Error in Dates of Data Analysis in the Methods: the Research Letter Titled

Letters

Author Affiliations: Renal Hypertension Center, Division of Nephrology, Table 2. Change in Daytime Systolic Blood Pressure University of Ottawa, Ottawa, Ontario, Canada (Ruzicka, Bugeja, Edwards, Mean (SD), mm Hg McCormick, Hiremath); Division of Cardiology, University of Ottawa Heart Institute, Ottawa, Ontario, Canada (Leenen); Ottawa Hospital Research With Daytime ABPM Mean Value Institute, Ottawa, Ontario, Canada (Ruzicka, Ramsay, Hiremath); Kidney SBP SBP Research Center, University of Ottawa, Ottawa, Ontario, Canada (Ruzicka, Time Point Overall <135 mm Hg ≥135 mm Hg Bugeja, Edwards, McCormick, Hiremath). Immediately after DOT −9.7 (17.3) −26.1 (19.9) −2.9 (10.4) (n = 48) Accepted for Publication: March 29, 2019. 1 mo After DOT −11.0 (15.8) −21.9 (14.0) −6.3 (14.2) Corresponding Author: Marcel Ruzicka, MD, PhD, FRCPC, Renal Hypertension (n = 46) Center, Division of Nephrology, University of Ottawa, 1967 Riverside, Rm 5-21, Ottawa, Ontario K1H 7W9, Canada ([email protected]). Abbreviations: ABPM, ambulatory blood pressure monitoring; DOT, directly Published Online: June 17, 2019. doi:10.1001/jamainternmed.2019.1455 observed therapy; SBP, systolic blood pressure. Author Contributions: Drs Ruzicka and Hiremath had full access to all of the data in the study and take responsibility for the integrity of the data and the accuracy of the data analysis. Results | A total of 60 consecutive patients (32 men [67%]; mean Concept and design: Ruzicka, Ramsay, McCormick, Hiremath. [SD] age, 62.1 [13.1] years) were enrolled in the study, and af- Acquisition, analysis, or interpretation of data: All authors. Drafting of the manuscript: Ruzicka, Leenen, Ramsay, Hiremath. ter exclusion of those who withdrew consent (n = 4), did not Critical revision of the manuscript for important intellectual content: Ruzicka, attend DOT (n = 4), or missed subsequent ABPM (n = 4), 48 par- Ramsay, Bugeja, Edwards, McCormick, Hiremath. ticipants completed this study for the primary outcome and Statistical analysis: Ruzicka, Ramsay, Hiremath. 46 for the secondary outcome. Baseline characteristics are re- Obtained funding: Ruzicka, Ramsay, Hiremath. Administrative, technical, or material support: Ruzicka, Leenen, Ramsay, ported in Table 1. After DOT, daytime systolic BP remained 135 McCormick, Hiremath. mm Hg or greater in 34 of 48 patients (71%) who experienced Supervision: Ruzicka, Ramsay, Bugeja, Edwards, McCormick, Hiremath. a mean (SD) decrease in systolic BP of 3 (10) mm Hg. In con- Conflict of Interest Disclosures: Dr Ruzicka reported receiving grants from trast, in 14 participants (29%), treatment-resistant hyperten- Physicians Services Incorporated and The Ottawa Hospital Academic Medical Organization–Innovation Fund Provincial Oversight Committee during the sion resolved and systolic BP decreased by 26 (20) mm Hg conduct of the study. Dr Hiremath reported receiving grants from the Canadian (Table 2). This proportion was similar at 1 month in 14 of 46 Institutes of Health Research, Physicians Services Incorporated, and The Ottawa patients (30%) who no longer had treatment-resistant hyper- Hospital Academic Medical Organization outside the submitted work. No other tension. disclosures were reported. Funding/Support: This study was funded by a peer-reviewed grant (Dr Ruzicka) provided by The Ottawa Hospital Academic Medical Organization and Discussion | The results suggest that nonadherence to BP- the Innovation Fund Provincial Oversight Committee (Ontario, Canada). lowering drug regimens is high among referred patients with Role of the Funder/Sponsor: The funding sources had no role in the design and apparent treatment-resistant hypertension, even among those conduct of the study; collection, management, analysis, and interpretation of the data; preparation, review, or approval of the manuscript; and decision to who said they were adherent on questioning before DOT, had submit the manuscript for publication. pristine pharmacy filling records, and had accurate pill counts. Additional Contributions: Peter Magner, MD, FRCPC, provided critical review Moreover, this apparent nonadherence occurred despite more of the concept and results of this study (not compensated); Valerie Cronin, RN, than 50% of these patients already having had an adverse vas- was the research coordinator for the study (compensated); and the Hypertension Clinic study staff conducted the direct observed therapy cular event related to uncontrolled hypertension. However, we procedures (compensated). cannot exclude the possibility that the process of being in the 1. Burnier M, Wuerzner G, Struijker-Boudier H, Urquhart J. Measuring, study or receiving treatment from a nurse in a clinic was as- analyzing, and managing drug adherence in resistant hypertension. Hypertension. sociated with lower BP for some patients. Of interest, most of 2013;62(2):218-225. doi:10.1161/HYPERTENSIONAHA.113.00687 those with markedly improved BP after DOT had a sustained 2. Fadl Elmula FE, Hoffmann P, Larstorp AC, et al. Adjusted drug treatment is superior to renal sympathetic denervation in patients with true improvement in BP control seen at 1 month. Limitations of the treatment-resistant hypertension. Hypertension. 2014;63(5):991-999. doi:10. study include that the patients were highly selected and likely 1161/HYPERTENSIONAHA.114.03246 do not represent most patients with hypertension in the com- 3. Ruzicka M, Hiremath S. Can drugs work in patients who do not take them? munity. The use of DOT as described here was strictly dichoto- the problem of non-adherence in resistant hypertension. Curr Hypertens Rep. mous (adherence vs nonadherence) and thus does not allow 2015;17(9):579. doi:10.1007/s11906-015-0579-4 4. Ruzicka M, McCormick B, Leenen FH, Froeschl M, Hiremath S. Adherence to for precise assessment of the degree of nonadherence (eg, par- blood pressure-lowering drugs and resistant hypertension: should trial of direct tial vs complete), as may be the case with therapeutic drug observation therapy be part of preassessment for renal denervation? Can J Cardiol. monitoring.5,6 Overall, the findings suggest that rigorous meth- 2013;29(12):1741.e1-1741.e3. doi:10.1016/j.cjca.2013.07.678 ods of adherence assessment and intervention such as DOT 5. Brinker S, Pandey A, Ayers C, et al. Therapeutic drug monitoring facilitates should be considered for patients with apparent treatment- blood pressure control in resistant hypertension. J Am Coll Cardiol. 2014;63(8): 834-835. doi:10.1016/j.jacc.2013.10.067 resistant hypertension. 6. Eskås PA, Heimark S, Eek Mariampillai J, Larstorp AC, Fadl Elmula FE, Høieggen A. Adherence to medication and drug monitoring in apparent Marcel Ruzicka, MD, PhD, FRCPC treatment-resistant hypertension. Blood Press. 2016;25(4):199-205. doi:10. Frans H. H. Leenen, MD, PhD, FRCPC 3109/08037051.2015.1121706 Tim Ramsay, PhD Ann Bugeja, MD, FRCPC Prevalence of Inappropriateness of Parenteral

Cedric Edwards, MD, FRCPC Vitamin B12 Administration in Ontario, Canada Brendan McCormick, MD, FRCPC Randomized clinical trials demonstrate that treating vitamin

Swapnil Hiremath, MD, MPH B12 (cobalamin, or hereinafter B12) deficiency with oral supple-

1434 JAMA Internal Medicine October 2019 Volume 179, Number 10 (Reprinted) jamainternalmedicine.com

Downloaded From: https://jamanetwork.com/ on 09/27/2021 Letters

mentation substantially increases serum B levels compared 12 Table. Characteristics of Patients Who Received an Intramuscular with intramuscular injections, with no difference in hemato- a Vitamin B12 Prescription logic or neuropsychiatric outcomes.1 Despite this, some pri- Patients, No. (%) mary care physicians still inappropriately administer B12 in- Characteristics (n = 146 850) jections to elderly patients.2 To our knowledge, there is no Age, y published literature characterizing prescribing patterns of in- Mean (SD) 76.5 (8.1) Median (IQR) [range] 76 (14.0) tramuscular B12 using laboratory data to document patient se- [65-110] rum levels. In this study, we assessed the prevalence of inap- Age categorized, y propriate B supplementation using population-based 12 65-69 36 866 (25.1) databases and estimated the associated cost. 70-74 28 196 (19.2) 75-79 28 014 (19.1) Methods | We performed a population-based, retrospective co- 80-84 26 055 (17.7) hort study using health system administrative databases within 85-89 18 384 (12.5) ICES, formerly the Institute for Clinical Evaluative Sciences, in ≥ 90 9335 (6.4) Ontario, Canada. Data sets were linked using unique, encoded Sex identifiers and analyzed at ICES. All persons 65 years or older Male 60 037 (40.9)

who received at least 1 intramuscular B12 prescription from Janu- Female 86 813 (59.1)

ary 1, 2011, to September 30, 2015 (data on B12 levels were not Location of residence available until January 1, 2010), were included. Data were ana- Rural 12 692 (8.6) lyzed from July 26, 2018, to November 22, 2018. The primary Urban 124 359 (84.7) Neighborhood income quintile outcome was the proportion of inappropriate B12 supplemen- Q1 (lowest) 32 905 (22.4) tation, defined as persons with either a normal serum B12 level Q2 32 230 (22.0) (≥ 221 pmol/L), or without a documented B12 level in the 12 Q3 29 828 (20.3) months prior to their first intramuscular B12 prescription. Vita- Q4 27 679 (18.9) min B12 supplementation was considered appropriate when per- Q5 (highest) 23 573 (16.1) sons had at least 1 documented level of marginal B12 deficiency Comorbidities (≤ 221 pmol/L) in the year prior to receiving their first B injec- 12 Crohn disease, ulcerative colitis, and malabsorption 9309 (6.3) tion. Annual cost of inappropriate, once-monthly injections was Pernicious anemia 40 908 (27.9) estimated in Canadian dollars using the amount paid for a phy- ADG comorbidity classification scheme sician visit ($33.70), intramuscular injection ($3.89), and pre- Low scores (0-5) 24 135 (16.4) scription cost ($6.74). Sunnybrook Health Sciences Centre’s re- Moderate scores (6-9) 51 920 (35.4) search ethics board approved this study and waived patient High scores (≥ 10) 70 795 (48.2) written informed consent for deidentified data. Dementia 14 844 (10.1) Neuropathy 2471 (1.7)

