIV as a risk-enhancing factor for ASCVD, several retrospective studies have shown statin therapy under-utilization in PLWH.^8-12 A retrospective cohort study compared prescribing rates of statin therapy in PLWH (n = 908) with uninfected (n = 2239) patients in two urban, academic medical system clinics.¹³ The PLWH were less likely to be prescribed a statin than uninfected patients [44% vs. 56%, respectively; adjusted odds ratio (OR) 0.79; 95% confidence interval (CI): 0.66–0.94]. Of those who did receive statin therapy, PLWH were more likely to receive a medium-intensity agent, whereas the uninfected patients received a high-intensity agent more often. However, previous studies fell short of assessing the appropriateness of statin intensity [i.e. evaluation of drug–drug interactions (DDIs), statin contraindications, and toxicities].

The reasons for the statin prescribing gap between PLWH and uninfected patients with ASCVD have not been well elucidated. One potential explanation is concern for clinically significant DDIs in those receiving ART and concomitant statin therapy. Most drug interactions with ART occur via the cytochrome P (CYP) 450 system, with protease inhibitors (PIs) inhibiting CYP3A4 and most nonnucleoside reverse transcriptase inhibitors (NNRTIs) inducing this isoenzyme.¹⁴ Depending on the interacting medication as well as patient-specific factors, alterations in serum statin concentrations may occur, leading to higher rates of adverse events or decreased statin efficacy.¹⁴ Specific statin agents are contraindicated or have dose restrictions with medications used to treat PLWH due to an increased risk of statin-related adverse events, including myopathy and elevated liver function tests (LFTs).^{15, 16} However, newer ART regimens are less likely to cause significant interactions, and unnecessary avoidance of statin therapy may lead to an increased risk of ASCVD in PLWH.¹⁷ Evaluation of the appropriateness of statin selection and dosing (ASD) is needed to optimize the management of ASCVD in PLWH, while minimizing the potential risk of adverse effects.

The primary objective of this study was to compare rates of ASD in PLWH vs. uninfected patients for prevention of ASCVD after adjusting for clinical- and patient-related variables. The secondary objective was to identify patient characteristics that were associated with inappropriate prescribing of statin and dose for both HIV-infected and uninfected patients with and without established ASCVD.

METHODS

This comparative, retrospective study evaluated the ASD in PLWH vs. uninfected patients at two outpatient clinics within one large, urban, academic medical centre in the United States between February and September 2017. This study was approved by the affiliated institutional review boards.

Patients over the age of 21 with an indication for statin therapy were included in this study. Statin indication was determined by the American College of Cardiology/American Heart Association (ACC/AHA) 2013 Guideline on the Treatment of Blood Cholesterol which included: (i) clinical ASCVD; (ii) low-density lipoprotein cholesterol (LDL-C) ≥ 190 mg/dL; (iii) age 40–75 years with diabetes and LDL-C in the range 70–189 mg/dL; or (iv) age 40–75 years with LDL-C in the range 70–189 mg/dL and an ASCVD 10-year risk ≥ 7.5%. Details regarding the inclusion criteria used for this study were previously published.^{5, 13}

Appropriateness of statin selection and dosing for each patient included in this study was determined based on the ACC/AHA Guideline recommended intensity after evaluation for statin allergy, autoimmune hepatitis, elevated liver enzymes, renal function, pregnancy and clinically significant DDIs. In cases where no statin was prescribed, these criteria were used to assess if it was appropriate to withhold statin therapy. Prescribing information, specific guidelines such as the ACC/AHA Guideline on the Treatment of Blood Cholesterol, Kidney Disease: Improving Global Outcomes (KDIGO) Guideline of Lipid Management in Chronic Kidney Disease, and Antiretroviral Guidelines for Adults and Adolescents, in addition to resources such as Lexicomp and Micromedex were used to determine ASD using the described criteria.^{5, 18, 19} Despite using noted resources, some patients were complex and lacked a concrete recommendation. A team of five clinical pharmacists, including three cardiology specialists, one infectious diseases specialist and one primary care specialist, reviewed these cases to reach a consensus regarding ASD.

