This population-based cohort study evaluates the association between being newly dispensed a calcium channel blocker and subsequent dispensing of a loop diuretic in older adults with hypertension.
Key Points
Question
Are older adults who begin taking a calcium channel blocker more likely to be subsequently prescribed a diuretic, leading to a prescribing cascade, than those who began taking other medications?
Findings
In a population-based cohort study of 41 086 older adults with hypertension, being newly dispensed a calcium channel blocker was associated with a statistically significantly higher rate of being subsequently dispensed a loop diuretic within 90 days compared with 2 groups (n = 66 494 and n = 231 439) who began taking other medications.
Meaning
Many older adults who begin taking a calcium channel blocker may subsequently experience a prescribing cascade; steps can be taken to avoid prescribing unnecessary medications that can cause harm and are costly.
Abstract
Importance
Calcium channel blockers (CCBs) are commonly prescribed agents for hypertension that can cause peripheral edema. A prescribing cascade occurs when the edema is misinterpreted as a new medical condition and a diuretic is subsequently prescribed to treat the edema. The extent to which this prescribing cascade occurs at a population level is not well understood.
Objective
To measure the association between being newly dispensed a CCB and subsequent dispensing of a loop diuretic in older adults with hypertension.
Design, Setting, and Participants
A population-based cohort study was performed using linked health administrative databases of community-dwelling adults 66 years or older with hypertension and new prescription drug claims from September 30, 2011, to September 30, 2016, in Ontario, Canada. The dates of analysis were September 1, 2018, to May 30, 2019.
Exposures
Individuals who were newly dispensed a CCB were compared with the following 2 groups: (1) individuals who were newly dispensed an angiotensin-converting enzyme inhibitor or angiotensin II receptor blocker and (2) individuals who were newly dispensed an unrelated medication.
Main Outcomes and Measures
Hazard ratios (HRs) with 95% CIs were estimated for individuals who were dispensed a loop diuretic within 90 days of follow-up using Cox proportional hazards regression models.
Results
The cohort included 41 086 older adults (≥66 years) with hypertension who were newly dispensed a CCB, 66 494 individuals who were newly dispensed another antihypertensive medication, and 231 439 individuals who were newly dispensed an unrelated medication. At index (ie, the dispensing date), the mean (SD) age was 74.5 (6.9) years, and 191 685 (56.5%) were women. Individuals who were newly dispensed a CCB had a higher cumulative incidence at 90 days of being dispensed a loop diuretic than individuals in both control groups (1.4% vs 0.7% and 0.5%, P < .001). After adjustment, individuals who were newly dispensed a CCB had increased relative rates of being dispensed a loop diuretic compared with individuals who were newly dispensed an angiotensin-converting enzyme inhibitor or angiotensin II receptor blocker (HR, 1.68; 95% CI, 1.38-2.05 in the first 30 days after index [days 1-30]; 2.26; 95% CI, 1.76-2.92 in the subsequent 30 days [days 31-60]; and 2.40; 95% CI, 1.84-3.13 in the third month of follow-up [days 61-90]) and individuals who were newly dispensed unrelated medications (HR, 2.51; 95% CI, 2.13-2.96 for 1-30 days after index; 2.99; 95% CI, 2.43-3.69 for 31-60 days after index; and 3.89; 95% CI, 3.11-4.87 for 61-90 days after index). This association persisted, although slightly attenuated, from 90 days to up to 1 year of follow-up and when restricted to a subgroup of individuals who were newly dispensed amlodipine.
Conclusions and Relevance
Many older adults with hypertension who are newly dispensed a CCB subsequently receive a loop diuretic. Given how widely CCBs are prescribed, interventions are needed to raise clinicians’ awareness of this common prescribing cascade to reduce the prescribing of potentially unnecessary medications that may cause harm.
Introduction
Calcium channel blockers (CCBs) are first-line agents for hypertension management1,2 and rank among the 10 most commonly used prescription medications in North America.3,4 Clinicians may preferentially prescribe CCBs, particularly in older adults, because of their favorable adverse event profile and the limited need for routine laboratory monitoring.5 However, CCBs commonly cause peripheral edema, with an incidence ranging from 2% to 25% depending on the CCB type, dosage, and duration of therapy.5,6,7 Amlodipine poses the greatest concern because it is the most widely used CCB and is more likely to lead to peripheral edema than nondihydropyridine (DHP) CCBs and newer lipophilic DHP CCBs.5 Peripheral edema can be distressing to patients and can alter their quality of life, prompting individuals to seek medical attention or discontinue therapy.5
The presentation of peripheral edema may lead a prescriber to manage the symptom with a diuretic,8 with preferential prescription of loop diuretics because they promote greater fluid loss than other diuretic classes.9 However, CCB-induced edema is not caused by fluid overload, and treating euvolemic individuals with a diuretic places them at increased risk of overdiuresis, leading to falls, urinary incontinence, acute kidney injury, electrolyte imbalances, and a cascade of other downstream consequences to which older adults are especially vulnerable.6,10,11,12 This risk, particularly for falls and related injuries, is greatest immediately after receipt of a new diuretic prescription.11 In addition to these harms, prescribing an additional and potentially unnecessary medication has costs for individuals and health systems.
Adverse drug events (ADEs) (eg, CCB-induced edema) that are misinterpreted as new medical conditions and can be associated with subsequent prescription of a potentially unnecessary drug therapy have been described as prescribing cascades.13,14,15 The prescribing cascade concept has gained international recognition as an important contributor to the global challenge of problematic polypharmacy.16 Characterizing prescribing cascades and their prevalence is important to reduce potentially inappropriate drug prescribing, adverse events, avoidable downstream medical conditions, and unnecessary costs.13,14,15 Recognizing prescribing cascades is also a key component of the deprescribing process.17 Although a 2016 case report18 and a recent cross-sectional observational study19 provide evidence of the CCB-diuretic prescribing cascade, these studies lack comparator groups, prospective follow-up, and data on the time to event. Therefore, the clinical importance and consequences of this prescribing cascade at a population level are unknown. In this study, we compared the rate at which a population-based cohort of older adults with hypertension and newly dispensed a CCB were subsequently dispensed a loop diuretic with 2 comparison groups who were newly dispensed other medications.
