Consensus recommendations for systematic evaluation of drug-drug interaction evidence for clinical decision support

BACKGROUND

Healthcare organizations, compendia, and drug knowledgebase vendors use varying methods to evaluate and synthesize evidence on drug-drug interactions (DDIs). This situation has a negative effect on electronic prescribing and medication information systems that warn clinicians of potentially harmful medication combinations.

OBJECTIVE

The aim of this study was to provide recommendations for systematic evaluation of evidence for DDIs from the scientific literature, drug product labeling, and regulatory documents.

METHODS

A conference series was conducted to develop a structured process to improve the quality of DDI alerting systems. Three expert workgroups were assembled to address the goals of the conference. The Evidence Workgroup consisted of 18 individuals with expertise in pharmacology, drug information, biomedical informatics, and clinical decision support. Workgroup members met via webinar 12 times from January 2013 to February 2014. Two in-person meetings were conducted in May and September 2013 to reach consensus on recommendations.

RESULTS

We developed expert consensus answers to the following three key questions. (i) What is the best approach to evaluate DDI evidence? (ii) What evidence is required for a DDI to be applicable to an entire class of drugs? (iii) How should a structured evaluation process be vetted and validated?

CONCLUSION

Evidence-based decision support for DDIs requires consistent application of transparent and systematic methods to evaluate the evidence. Drug compendia and clinical decision support systems in which these recommendations are implemented should be able to provide higher-quality information about DDIs.

Proximal Roux-en-Y gastric bypass alters drug absorption pattern but not systemic exposure of CYP3A4 and P-glycoprotein substrates

STUDY OBJECTIVES

To evaluate the effect of Roux-en-Y gastric bypass surgery (RYGB) on the pharmacokinetics of midazolam (a CYP3A4 substrate) and digoxin (a P-glycoprotein substrate).

DESIGN

Prospective, nonblinded, longitudinal, single-dose pharmacokinetic study in three phases: presurgery baseline and postoperative assessments at 3 and 12 months.

PATIENTS

Twelve obese patients meeting current standards for bariatric surgery.

MEASUREMENTS AND MAIN RESULTS

At each study visit, patients received a single dose of oral digoxin and midazolam at 8 a.m. Blood samples were collected at regular intervals for 24 hours after dosing. Continuous 12-lead electrocardiogram (EKG), heart rate, blood pressure, and respiratory rate were monitored, and pharmacokinetic parameters from the three visits were compared. The peak plasma concentration (Cmax ) of midazolam increased by 66% and 71% at 3- and 12-month post-RYGB (p=0.017 and p=0.001, respectively), whereas the median time to peak concentration (Tmax ) was reduced by 50%. The mean Cmax for 1'-hydroxymidazolam increased by 87% and 80% at 3 and 12 months (p=0.001 and p<0.001, respectively). However, neither the area under the concentration-time curve (AUC) for midazolam nor the metabolite-to-parent AUC ratio changed significantly over time. For digoxin, the median Tmax decreased from 40 minutes at baseline to 30 and 20 minutes at 3 and 12 months, respectively. The mean AUC for digoxin, heart rate, and EKG patterns were similar across the three study phases.

CONCLUSION

Contemporary proximal RYGB increases the rate of drug absorption without significantly changing the overall exposure to midazolam and digoxin. The Cmax of a CYP3A4 substrate with a high extraction ratio was substantially increased after RYGB.

An evaluation of the potential drug interaction between warfarin and levothyroxine

BACKGROUND

Drug interaction references report that initiation of levothyroxine potentiates the effects of warfarin, and recommend more frequent International Normalized Ratio (INR) monitoring, but the mechanism is not well understood.

OBJECTIVE

To assess the impact of levothyroxine initiation on INR response.