Results | A total of 405 469 intramuscular B12 prescriptions were Abbreviations: ADG, Aggregated Diagnosis Group; IQR, interquartile range. dispensed to 146 850 persons (Table); the majority (63.7%; a Some totals may not add up owing to missing data. n = 93 615) of these were inappropriate (Figure). In the year

Figure. Prevalence of Inappropriate Intramuscular Vitamin B12 Prescribing in Ontario From January 1, 2011, to September 30, 2015

146 850 Patients ≥65 y who received an intramuscular vitamin B12 prescription from January 1, 2011, to September 30, 2015

93 615 Inappropriate 53 235 Appropriate

56 128 37 487 27 200 6035 No documented vitamin Vitamin B12 levels Vitamin B12 levels Vitamin B12 levels B12 levels in the past 12 mo >221 pmol/L 149-221 pmol/L <148 pmol/L

24 175 Vitamin B12 levels 1461 Patients with malabsorptive 1037 Patients with malabsorptive documented >12 mo disorders disorders prior to prescription

15 631 8539 Vitamin B12 levels Vitamin B12 levels >221 pmol/L <221 pmol/L

jamainternalmedicine.com (Reprinted) JAMA Internal Medicine October 2019 Volume 179, Number 10 1435

Downloaded From: https://jamanetwork.com/ on 09/27/2021 Letters

preceding persons’ first intramuscular B12 injection, 25.5% Corresponding Author: William K. Silverstein, MD, Core Internal Medicine, Faculty of Medicine, University of Toronto, Department of Medicine, University (n = 37 487) had a normal B12 level, whereas 38.2% (n = 56 128) Health Network, 200 Elizabeth St, Eaton Building 14-217, Toronto, Ontario M5G did not have a B12 level documented. Findings were similar over 2C4, Canada ([email protected]). a 24-month look-back period (data not shown). Only 43.1% Author Contributions: Drs Cheung, Croxford, and Dharma had full access to all (n = 24 175) of the 56 128 people without a B12 level docu- of the data in the study and take responsibility for the integrity of the data and

mented in the year preceding their first B12 prescription had the accuracy of the data analysis. Drs Lin and Cheung contributed equally as ever had one measured. This was performed a mean (SD) 1033.5 co–senior authors to this study. Study concept and design: Silverstein, Lin, Dharma, Cheung. (488.1) days prior to their first prescription (range, 366-2801 Acquisition, analysis, or interpretation of data: Lin, Dharma, Croxford, days). Only 35.3% (n = 8539) of these 24 175 persons had mar- Earle, Cheung. Drafting of the manuscript: Silverstein, Cheung. ginally deficient B12 levels. The estimated annual cost of in- Critical revision of the manuscript for important intellectual content: Lin, appropriate B prescribing was $45.6 million, assuming a 64% 12 Dharma, Croxford, Earle, Cheung. inappropriate prescription rate. Finally, only 1.7% (n = 2498) Statistical analysis: Dharma, Croxford, Cheung.

of persons prescribed intramuscular B12 demonstrated any Obtained funding: Cheung. deficiency with a malabsorptive indication. Administrative, technical, or material support: Silverstein, Dharma, Cheung. Study supervision: Lin, Cheung. Conflict of Interest Disclosures: Dr Lin reported that she was a consultant for Discussion | Most parenteral B in Ontario was prescribed to 12 Pfizer and that she was on the advisory board of Amgen. No other disclosures persons without evidence of deficiency in the year preced- were reported.

ing their first B12 prescription. Potential drivers of this Funding/Support: This study was supported by research funding from the include patient demands and poor physician awareness of Sunnybrook Hospital Foundation. This study was also supported by ICES, which 3,4 is funded by an annual grant from the Ontario Ministry of Health and Long-Term the evidence informing B12 supplementation. It is also Care (MOHLTC). questionable whether parenteral supplementation is Role of the Funder/Sponsor: The funders had no role in the design and required over oral supplementation because oral B raises 12 conduct of the study; collection, management, analysis, and interpretation of B12 serum levels and improves sequelae of deficiency as well the data; preparation, review, or approval of the manuscript; and decision to

as, if not better than, intramuscular B12, even for pernicious submit the manuscript for publication. anemia.1 Plausible reasons why physicians prefer parenteral Disclaimer: No endorsement by ICES or the Ontario MOHLTC is intended or should be inferred. Parts of this material are based on data and information B12 include low quality of evidence supporting oral B12, soci- compiled and provided by the Canadian Institute for Health Information (CIHI). ety guidelines recommending intramuscular B12 for all The opinions, results and, conclusions reported in this article are those of the patients, poor physician understanding of how to prescribe authors and are independent from the funding sources and CIHI.

oral B12, and physician misperception that patients prefer Additional Contributions: We thank IMS Brogan Inc for use of their Drug 1,3,5,6 parenteral over oral B12. Information Database. No contributors received compensation for their Our study’s limitations include only looking 2 years be- assistance. fore a person’s first documented prescription; using this 1. Chan CQH, Low LL, Lee KH. Oral vitamin B12 replacement for the treatment of pernicious anemia. Front Med (Lausanne). 2016;3(38):38. abridged period might have misclassified persons undergo- 2. van Walraven CG, Naylor CD. Use of vitamin B injections among elderly ing treatment for chronic B deficiency, and so with normal 12 12 patients by primary care practitioners in Ontario. CMAJ. 1999;161(2):146-149. B levels, as receiving inappropriate supplementation. We were 12 3. Graham ID, Jette N, Tetroe J, Robinson N, Milne S, Mitchell SL. Oral also unable to access information on oral B12, and could not cobalamin remains medicine’s best kept secret. Arch Gerontol Geriatr.2007;44

understand why B12 was prescribed without laboratory evi- (1):49-59. doi:10.1016/j.archger.2006.02.003 dence of deficiency. Further studies should examine this 4. Tilburt JC, Wynia MK, Sheeler RD, et al. Views of US physicians about issue, to inform quality improvement initiatives aimed at controlling health care costs. JAMA. 2013;310(4):380-388. doi:10.1001/jama. 2013.8278 reducing this unnecessary care. 5. Devalia V, Hamilton MS, Molloy AM; British Committee for Standards in Haematology. Guidelines for the diagnosis and treatment of cobalamin and William K. Silverstein, MD folate disorders. Br J Haematol. 2014;166(4):496-513. doi:10.1111/bjh.12959 Yulia Lin, MD, FRCPC 6. Kwong JC, Carr D, Dhalla IA, Tom-Kun D, Upshur RE. Oral vitamin B12 therapy Christoffer Dharma, MSc in the primary care setting: a qualitative and quantitative study of patient Ruth Croxford, MSc perspectives. BMC Fam Pract. 2005;6(1):8. doi:10.1186/1471-2296-6-8 Craig C. Earle, MD, MSc, FRCPC Matthew C. Cheung, MD, SM, FRCPC Association of Mirabegron With the Risk of Arrhythmia in Adult Patients 66 Years or Older— Author Affiliations: Department of Medicine, University of Toronto, Toronto, Ontario, Canada (Silverstein, Lin, Earle, Cheung); Division of Medical Oncology A Population-Based Cohort Study & Hematology, Sunnybrook Health Sciences Centre, Toronto, Ontario, Canada Recently, mirabegron, the first β3-adrenoceptor agonist, has (Lin, Earle, Cheung); Department of Laboratory Medicine & Molecular been prescribed to treat overactive bladder (OAB) more Diagnostics, Sunnybrook Health Sciences Centre, Toronto, Ontario, Canada frequently than antimuscarinic agents.1 The β-3 agonist medi- (Lin); Department of Laboratory Medicine and Pathobiology, University of Toronto, Toronto, Ontario, Canada (Lin); ICES, Toronto, Ontario, Canada cations have limited adverse (Dharma, Croxford, Earle, Cheung). effects compared with anti- 2,3 Accepted for Publication: April 20, 2019. Supplemental content muscarinic agents. How-

Published Online: July 15, 2019. doi:10.1001/jamainternmed.2019.1859 ever, β3-adrenoreceptors are Correction: This article was corrected on August 26, 2019, to fix errors in the associated with increases in contractile force and reductions dates of data analysis in the Methods section. in inotropic effects,4 actions which raise concerns of cardio-

1436 JAMA Internal Medicine October 2019 Volume 179, Number 10 (Reprinted) jamainternalmedicine.com

Downloaded From: https://jamanetwork.com/ on 09/27/2021 Letters

Table 1. Baseline Characteristics of Users by Overactive Bladder-Treatment Group

No. (%) Standardized Characteristic Mirabegron Other OAB Drugs Differencea New users, No. 16 948 21 870 NA Age, median (IQR), y 76 (71-82) 76 (71-82) NA Male 6342 (37.4) 7287 (33.3) NA Neighborhood income quintileb 1 2907 (17.2) 4100 (18.7) 0.05 2 3241 (19.1) 4417 (20.2) 0.02 3 3353 (19.8) 4338 (19.8) 0.01 4 3629 (21.4) 4543 (20.8) 0.02 5 3745 (22.1) 4363 (19.9) 0.05 Living in urban community 14 867 (87.7) 18 911 (86.5) 0.01 Living in long-term care 351 (2.1) 511 (2.3) 0.02 Charlson comorbidity score No hospitalization 11 213 (66.2) 14 759 (67.5) 0.02 0 2565 (15.1) 3113 (14.2) 0.02 1 1196 (7.1) 1549 (7.1) <0.01 2+ 1974 (11.6) 2449 (11.2) <0.01 Comorbidities Frailty 2910 (17.2) 3398 (15.5) 0.03 Chronic obstructive pulmonary disease 4560 (26.9) 5819 (26.6) <0.01 Diabetes 5951 (35.1) 7806 (35.7) 0.02 Hypertension 13 190 (77.8) 17 203 (78.7) 0.02 CV events in the past 5 y Acute myocardial infarction 425 (2.5) 531 (2.4) <0.01 Transient ischemic attack 1015 (6.0) 1215 (5.6) 0.02 Peripheral vascular disease 195 (1.2) 244 (1.1) 0.01 Atrial fibrillation 1098 (6.5) 1469 (6.7) 0.02 Angina 484 (2.9) 675 (3.1) 0.02 Ventricular arrhythmia 35 (0.2) 36 (0.2) <0.01 Congestive heart failure 1766 (10.4) 2347 (10.7) 0.02 Health care use in past year, visits, median (IQR) Emergency department 0 (0-1) 0 (0-1) 0.03 Physician 8 (4-15) 8 (4-14) 0.01 Cardiologist 0 (0-0) 0 (0-0) 0.01 Urologist 0 (0-1) 0 (0-0) 0.10 Different medications, No., median (IQR), No. 8 (5-11) 8 (5-11) 0.03 Concurrent recent CV medication use Angiotensin-converting enzyme 4033 (23.8) 5160 (23.6) <0.01 Angiotensin receptor blocker 3667 (21.6) 4611 (21.1) 0.01 Statin 7729 (45.6) 9641 (44.1) 0.01 Other lipid lowering agents 950 (5.6) 1148 (5.2) 0.01 β-Blocker 3606 (21.3) 4684 (21.4) 0.02 Abbreviations: CV, cardiovascular; Calcium-channel blocker 4360 (25.7) 5580 (25.5) <0.01 IQR, interquartile range; NA, not Diuretics 4380 (25.8) 5839 (26.7) 0.02 applicable; OAB, overactive bladder. Platelet inhibitor 1066 (6.3) 1287 (5.9) 0.01 a Standardized difference of >0.1 used Benign prostatic hyperplasia treatments 3197 (18.9) 3217 (14.7) 0.06 to signify a meaningful difference. b Switched therapy within 1 y 960 (5.7) 1306 (6.0) 0.023 Neighborhood income quintile is based on each participant’s postal Days on therapy, median (IQR) 104 (30-249) 81 (30-199) 0.194 code.