Methods
Study Design and Setting
We conducted a population-based, retrospective cohort study using health administrative data collected as part of the publicly funded universal health insurance program in Ontario, Canada. Ontario is Canada’s most populous province, with an estimated 2.3 million older adults (≥65 years).20 Data sources on physician services, ambulatory and hospital care, and prescription medications for adults 66 years or older in Ontario, Canada, are listed in eTable 1 in the Supplement. These data sets were linked using unique encoded identifiers and analyzed at ICES, Toronto, Ontario, Canada. The use of data in this project was authorized under section 45 of Ontario’s Personal Health Information Protection Act, which does not require review by a research ethics board. All analyses were performed using SAS, version 9.4 (SAS Institute Inc).
Cohort Creation
First, we identified residents of Ontario, Canada, with prevalent hypertension21 as of September 30, 2016, using a validated algorithm (eTable 2 in the Supplement). The study flowchart is shown in the eFigure in the Supplement. This algorithm was applied to health records dating back as early as 1988. Next, we identified the exposed and comparison groups by searching for new prescription drug claims within the Ontario Drug Benefit (ODB) claims database between September 30, 2011, and September 30, 2016 (the accrual period). The dates of analysis were September 1, 2018, to May 30, 2019. Individuals whose relevant drug claim date preceded their hypertension diagnosis date were excluded. Once the groups based on prescription drug use were defined, a set of exclusion criteria were applied. Individuals were excluded if they (1) had died as of their index date (ie, the dispensing date), (2) were 65 years or younger or 110 years or older, (3) were not a resident of Ontario in the 2 years prior, (4) were ineligible for publicly funded health insurance at any point in the preceding year, and (5) did not have at least 1 ODB claim within the 2 preceding years to ensure they were using the ODB. Individuals were also excluded if they (6) had a diagnosis of heart failure or end-stage renal disease in the prior year (defined in eTable 2 in the Supplement) because these conditions are associated with peripheral edema; (7) had been hospitalized within 1 month preceding their index date because we are unable to track medication use during hospitalization stays; (8) were long-term care residents in the prior 6 months to restrict the cohort to community-dwelling older adults; and (9) were dispensed an antihypertensive medication or a diuretic in the prior 12 months to improve the likelihood that diuretic dispensing during the follow-up period was related to the exposure.
CCB Exposure
Individuals were classified as exposed if they had a new prescription claim for any of the CCBs available in the provincial drug formulary (ie, amlodipine, felodipine, nifedipine, diltiazem hydrochloride, and verapamil hydrochloride) within the ODB claims database during the 5-year accrual period (September 30, 2011-September 30, 2016) (eTable 2 in the Supplement). The first date of the first prescription claim (the dispensing date) was used as the index date. New use was defined by the absence of a CCB claim in the prior year.
Comparison Groups
Individuals with hypertension and no claims for a CCB during the accrual period or in the year before their index date were eligible as comparators. Two comparison groups were identified: one that comprised new users of other antihypertensive medications and a second that acted as a more general comparator. Individuals in the other antihypertensive medication comparison group were newly dispensed an angiotensin-converting enzyme inhibitor (ACEI) or angiotensin II receptor blocker (ARB) during the accrual period, with no ACEI or ARB claims in the year before their index date (ie, first dispensing date of ACEI or ARB during the accrual period) (eTable 2 in the Supplement). The general comparison group had at least 1 unrelated (ie, non-CCB) new drug claim during the accrual period. If there were multiple claims, 1 drug subclass was randomly selected, and a 1-year look-back period was used to ensure there were no claims for the same drug subclass. An index date was assigned based on the selected claim’s dispensing date.
Observation Period and Outcome Measurement
Individuals were observed for 90 days after their index date for evidence of the development of a prescribing cascade. The primary outcome was being dispensed a loop diuretic (ie, furosemide) (eTable 2 in the Supplement). Loop diuretics were selected as the primary outcome because they are more likely than other diuretics to be used to treat edema and are not generally recommended for management of hypertension, unlike thiazide or thiazide-like diuretics.1,2 Individuals were followed up until they were dispensed a loop diuretic, discontinued CCB treatment (the time to next prescription exceeding 1.5 times the total days supplied of the current medication prescription22), were hospitalized, died, or until the end of the observation period. Because other diuretics may also be used to treat edema, we considered any diuretic use as a secondary outcome (ie, amiloride, chlorthalidone, eplerenone, hydrochlorothiazide, indapamide, spironolactone, or triamterene, in addition to furosemide) (eTable 2 in the Supplement). As an a priori sensitivity analysis, the observation period was extended to 12 months to account for delays in health care–seeking behavior and/or delayed onset of edema.5 Post hoc, a sensitivity analysis was conducted in which other antihypertensive medication comparators were censored if individuals discontinued ACEI or ARB treatment to address the potential for differential follow-up in exposed and comparator groups because of discontinuation of treatment.
Covariates
Baseline characteristics included sociodemographic variables (eg, age, sex, and income), medical history (eg, duration of hypertension and comorbidities), health system use (eg, hospitalizations and emergency department visits), and concurrent drug therapies (Table 1 and eTable 2 in the Supplement). The covariates were selected because they have been shown to be associated with the development of edema or diuretic use.5,23,24,25
Table 1. Characteristics of Study Population of Older Adults With Hypertension.