PATIENTS/METHODS

A retrospective, self-controlled study was performed on patients aged ≥ 18 years receiving chronic warfarin therapy who were started on levothyroxine between 1 January 2006 and 30 June 2013, and who were followed for 90 days prior to and after levothyroxine initiation. The included patients had at least one elevated thyroid-stimulating hormone laboratory value in the pre-period, continuous warfarin therapy for 100 days prior to levothyroxine initiation, no purchases of medications known to interact with warfarin, no procedures requiring warfarin interruption, and no bleeding or thromboembolic event during the study period. The primary outcome was a comparison of the warfarin dose/INR ratio recorded before the initiation of levothyroxine with the ratio recorded during the post-period after two consecutive INRs with no warfarin dose change.

RESULTS

One hundred and two patients were included in the primary outcome. The mean warfarin dose/INR ratios in the pre-period and post-period were equivalent (P = 0.825). Although the mean warfarin dose was numerically lower in the post-period than in the pre-period, this difference did not reach statistical significance (P = 0.068).

CONCLUSION

No difference in the mean warfarin dose/INR ratio before and after initiation of levothyroxine was detected. The results suggest that there is not a clinically significant interaction between warfarin and levothyroxine, and so additional monitoring may not be necessary.

Dynamic enhancement of drug product labels to support drug safety, efficacy, and effectiveness

Out-of-date or incomplete drug product labeling information may increase the risk of otherwise preventable adverse drug events. In recognition of these concerns, the United States Federal Drug Administration (FDA) requires drug product labels to include specific information. Unfortunately, several studies have found that drug product labeling fails to keep current with the scientific literature. We present a novel approach to addressing this issue. The primary goal of this novel approach is to better meet the information needs of persons who consult the drug product label for information on a drug's efficacy, effectiveness, and safety. Using FDA product label regulations as a guide, the approach links drug claims present in drug information sources available on the Semantic Web with specific product label sections. Here we report on pilot work that establishes the baseline performance characteristics of a proof-of-concept system implementing the novel approach. Claims from three drug information sources were linked to the Clinical Studies, Drug Interactions, and Clinical Pharmacology sections of the labels for drug products that contain one of 29 psychotropic drugs. The resulting Linked Data set maps 409 efficacy/effectiveness study results, 784 drug-drug interactions, and 112 metabolic pathway assertions derived from three clinically-oriented drug information sources (ClinicalTrials.gov, the National Drug File - Reference Terminology, and the Drug Interaction Knowledge Base) to the sections of 1,102 product labels. Proof-of-concept web pages were created for all 1,102 drug product labels that demonstrate one possible approach to presenting information that dynamically enhances drug product labeling. We found that approximately one in five efficacy/effectiveness claims were relevant to the Clinical Studies section of a psychotropic drug product, with most relevant claims providing new information. We also identified several cases where all of the drug-drug interaction claims linked to the Drug Interactions section for a drug were potentially novel. The baseline performance characteristics of the proof-of-concept will enable further technical and user-centered research on robust methods for scaling the approach to the many thousands of product labels currently on the market.

Inhibitory metabolic drug interactions with newer psychotropic drugs: inclusion in package inserts and influences of concurrence in drug interaction screening software

BACKGROUND

Food and Drug Administration (FDA) regulations mandate that package inserts (PIs) include observed or predicted clinically significant drug-drug interactions (DDIs), as well as the results of pharmacokinetic studies that establish the absence of effect.

OBJECTIVE

To quantify how frequently observed metabolic inhibition DDIs affecting US-marketed psychotropics are present in FDA-approved PIs and what influence the source of DDI information has on agreement between 3 DDI screening programs.

METHODS

The scientific literature and PIs were reviewed to determine all drug pairs for which there was rigorous evidence of a metabolic inhibition interaction or noninteraction. The DDIs were tabulated noting the source of evidence and the strength of agreement over chance. Descriptive statistics were used to examine the influence of source of DDI information on agreement among 3 DDI screening tools. Logistic regression was used to assess the influence of drug class, indication, generic status, regulatory approval date, and magnitude of effect on agreement between the literature and PI as well as agreement among the DDI screening tools.