vascular (CV) adverse effects. These adverse effects have been Methods | We used health care administrative data from 38 818 reinforced by trials, finding a small increase in heart rate, blood patients 66 years or older who initiated treatment between June pressure, and QTc intervals.2-4 Real-world data among older 1, 2015, and March 31, 2017 (eTable 1 in the Supplement). Data patients with CV comorbidities are lacking.5 We conducted a were analyzed from June 1, 2015, to March 31, 2017. In On- population-based cohort study to evaluate the risk of cardiac tario, all health care services and medications are publicly arrhythmias and other CV events in a population of patients funded for individuals 65 years and older. Information is re- 66 years and older receiving mirabegron. corded by the Ministry of Health and Long-term Care and data

jamainternalmedicine.com (Reprinted) JAMA Internal Medicine October 2019 Volume 179, Number 10 1437

Downloaded From: https://jamanetwork.com/ on 09/27/2021 Letters

Table 2. Results of Primary and Secondary Outcomes

Incidence Rate Events, No. (%) per 100 Person-Years Mirabegron Other OAB Drugs Other Hazard Ratio Variable (n = 16 948) (n = 21 870) Mirabegron OAB Drugs (95% CI) Primary outcome (hospitalization for arrhythmia 284 (1.7) 340 (1.6) 3.6 3.8 0.93 (0.80-1.09) or tachycardia) Secondary outcome (hospitalization for myocardial 233 (1.4) 247 (1.1) 3.0 2.8 1.06 (0.89-1.27) infarction or stroke) Subgroup analysis (primary outcome: no previous 177 (1.1) 190 (0.9) 2.4 2.3 0.95 (0.75-1.20) atrial fibrillation or ventricular arrhythmias at baseline) Aged 75 y or older 221 (2.5) 255 (2.2) 5.2 5.3 0.93 (0.76-1.13)

Abbreviation: OAB, overactive bladder.

are securely held in databases at ICES6 and are linked using with higher prevalence of comorbidities than in previous clini- unique identifiers (eAppendix in the Supplement). Use of these cal trials.2,3 In previous studies, the prevalence of diabetes data was authorized under section 45 of the Ontario Personal ranged between 6% and 9% and that of hypertension ranged Health Information Protection Act, which does not require between 10% and 29%.2,3 In contrast, a 79% prevalence of hy- review by a research ethics board. pertension and 36% prevalence of diabetes were found in the We identified new users of OAB treatments and followed present study. Moreover, the mean age of patients in this co- up for 1 year or until they discontinued or switched therapy hort, 77 years, was higher than that of previous trials, where (eAppendix in the Supplement). New users were defined as mean ages ranged from 58 to 60. having no OAB treatment in the previous year. The primary out- Our study was limited by our inability to ascertain life- come was any hospitalization or emergency department visit style factors and over-the-counter medication use, and the po- for arrhythmia or tachycardia (eTable 2 in the Supplement). tential for confounding by indication owing to prescriber per- The secondary outcome was any hospitalization or visit for ceptions of mirabegron risk. However, we anticipate these myocardial infarction (MI) or stroke. Patients receiving mira- factors have minimal consequences given the use of strin- begron were matched with up to 4 patients receiving other OAB gent matching criteria and the HDPS. agents (anticholinergic agents) on age (±3 y), sex, index date The findings of this study suggest that mirabegron was not (±3 mo), and high-dimensional propensity score (HDPS; within associated with a higher risk of CV events compared with other 0.2 SD). We compared characteristics between groups using treatments. Our findings are not meant to endorse preferen- standardized differences and used Cox proportional hazards tial use of mirabegron but to support a growing body of evi- regression models accounting for the matched nature of the dence that mirabegron is not associated with an excess risk of data to compare mirabegron with other agents. We con- CV events compared with other treatments in older patients. ducted 2 subgroup analyses limited to (1) those with no pre- These results appear to support current prescribing patterns vious atrial fibrillation or ventricular arrhythmias at baseline, and give a balanced view of real-world safety. and (2) those older than 75 years. Analyses were conducted using SAS, version 9.2 (SAS Institute Inc). All outcomes are Mina Tadrous, PharmD, PhD based on 95% CIs from Cox proportional hazards regression Rano Matta, MD, MASc model hazard ratios (HRs). Simon Greaves, MSc Sender Herschorn, MD Results | We matched 16 948 patients who received mirabegron Muhammad M. Mamdani, PharmD, MA, MPH to 21 870 patients who received of other OAB drugs (Table 1). David N. Juurlink, MD, PhD The median age was 76 (interquartile range, 71-83), and 64.9% Tara Gomes, MHSc, PhD were female (25 189). Hypertension (30 393 [78.3%]) and diabetes (13 757 [35.4%]) were highly prevalent in the cohort. The Author Affiliations: Women's College Research Institute, Women's College Hospital, Toronto, Ontario, Canada (Tadrous); Institute for Clinical Evaluative 1-year cumulative incidence (adjusted for person-years) of ar- Sciences (ICES), Toronto, Ontario, Canada (Tadrous, Matta, Greaves, Mamdani, rhythmia or tachycardia events was similar between exposure Juurlink, Gomes); Leslie Dan Faculty of Pharmacy, University of Toronto, groups (3.6% for mirabegron vs 3.8% for other OAB drugs; HR, Toronto, Ontario, Canada (Tadrous, Mamdani, Gomes); Li Ka Shing Knowledge Institute, St. Michael’s Hospital, Toronto, Ontario, Canada (Tadrous, Gomes); 0.93; 95% CI, 0.80-1.09; Table 2). Mirabegron was not associ- Division of Urology, University of Toronto, Toronto, Ontario, Canada (Matta, ated with an increased risk of MI or stroke compared with other Herschorn); Institute of Health Policy, Management and Evaluation, University OAB drugs (HR 1.06; 95% CI, 0.89-1.27). Results were consis- of Toronto, Toronto, Ontario, Canada (Matta, Mamdani, Juurlink, Gomes); tent in subgroup analysis (2.4% for mirabegron vs 2.3% for other Sunnybrook Research Institute, Toronto, Ontario, Canada (Herschorn, Juurlink); Li Ka Shing Centre for Healthcare Analytics Research and Training, St Michael’s OAB drugs; HR, 0.95; 95% CI, 0.75-1.20). Hospital, Toronto, Ontario, Canada (Mamdani). Accepted for Publication: April 27, 2019. Discussion | A lack of evidence exists for assessing the safety of Corresponding Author: Mina Tadrous, PharmD, PhD, Women's College mirabegron in older patients with CV risk factors, and our work Research Institute, Women's College Hospital, 76 Grenville St, highlights the CV safety of mirabegron in a cohort of patients Toronto, ON M5S 1B2 ([email protected]).