| Characteristic | Exposure to Newly Dispensed CCB, No. (%) (n = 41 086) | Other Antihypertensive Medication Comparator, No. (%) (n = 66 494) | Standardized Differencea | General Comparator, No. (%) (n = 231 439) | Standardized Differencea |
|---|---|---|---|---|---|
| Sociodemographics | |||||
| Age, mean (SD), y | 74.8 (7.0) | 74.0 (6.6) | 0.1192 | 74.6 (7.0) | 0.0281 |
| Female sex | 24 384 (59.3) | 35 773 (53.8) | 0.1121 | 131 528 (56.8) | 0.0511 |
| Low-income older adult | 6632 (16.1) | 8874 (13.3) | 0.0789 | 27 350 (11.8) | 0.1249 |
| Neighborhood income quintile | |||||
| 1, Lowest | 7790 (19.0) | 11 766 (17.7) | 0.0327 | 38 490 (16.6) | 0.0609 |
| 2 | 8605 (20.9) | 13 525 (20.3) | 0.0149 | 45 726 (19.8) | 0.0295 |
| 3 | 8288 (20.2) | 13 181 (19.8) | 0.0087 | 45 648 (19.7) | 0.0112 |
| 4 | 8309 (20.2) | 13 943 (21.0) | 0.0184 | 49 964 (21.6) | 0.0336 |
| 5, Highest | 7974 (19.4) | 13 867 (20.9) | 0.0361 | 50 940 (22.0) | 0.0642 |
| Missing data | 120 (0.3) | 212 (0.3) | 0.0048 | 671 (0.3) | 0.0004 |
| Rural residence | 4139 (10.1) | 7810 (11.7) | 0.0536 | 28 629 (12.4) | 0.0728 |
| Medical History 1 y Before Index Date | |||||
| Duration of hypertension at index, mean (SD), y | 10.9 (7.1) | 9.5 (7.0) | 0.1926 | 10.2 (6.7) | 0.1043 |
| Chronic disease burden, mean (SD), No. | 3.2 (1.5) | 3.1 (1.4) | 0.0218 | 3.2 (1.4) | 0.0132 |
| 1 | 4494 (10.9) | 6576 (9.9) | 0.0343 | 22 303 (9.6) | 0.0428 |
| 2 | 10 536 (25.6) | 17 438 (26.2) | 0.0133 | 61 382 (26.5) | 0.0200 |
| 3 | 11 210 (27.3) | 19 457 (29.3) | 0.0439 | 66 755 (28.8) | 0.0347 |
| 4 | 7749 (18.9) | 12 615 (19.0) | 0.0028 | 43 958 (19.0) | 0.0034 |
| ≥5 | 7097 (17.3) | 10 408 (15.7) | 0.0437 | 37 041 (16.0) | 0.0341 |
| Comorbid chronic conditions that can cause peripheral edema | |||||
| Cancer | 6432 (15.7) | 10 444 (15.7) | 0.0014 | 41 941 (18.1) | 0.0659 |
| Diabetes | 10 812 (26.3) | 20 179 (30.3) | 0.0896 | 51 431 (22.2) | 0.0956 |
| Chronic liver disease | 504 (1.2) | 827 (1.2) | 0.0015 | 3009 (1.3) | 0.0066 |
| Chronic kidney disease | 3070 (7.5) | 3494 (5.3) | 0.0909 | 10 141 (4.4) | 0.1312 |
| Stroke | 1494 (3.6) | 2298 (3.5) | 0.0098 | 6215 (2.7) | 0.0544 |
| Health System Use 1 y Before Index Date | |||||
| ≥1 Primary care visit | 38 902 (94.7) | 62 972 (94.7) | 0.0008 | 209 206 (90.4) | 0.1638 |
| ≥1 Specialist visitb | 16 005 (39.0) | 23 322 (35.1) | 0.0804 | 65 834 (28.4) | 0.2237 |
| ≥1 Home care service | 3618 (8.8) | 5147 (7.7) | 0.0387 | 22 782 (9.8) | 0.0357 |
| ≥1 Hospitalization | 4101 (10.0) | 5650 (8.5) | 0.0513 | 18 603 (8.0) | 0.0679 |
| ≥1 Emergency department visit | 13 290 (32.3) | 19 104 (28.7) | 0.0786 | 61 608 (26.6) | 0.1259 |
| Drug Therapies | |||||
| Distinct drugs claimed in previous year, mean (SD) | 5.5 (4.4) | 5.3 (4.1) | 0.0440 | 5.0 (3.9) | 0.1326 |
| Concurrent medications, excluding CCB, mean (SD) | 3.1 (2.5) | 3.8 (2.3) | 0.2941 | 3.7 (2.2) | 0.2515 |
| Concurrent antihypertensive medications, mean (SD) | 0.1 (0.4) | 1.0 (0.2) | 2.9846 | 0.1 (0.3) | 0.1614 |
| ACEI | 1890 (4.6) | 42 192 (63.5) | 1.5849 | 8340 (3.6) | 0.0503 |
| ARB | 1585 (3.9) | 24 320 (36.6) | 0.8920 | 4432 (1.9) | 0.1162 |
| β-Blocker | 1490 (3.6) | 1973 (3.0) | 0.0369 | 3786 (1.6) | 0.1246 |
| α1-Adrenergic antagonist | 111 (0.3) | 82 (0.1) | 0.0331 | 411 (0.2) | 0.0196 |
| α2-Adrenergic agonist | 17 (0) | 7 (0) | 0.0192 | 57 (0) | 0.0092 |
| Vasodilator | 181 (0.4) | 101 (0.2) | 0.0531 | 154 (0.1) | 0.0744 |
| Concurrent medication classes known to cause edema | |||||
| NSAIDs | 3558 (8.7) | 5231 (7.9) | 0.0288 | 29 257 (12.6) | 0.1293 |
| Corticosteroids | 1225 (3.0) | 1626 (2.4) | 0.0330 | 11 933 (5.2) | 0.1102 |
| Gabapentinoids | 859 (2.1) | 1484 (2.2) | 0.0097 | 6965 (3.0) | 0.0583 |
| Dopamine agonists | 177 (0.4) | 233 (0.4) | 0.0129 | 1157 (0.5) | 0.0102 |
Abbreviations: ACEI, angiotensin-converting enzyme inhibitor; ARB, angiotensin II receptor blocker; CCB, calcium channel blocker; NSAIDs, nonsteroidal anti-inflammatory drugs.
Standardized difference relative to the exposed group.