RESULTS

Thirty percent (13/44) of the metabolic inhibition DDIs affecting newer psychotropics were not mentioned in PIs. Drug class, indication, regulatory approval date, generic status, or magnitude of effect did not appear to be associated with more complete DDI information in PIs. DDIs found exclusively in PIs were 3.25 times more likely to be agreed upon by all 3 DDI screening tools than were those found exclusively in the literature. Generic status was inversely associated with agreement among the DDI screening tools (odds ratio 0.11; 95% CI 0.01 to 0.89).

CONCLUSIONS

The presence in PIs of DDI information for newer psychotropics appears to have a strong influence on agreement among DDI screening tools. Users of DDI screening software should consult more than 1 source when considering interactions involving generic psychotropics.

The risk of hypotension following co-prescription of macrolide antibiotics and calcium-channel blockers

BACKGROUND

The macrolide antibiotics clarithromycin and erythromycin may potentiate calcium-channel blockers by inhibiting cytochrome P450 isoenzyme 3A4. However, this potential drug interaction is widely underappreciated and its clinical consequences have not been well characterized. We explored the risk of hypotension or shock requiring hospital admission following the simultaneous use of calcium-channel blockers and macrolide antibiotics.

METHODS

We conducted a population-based, nested, case-crossover study involving people aged 66 years and older who had been prescribed a calcium-channel blocker between Apr. 1, 1994, and Mar. 31, 2009. Of these patients, we included those who had been admitted to hospital for the treatment of hypotension or shock. For each antibiotic, we estimated the risk of hypotension or shock associated with the use of a calcium blocker using a pair-matched analytic approach to contrast each patient's exposure to each macrolide antibiotic (erythromycin, clarithromycin or azithromycin) in a seven-day risk interval immediately before admission to hospital and in a seven-day control interval one month earlier.

RESULTS

Of the 7100 patients admitted to hospital because of hypotension while receiving a calcium-channel blocker, 176 had been prescribed a macrolide antibiotic during either the risk or control intervals. Erythromycin (the strongest inhibitor of cytochrome P450 3A4) was most strongly associated with hypotension (odds ratio [OR] 5.8, 95% confidence interval [CI] 2.3-15.0), followed by clarithromycin (OR 3.7, 95% CI 2.3-6.1). Azithromycin, which does not inhibit cytochrome P450 3A4, was not associated with an increased risk of hypotension (OR 1.5, 95% CI 0.8-2.8). We found similar results in a stratified analysis of patients who received only dihydropyridine calcium-channel blockers.

INTERPRETATION

In older patients receiving a calcium-channel blocker, use of erythromycin or clarithromycin was associated with an increased risk of hypotension or shock requiring admission to hospital. Preferential use of azithromycin should be considered when a macrolide antibiotic is required for patients already receiving a calcium-channel blocker.

What Do Medical Records Tell Us About Potentially Harmful Co-Prescribing?

BACKGROUND

Previous efforts document drug-drug interactions in ambulatory care. Yet little is known about medical record documentation or clinical management when interacting medications are received.

METHODS

The study population was identified from the HMO Research Network's Centers for Education and Research on Therapeutics (n = 2,020,037). A random subsample of patients > or = 18 years of age with drug coverage in 2000 initiating a co-dispensing for (1) warfarin with a nonsteroidal anti-inflammatory drug (n = 97), (2) digoxin with verapamil or diltiazem (n = 100), or (3) lovastatin/simvastatin with diltiazem or verapamil (n = 89) was identified.

RESULTS

The majority (63%-74%) of patients had documentation indicating receipt of both drugs during a single office visit. Documentation of risks and patient education was less common (< or = 14%, with all corresponding upper bounds of the 95% CIs < 23%). Clinical management changes were more frequently documented, ranging from 64% (95% CI: 47-81%) for lovastatin/simvastatin patients to 79% (95% CI: 60-99%) for warfarin patients.