1438 JAMA Internal Medicine October 2019 Volume 179, Number 10 (Reprinted) jamainternalmedicine.com

Downloaded From: https://jamanetwork.com/ on 09/27/2021 Letters

Published Online: July 15, 2019. doi:10.1001/jamainternmed.2019.2011 reductase inhibitors (5-ARIs). As noted, the prediagnostic use AuthorContributions:MrGreaveshadfullaccesstoallthedatainthestudyandtakes of 5-ARIs is associated with delayed prostate cancer diagno- responsibility for the integrity of the data and the accuracy of the data analysis. sis, more advanced disease, and worse prostate cancer– Concept and design: Tadrous, Matta, Mamdani, Gomes. specific and all-cause mortality compared with nonusers who Acquisition, analysis, or interpretation of data: Tadrous, Greaves, Herschorn, Mamdani, Juurlink, Gomes. underwent prostate-specific antigen (PSA) screening. Drafting of the manuscript: Tadrous, Matta. Although the recommendation of how to make the adjust- Critical revision of the manuscript for important intellectual content: All authors. ment of PSA suppression by 5-ARIs is not completely clear,2,3 Statistical analysis: Tadrous, Matta, Greaves, Mamdani. Obtained funding: Gomes. the association of corrected PSA to prostate volume (PSA den- Administrative, technical, or material support: Tadrous, Mamdani, Gomes. sity [PSAD]) is a tool that has been recognized as having greater Supervision: Tadrous, Mamdani, Juurlink, Gomes. specificity compared with using total PSA and has shown Conflict of Interest Disclosures: Dr Tadrous reported receiving grants from to be a significant predictor of different indices of cancer Ministry of Health and Long-term Care during the conduct of the study. aggressiveness.4 Dr Herschorn reported receiving grants and personal fees from Astellas, personal fees from Pfizer, grants from Ipsen and Ixaltis, and grants and personal Moreover, several studies have shown prostate volume to fees from Allergan outside the submitted work. Dr Mamdani reported other have a significant role predicting prostate cancer with PSA val- support from NovoNordisk, Allergan, and other support from Amgen outside ues in the gray zone and thus propose it as a tool that must be the submitted work. Dr Gomes reported receiving grants from Ontario Ministry taken into consideration in the diagnostic approach.5 It is pre- of Health and Long-Term Care during the conduct of the study. No other disclosures were reported. sumed that users of 5-ARIs are patients with larger prostates, and Funding/Support: This study was funded by a grant (HSRF; 06673) from the a fundamental tool to avoid misinterpretation could be PSAD. Ontario Ministry of Health and Long-Term Care (MOHLTC) Health System Although Sarkar and colleagues1 described multiple vari- Research Fund and supported by the Institute for Clinical Evaluative Sciences ables listed in Table 1 of their article, no mention is made of (ICES), a nonprofit research institute sponsored by the Ontario MOHLTC. these fundamental clinical parameters at the time of diagno- Role of the Funder/Sponsor: The funding organizations had no role in the sis. This information could help elucidate differences be- design and conduct of the study; collection, management, analysis, and interpretation of the data; preparation, review, or approval of the manuscript; tween groups by (1) assessing the degree of PSA misinterpre- and decision to submit the manuscript for publication. tation in real-life clinical practice, (2) knowing if for the same Disclaimer: The opinions, results and conclusions reported in this article are comparative PSAD there are still worse prostate cancer char- those of the authors and are independent from the funding sources. No acteristics in 5-ARI users, or (3) identifying a subgroup for which endorsement by ICES or the Ontario MOHLTC is intended or should be inferred. Parts of this material are based on data and information compiled and provided there is not an increased risk of delayed or worst prostate can- by CIHI. However, the analyses, conclusions, opinions and statements cer with the use of this medication. expressed herein are those of the author, and not necessarily those of CIHI. In summary, we believe PSAD is a determining factor to Additional Contributions: We thank Brogan Inc, Ottawa for use of their Drug be considered and it should be used to correlate to other clini- Product and Therapeutic Class Database. cal and laboratory factors to accurately describe differences 1. Tadrous M, Elterman D, Khuu W, Mamdani MM, Juurlink DN, Gomes T. between groups in the study by Sarkar and colleagues. Publicly funded overactive bladder drug treatment patterns in Ontario over 15 years: an ecological study. Can Urol Assoc J. 2018;12(3):E142-E145. doi:10.5489/ cuaj.4541 Camilo A. Medina, MD 2. Kelleher C, Hakimi Z, Zur R, et al. Efficacy and tolerability of mirabegron Laura Zuluaga, MD compared with antimuscarinic monotherapy or combination therapies for Mauricio Plata, MD, MSc overactive bladder: a systematic review and network meta-analysis. Eur Urol. 2018;74(3):324-333. doi:10.1016/j.eururo.2018.03.020 Author Affiliations: Department of Urology, Fundación Santa Fe de Bogotá, 3. Maman K, Aballea S, Nazir J, et al. Comparative efficacy and safety of medical Bogotá DC, Colombia. treatments for the management of overactive bladder: a systematic literature review and mixed treatment comparison. Eur Urol. 2014;65(4):755-765. doi:10. Corresponding Author: Camilo A. Medina, MD, Department of Urology, 1016/j.eururo.2013.11.010 Fundación Santa Fe de Bogotá, Cra 7 No 118-09, Bogotá DC, Colombia ([email protected]). 4. Gauthier C, Leblais V, Kobzik L, et al. The negative inotropic effect of beta3-adrenoceptor stimulation is mediated by activation of a nitric oxide Conflict of Interest Disclosures: Dr Plata reports receiving grants from synthase pathway in human ventricle. J Clin Invest. 1998;102(7):1377-1384. GlaxoSmithKline outside of the submitted work. No other disclosures doi:10.1172/JCI2191 are reported. 5. Rosa GM, Ferrero S, Nitti VW, Wagg A, Saleem T, Chapple CR. Cardiovascular 1. Sarkar RR, Parsons JK, Bryant AK, et al. Association of treatment with safety of β3-adrenoceptor agonists for the treatment of patients with 5α-reductase inhibitors with time to diagnosis and mortality in prostate cancer. overactive bladder syndrome. Eur Urol. 2016;69(2):311-323. doi:10.1016/ JAMA Intern Med. 2019;179(6):812-819. doi:10.1001/jamainternmed.2019.0280 j.eururo.2015.09.007 2. GrossmanDC,CurrySJ,OwensDK,etal;USPreventiveServicesTaskForce.Screen- 6. Data & Analytic Services. ICES web site. https://www.ices.on.ca/. Accessed ingforprostatecancer:USPreventiveServicesTaskForceRecommendationStatement. July 9, 2018. JAMA. 2018;319(18):1901-1913. doi:10.1001/jama.2018.3710 3. NCCN Clinical Practice Guidelines in Oncology. Prostate cancer early detection recommendations. National Comprehensive Cancer Network. COMMENT & RESPONSE https://www.nccn.org/professionals/physician_gls/pdf/prostate_detection.pdf. Published 2019. 5α-Reductase Inhibitor Use in Patients 4. Verma A, St Onge J, Dhillon K, Chorneyko A. PSA density improves With Prostate Cancer prediction of prostate cancer. Can J Urol. 2014;21(3):7312-7321. To the Editor The article by Sarkar and colleagues1 recently pub- 5. Erdogan A, Polat S, Keskin E, Turan A. Is prostate volume better than PSA density and free/total PSA ratio in predicting prostate cancer in patients with lished in JAMA Internal Medicine contributes to deeper knowl- PSA 2.5-10 ng/mL and 10.1-30 ng/mL? [published online March 12, 2019] Aging edge of prostate cancer diagnosis in patients who are using 5-α Male. doi:10.1080/13685538.2019.1578741

jamainternalmedicine.com (Reprinted) JAMA Internal Medicine October 2019 Volume 179, Number 10 1439

Downloaded From: https://jamanetwork.com/ on 09/27/2021 Letters

To the Editor Interpretation of observational studies is challeng- Corresponding Author: Ian M. Thompson Jr, MD, CHRISTUS Santa Rosa ing when results are different from randomized trials.1 The re- Hospital Medical Center, 2827 Babcock Rd, San Antonio, TX 78229 ([email protected]). cently published report by Sarkar and colleagues2 illustrates this Conflict of Interest Disclosures: None reported. challenge. A retrospective cohort of patients diagnosed with 1. Soni PD, Hartman HE, Dess RT, et al. Comparison of population-based prostate cancer between 2001 and 2005 was described, and it observational studies with randomized trials in oncology. J Clin Oncol. 2019;37 was found that the use of 5-α reductase inhibitors (5-ARIs) was (14):1209-1216. doi:10.1200/JCO.18.01074 associated with a delayed diagnosis and a 39% increase in pros- 2. Sarkar RR, Parsons JK, Bryant AK, et al. Association of treatment with tate cancer mortality. Initially, this conclusion seems at vari- 5α-reductase inhibitors with time to diagnosis and mortality in prostate cancer. ance with the follow-up report of the randomized Prostate JAMA Intern Med. 2019;179(6):812-819. doi:10.1001/jamainternmed.2019.0280 Cancer Prevention Trial, which showed no increase in prostate 3. Goodman PJ, Tangen CM, Darke AK, et al. Long-term effects of finasteride on prostate cancer mortality. N Engl J Med. 2019;380(4):393-394. doi:10.1056/ cancer mortality in men receiving finasteride compared NEJMc1809961 with placebo.3 4. Andriole GL, Crawford ED, Grubb RL III, et al; PLCO Project Team. Mortality Observational studies are prone to major biases. In the study results from a randomized prostate-cancer screening trial. N Engl J Med. 2009; by Sarkar and colleagues,2 the 39% improvement in prostate can- 360(13):1310-1319. doi:10.1056/NEJMoa0810696 cer mortality in men with a 2-year earlier diagnosis of prostate 5. AUA Practice Guidelines Committee. AUA guideline on management of cancer and median 5.9 years of follow-up is implausible. Two benign prostatic hyperplasia (2003). Chapter 1: diagnosis and treatment recommendations. J Urol. 2003;170(2 pt 1):530-547. large randomized trials of prostate-specific antigen (PSA) screening found only a 20% reduction and no reduction in prostate cancer mortality, respectively.4 As accurate assessment of 5-ARI In Reply We would like to thank Medina and colleagues and medication adherence was not possible, the study by Sarkar and Thompson Jr and colleagues for their interest and thoughtful colleagues2 was subject to misadjustment of the PSA. comments on our article.1 We completely agree that the goal Men being treated with 5-ARIs are inherently different, of observational studies is not to refute clinical trials, but rather which is illustrated by important differences in baseline char- to complement them through assessment of real-world prac- acteristics in the study by Sarkar and colleagues,2 for which tice. In the Prostate Cancer Prevention Trial, prostate- adequate adjustment is difficult. Also, men diagnosed with specific antigen (PSA) values for men being treated with fin- prostate cancer who received a 5-ARI will have fewer low- asteride (a 5-α reductase inhibitor [5-ARI]) were automatically grade prostate cancers, enriching for more high-grade tu- adjusted for the effect of the medication by a relatively com- mors. National guidelines for treatment of benign prostatic hy- plicated formula where the PSA was doubled for the first couple perplasia recommend use of 5-ARIs for men with prostate of years, then multiplied by 2.3 for subsequent years.2 enlargement.5 Because prostate size is directly related to PSA, Our study was motivated by the question of what hap- and aggressive prostate cancers are often missed on initial bi- pens in real-world practice if PSA values are not automati- opsy in large prostates, it is possible that men treated with cally adjusted. The data in our study suggest that PSA adjust- 5-ARIs had a first biopsy that was benign but a biopsy years ment is not routinely performed, which may lead to substantial later showing cancer. Without knowing the biopsy and PSA his- delays in diagnosis, more advanced disease, and a small tory, these confounding factors cannot be adjusted. Finally, the increase in the risk of prostate cancer mortality.1 seismic changes in PSA screening guidelines in the United Our colleagues bring up important considerations. First, States between 2001 and 2015 likely influenced screening and they mention that the magnitude of effect seems implausible biopsy patterns, and are not included in the analysis. given the small benefit from PSA screening trials.3,4 The authors’ analysis2 did not adjust for survival bias: as However, one trial had very high rates of PSA screening in the the prostate cancer survival clock started earlier in men not control arm, which substantially dilutes the expected effect. receiving a 5-ARI, there is an associated longer survival in them, We believe the best estimate comes from a modeling study exacerbated by the significant baseline age differences (65 years using data from the randomized screening trials that showed vs 70 years). a reduction in prostate cancer mortality of 7% to 9% for every Adjustment for these powerful study biases is beyond the year of earlier diagnosis.5 Our study suggested a potential de- reach of the most sophisticated statistical approaches. Be- lay in diagnosis of 2 to 3 years, which would be expected to cause a key study hypothesis was that lack of PSA adjustment increase the risk of death by 14% to 27%, similar to the ob- for 5-ARIs led to delayed diagnosis, a superior approach would served effect size. Also, the absolute increase risk in prostate have been to examine the time between first PSA screening re- cancer mortality (5%) is very similar to the increase in stage sults greater than 4.0 and biopsy in men receiving or not IV cancers in the 5-ARI group (5%), which supports the plau- receiving a 5-ARI. sibility of our findings. It is true that use of 5-ARIs have consistently been shown Ian M. Thompson Jr, MD to reduce the incidence of Gleason grade 6 cancers but not Catherine M. Tangen, DrPH Gleason grade 7 and higher cancers.6 To avoid the problem of Barnett Kramer, MD, MPH enriching for high-grade prostate cancer in the 5-ARI group, we did an analysis where Gleason grade 6 cancers were excluded Author Affiliations: CHRISTUS Santa Rosa Hospital Medical Center, San Antonio, from both groups. The results were largely unchanged. Texas (Thompson); Fred Hutchinson Cancer Research Center, Seattle, Washington (Tangen); Division of Cancer Control and Population Sciences, National Cancer The survival clock started at the time of diagnosis of pros- Institute, National Institutes of Health, Washington, DC (Kramer). tate cancer in both groups. The potential for differences in lead