Cardiologist or nephrologist.
Statistical Analysis
Descriptive statistics were used to evaluate the sociodemographic and clinical characteristics of cohort members at their index date. Standardized differences exceeding 0.10 (or 10%) were used to identify differences in baseline covariates between the exposed and comparison groups.26
Unadjusted cumulative incidence functions between the exposed and comparison groups were compared using the Gray test (2-sided). Multivariable Cox proportional hazards regression models were used to estimate hazard ratios (HRs) and 95% CIs for receipt of a loop diuretic, adjusting for a set of covariates assessed a priori, including age, sex, number of comorbidities (to account for variations in risk), and 6-month categories of the index date (to account for changes in drug trends or patient care over the study period). This analysis was repeated for the secondary outcome of any diuretic use. To be consistent with the definition of a prescribing cascade, we excluded individuals from our analysis who were dispensed a loop diuretic on their index date (for the primary outcome) (ie, 316 individuals [0.8%] who were newly dispensed a CCB, 760 individuals [1.1%] in the other antihypertensive medication comparator group, and 1872 individuals in the general comparator group [0.8%]) or who were dispensed any diuretic on their index date (for the secondary outcome) (ie, 1638 [4.0%] in the loop diuretic group, 4673 [7.0%] in the other antihypertensive group, and 5761 [2.5%] in the general comparator group).
We found that the association of new exposure to a CCB with the hazard of receiving a loop diuretic was time dependent using Shoenfeld residual plots. Therefore, we estimated HRs within 3 strata of follow-up time for the main analysis (1-30, 31-60, and 61-90 days) and in 2 strata of 91 to 180 days and 181 to 365 days for the extended observation period. Statistical tests were 2 sided, with P < .05 interpreted as statistically significant.
Subgroup Analyses
All subgroup analyses were defined a priori. We stratified analyses by sex given that women have been reported to be more likely than men to experience CCB-related edema.27 Because amlodipine is more likely to result in edema than other CCB types, we also stratified analyses by type of CCB (amlodipine or nonamlodipine CCBs). In addition, we examined whether there was a dose-dependent response.5 We categorized CCB dosage as low or high based on the last dosage prescribed during the follow-up period. Low dosages were considered to be one-half of the maximum daily dose or less (ie, ≤5 mg of amlodipine and felodipine, ≤30 mg of nifedipine, or ≤240 mg of diltiazem hydrochloride extended release or verapamil hydrochloride sustained release) as described in prior studies.5,28,29 Subgroup analyses were conducted over the 90-day and extended 12-month observation periods.
Results
Study Cohort
The final study cohort included 41 086 older adults with hypertension who were newly dispensed a CCB, 66 494 individuals in the other antihypertensive medication group, and 231 439 individuals in the general comparator group (eFigure in the Supplement). At the index date (ie, the dispensing date), the mean (SD) age was 74.5 (6.9) years, and 191 685 (56.5%) were women. Compared with individuals in the 2 comparison groups, those who were newly dispensed a CCB were more likely to be low-income older adults and to have a longer duration of hypertension at index but were similar with regard to other measured covariates (Table 1 and eTable 3 in the Supplement [stratified by sex]). Users of CCBs were primarily prescribed amlodipine (79.6%), followed by diltiazem (9.6%), nifedipine (9.1%), verapamil (0.9%), and felodipine (0.9%).
Primary Outcome of Being Prescribed a Loop Diuretic Within 90 Days
By 90 days, individuals newly dispensed a CCB had a higher cumulative incidence of receiving a loop diuretic than individuals in the comparison groups (1.4% vs 0.7% [other antihypertensive medication comparators] and 0.5% [general comparators], P < .001) (Figure). The mean (SD) time to receipt of a loop diuretic was 69 (29) days for individuals who were newly dispensed a CCB, 87 (15) days for those in the other antihypertensive medication comparator group, and 87 (13) days for those in the general comparator group. After adjustment, individuals who were newly dispensed a CCB were dispensed a loop diuretic at higher rates than those in the other antihypertensive medication comparator group over the 3 periods (HR, 1.68; 95% CI, 1.38-2.05 for 1-30 days; 2.26; 95% CI, 1.76-2.92 for 31-60 days; and 2.40; 95% CI, 1.84-3.13 for 61-90 days) (Table 2). Being newly dispensed a CCB more than doubled the hazards of receiving a loop diuretic compared with the general comparators at all intervals (HR, 2.51; 95% CI, 2.13-2.96 in the first 30 days after index [days 1-30]; 2.99; 95% CI, 2.43-3.69 in the subsequent 30 days [days 31-60]; and 3.89; 95% CI, 3.11-4.87 in the third month of follow-up [days 61-90]) (Table 2).
Figure. Cumulative Incidence of Being Dispensed a Loop Diuretic Among Older Adults With Hypertension Who Were Newly Dispensed a Calcium Channel Blocker (CCB) Compared With Other Antihypertensive Medication Comparators and General Comparators.
By 90 days, individuals newly dispensed a CCB had a higher cumulative incidence of receiving a loop diuretic (1.4%) compared with the other antihypertensive medication comparator group (0.7%) and the general comparator group (0.5%) (P < .001).
Table 2. Associations Between Being Dispensed a Calcium Channel Blocker (CCB) and Being Dispensed a Loop Diuretic in a Cohort of Older Adults With Hypertension.
| Comparison Group | HR (95% CI) | ||||
|---|---|---|---|---|---|
| 1-30 d | 31-60 d | 61-90 d | 91-180 d | 181-365 d | |
| CCB exposure vs other antihypertensive medication comparator | |||||
| Crude | 1.71 (1.41-2.08) | 2.32 (1.80-2.99) | 2.45 (1.88-3.19) | 2.29 (1.90-2.76) | 1.67 (1.41-1.98) |
| Adjusteda | 1.68 (1.38-2.05) | 2.26 (1.76-2.92) | 2.40 (1.84-3.13) | 2.24 (1.86-2.71) | 1.64 (1.38-1.94) |
| CCB exposure vs general comparator | |||||
| Crude | 2.37 (2.01-2.79) | 2.86 (2.32-3.53) | 3.72 (2.97-4.65) | 3.08 (2.62-3.62) | 2.15 (1.85-2.51) |
| Adjusteda | 2.51 (2.13-2.96) | 2.99 (2.43-3.69) | 3.89 (3.11-4.87) | 3.20 (2.72-3.76) | 2.22 (1.90-2.60) |
Abbreviation: HR, hazard ratio for time (in days) from index date.