Contraindications to statin use were considered if patients had a documented statin allergy in the electronic medical record (EMR), autoimmune hepatitis noted in the EMR problem list, pregnancy determined by human chorionic gonadotrophin (hCG) levels, creatine kinase (CK) levels more than five times the upper limit of normal (ULN), and AST/ALT levels more than three times the ULN.²⁰ As the true incidence of statin allergy/intolerance is low in the general population, if a patient was continued on a statin therapy with a documented statin allergy/intolerance in the EMR, the appropriateness of the statin was based on the dosing and not on the contraindication. If a statin allergy was documented and no statin medication was prescribed, this was considered appropriate.

Data were analysed using descriptive statistics for frequencies and ranges. A t-test or Wilcoxon rank-sum test and χ² test or Fisher's exact test were used to check any differences of characteristics between cohorts. The associations between the outcome variable, statin appropriateness and various factors were analysed using multivariate analysis. For multivariate analysis, a logistic regression model was conducted to evaluate for independent factors. The relative risk (RR) with 95% CI was used to measure the magnitude of association. The variance inflation factor (VIF) and correlation coefficients were used to identify multicollinearity. In addition, as there was a greater sample size of uninfected subjects than PLWH, the propensity score (PS) matching method was applied, and the results of multivariate analysis were compared with those of PS matching score. The PS matching was performed using all variables from Table 1. Matched sets of PLWH and uninfected subjects who share a similar value of the propensity score were formed. Optimized PS matching within specified absolute differences of PS was used with random matching of patients to controls within the maximum radius. The matching ratio was applied 1:1 with 0.05 absolute score difference between PSs across cohorts. Also, bivariate analysis was performed using χ² test or Fisher's exact test and Wilcoxon rank-sum test to detect any associations between each possible predictor and the outcome. For multiple comparisons, testing each level vs. the sum of all other levels using Fisher's exact test with Bonferroni correction was applied. The results of the PS matched analysis are presented here. Non-PS matched data are available in the Supporting Information (Appendix 1). Analyses were performed with SAS v.9.4 (SAS Institute, Cary, NC, USA).

RESULTS

A total of 3145 patients met the inclusion criteria, of whom 908 (28.9%) were PLWH and 2237 (71.1%) were uninfected (Appendix 1). After PS matching, 1758 patients (55.9% of total cohort) were included for analysis [n = 879 (50%) PLWH; n = 879 (50%) uninfected] (Table 1). The mean (SD) age was 57.2 (9) years. The majority of patients were black (n = 1577, 89.7%) and male (n = 1122, 63.8%). After propensity matching there were no differences in baseline characteristics between groups.

Fewer PLWH were prescribed an appropriate statin dose (n = 244, 27.8%) compared with uninfected patients (n = 356, 40.5%, P < 0.0001) (Table 2). This was primarily driven by inappropriately low statin doses (19.5% PLWH, 12.7% uninfected, P = 0.000 64). There were no differences in the proportion of omitted statins or statins dosed too high between PLWH and uninfected patients.

Lower rates of appropriate statin prescribing in PLWH were seen in subgroups of patients with clinical ASCVD (37.2% PLWH, 55.3% uninfected, P = 0.000 13) and 10-year ASCVD risk ≥ 7.5% (22.2% PLWH, 34.0% uninfected, P = 0.000 55), but not in patients with LDL-C ≥ 190 mg/dL or diabetes.

Multivariate analysis revealed that PLWH were 32% less likely to be prescribed an appropriately dosed statin [adjusted relative risk (aRR) = 0.68, 95% CI: 0.58–0.78]. HIV status remained significant for patients included based on indications of clinical ASCVD (aRR = 0.62, 95% CI: 0.45–0.80), diabetes (aRR = 0.70, 95% CI: 0.41–0.99), and ASCVD risk ≥ 7.5% (aRR = 0.66, 95% CI: 0.50–0.83), but was not significant for patients included for LDL-C ≥ 190 mg/dL (aRR = 0.89, 95% CI: 0.41–1.38). Smoking was associated with lower ASD rates in the full cohort (aRR = 0.85, 95% CI: 0.73–0.96). Patients included based on ASCVD risk ≥ 7.5% were less likely to receive ASD if they smoked (aRR = 0.78, 95% CI: 0.61–0.96) or were black (aRR = 0.54, 95% CI: 0.33–0.75), and more likely to receive ASD if they had a diagnosis of hypertension (aRR = 1.59, 95% CI: 1.16–2.02).