CONCLUSIONS

The findings, although indicating that clinicians are likely aware of concomitant receipt of interacting medications, call into question the adequacy of medical record documentation as well as clinical management when interacting drugs are co-prescribed in the ambulatory setting.

Potential Drug–Drug Interactions in the Outpatient Setting

BACKGROUND

Although medication safety research has tended to focus on inpatients, the safety of drug use among outpatients is also a concern.

OBJECTIVE

We estimate the frequency of potentially interacting concomitant medication dispensing among outpatients.

RESEARCH DESIGN

We report the number and percent of patients annually dispensed an object drug of interest (ie, warfarin, digoxin, cyclosporine, or lovastatin/simvastatin) with a potentially interacting drug among a random sample of insured adults receiving care from 10 integrated delivery systems. We use 2 definitions of concomitant dispensing: medications dispensed: 1) during the time period for which the patient had the other medication available ('days supply') and 2) on the same day. We also estimate the number of insured U.S. population codispensed these medication pairs.

RESULTS

Among patients dispensed a drug of interest, between 17.8% (95% confidence interval [CI]=17.1-18.6%) and 28.0% (95% CI=24.0-32.1%) were concomitantly dispensed a potentially interacting drug using the "days supply" definition, and between 7.1% (95% CI=6.6-7.7%) and 17.7% (95% CI=14.4-21.1%) using the "same day" definition. Extrapolating to the insured U.S. population, between 1.29 (95% CI=1.25-1.33; same day) and 2.67 (95% CI=2.62-2.72; days supply) million insured adults are dispensed 1 of these 4 potentially interacting pairs.

CONCLUSIONS

We found evidence of potentially interacting concomitant medication dispensing among outpatients. An opportunity exists to better understand how such dispensing translates into adverse events and ultimately to improved medication safety.

Review article: similarities and differences among delayed-release proton-pump inhibitor formulations

Proton-pump inhibitors are acid-labile, and require an enteric coating to protect them from degradation in the stomach when given orally. However, this leads to delayed absorption and onset of action of the proton-pump inhibitor. This article aims to review the similarities and differences between the various formulations of delayed release proton-pump inhibitors. Delayed-release omeprazole and delayed-release lansoprazole have been suspended in sodium bicarbonate for tube administration; however, for omeprazole, absorption is further impaired and antisecretory effects are disappointing. Although such formulations may be more convenient for clinical use in certain patient groups, absorption of the proton-pump inhibitor is still influenced by residual enteric coating. There are few differences among the currently available delayed-release proton-pump inhibitors with respect to their pharmacodynamic effects during chronic administration. There are minor formulation-based pharmacokinetic differences among these agents, primarily reflected in their bioavailability following the first few doses. Differences in bioavailability may explain slight differences in the rate of onset of maximal antisecretory effect. However, minor pharmacodynamic and pharmacokinetic differences are not associated with meaningful differences in clinical outcomes.

Comparative pharmacology of proton pump inhibitors

The PPIs are the most effective therapy to suppress gastric acid secretion. These agents decrease acid secretion by inhibiting parietal cell proton pumps. From chemical and pharmacodynamic points of view, subtle differences that exist among the PPIs may influence clinical activity.

Performance of community pharmacy drug interaction software

OBJECTIVE

To evaluate the performance of computerized drug-drug interaction (DDI) software in identifying clinically important drug-drug interactions.

DESIGN

One-time performance test of computer systems using a standard set of prescriptions.

SETTING

Community pharmacies or central corporate locations with pharmacy terminals identical to those used in actual pharmacies.

PARTICIPANTS

Chain and health maintenance organization (HMO) pharmacies with seven or more practice sites in Washington State. A total of nine different DDI software programs were installed in 516 community pharmacies represented by these chains and HMOs.

MAIN OUTCOME MEASURES

Sensitivity, specificity, and positive and negative predictive values of software in detecting 16 well-established DDIs contained within six fictitious patient profiles.