Adjusted for age, sex, low-income status, rural residence, duration of hypertension, chronic disease burden, concurrent antihypertensive medications, concomitant use of medication classes that may also produce peripheral edema, and index date.
Secondary Outcome of Any Diuretic Use Within 90 Days
By 90 days, individuals who were newly dispensed a CCB had a higher cumulative incidence of receiving any diuretic vs those in the comparison groups (4.5% vs 3.4% and 1.0%, P < .001); these proportions increased to 9.5% vs 7.3% and 3.0% by 1 year. The mean time to being dispensed any diuretic was similar to the time to loop diuretic dispensing. Individuals who were newly dispensed a CCB subsequently were prescribed any diuretic at higher rates than individuals in both of the comparison groups (eTable 4 in the Supplement).
Sensitivity Analyses
By 1 year of follow-up, 3.5% of individuals newly exposed to a CCB were dispensed a loop diuretic compared with 1.8% of those in the other antihypertensive medication comparator group and 1.4% of those in the general comparator group. Individuals who were newly dispensed a CCB had a higher rate of being dispensed a loop diuretic and any diuretic vs those in both comparison groups across all intervals when follow-up was extended to 12 months (Table 2 and eTable 4 in the Supplement). Censoring individuals in the other antihypertensive medication comparator group if they discontinued ACEI or ARB treatment reduced the mean length of follow-up by 14 days but did not appreciably change effect estimates (eTable 5 in the Supplement).
Subgroup Analyses
No sex differences in the association between new CCB exposure and subsequent diuretic dispensing were observed (eTable 6 in the Supplement). Individuals newly dispensed a CCB had a greater hazard of receiving a loop diuretic vs those in the comparison groups regardless of CCB type (ie, amlodipine vs nonamlodipine CCBs) (Table 3). The cumulative incidence of being dispensed a loop diuretic within 90 days was 1.2% for individuals newly dispensed amlodipine and 2.3% for those dispensed nonamlodipine CCBs. Although individuals newly prescribed amlodipine had a higher rate of being dispensed a loop diuretic up to 90 days after the index date than did those in the other antihypertensive medication comparator group (HR, 1.42; 95% CI, 1.14-1.77 for 1-30 days; 1.82; 95% CI, 1.36-2.42 for 31-60 days; and 2.18; 95% CI 1.63-2.91 for 61-90 days), the magnitude of association was greater in the smaller subgroup of individuals newly prescribed other CCBs (Table 3). The same pattern was observed for any diuretic use (eTable 7 in the Supplement). The magnitude of association was generally greater with high-dose CCBs (Table 3); however, a dose-dependent response was only observed for the secondary outcome of any diuretic use (eTable 7 in the Supplement).
Table 3. Associations Between Being Dispensed a Calcium Channel Blocker (CCB) and Being Dispensed a Loop Diuretic, by Type and Dosage of CCB Dispenseda.
| Comparison Group | Adjusted HR (95% CI) | ||||
|---|---|---|---|---|---|
| 1-30 d | 31-60 d | 61-90 d | 91-180 d | 181-365 d | |
| CCB exposure vs other antihypertensive medication comparator | |||||
| Amlodipine | 1.42 (1.14-1.77) | 1.82 (1.36-2.42) | 2.18 (1.63-2.91) | 2.03 (1.65-2.50) | 1.56 (1.29-1.89) |
| Nonamlodipine CCB | 2.70 (2.05-3.56) | 4.03 (2.85-5.71) | 3.30 (2.21-4.92) | 3.14 (2.33-4.21) | 1.98 (1.46-2.70) |
| Low-dose CCB | 1.71 (1.39-2.10) | 2.11 (1.60-2.79) | 2.42 (1.82-3.21) | 1.90 (1.54-2.36) | 1.43 (1.17-1.74) |
| High-dose CCB | 1.70 (1.21-2.39) | 2.96 (2.01-4.36) | 2.59 (1.68-4.00) | 3.55 (2.69-4.69) | 2.66 (2.01-3.52) |
| CCB exposure vs general comparator | |||||
| Amlodipine | 2.11 (1.74-2.55) | 2.40 (1.87-3.08) | 3.53 (2.74-4.55) | 2.89 (2.41-3.48) | 2.12 (1.78-2.52) |
| Nonamlodipine CCB | 4.02 (3.13-5.18) | 5.31 (3.87-7.29) | 5.32 (3.66-7.74) | 4.46 (3.37-5.89) | 2.67 (1.98-3.61) |
| Low-dose CCB | 2.55 (2.14-3.04) | 2.78 (2.20-3.53) | 3.91 (3.06-5.00) | 2.71 (2.24-3.28) | 1.93 (1.61-2.32) |
| High-dose CCB | 2.57 (1.86-3.55) | 3.97 (2.77-5.68) | 4.26 (2.82-6.43) | 5.14 (3.96-6.67) | 3.67 (2.81-4.80) |
Abbreviation: HR, hazard ratio for time (in days) from index date.
Fully adjusted for age, sex, low-income status, rural residence, duration of hypertension, chronic disease burden, concurrent antihypertensive medications, concomitant use of medication classes that may also produce peripheral edema, and index date.
Discussion
In a large, population-based cohort of older adults with hypertension, individuals who were newly dispensed a CCB experienced more than a 60% higher rate of being subsequently dispensed a loop diuretic compared with those who began taking other antihypertensive medications. Rates of loop diuretic dispensing increased within the first 30 days and remained elevated throughout 1 year of follow-up. Findings were robust across CCB type and dosage.