DISCUSSION

This study confirms the observed gap in statin prescribing described previously in the literature, and highlights that even when PLWH are prescribed statin therapy, it is often underdosed. The PLWH were 1.5 times more likely to be prescribed a suboptimal statin dose than were uninfected patients. To our knowledge, no other studies have characterized ASD based on DDIs, comorbidities and laboratory findings. The findings of this study suggest that when ASD is included, the statin gap may represent a larger disparity in ASCVD risk reduction therapy than previously described.

These findings support the results of the Prognosis of Acute Coronary Syndrome in HIV-infected patients (PACS-HIV) lipids study, a prospective observational study that found lower rates of high-intensity statin prescribing across the 6 months following a first acute coronary syndrome event in PLWH than in uninfected controls. This resulted in worse lipid profiles in the PLWH cohort despite similar levels at baseline.²¹ Patients included in our study may not have experienced a recent acute coronary syndrome event, but were similarly observed to be prescribed ASD less frequently, often due to underdosing.

Although the aim of this study was to compare the rates of ASD between PLWH and uninfected patients, it should be noted that after accounting for potential DDIs and contraindications, only 41% of uninfected patients were prescribed an appropriate statin. This was largely due to omissions in almost half (46%) of patients despite indications for therapy. In uninfected patients, rates of omissions were highest in those with indications based on elevated 10-year ASCVD risk (64%) and elevated LDL-C levels (50%), but were notable for all risk groups.

Nevertheless, the gap in ASD was larger in PLWH, including those with clinical ASCVD, diabetes or a high 10-year ASCVD risk. The benefits of statin therapy for primary prevention of ASCVD are well established and include reduction in the burden of disability for non-fatal ASCVD events and healthcare costs.⁵ For patients with a 10-year ASCVD risk of 5–7.5%, LDL-C in the range 70–189 mg/dL or aged < 40 or > 75 years without clinical ASCVD, there is moderate evidence to support benefit, but individualized treatment decisions are needed. These patients were not included in this study in an effort to limit included patients to those with clear indications for statin therapy. People living with HIV have an independent risk factor for the development of ASCVD, and probably derive an even greater benefit from statin therapy. This was underscored by the addition of HIV infection as a risk-enhancing factor in recent revisions to blood cholesterol management guidelines.⁷ The guidelines recommend considering presence of HIV infection for decisions surrounding statin initiation in patients with ‘borderline’ or ‘intermediate’ risk and for dose intensity in patients with diabetes. Although these updated guidelines were not yet available during the study time frame, the correlation between risk of ASCVD and HIV was first described several years prior.^2-4 Specifically for PLWH, where virus-induced inflammation is a proposed cause of increased ASCVD risk, the anti-inflammatory effects of statin therapy are a particularly appealing risk-reduction strategy.

Uninfected patients with clinical ASCVD were the only population where more than half the patients were treated with ASD. The clinical ASCVD benefit group also had the highest ASD rates for PLWH, although this rate remained low. Conversely, patients indicated based on high 10-year ASCVD risk had the lowest ASD rates in the total population and both PLWH and uninfected cohorts. Higher prescribing rates in uninfected patients persisted for the elevated LDL-C group, but did not reach statistical difference. This is most likely due to smaller numbers of patients comprising these groups. As such, larger studies with adequate statistical power may provide additional insights into these patients.