1440 JAMA Internal Medicine October 2019 Volume 179, Number 10 (Reprinted) jamainternalmedicine.com

Downloaded From: https://jamanetwork.com/ on 09/27/2021 Letters

time would only exist if use of 5-ARIs was associated with sig- with newer biomarkers that are less sensitive to alterations in nificant delays in diagnosis, which the trials suggest may lead to androgen levels, such as urinary PCA3 or ERG fusion assays. poorer outcomes,5 providing external support for our findings. Second, the article that was cited to represent the US Food We agree that the PSA density can be a helpful parameter. and Drug Administration’s argument that 5-ARIs may cause Unfortunately,the prostate volume was not consistently known high-grade prostate cancer is a frequently cited Perspectives for patients in the study and therefore could not be included publication that was not subjected to full peer review, de- in the analysis. Similarly, prostate volume is not typically spite its inclusion of reanalyzed data.2 The US Food and Drug known by primary care physicians when they would need to Administration analysts reproduced results we obtained in the interpret the PSA values. However, PSA density could be used Prostate Cancer Prevention Trial from a binary logistic model by urologists at the time of transrectal ultrasound before de- adjusting for gland volume and number of biopsy cores (sam- ciding to proceed with biopsy. This may lead to fewer unnec- pling density), in which risk of high-grade cancer (Gleason essary biopsies and perhaps less overdiagnosis of clinically score, 7-10) is compared with a reference group with low- insignificant prostate cancers. grade and no cancer combined.3 A more important polyto- mous model in that trial estimated separate relative risks for Reith R. Sarkar, MAS, MD high-grade cancer and low-grade cancer vs no cancer while ad- Brent S. Rose, MD justing for sampling density. This model yielded odds ratios for finasteride vs placebo of 0.47 (95% CI, 0.41-0.54) for low- Author Affiliations: Department of Radiation Medicine and Applied Sciences, grade cancer and 0.88 (95% CI, 0.72-1.09) for high-grade can- University of California, San Diego, La Jolla. cer. Other authors found similar results using alternate meth- Corresponding Author: Brent S. Rose, MD, Department of Radiation Medicine ods for sampling density adjustment.4 and Applied Sciences, University of California, San Diego, 9452 Medical Center Dr, La Jolla, CA 92037 ([email protected]). There is now ample evidence that 5-ARIs, under reason- Conflict of Interest Disclosures: None reported. able PSA interpretation guidelines, make biopsies more effi- cient and therefore reduce risk of low-grade cancers while ac- 1. Sarkar RR, Parsons JK, Bryant AK, et al. Association of treatment with 5α-reductase inhibitors with time to diagnosis and mortality in prostate cancer. celerating detection of cancers that are more likely to require JAMA Intern Med. 2019;179(6):812-819. doi:10.1001/jamainternmed.2019.0280 treatment. Several recent long-term follow-up studies of high- 2. Goodman PJ, Tangen CM, Darke AK, et al. Long-term effects of finasteride on quality cohorts, including the Prostate Cancer Prevention Trial, prostate cancer mortality. N Engl J Med. 2019;380(4):393-394. doi:10.1056/ reported not increases but nonsignificant decreases in lethal NEJMc1809961 prostate cancer associated with 5-ARI use, despite the likely 3. Andriole GL, Crawford ED, Grubb RL III, et al; PLCO Project Team. Mortality effect of diagnostic delay when care is received outside of a results from a randomized prostate-cancer screening trial. N Engl J Med. 2009; 5 360(13):1310-1319. doi:10.1056/NEJMoa0810696 trial protocol. This is not a blanket endorsement of 5-ARIs 4. Schröder FH, Hugosson J, Roobol MJ, et al; ERSPC Investigators. Screening for chemoprevention; however, targeted use of these agents and prostate cancer mortality: results of the European Randomised Study of in carefully selected men at high risk is reasonable after Screening for Prostate Cancer (ERSPC) at 13 years of follow-up. Lancet. 2014; weighing the established risks and benefits, and will become 384(9959):2027-2035. doi:10.1016/S0140-6736(14)60525-0 increasingly reasonable as risk-assessment tools continue 5. Tsodikov A, Gulati R, Heijnsdijk EAM, et al. Reconciling the effects of to improve. screening on prostate cancer mortality in the ERSPC and PLCO trials. Ann Intern Med. 2017;167(7):449-455. doi:10.7326/M16-2586 6. US Food and Drug Administration. FDA drug safety communication: 5-alpha Peter H. Gann, MD, ScD reductase inhibitors (5-ARIs) may increase the risk of a more serious form of prostate cancer. https://www.fda.gov/drugs/drug-safety-and-availability/fda- Author Affiliation: Department of Pathology, College of Medicine, University of drug-safety-communication-5-alpha-reductase-inhibitors-5-aris-may-increase- Illinois at Chicago. risk-more-serious. Updated February 8, 2018. Accessed June 18, 2019. Corresponding Author: Peter H. Gann, MD, ScD, Department of Pathology, College of Medicine, University of Illinois at Chicago, 840 S Wood St, Chicago, IL 60612 ([email protected]). α Benefits of Targeted Use of 5 -Reductase Inhibitors Conflict of Interest Disclosures: None reported. in Patients With Prostate Cancer 1. Sarkar RR, Parsons JK, Bryant AK, et al. Association of treatment with To the Editor I would like to offer 2 points to expand on the im- 5α-reductase inhibitors with time to diagnosis and mortality in prostate cancer. portant analysis by Sarkar and colleagues.1 First, use of a single JAMA Intern Med. 2019;179(6):812-819. doi:10.1001/jamainternmed.2019.0280 prostate-specific antigen (PSA) multiplier for all men using 5α- 2. Theoret MR, Ning YM, Zhang JJ, Justice R, Keegan P, Pazdur R. The risks and reductase inhibitors (5-ARIs) is probably not optimal. The goal benefits of 5α-reductase inhibitors for prostate-cancer prevention. N Engl J Med. 2011;365(2):97-99. doi:10.1056/NEJMp1106783 is to estimate what each man’s PSA would have been in the ab- sence of drug treatment. However, because 5-ARIs inhibit the 3. Cohen YC, Liu KS, Heyden NL, et al. Detection bias due to the effect of finasteride on prostate volume: a modeling approach for analysis of the common age-associated increase in gland volume and PSA, the Prostate Cancer Prevention Trial. J Natl Cancer Inst. 2007;99(18):1366-1374. PSA trajectory for a man being treated with a drug vs his coun- doi:10.1093/jnci/djm130 terfactual trajectory with no drug diverge over time. Further- 4. Redman MW, Tangen CM, Goodman PJ, Lucia MS, Coltman CA Jr, Thompson more, the benign prostatic hyperplasia component of total PSA IM. Finasteride does not increase the risk of high-grade prostate cancer: a bias-adjusted modeling approach. Cancer Prev Res (Phila). 2008;1(3):174-181. would be higher in an older man vs a younger one. Therefore, doi:10.1158/1940-6207.CAPR-08-0092 a larger multiplier may be optimal in older men with longer- 5. Goodman PJ, Tangen CM, Darke AK, et al. Long-term effects of finasteride on term 5-ARI exposure to further avoid delayed diagnosis. Al- prostate cancer mortality. N Engl J Med. 2019;380(4):393-394. doi:10.1056/ ternatively, men taking 5-ARIs could benefit from monitoring NEJMc1809961

jamainternalmedicine.com (Reprinted) JAMA Internal Medicine October 2019 Volume 179, Number 10 1441

