Polypharmacy and Potential Drug–Drug Interactions for People With HIV in the UK From the Climate-HIV Database

C Okoli; A Schwenk; M Radford; M Myland; S Taylor; A Darley; J Barnes; A Fox; F Grimson; I Reeves; S Munshi; A Croucher; N Boxall; P Benn; A Paice; J van Wyk; S Khoo


HIV Medicine. 2020;21(8):471-480. 

In This Article


Patient Characteristics

The study population included 4630 PWHIV (Table 1); median time since HIV diagnosis or transfer into an HIV clinic using Climate-HIV was 11 years (IQR: 5.7–15.7), and patients had received a median of 3 (IQR: 2–5) previous ARV regimens.

Antiretroviral Medications and Regimens. Overall, 96% of patients had three or more ARV medications in their regimen, excluding ritonavir or cobicistat [median = 3 (IQR: 3–3)], and all regimens were composed of two nucleoside reverse transcriptase inhibitors (NRTIs) plus a third agent. When ARV regimens were grouped by the third agent, the largest proportion of patients were taking EFV + two NRTIs (23%, n = 1049), followed by DTG + two NRTIs (18%, n = 826). Overall, the most frequently recorded ARV regimens were EFV/emtricitabine (FTC)/tenofovir disoproxil fumarate (TDF) (18%, n = 811); DTG/abacavir (ABC)/lamivudine (3TC) (14%, n = 644); and rilpivirine (RPV)/FTC/TDF (7%, n = 308). A larger proportion of patients aged < 50 years were taking DTG/ABC/3TC than patients aged ≥ 50 years (18% vs. 8%; Table 2).

Concomitant Medications. Overall, 65% of patients (n = 2992) were taking ≥ 1 non-ARV medication, and 17% (n = 770) were taking ≥ 5 non-ARV medications (Table 3). Among patients aged ≥ 50 years, 27% (n = 508) were taking ≥ 5 non-ARV medications, compared with 10% (n = 262) of patients aged < 50 years (P < 0.0001). Similarly, a higher proportion of patients aged < 50 years were taking no comedications compared with those aged ≥ 50 years (43% vs. 24%). The most frequently recorded non-ARV medications were cholecalciferol (25%), atorvastatin (9%), and co-trimoxazole (9%). Differences in non-ARV medications were noted between patients aged < 50 vs. ≥ 50 years, where the use of atorvastatin (3% vs. 19%), amlodipine (4% vs. 12%), and ramipril (2% vs. 11%) increased with age (Table 3).

Comorbid or Concurrent Health Conditions. Overall, 40% (n = 1838) of patients had one or more comorbid or concurrent health conditions, with a lower proportion of patients aged < 50 years recording one or more comorbidity/health conditions compared with those aged ≥ 50 years (32% vs. 50%). The median number of health conditions in addition to HIV was 0 (IQR: 0–1) for patients aged < 50 years and 1 (IQR: 0–1) for those aged ≥ 50 years. Among selected health conditions, the most prevalent were hepatitis B, mental health disorders, and hypertension (6% each). The prevalence of hypertension demonstrated the most pronounced difference between patients aged < 50 years and those aged ≥ 50 years, affecting 3% of younger patients vs. 10% of older patients (Table 3). Assessment of combinations of comorbid/concurrent health conditions indicated that diabetes and hypertension were the most frequent combination (1% of patients overall), increasing from < 1% in patients aged < 50 years to 2% in those aged ≥ 50 years.

Patient-level DDIs

Drug–drug interactions were assessed at the patient level using the University of Liverpool DDI tool and categorized by colour. In total, 40 red and 1458 amber DDIs were observed in this patient population. Sildenafil and quetiapine were most frequently associated with red interactions, with sildenafil resulting in a red interaction for 18 (5.3%) patients taking ritonavir- or cobicistat-boosted darunavir (DRV/r/c) + two NRTIs and quetiapine resulting in a red interaction for four (1.2%) patients taking the same regimen. Red interactions were also identified with lansoprazole for three (5.5%) patients taking a regimen of RPV + two NRTIs; with omeprazole for two (2.7%) patients taking a regimen of ATV/r/c + two NRTIs; and with norethisterone, desogestrel, and etonogestrel, each for two (0.4%) patients taking a regimen of EFV + two NRTIs. Cholecalciferol and atorvastatin accounted for the greatest proportion of amber interactions.