RESULTS

The software systems failed to detect clinically relevant DDIs one-third of the time. Sensitivity of the software programs ranged from 0.44 to 0.88, with 1.00 being perfect; specificity ranged from 0.71 to 1.00; positive predictive value ranged from 0.67 to 1.00; and negative predictive value ranged from 0.69 to 0.90. For software packages that were installed at different locations, between-installation differences were observed.

CONCLUSION

The performance of most DDI-detecting software programs tested in this study was suboptimal. Improvement is needed to advance their contribution to detection of DDIs.

Van't Hoff and calorimetric enthalpies from isothermal titration calorimetry: are there significant discrepancies?

The enthalpy of a reaction is most often determined through one of two means; it can be determined directly using calorimetry or indirectly by measuring the temperature dependence of the equilibrium constant (i.e., the van't Hoff method). Recently, discrepancies have been noted between the enthalpy measured by calorimetry, and the enthalpy determined by the van't Hoff method,. This has been suggested to indicate that the binding reaction is more complex than the simple one-to-one binding model used to describe the data. To better understand possible discrepancies between and, we have undertaken both experimental studies using isothermal titration calorimetry to measure the binding energetics of Ba(2+) binding 18-crown-6 ether and 2'-CMP binding RNase A, along with a simulation of a system involving a molecule in conformational equilibrium coupled with binding. We find that when experimental setup and analysis are correctly performed, no statistically significant discrepancies between and exist even for the linked system.

Cyclosporine-drug interactions and the influence of patient age

Cyclosporine-drug interactions in adult transplant patients and the impact of age were studied. The medical records of transplant patients receiving cyclosporine therapy were identified. Data on patient demographics, cyclosporine dosages, dosage form, blood trough concentrations, clinical laboratory test values, and concurrent medications were collected. One-compartment models for oral and i.v. administration were used to fit cyclosporine concentration data to population pharmacokinetic and statistical models. Nonlinear mixed-effect modeling (NONMEM) software was used. The influence of covariates, including but not limited to concomitant medications and age, on cyclosporine pharmacokinetics was evaluated. The records of 100 patients (36 women and 64 men) were reviewed. A mean +/- S.D. of 9 +/- 2 and 9 +/- 1 medications per day were consumed by patients < 60 and > or = 60 years old, respectively. Mean population pharmacokinetic values of 0.407 L/hr/kg for clearance, 4.0 L/kg for volume of distribution, 31% for bioavailability, and 10.6 hours for half-life were determined on the basis of 569 blood cyclosporine levels. Twelve medications (sertraline, losartan, valsartan, quinine, atorvastatin, simvastatin, pravastatin, fluvastatin, alendronate, digoxin, acyclovir, and oxycodone) with previously unconfirmed pharmacokinetic interactions with cyclosporine were identified as interacting. There was no correlation between age and interactions. Patients taking cyclosporine were at risk for pharmacokinetic drug interactions when cyclosporine was used in combination with sertraline, losartan, valsartan, quinine, atorvastatin, simvastatin, pravastatin, fluvastatin, alendronate, digoxin, acyclovir, and oxycodone. Transplant patients 60-75 years of age had cyclosporine-drug interactions similar to those in younger patients.

Effect of metoprolol and verapamil administered separately and concurrently after single doses on liver blood flow and drug disposition