Downloaded From: https://jamanetwork.com/ on 09/27/2021 Letters

In Reply We would like to thank Gann for his important per- Prostate Cancer Prevention Trial. J Natl Cancer Inst. 2007;99(18):1366-1374. spective on this topic and for his interest in our study.1 In his doi:10.1093/jnci/djm130 letter, Gann notes an important modeling article that takes the size of the prostate into account when determining the effect State-Level Approaches to Expanding of use of a 5-α reductase inhibitor (5-ARI) on the develop- Pharmacists’ Authority to Dispense Naloxone ment of low-grade and high-grade prostate cancers.2 There is May Affect Accessibility extensive debate in the literature on whether 5-ARIs directly To the Editor We read with great interest the recently pub- increase the risk of high-grade cancers through some biologic lished study by Abouk and colleagues1 that identified re- mechanism. Although this is still an open question, the ma- duced opioid-related mortality in states granting pharma- jority of the data seems to suggest that use of 5-ARIs does not cists direct authority to dispense naloxone. We agree that directly increase this risk. Owing to the inherent limitations empowering pharmacists to dispense naloxone can pro- in observational data, we do not feel that our study meaning- foundly influence overdose risk and concur that the method fully contributes to this question. of implementing such policies can greatly influence their ef- Our study focused specifically on whether the prostate- fectiveness. However, categorizing naloxone access laws (NALs) specific antigen (PSA) suppression from use of 5-ARIs was as- as either providing pharmacists direct or indirect authority to sociated with later diagnoses, more advanced disease, and dispense naloxone is an oversimplification. Based on this meth- worse overall outcomes. It appears that in routine practice, odology, most states were grouped in the indirect category de- 5-ARI–induced PSA suppression does affect early prostate can- spite noteworthy state-to-state heterogeneity in implemen- cer diagnosis. Gann nicely explains some of the factors that can tation. Two recent studies, similarly designed and concurrently affect the degree of PSA suppression from 5-ARI use. He notes published in JAMA,2,3 assessed the influence of Texas’ and that age, prostate volume, and duration of 5-ARI use could California’s NALs on naloxone accessibility from community be included in the interpretation of the PSA. Unfortunately, pharmacies. The surprisingly divergent results highlight the incorporating all of these factors into a decision-making importance of how NALs are implemented. process is difficult within the constraints of a busy clinic. Pri- The Texas NAL provides only indirect authority to phar- mary care physicians will usually not know how large a man’s macists, and the Texas Health and Human Services Commis- prostate is or how much to increase the multiplier for a given sion has not issued a statewide standing order to facilitate age and duration of treatment. Compliance with the medica- access.4 Thus, pharmacists in Texas are unable to autono- tion would also need to be considered. mously dispense naloxone unless they or their pharmacies Because this is effectively the status quo that led to the develop a unique standing order with a prescriber. While findings of our study, we would propose 2 possible solutions. this extra step of developing unique standing orders repre- First, one could build a decision aid into the electronic medi- sents a potential barrier to pharmacy-based naloxone access, cal record that provides an automatic correction for age and robust chain-level implementation, accompanied by internal duration of 5-ARI usage. Second, one could default to the con- pharmacist education, appears to have increased readiness ventional doubling rule in which the PSA is multiplied by 2 to to dispense naloxone. In the Texas study,2 naloxone was get the corrected value. The doubling rule may oversimplify accessible via standing order from 83.7% of pharmacies the decision, but it is probably the most feasible recommen- audited. dation for routine practice. There are certainly other possible In contrast, the California NAL provides pharmacists di- solutions. The main goal of our study was to raise awareness rect authority to dispense naloxone. However, the protocol re- of the issue and to stimulate discussion among the commu- quires pharmacists to complete several time-consuming tasks, nity on the best way to deal with 5-ARI–associated PSA sup- which constitute substantial barriers to participation.5 Phar- pression. Hopefully, new screening tests will allow us to move macists must complete 1 hour of naloxone-centered educa- away from PSA in the future; however, their utility will need tion, and with each dispensing, they must administer a screen- to be demonstrated before they are broadly accepted. ing questionnaire, provide patient education with 19 legally mandated components, and notify the patient’s physician. Reith R. Sarkar, MD, MAS Without encouragement from management for implementa- Brent S. Rose, MD tion of a chain-specific standing order, California pharmacists may be less motivated to begin dispensing naloxone Author Affiliations: Department of Radiation Medicine and Applied Sciences, under the statewide protocol. Thus, despite providing direct University of California, San Diego, La Jolla. authority to dispense naloxone, this method of implementa- Corresponding Author: Brent S. Rose, MD, Department of Radiation Medicine and Applied Sciences, University of California, San Diego, 9452 Medical Center tion appears to limit the influence of California’s NAL be- Dr, La Jolla, CA 92037 ([email protected]). cause naloxone was accessible in only 23.5% of pharmacies Conflict of Interest Disclosures: Dr Sarkar reports receiving personal fees from audited in the California study.3 Boston Consulting Group outside of the submitted work. No other disclosures In conclusion, we agree with Abouk and colleagues’ are reported. assertion1 that how NALs are implemented can substantially 1. Sarkar RR, Parsons JK, Bryant AK, et al. Association of treatment with influence their real-world effectiveness, but we assert that clas- 5α-reductase inhibitors with time to diagnosis and mortality in prostate cancer. JAMA Intern Med. 2019;179(6):812-819. doi:10.1001/jamainternmed.2019.0280 sifying these state-level policies as granting pharmacists 2. Cohen YC, Liu KS, Heyden NL, et al. Detection bias due to the effect of direct or indirect authority is an oversimplification. Further finasteride on prostate volume: a modeling approach for analysis of the research is needed to identify the optimal approach for

1442 JAMA Internal Medicine October 2019 Volume 179, Number 10 (Reprinted) jamainternalmedicine.com

Downloaded From: https://jamanetwork.com/ on 09/27/2021 Letters

empowering pharmacists to reduce opioid-related mortality to strongly predict reductions in overdoses. Future work should by distributing naloxone. continue to disaggregate this relationship further as more states adopt these policies and there is sufficient power to consider Lucas G. Hill, PharmD timing of implementation. Kelly R. Reveles, PharmD, PhD Kirk E. Evoy, PharmD Rahi Abouk, PhD Rosalie Liccardo Pacula, PhD Author Affiliations: College of Pharmacy, The University of Texas at Austin. David Powell, PhD Corresponding Author: Lucas G. Hill, PharmD, College of Pharmacy, The University of Texas at Austin, 2409 University Ave, Stop A1910, PHR 2.222G, Author Affiliations: William Paterson University, Wayne, New Jersey (Abouk); Austin, TX 78712 ([email protected]). RAND Corporation, Arlington, Virginia (Pacula, Powell); National Bureau of Conflict of Interest Disclosures: None reported. Economic Research, Cambridge, Massachusetts (Pacula). 1. Abouk R, Pacula RL, Powell D. Association between state laws facilitating Corresponding Author: Rahi Abouk, PhD, William Paterson University, 300 pharmacy distribution of naloxone and risk of fatal overdose. JAMA Intern Med. Pompton Rd, Wayne, NJ 07410 ([email protected]). 2019;179(6):805-811. doi:10.1001/jamainternmed.2019.0272 Conflict of Interest Disclosures: Dr Pacula reports receiving grants from the 2. Evoy KE, Hill LG, Groff L, Mazin L, Carlson CC, Reveles KR. Naloxone National Institute on Drug Abuse. Dr Powell reports receiving grants from the accessibility without a prescriber encounter under standing orders at Centers for Disease Control and Prevention and the National Institute on Drug community pharmacy chains in Texas. JAMA. 2018;320(18):1934-1937. Abuse. No other disclosures are reported. doi:10.1001/jama.2018.15892 1. Abouk R, Pacula RL, Powell D. Association between state laws facilitating 3. Puzantian T, Gasper JJ. Provision of naloxone without a prescription by pharmacy distribution of naloxone and risk of fatal overdose. JAMA Intern Med. California pharmacists 2 years after legislation implementation. JAMA. 2018; 2019;179(6):805-811. doi:10.1001/jamainternmed.2019.0272 320(18):1933-1934. doi:10.1001/jama.2018.12291 4. S 1462, 84th Leg, 1st Sess (Tx 2015). 5. Cal Reg Code §1746. An Elderly Bias, Nocturia, and Adverse Effects of Sedative-Hypnotic Medication To the Editor We read with great interest the recently pub- In Reply We appreciate the comments by Hill and colleagues lished Special Communication by Soong and colleagues.1 They concerning the way we categorized naloxone access laws persuasively showed why the initiation of sedative-hypnotic (NALs) in our article.1 We acknowledge the oversimplifica- medications should be minimized along with the strategies that tion issue in our investigation of the effect of NALs on opioid can be exploited. However, we raise 2 concerns, which were overdose deaths. The fact is that all NALs are different in some not adequately mentioned in the article. manner, whether it is by statute or by implementation. It would First, the evidence regarding the harmful effects of seda- be paralyzing to study every dimension of every state policy tive-hypnotic drug use and the effectiveness of nonpharma- independently because it would be difficult to determine what cological strategies relied too much on results from studies on common features appear to be most effective. Some type of the elderly population. By sheer numbers alone, 27 of 45 (60%) aggregation is necessary if we are trying to learn about these references in the article1 predominantly focused on the el- laws, what they do in practice, and what works. This trade- derly. Regarding the nonpharmacological interventions, which off likely resonates with many researchers. Studying too many are used to reduce sedative-hypnotic use among inpatients, dimensions at once reduces power and makes it difficult to 10 of 13 (77%) studies were based on elderly patients. This bi- learn about the effects of potential common features across ased content naturally leads to a question of whether the sug- policies. On the other hand, aggregating too much masks the gested evidence can be generalized to all adults. Unfortu- effects of dimensions that matter and dimensions that do not. nately,the limitations resulting from these unbalanced citations Our article was motivated by concerns that the literature has were not mentioned in the Discussion section. In addition, sev- often aggregated too much and that the details of NALs mat- eral studies involving elderly people were cited without being ter. Hill and colleagues have suggested that some other di- mentioned, which may mislead the readers about the gener- mensions not captured in our models may strengthen or di- alizability of the results. For example, Choosing Wisely Canada2 minish the effects that we estimated, which could certainly be recommended against the use of sedative-hypnotic medica- the case. Given our focus on the direct dispensing dimen- tions as a first-line therapy for elderly patients. It is an over- sion, we replicated our analysis while dropping California to generalization to interpret such recommendations as indica- evaluate the implications of Hill and colleagues’ point. We ob- tions to avoid the use of sedative-hypnotic drugs and to served California in its first and second full years of adoption consider them unnecessary in all adults. in the data. Our estimated effect in the published article using Second, the adverse reactions associated with zolpidem all states for the second year postadoption was −0.313. When were not sufficiently mentioned. It has been widely accepted we exclude California, we estimate a nearly identical effect of that zolpidem contributes to visual hallucinations, sleepwalk- −0.309, suggesting that California is not driving the main re- ing, night eating, and sleep driving.3 Whereas falls and frac- sult of the article, but also not attenuating it. tures are the main concerns in the elderly,4 adverse reactions As states continue to pass these policies, it will become to zolpidem can occur in any age group. This may be associ- more appropriate to add more dimensions in these models and ated with the distorted prescription patterns of zolpidem. play close attention to actual implementation on the ground. Zolpidem is generally prescribed at a higher dosage and For now, our work found a shared characteristic that appears for longer periods than recommended.5 Hence, relevant

jamainternalmedicine.com (Reprinted) JAMA Internal Medicine October 2019 Volume 179, Number 10 1443