In a logistic regression model, compared with DTG + two NRTIs, all regimens [aside from RPV + two NRTIs (data not shown)] significantly increased the risk of red or amber DDIs (P < 0.002; Figure 1). The most pronounced increased risk was for elvitegravir [EVG/c; odds ratio (OR) = 8.5 (95% CI: 3.9–18.8)] and atazanavir [ATV/r/c; OR = 5.8 (95% CI: 3.4–10.0)]. Addition of a single non-ARV medication [OR = 1.3 (95% CI: 1.2–1.3)] and a comorbid diagnosis [OR = 1.4 (95% CI: 1.2–1.7)] also increased the risk of DDIs by at least 28%, irrespective of ARV regimen. Age and sex did not influence the risk of DDI; however, the number of concomitant non-ARV medications did increase with age, and this factor was significant but not collinear [OR = 1.3 (95% CI: 1.2–1.3)].

Figure 1.

Logistic regression analysis assessing factors influencing the risk of red or amber drug–drug interactions relative to regimens based on dolutegravir. For odds ratios (ORs) associated with antiretroviral (ARV) regimens, the reference regimen is dolutegravir. The symbol size represents the relative number of patients. c, cobicistat; CL, confidence level; NRTI, nucleoside reverse transcriptase inhibitor; r, ritonavir.

Interaction-level DDIs

A separate analysis of DDIs was conducted at the interaction level based on the observed combinations of ARV/non-ARV medications. The prescribed non-ARV medications observed for all patients taking each ARV regimen were cross-tabulated to summarize the proportion of colour classifications. Figure 2 shows the proportion of non-ARV medications that had an interaction or did not interact with the ARV medication of interest. The denominators are the number of different non-ARV medications recorded by patients taking each of the regimens. The highest proportion of red or amber interactions occurred with ATV/r/c + two NRTIs (58%, n = 57), followed by DRV/r/c + two NRTIs (48%, n = 128). DTG + two NRTIs (87%, n = 194) and RAL + two NRTIs (84%, n = 147) had the greatest proportion of green medication combinations (Figure 2a).

Figure 2.

Proportion of potential drug-drug interactions based on observed non-antiretroviral (non-ARV) medications taken with ARV regimens (a) in the overall population and (b) stratified by patient age. All regimens listed are taken with two nucleoside reverse transcriptase inhibitors (NRTIs). N represents the number of non-ARV medications recorded by the population using each specific regimen. ATV, atazanavir; c, cobicistat; DRV, darunavir; DTG, dolutegravir; EFV, efavirenz; EVG, elvitegravir; NVP, nevirapine; r, ritonavir; RAL, raltegravir; RPV, rilpivirine.

Rates of DDIs with DTG were similar between patients aged ≥ 50 years and those aged < 50 years (14% each; Figure 2b). Results were comparable for patients taking RAL-based regimens, where DDIs were identified for 14% of the comedication pairings for patients aged ≥ 50 years and for 18% of those aged < 50 years; however, there was a higher proportion of DDIs classified as amber (drugs that potentially interact) among patients aged ≥ 50 years than among those aged < 50 years (14% vs. 10%; Figure 2b). Overall, the slightly higher rates of observed amber and yellow DDIs with RAL compared with DTG (13% vs. 15%; Figure 2a) can be attributed to the more frequent prescription of TDF/FTC with RAL vs. ABC/3TC with DTG.

Propensity for DDIs With Common Comorbidities/Concurrent Health Conditions

Table 4 illustrates that for the common comorbidities/concurrent health conditions, adherence to NICE-recommended guidelines yielded the greatest potential risk of DDIs in patients with hepatitis C co-infection, with red (22%) and amber (58%) interactions across all current ARV regimens. Additionally, 50% of treatment combinations for malignancy and 32% for mental health conditions resulted in amber interactions.