Nine healthy males participated in a double-blind, placebo-controlled, randomized, crossover study to determine the effects of verapamil and metoprolol administered alone and concurrently on blood flow through the hepatic artery and portal and hepatic veins and to detect a possible drug interaction between the two agents. Single oral doses of placebo/placebo, metoprolol (50 mg)/placebo, verapamil (80 mg)/placebo, or verapamil/metoprolol were separated by at least 14 days. Liver blood flow through individual hepatic vessels was measured up to 8 hours after dosage administration using a duplex Doppler ultrasound technique. Cardiac output, heart rate, blood pressure, stroke volume, and total peripheral resistance were measured for 3 hours after drug doses were given. In 5 subjects, pharmacokinetic parameters for total drug as well as S- and R-enantiomers were also measured. Verapamil given alone caused a rapid and intense increase in liver blood flow (hepatic artery = 50%, portal vein = 42%, hepatic vein = 55%) 0.75 to 1 hour after administration because of a decrease in total peripheral resistance and an increase in heart rate, stroke volume, and cardiac output. Metoprolol given alone caused a slow but prolonged decrease in liver blood flow (maximum decrease: hepatic artery = -54%, portal vein = -21%, hepatic vein = -27%) 4 hours after administration because of a decrease in heart rate and cardiac output. When the two agents were given together, a composite of the changes noted after separate administration was noted: a brief peak increase in liver blood flow at 0.33 to 1 hour followed by a slow, prolonged decrease that reached its maximum decline 4 to 5 hours postdose. During the combined phase, metoprolol and its enantiomers had an increased AUC and Cmax, while verapamil and its enantiomers had an increased AUC and t1/2. These pharmacokinetic changes were consistent with the magnitude and time course of liver blood flow changes through the hepatic artery and portal or hepatic veins.

Pharmacokinetics of diclofenac sodium in chronic active hepatitis and alcoholic cirrhosis

The objective of this study was to assess the pharmacokinetics of diclofenac sodium and its five metabolites following administration of a 150 mg oral dose to healthy subjects and patients with either chronic active hepatitis of varying morphology or alcoholic cirrhosis. Six healthy subjects, 6 chronic active hepatitis patients, and 6 alcoholic cirrhosis patients were enrolled in this prospective, open-label, parallel study. Blood samples were drawn at 0, 0.25, 0.5, 0.75, 1, 2, 4, 6, 8, 12, 24, 48, 72, 144, 312, and 480 hours, and urine samples were collected for 144 hours after administration of a single oral dose of diclofenac sodium. The mean area under the serum concentration-time curve extrapolated to infinity, oral clearance, half-life, maximal concentration, and time to peak concentration for diclofenac and its metabolites were determined and compared using analysis of variance. Cirrhotics had a mean +/- SD diclofenac AUC value (19,114 +/- 6806 ng.h/ml) significantly different (p < 0.02) from hepatitis patients (6071 +/- 1867 ng.h/ml) and healthy subjects (7008 +/- 2006 ng.h/ml), whereas healthy subjects and hepatitis patients had similar values. Comparable results were found for 4'-hydroxydiclofenac. The AUC values for 3'-hydroxydiclofenac and 3'-hydroxy-4'methoxydiclofeanc were significantly different when healthy subjects were compared to cirrhotics. However, hepatitis subjects were not significantly different from either group. The results indicate that hepatitis does not alter the pharmacokinetics of diclofenac. Alcoholic cirrhosis increased the mean diclofenac AUC approximately three times compared to normal subjects, indicating that one-third of the usual dose in cirrhotics would produce equivalent AUC values in normal subjects and subjects with alcoholic cirrhosis. However, since pharmacodynamic measurements were not made and no increase in untoward or side effects was noted in the alcoholic cirrhosis patients after a single dose, maintenance doses should be titrated to patients response.

Life-threatening interactions between HIV-1 protease inhibitors and the illicit drugs MDMA and gamma-hydroxybutyrate

Human immunodeficiency virus 1 (HIV-1) protease inhibitors have dramatically reduced the morbidity and mortality due to HIV-1 infection. However, most of these antiretrovirals are also potent inhibitors (and occasionally inducers) of hepatic and intestinal cytochrome P450 systems and, therefore, have the potential to alter the elimination of any substance that utilizes these metabolic pathways. We describe a patient infected with HIV-1 who was treated with ritonavir and saquinavir and then experienced a prolonged effect from a small dose of methylenedioxymetamphetamine (MDMA or ecstacy) and a nearly fatal reaction from a small dose of gamma-hydroxybutyrate (GHB). We also discuss the potential for HIV-1 protease inhibitors to alter the metabolism of other abusable prescribed and illicit substances.