This study was not designed to identify causes of statin omission, but concern for DDIs with ARTs has previously been proposed.^9-11 Indeed, specific ARTs are known to increase serum statin concentrations and risk of ADEs significantly. However, only 193 clinically relevant DDIs were identified, which did not account for the low ASD rates. These interactions included many non-ART medications (e.g. verapamil, rifampin, fluconazole, direct-acting antivirals for hepatitis C treatment). With the availability of newer ARTs with fewer statin interactions, appropriate product selection and dose adjustment can allow for statin treatment in most patients. Given the increased risk of ASCVD in PLWH and the known benefit of statin therapy, overstated concerns for DDIs should not be a reason to withhold therapy from indicated patients.

Statin allergies were documented in a small proportion (4%) of the study population. In clinical practice, it is known that documented statin allergies often describe an intolerance or perceived side effect to the agent instead of a true allergy. Given the ambiguity in this documentation, we were permissive in determining ASD for patients with a documented statin allergy. Therapy was deemed appropriate if no statin was prescribed. Additionally, many patients with a listed allergy were successfully prescribed a statin, usually an alternate agent from the class. This further highlights the conservative nature of allowing for no statin therapy in patients with a documented allergy and was also deemed appropriate for the purposes of our analysis. Similarly, guidance on statin therapy in patients with renal disease is nuanced. Risk reduction for ASCVD has not been demonstrated for patients on haemodialysis and the Kidney Disease Improving Global Outcomes (KDIGO) Clinical Practice Guideline for Lipid Management in Chronic Kidney Disease advise against statin initiation in this population without established ASCVD, although statins initiated prior to dialysis may be continued.^{5, 18} To determine ASD for patients with renal disease we consulted prescribing information for recommended dose adjustments or contraindications, but were permissive regarding decision to prescribe or omit statin in patients with end-stage renal disease on dialysis. Given our conservative approach in areas of clinical uncertainty, ASD rates may be modestly inflated, biasing comparisons towards null. Overall, these populations were small and, given the low rates of appropriate statin prescribing found in this study, do not appear to have impacted findings to a large degree.

There are several limitations to this study. The retrospective nature made it difficult to verify accuracy of medication lists, interpret allergy documentation, and determine duration and timing of interacting medications. The included population was mostly black and male, which may not be representative of all patients. There is also a degree of subjectivity to any assessment of appropriateness. To apply this determination fairly between cohorts, a consensus was reached on criteria used and applied to all patients. The sub-groups may not be powered to show difference in ASD rates. Additionally, the blood cholesterol management guidelines had been updated since the onset of this project. While the recommendations regarding use of statins and specified benefit groups are not substantially different, HIV was added as a risk-enhancing factor. The effect of this addition on prescribing habits is not assessed in this study. If anything, however, this further emphasizes the need for improved statin prescribing in this population. We also did not assess the appropriateness of statin dosing based on LDL-C targets, which has been further clarified in more recent guidelines. Nonetheless, it is unlikely that the large percentages of patients with under-optimized statin therapy were as a result of already low LDL-C levels.

Despite these limitations, this study did consider potential DDIs and contraindications for statin therapy, which are known barriers to statin prescribing.²² This adds to the existing body of literature related to the gap in statin prescribing for PLWH, adding insights into how appropriately these agents are dosed when used. Our findings confirm that PLWH are prescribed appropriate statins less often than uninfected patients, largely due to underdosing. In addition, these findings continue to highlight the need for further education on statin indication and dosing for both uninfected and PLWH populations.

CONCLUSIONS

While less than half of all patients indicated for statin therapy receive ASD, the rates are significantly lower for PLWH than for matched uninfected patients, largely driven by underdosing. Presence of DDIs and contraindications did not account for the differences in prescribing observed. Given the known increased risk of ASCVD for PLWH and the safety of concomitant use of statins with newer ARTs, improvement in ASD rates is imperative to improve patient outcomes.

AUTHOR CONTRIBUTIONS

KJP, AlB and NSP contributed to the development of this research project. SYS and ALB contributed to data collection. KJP, NSP, ZRN, SG and SD reviewed and came to a consensus on any discrepancies related to ASD determinations. HS provided statistical analysis. All authors contributed to the development and revision of the manuscript.