Given how widely CCBs are prescribed, the results of the present study highlight a prescribing cascade that occurs in a large number of adults. In this study, 3.5% of older adults who were newly dispensed a CCB subsequently were prescribed a loop diuretic within 1 year; this proportion rose to 9.5% for any diuretic use. With more than 14 million people receiving an amlodipine prescription in the United States in 2016, this finding reflects 500 000 to 1.3 million new, potentially unnecessary diuretic prescriptions each year.30 We believe that our findings corroborate those from a 2016 case report18 and a 2018 cross-sectional study19 of US patient visits, which found that a loop diuretic was continued or newly prescribed in 4.6% of visits at which a DHP CCB was continued, but the present study improves on this evidence by establishing temporality and examining the extent of this practice at a population level.
Because peripheral edema occurs more frequently in patients taking DHP CCBs compared with non-DHP CCBs (eg, verapamil and diltiazem),5 we expected a stronger association between individuals who were newly dispensed a CCB and subsequent dispensing of a loop diuretic in the subgroup of individuals who were prescribed amlodipine vs those who were prescribed nonamlodipine CCBs. That we did not see this association may be attributed to differences in how edema was managed and/or the indication for CCB drug therapy. For example, individuals taking verapamil may have been prescribed a CCB to provide rate control for atrial fibrillation or tachyarrhythmia rather than solely to treat hypertension.31 Therefore, higher rates of edema and resulting diuretic treatment may have been attributable to underlying cardiovascular illness. In this subgroup, diuretic therapy may have been appropriate. We excluded individuals with diagnosed heart failure and those with a history of antihypertensive or diuretic use in the preceding year to minimize these potential confounders, although undiagnosed individuals with heart failure or those who developed heart failure after their index date would have been included.
Diuretic therapy may have also been appropriate among individuals requiring multiple agents for blood pressure control.2 As mentioned previously, we excluded individuals with a recent history of antihypertensive or diuretic use in the preceding year. As such, the cohort included individuals with new-onset or mild hypertension for whom diuretics would unlikely be prescribed as part of guideline-based hypertension management.
No sex differences were observed in the association between new CCB exposure and subsequent diuretic dispensing, similar to findings reported in a recent US study,19 despite the fact that women have higher rates of CCB-induced edema than men.27 The lack of sex differences observed in this study may be attributable to differences in edema management or variations in CCB prescribing practices (ie, women may receive lower dosages or for shorter duration). Previous studies32,33,34,35 failed to report results by sex; however, we believe that patient sex should be an ongoing consideration in future research on hypertension treatment, ADEs, and subsequent management.
Implications
Given the harms and costs associated with prescribing diuretics to treat CCB-induced edema, especially for vulnerable older adults, clinicians need to be aware of the prescribing cascade of CCBs and diuretics in adults with hypertension and how it can be avoided. At the outset, clinicians should consider whether an antihypertensive drug therapy is needed for blood pressure control in an older patient given the potential for many of these medications to increase the risk of falls and associated hip fractures.36 If CCB therapy for management of hypertension is warranted and peripheral edema occurs, clinicians should consider whether peripheral edema is an ADE, even if the edema occurs later (ie, weeks to months) in the patient’s course of treatment. Before prescribing a diuretic to manage the edema, clinicians should consider whether the CCB is still necessary, whether it could be discontinued or the dosage could be reduced, or whether the patient can be switched to another therapy.13,15 Nonpharmacologic strategies to address peripheral edema should also be considered.
As described earlier, a CCB-diuretic prescribing cascade may be appropriate based on an individual’s unique circumstances. Even when deemed appropriate, frequent reevaluation of the goals for care and ongoing assessments are recommended because the appropriateness of the treatment may change over time.13
Limitations
This study has limitations. The indication for prescribed medications is not included in the ODB nor is a standardized diagnostic code available for peripheral edema; as a result, we could not confirm that a dispensed diuretic was used to treat CCB-induced edema. By excluding patients with heart failure and end-stage renal disease, selecting loop diuretics as a primary outcome, and controlling for conditions and drug therapies known to lead to edema, we minimized other indications for diuretic prescribing. Prescribing cascades may precipitate other potentially harmful and costly actions beyond prescribing a second drug therapy.15 Unnecessary diagnostic tests may be ordered if the edema is incorrectly thought to have another source (eg, heart failure) requiring further investigation. Because we were unable to measure these practices, the true burden of the CCB-diuretic prescribing cascade at a population level was likely underestimated. Future studies could examine the use of novel data sources to examine the broader-reaching consequences of prescribing cascades.
Conclusions
We observed that older adults with hypertension who were newly dispensed a CCB subsequently were dispensed a loop diuretic at higher rates than those who began taking other antihypertensive medications or unrelated medications. We believe that the results of the present study stress the need to raise awareness of this prescribing cascade and call for vigilance in preventing the cascade and its related harms.
eTable 1. Description of ICES administrative health care databases used in the study
eTable 2. Definitions of key variables
eTable 3. Characteristics of study population by sex
eTable 4. Hazard ratios with 95% CIs for associations between dispensed calcium channel blockers (CCB) and any diuretic (secondary outcome)
eTable 5. Associations between dispensed calcium channel blockers (CCB) and diuretics, censoring other antihypertensive medication comparators for discontinued angiotensin-converting enzyme inhibitor (ACEI) or angiotensin II receptor blocker (ARB) treatment
eTable 6. Hazard ratios with 95% CIs for associations between being newly dispensed a calcium channel blocker (CCB) and subsequent dispensing of a loop diuretic (primary) or any diuretic (secondary), by sex
eTable 7. Fully adjusted hazard ratios and 95% cis for associations between dispensed calcium channel blockers (CCB) and any diuretics (secondary outcome), by type and dose of CCB dispensed
eFigure. Overall study flow diagram
eReferences.