Downloaded From: https://jamanetwork.com/ on 09/27/2021 Letters

comments on zolpidem-related adverse reactions and unsafe perhaps more attention ought to be focused on causes of the use may motivate health care providers to avoid prescribing latter. I agree with Soong and colleagues1 that focusing more the so-called sleep pills. precisely on why awakenings occur and what happens after In summary, clarifications about the target population and they do may broaden the perspective on nocturnal treat- emphasis on the known adverse reactions associated with ments to promote sleep. A complementary perspective is that zolpidem will improve this short communication. the target organ for such interventions may not always be brain, but rather bladder and/or kidney. Seo-Eun Cho, MD Seung-Gul Kang, MD, PhD Donald L. Bliwise, PhD Kyoung-Sae Na, MD Author Affiliation: Sleep Center, Emory University School of Medicine, Atlanta, Author Affiliations: Gil Medical Center, Gachon University, Incheon, Republic of Georgia. Korea (Cho); College of Medicine, Gachon University, Incheon, Republic of Corresponding Author: Donald L. Bliwise, PhD, Sleep Center, Emory University Korea (Kang, Na). School of Medicine, 12 Executive Park Dr NE, Rm 435, Atlanta, GA 30329 Corresponding Author: Kyoung-Sae Na, MD, College of Medicine, Gachon ([email protected]). University, 21 Namdong-daero, 774 beon-gil, Namdong-gu, Incheon 21565, Conflict of Interest Disclosures: Dr Bliwise reports receiving personal fees Republic of Korea ([email protected]). from Merck, Ferring, Jazz, Eisai, and Respicardia. Conflict of Interest Disclosures: Dr Na reports receiving grants from the 1. Soong C, Burry L, Cho HJ, et al. An implementation guide to promote sleep National Research Foundation of Korea. and reduce sedative-hypnotic initiation for noncritically ill inpatients. JAMA 1. Soong C, Burry L, Cho HJ, et al. An implementation guide to promote sleep Intern Med. 2019;179(7):965-972. doi:10.1001/jamainternmed.2019.1196 and reduce sedative-hypnotic initiation for noncritically ill inpatients. JAMA 2. Monaghan TF, Bliwise DL, Suss NR, et al. Overnight urge perception in Intern Med. 2019;179(7):965-972. doi:10.1001/jamainternmed.2019.1196 nocturia is independent of depression, PTSD, or anxiety in a male Veterans 2. Choosing Wisely Canada. Hospital medicine: five things physicians and Administration population. J Clin Sleep Med. 2019;15(4):615-621. doi:10.5664/ patients should question. https://choosingwiselycanada.org/hospital- jcsm.7726 medicine/. Updated June 2017. Accessed June 7, 2019. 3. Araujo AB, Yaggi HK, Yang M, McVary KT, Fang SC, Bliwise DL. Sleep related 3. Inagaki T, Miyaoka T, Tsuji S, Inami Y, Nishida A, Horiguchi J. Adverse problems and urological symptoms: testing the hypothesis of bidirectionality in reactions to zolpidem: case reports and a review of the literature. Prim Care a longitudinal, population based study. J Urol. 2014;191(1):100-106. doi:10. Companion J Clin Psychiatry. 2010;12(6):PCC.09r00849. doi:10.4088/PCC. 1016/j.juro.2013.07.011 09r00849bro 4. Bliwise DL, Foley DJ, Vitiello MV, Ansari FP, Ancoli-Israel S, Walsh JK. 4. Glass J, Lanctôt KL, Herrmann N, Sproule BA, Busto UE. Sedative hypnotics Nocturia and disturbed sleep in the elderly. Sleep Med. 2009;10(5):540-548. in older people with insomnia: meta-analysis of risks and benefits. BMJ. 2005; doi:10.1016/j.sleep.2008.04.002 331(7526):1169. doi:10.1136/bmj.38623.768588.47 5. Nakagawa H, Niu K, Hozawa A, et al. Impact of nocturia on bone fracture and 5. Moore TJ, Mattison DR. Assessment of patterns of potentially unsafe use of mortality in older individuals: a Japanese longitudinal cohort study. J Urol. 2010; zolpidem. JAMA Intern Med. 2018;178(9):1275-1277. doi:10.1001/ 184(4):1413-1418. doi:10.1016/j.juro.2010.05.093 jamainternmed.2018.3031

In Reply We thank Bliwise for his comments on our article1 and To the Editor In their extraordinarily important article, Soong for highlighting the role of nocturia in disordered sleep. Rou- and colleagues1 emphasize the significance of reducing tinely in hospital settings, diuretics such as furosemide may sedative-hypnotic medication usage in noncritically ill be unnecessarily dosed in the late evening (known as q 12 inpatients. Falls and fractures represent major morbidities hours, or 9 AM and 9 PM), contributing to nocturia, inter- associated with the usage of such medications, and there can rupted sleep, and increasing the risk of falls. Unnecessary con- be little doubt that numerous nonpharmacologic approaches tinuous intravenous maintenance fluids that are not reas- for poor sleep remain vastly underutilized in the inpatient set- sessed may also contribute and are important to address. ting. As the authors note, however, addressing the underly- Additionally, according to Pressman and colleagues,2 awak- ing cause(s) of such sleep disturbance is a necessary first step enings perceived to be stimulated by a need to void are actu- for interventions that promote better sleep. ally caused by undiagnosed obstructive sleep apnea, which Patients arise from bed for many reasons, but voiding ur- suggests that frequent nocturia should prompt solicitation of gency (nocturia),2 either as a cause or an effect of awakening at other symptoms of sleep apnea and subsequent study. Noc- night,3 has been recognized widely as a condition strongly as- turia has been found to be comparable with snoring as a screen- sociated with disordered sleep and occurs perhaps with higher ing tool for sleep apnea.3 Treatment of underlying sleep ap- prevalence in this regard than either pain or even psychiatric nea can reduce the frequency of awakening to void by 50%.4 conditions.4 Urinary urgency at any time, but particularly at We also thank Cho and colleagues for raising the impor- night, has been associated with both falls and fractures, and even tant issues on the limitations of available evidence around the mortality,5 and observational evidence suggests that some of adverse effects of sedative-hypnotic medications and inter- these effects may be mediated by poor sleep quality. ventions to reduce their use. We agree that the majority of cited Nocturia is so mundane and ubiquitous in middle-aged and literature in our article directly involved the elderly popula- elderly populations so as to defy its recognition as a clinically tion. However, we directed our literature search to any adult significant issue. Yet every bed-rise episode represents an population and did not limit to any particular age group. The incremental exposure for a fall. Some sedative-hypnotic findings of our study reflect the available published litera- medications may well contribute to this added risk, but if the ture, some of which was not limited to age. For example, Kolla ultimate reason for leaving the bed is to use the bathroom, and colleagues5 included the entire hospitalized population and

1444 JAMA Internal Medicine October 2019 Volume 179, Number 10 (Reprinted) jamainternalmedicine.com

Downloaded From: https://jamanetwork.com/ on 09/27/2021 Letters

showed the increased risks of falls related to zolpidem re- 6. US Food and Drug Administration. FDA adds Boxed Warning for risk of mained present even after adjusting for age. Similarly, we cited serious injuries caused by sleepwalking with certain prescription insomnia medicines. https://www.fda.gov/drugs/drug-safety-and-availability/fda-adds- several other studies that included adult inpatients of all age boxed-warning-risk-serious-injuries-caused-sleepwalking-certain-prescription- groups. It may be reasonable to state that the adverse events insomnia. Published April 30, 2019. Accessed July 4, 2019. associated with sedative-hypnotic medications, or the interventions suggested in these studies, lack ample evidence to What If Miró’s Dots and Lines Are Not Simply extend to younger populations. However, as a corollary, there Dots and Lines? is also no physiological reason that risks or benefits would re- To the Editor We read with interest the recently published Per- 1 verse in a younger population. Because we are recommend- spective by Taran and Detsky that explored how diagnosing ing minimizing use of a medication that might increase risks disease may be thought of in similar terms as interpreting mod- (and is prescribed without appropriate indications in the first ern art. The authors compared diagnosing with the experi- place), it is unlikely that deprescribing sedative-hypnotic ence of finding paintings’ meaning through the impressions medications will result in greater harm in younger popula- that Catalan painter Joan Miró’s paintings Bleu I, Bleu II, and tions. In fact, given that sedative-hypnotic coingestion is as- Bleu III transmit to the viewer. After different interpreta- sociated with increased lethality of opioid overdoses, it would tions, they suggest the following question: What if the paint- be reasonable to anticipate harm reduction in any efforts and ings were simply dots and lines? 1 curb inappropriate sedative-hypnotic prescribing. We would According to Taran and Detsky, these paintings were com- welcome more evidence and study regarding the younger posed with a technique called automatic drawing (psychic population. automatism), in which Miró allowed his hand to roam freely Regarding the association of zolpidem with complex and on the canvas without knowing how the piece would unfold. sometimes amnestic behaviors, there is sufficient validation of In the 1920s, Miró was among the first artists to develop those adverse effects that a new black-box warning has been automatic drawing: it represented the beginning of Surreal- 2 issued.6 However, the cited article focuses on hypnotics in ism as an art movement. However, Miró always balanced hospitalized patients, and our broad search strategy did not iden- the type of spontaneity and automatism encouraged by the tify publications specifically addressing these complex behav- surrealists with meticulous planning and execution to achieve iors in hospitalized patients. Nonetheless, we agree that cau- finished works that, owing to their precision, seemed repre- tion is needed, carefully weighing risks and benefits prior to sentative despite their considerable level of abstraction. prescribing zolpidem. Additionally, if it must be prescribed, In February 1960, Miró began to think of a series of blue it should be at the minimal effective and approved dose. paintings. He drew tiny sketches in ink, colored pencil, or ballpoint pen—lines, spots, and traces. Then he meditated. Christine Soong, MD, MSc Finally, after a long preparation, he attacked the first paint- Hyung J. Cho, MD ing: “The most recent works are the three large blue can- Timothy Morgenthaler, MD vases, and it took me a long time to do them, not to paint them, but to meditate on them. It took a tremendous effort, a great