References
- 1.Whelton PK, Carey RM, Aronow WS, et al. 2017 ACC/AHA/AAPA/ABC/ACPM/AGS/APhA/ASH/ASPC/NMA/PCNA guideline for the prevention, detection, evaluation, and management of high blood pressure in adults: a report of the American College of Cardiology/American Heart Association Task Force on Clinical Practice Guidelines [published correction appears in J Am Coll Cardiol. 2018;71(19):2275-2279]. J Am Coll Cardiol. 2018;71(19):e127-e248. doi: 10.1016/j.jacc.2017.11.006 [DOI] [PubMed] [Google Scholar]
- 2.Nerenberg KA, Zarnke KB, Leung AA, et al. ; Hypertension Canada . Hypertension Canada’s 2018 guidelines for diagnosis, risk assessment, prevention, and treatment of hypertension in adults and children. Can J Cardiol. 2018;34(5):506-525. doi: 10.1016/j.cjca.2018.02.022 [DOI] [PubMed] [Google Scholar]
- 3.Kantor ED, Rehm CD, Haas JS, Chan AT, Giovannucci EL. Trends in prescription drug use among adults in the United States from 1999-2012. JAMA. 2015;314(17):1818-1831. doi: 10.1001/jama.2015.13766 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 4.Canadian Institute for Health Information . Drug use among seniors on public drug programs in Canada, 2012. https://secure.cihi.ca/free_products/Drug_Use_in_Seniors_on_Public_Drug_Programs_2012_EN_web.pdf. Published 2014. January 13, 2020.
- 5.Makani H, Bangalore S, Romero J, et al. Peripheral edema associated with calcium channel blockers: incidence and withdrawal rate: a meta-analysis of randomized trials. J Hypertens. 2011;29(7):1270-1280. doi: 10.1097/HJH.0b013e3283472643 [DOI] [PubMed] [Google Scholar]
- 6.Sica D. Calcium channel blocker–related periperal edema: can it be resolved? J Clin Hypertens (Greenwich). 2003;5(4):291-294, 297. doi: 10.1111/j.1524-6175.2003.02402.x [DOI] [PMC free article] [PubMed] [Google Scholar]
- 7.Leonetti G, Magnani B, Pessina AC, Rappelli A, Trimarco B, Zanchetti A; COHORT Study Group . Tolerability of long-term treatment with lercanidipine versus amlodipine and lacidipine in elderly hypertensives. Am J Hypertens. 2002;15(11):932-940. doi: 10.1016/S0895-7061(02)03000-5 [DOI] [PubMed] [Google Scholar]
- 8.Weir MR, Rosenberger C, Fink JC. Pilot study to evaluate a water displacement technique to compare effects of diuretics and ACE inhibitors to alleviate lower extremity edema due to dihydropyridine calcium antagonists. Am J Hypertens. 2001;14(9, pt 1):963-968. doi: 10.1016/S0895-7061(01)02167-7 [DOI] [PubMed] [Google Scholar]
- 9.Brater DC. Pharmacology of diuretics. Am J Med Sci. 2000;319(1):38-50. doi: 10.1016/S0002-9629(15)40678-0 [DOI] [PubMed] [Google Scholar]
- 10.Sica DA. Diuretic-related side effects: development and treatment. J Clin Hypertens (Greenwich). 2004;6(9):532-540. doi: 10.1111/j.1524-6175.2004.03789.x [DOI] [PMC free article] [PubMed] [Google Scholar]
- 11.Berry SD, Mittleman MA, Zhang Y, et al. New loop diuretic prescriptions may be an acute risk factor for falls in the nursing home. Pharmacoepidemiol Drug Saf. 2012;21(5):560-563. doi: 10.1002/pds.3256 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 12.Kelly J, Chamber J. Inappropriate use of loop diuretics in elderly patients. Age Ageing. 2000;29(6):489-493. doi: 10.1093/ageing/29.6.489 [DOI] [PubMed] [Google Scholar]
- 13.McCarthy LM, Visentin JD, Rochon PA. Assessing the scope and appropriateness of prescribing cascades. J Am Geriatr Soc. 2019;67(5):1023-1026. doi: 10.1111/jgs.15800 [DOI] [PubMed] [Google Scholar]
- 14.Rochon PA, Gurwitz JH. Optimising drug treatment for elderly people: the prescribing cascade. BMJ. 1997;315(7115):1096-1099. doi: 10.1136/bmj.315.7115.1096 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 15.Rochon PA, Gurwitz JH. The prescribing cascade revisited. Lancet. 2017;389(10081):1778-1780. doi: 10.1016/S0140-6736(17)31188-1 [DOI] [PubMed] [Google Scholar]
- 16.Brath H, Mehta N, Savage RD, et al. What is known about preventing, detecting, and reversing prescribing cascades: a scoping review. J Am Geriatr Soc. 2018;66(11):2079-2085. doi: 10.1111/jgs.15543 [DOI] [PubMed] [Google Scholar]
- 17.Scott IA, Hilmer SN, Reeve E, et al. Reducing inappropriate polypharmacy: the process of deprescribing. JAMA Intern Med. 2015;175(5):827-834. doi: 10.1001/jamainternmed.2015.0324 [DOI] [PubMed] [Google Scholar]
- 18.Nguyen PV, Spinelli C. Prescribing cascade in an elderly woman. Can Pharm J (Ott). 2016;149(3):122-124. doi: 10.1177/1715163516640811 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 19.Vouri SM, van Tuyl JS, Olsen MA, Xian H, Schootman M. An evaluation of a potential calcium channel blocker–lower-extremity edema–loop diuretic prescribing cascade. J Am Pharm Assoc (2003). 2018;58(5):534-539.e4. doi: 10.1016/j.japh.2018.06.014 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 20.Statistics Canada . Census profile, 2016 census: Ontario [province] and Canada [country]. https://www12.statcan.gc.ca/census-recensement/2016/dp-pd/prof/details/Page.cfm?Lang=E&Geo1=PR&Code1=35&Geo2=&Code2=&Data=Count&SearchText=Ontario&SearchType=Begins&SearchPR=01&B1=All&GeoLevel=PR&GeoCode=35. Published 2017. Accessed January 13, 2020.
- 21.Tu K, Campbell NR, Chen ZL, Cauch-Dudek KJ, McAlister FA. Accuracy of administrative databases in identifying patients with hypertension. Open Med. 2007;1(1):e18-e26. [PMC free article] [PubMed] [Google Scholar]
- 22.Vigod SN, Gomes T, Wilton AS, Taylor VH, Ray JG. Antipsychotic drug use in pregnancy: high dimensional, propensity matched, population based cohort study. BMJ. 2015;350:h2298. doi: 10.1136/bmj.h2298 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 23.Messerli FH. Vasodilatory edema: a common side effect of antihypertensive therapy. Am J Hypertens. 2001;14(9, pt 1):978-979. doi: 10.1016/S0895-7061(01)02178-1 [DOI] [PubMed] [Google Scholar]
- 24.Ely JW, Osheroff JA, Chambliss ML, Ebell MH. Approach to leg edema of unclear etiology. J Am Board Fam Med. 2006;19(2):148-160. doi: 10.3122/jabfm.19.2.148 [DOI] [PubMed] [Google Scholar]
- 25.Epstein BJ, Roberts ME. Managing peripheral edema in patients with arterial hypertension. Am J Ther. 2009;16(6):543-553. doi: 10.1097/MJT.0b013e3181afbf9f [DOI] [PubMed] [Google Scholar]
- 26.Austin PC. An introduction to propensity score methods for reducing the effects of confounding in observational studies. Multivariate Behav Res. 2011;46(3):399-424. doi: 10.1080/00273171.2011.568786 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 27.Pfizer. NORVASC (amlodipine besylate). https://www.accessdata.fda.gov/drugsatfda_docs/label/2011/019787s047lbl.pdf. Revised May 2011. Accessed January 13, 2020.
- 28.Canadian Pharmacists Association . Calcium channel blockers. In: RxTx. Hudson, OH: Canadian Pharmacists Association; 2018. [Google Scholar]
- 29.Marras C, Austin PC, Bronskill SE, Diong C, Rochon PA. Antipsychotic drug dispensing in older adults with parkinsonism. Am J Geriatr Psychiatry. 2018;26(12):1244-1257. doi: 10.1016/j.jagp.2018.08.003 [DOI] [PubMed] [Google Scholar]
- 30.Agency for Healthcare Research and Quality . Total purchases in thousands by prescribed drug, United States, 1996-2016. https://meps.ahrq.gov/mepstrends/hc_pmed/. Accessed January 13, 2020.
- 31.January CT, Wann LS, Calkins H, et al. 2019 AHA/ACC/HRS focused update of the 2014 AHA/ACC/HRS guideline for the management of patients with atrial fibrillation: a report of the American College of Cardiology/American Heart Association Task Force on Clinical Practice Guidelines and the Heart Rhythm Society in collaboration with the Society of Thoracic Surgeons. Circulation. 2019;140(2):e125-e151. doi: 10.1161/CIR.0000000000000665 [DOI] [PubMed] [Google Scholar]
- 32.Gradman AH, Cutler NR, Davis PJ, Robbins JA, Weiss RJ, Wood BC; Enalapril-Felodipine ER Factorial Study Group . Combined enalapril and felodipine extended release (ER) for systemic hypertension. Am J Cardiol. 1997;79(4):431-435. doi: 10.1016/S0002-9149(96)00781-3 [DOI] [PubMed] [Google Scholar]
- 33.Pepine CJ, Cooper-DeHoff RM, Weiss RJ, et al. ; Comparative Efficacy and Safety of Nisoldipine and Amlodipine (CESNA-II) Study Investigators . Comparison of effects of nisoldipine–extended release and amlodipine in patients with systemic hypertension and chronic stable angina pectoris. Am J Cardiol. 2003;91(3):274-279. doi: 10.1016/S0002-9149(02)03154-5 [DOI] [PubMed] [Google Scholar]
- 34.Toal CB, Mahon WA, Barnes C, Burelle D. Nifedipine gastrointestinal therapeutic system (GITS) for hypertensive patients in a primary care setting: results of the Extended Release Adalat Canadian Trial (EXACT). Clin Ther. 1997;19(5):924-935. doi: 10.1016/S0149-2918(97)80046-X [DOI] [PubMed] [Google Scholar]
- 35.Kloner RA, Weinberger M, Pool JL, et al. ; Comparison of Candesartan and Amlodipine for Safety, Tolerability and Efficacy (CASTLE) Study Investigators . Comparative effects of candesartan cilexetil and amlodipine in patients with mild systemic hypertension. Am J Cardiol. 2001;87(6):727-731. doi: 10.1016/S0002-9149(00)01491-0 [DOI] [PubMed] [Google Scholar]
- 36.Butt DA, Mamdani M, Austin PC, Tu K, Gomes T, Glazier RH. The risk of hip fracture after initiating antihypertensive drugs in the elderly. Arch Intern Med. 2012;172(22):1739-1744. doi: 10.1001/2013.jamainternmed.469 [DOI] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.
Supplementary Materials
eTable 1. Description of ICES administrative health care databases used in the study
eTable 2. Definitions of key variables
eTable 3. Characteristics of study population by sex
eTable 4. Hazard ratios with 95% CIs for associations between dispensed calcium channel blockers (CCB) and any diuretic (secondary outcome)
eTable 5. Associations between dispensed calcium channel blockers (CCB) and diuretics, censoring other antihypertensive medication comparators for discontinued angiotensin-converting enzyme inhibitor (ACEI) or angiotensin II receptor blocker (ARB) treatment
eTable 6. Hazard ratios with 95% CIs for associations between being newly dispensed a calcium channel blocker (CCB) and subsequent dispensing of a loop diuretic (primary) or any diuretic (secondary), by sex
eTable 7. Fully adjusted hazard ratios and 95% cis for associations between dispensed calcium channel blockers (CCB) and any diuretics (secondary outcome), by type and dose of CCB dispensed
eFigure. Overall study flow diagram
eReferences.