Author Affiliations: Division of General Internal Medicine, Sinai Health System, deal of inner tension…. The preliminary stage was intellec- Toronto, Ontario, Canada (Soong); Institute of Health Policy, Management and tual…I began by drawing in charcoal, with a lot of precision Evaluation, Toronto, Ontario, Canada (Soong); Centre for Quality and Patient (I always get to work very early in the morning). In the after- Safety, University of Toronto, Toronto, Ontario, Canada (Soong); Department of noon, I only looked at what I had drawn. All the rest of the day, Medicine, Icahn School of Medicine at Mount Sinai, New York, New York (Cho); Department of Quality, New York Health and Hospitals, New York (Cho); Lown I was preparing myself. Finally, I began to paint: first the bot- Institute, Brookline, Massachusetts (Cho); Division of Pulmonary, Critical Care, tom, all blue; but it was not simply a matter of posing in color, and Sleep Medicine, Mayo Clinic, Rochester, Minnesota (Morgenthaler). like a house painter: all the movements of the brush, those of Corresponding Author: Christine Soong, MD, MSc, Division of General Internal the wrist, the breathing of a hand also intervened. ‘Perfect’ the Medicine, Mount Sinai Hospital, 433-600 University Ave, Toronto, ON M5G 1X5, bottom put me in condition to continue the rest. This fight ex- Canada ([email protected]). hausted me; I have not painted anything since. These paint- Conflict of Interest Disclosures: None reported. ings are the culmination of everything I had tried to do.”3(p3) 1. Soong C, Burry L, Cho HJ, et al. An implementation guide to promote sleep and reduce sedative-hypnotic initiation for noncritically ill inpatients. JAMA Miró completed the triptych on March 4, 1961. Intern Med. 2019;179(7):965-972. doi:10.1001/jamainternmed.2019.1196 After reading this small part of the longest of Miró’s ex- 2. Pressman MR, Figueroa WG, Kendrick-Mohamed J, Greenspon LW, Peterson planations, it is difficult to think that the paintings were sim- DD. Nocturia: a rarely recognized symptom of sleep apnea and other occult ply dots and lines. sleep disorders. Arch Intern Med. 1996;156(5):545-550. doi:10.1001/archinte. 1996.00440050103011 Elena Guardiola, MD, PhD 3. Romero E, Krakow B, Haynes P, Ulibarri V. Nocturia and snoring: predictive symptoms for obstructive sleep apnea. Sleep Breath. 2010;14(4):337-343. Josep-E. Baños, MD, PhD doi:10.1007/s11325-009-0310-2 School of Medicine, Universitat de Vic–Universitat Central 4. Margel D, Shochat T, Getzler O, Livne PM, Pillar G. Continuous positive Author Affiliations: airway pressure reduces nocturia in patients with obstructive sleep apnea. Urology. de Catalunya, Vic, Spain. 2006;67(5):974-977. doi:10.1016/j.urology.2005.11.054 Corresponding Author: Josep-E. Baños, MD, PhD, School of Medicine, Universitat de Vic–Universitat Central de Catalunya, Casa de Convalescència Dr 5. Kolla BP, Lovely JK, Mansukhani MP, Morgenthaler TI. Zolpidem is independently associated with increased risk of inpatient falls. JHospMed. Junyent 1, 08500 Vic, Spain ([email protected]). 2013;8(1):1-6. doi:10.1002/jhm.1985 Conflict of Interest Disclosures: None reported.

jamainternalmedicine.com (Reprinted) JAMA Internal Medicine October 2019 Volume 179, Number 10 1445

Downloaded From: https://jamanetwork.com/ on 09/27/2021 Letters

1. Taran S, Detsky AS. Miro’s dots and lines. JAMA Intern Med. 2019;179(8):1019- The authors should look more closely at these pairs of studies 1020. doi:10.1001/jamainternmed.2019.1922 to determine whether they used the same study population and 2. Soby JT. Joan Miró. New York, NY: The Museum of Modern Art; 1959. then select the ones with the best quality or the largest sample 3. Rowell M. Joan Miró, les Trois Bleus. Paris, France: Centre Georges Pompidou; size for analysis to ensure the accuracy of their analysis. 1994.

Mingyang Jiang, MD In Reply We thank Guardiola and Baños for their comments on Yunni Lin our article1 and for showing us the quote attributed to Joan Shanggui Su, MD Miró. It helps explain the process he used to create the triptych. But it still does not explain what the paintings mean, at Author Affiliations: Basic Medical College, Department of Biochemistry and least not to the artist. As we say in our essay, that’s part of the Molecular Biology, Guangxi Medical University, Nanning, Guangxi, China. beauty of art: interpretation is subjective and cannot be Corresponding Author: Shanggui Su, MD, Basic Medical College, Department of Biochemistry and Molecular Biology, Guangxi Medical University, imposed on others. Shuangyong 22 St, Nanning 530021, Guangxi, China ([email protected]). Conflict of Interest Disclosures: None reported. Shaurya Taran, MD Funding/Support: The work for this letter was supported by Guangxi Medical Allan S. Detsky, MD, PhD University. Role of the Funder/Sponsor: Guangxi Medical University had no role in the Author Affiliations: Department of Medicine, University of Toronto, Toronto, design and conduct of the study; collection, management, analysis, and Ontario, Canada (Taran); Department of Medicine, Institute for Health Policy, interpretation of the data; preparation, review, or approval of the manuscript; Management, and Evaluation, University of Toronto, Toronto, Ontario, Canada and decision to submit the manuscript for publication. (Detsky); Department of Medicine, Mount Sinai Hospital, University Health Network, Toronto, Ontario, Canada (Detsky). 1. Smyth B, Haber A, Trongtrakul K, et al. Representativeness of randomized clinical trial cohorts in end-stage kidney disease: a meta-analysis [published Corresponding Author: Allan S. Detsky, MD, PhD, Mount Sinai Hospital, 600 online July 8, 2019]. JAMA Intern Med. University Ave, Rm 429, Toronto, ON M5G 1X5, Canada (allan.detsky@ sinaihealthsystem.ca). 2. Martin KJ. The first human cell line-derived erythropoietin, epoetin-delta (Dynepo), in the management of anemia in patients with chronic kidney Conflict of Interest Disclosures: None reported. disease. Clin Nephrol. 2007;68(1):26-31. doi:10.5414/CNP68026 1. Taran S, Detsky AS. Miro’s dots and lines. JAMA Intern Med. 2019;179(8):1019- 3. Martin KJ; Epoetin Delta 3001 Study Group. Epoetin delta in the 1020. doi:10.1001/jamainternmed.2019.1922 management of renal anaemia: results of a 6-month study. Nephrol Dial Transplant. 2007;22(10):3052-3054. doi:10.1093/ndt/gfm490 4. Chen N, Wu X, Ding X, et al. Sevelamer carbonate lowers serum phosphorus Studies Making Use of the Same Randomized effectively in haemodialysis patients: a randomized, double-blind, Clinical Trial Cohorts placebo-controlled, dose-titration study. Nephrol Dial Transplant. 2014;29(1): To the Editor We read the article by Smyth et al1 with interest. 152-160. doi:10.1093/ndt/gft232 The authors performed a large meta-analysis to determine 5. Chen JB, Chiang SS, Chen HC, et al. Efficacy and safety of SBR759, a novel whether participants in large, multicenter dialysis trials were calcium-free, iron(III)-based phosphate binder, in Asian patients undergoing hemodialysis: a 12-week, randomized, open-label, dose-titration study versus similar to the general population undergoing dialysis in terms sevelamer hydrochloride. Nephrology (Carlton). 2011;16(8):743-750. doi:10.1111/ of age, comorbidities, and mortality rate. They concluded that j.1440-1797.2011.01509.x participants in large, multicenter randomized clinical trials of 6. Kauric-Klein Z, Peters RM, Yarandi HN. Self-efficacy and blood pressure patients with end-stage kidney disease undergoing dialysis are self-care behaviors in patients on chronic hemodialysis. West J Nurs Res.2017; 39(7):886-905. doi:10.1177/0193945916661322 younger, have a different pattern of comorbidities, and have a lower mortality rate than the general population of patients 7. Kauric-Klein Z. Improving blood pressure control in end stage renal disease through a supportive educative nursing intervention. Nephrol Nurs J. 2012;39 undergoing dialysis. This finding has implications for the gen- (3):217-228. eralization of trial results to the broader patient population and 8. Locatelli F, Dimkovic N, Spasovski G. Evaluation of colestilan in chronic for future trial design. However, after carefully reading the kidney disease dialysis patients with hyperphosphataemia and dyslipidaemia: article, we are concerned that studies making use of the same a randomized, placebo-controlled, multiple fixed-dose trial. Nephrol Dial Transplant. 2013;28(7):1874-1888. doi:10.1093/ndt/gft064 cohort of patients were included multiple times in the analysis. There were several studies described in eTable 1 in the 9. Locatelli F, Spasovski G, Dimkovic N, Wanner C, Dellanna F, Pontoriero G. The effects of colestilan versus placebo and sevelamer in patients with CKD 5D and Supplement that look quite likely to be making use of the same hyperphosphataemia: a 1-year prospective randomized study. Nephrol Dial cohort of patients. Transplant. 2014;29(5):1061-1073. doi:10.1093/ndt/gft476 For instance, the 2 studies numbered 120 and 1212,3 in 10. Moustafa M, Lehrner L, Al-Saghir F, et al. A randomized, double-blind, eTable 1 of the Supplement were conducted by the same group placebo-controlled, dose-ranging study using Genz-644470 and sevelamer carbonate in hyperphosphatemic chronic kidney disease patients on of authors, and both of these studies had the same number of hemodialysis. Int J Nephrol Renovasc Dis. 2014;7:141-152. doi:10.2147/IJNRD.S56217 participants from the same hospitals, with similar baseline char- 11. Moustafa M, Aronoff GR, Chandran C, et al. Phase IIa study of the acteristics. Hence, we suspect that these studies made use of immunogenicity and safety of the novel Staphylococcus aureus vaccine V710 in the same cohorts, and thus one of these studies should have adults with end-stage renal disease receiving hemodialysis. Clin Vaccine Immunol. been removed as a duplicate of the other. Other pairs of stud- 2012;19(9):1509-1516. doi:10.1128/CVI.00034-12 ies in eTable 1 of the Supplement that deserve further scrutiny 12. Shigematsu T; Lanthanum Carbonate Group. Multicenter prospective randomized, double-blind comparative study between lanthanum carbonate include the pairs numbered 27 and 28,4,5 the pairs numbered and calcium carbonate as phosphate binders in Japanese hemodialysis patients 6,7 8,9 78 and 80, the pairs numbered 111 and 112, the pairs num- with hyperphosphatemia. Clin Nephrol. 2008;70(5):404-410. doi:10.5414/ bered 130 and 131,10,11 and the pairs numbered 150 and 151.12,13 cnp70404

1446 JAMA Internal Medicine October 2019 Volume 179, Number 10 (Reprinted) jamainternalmedicine.com

Downloaded From: https://jamanetwork.com/ on 09/27/2021 Letters

13. Shigematsu T; Lanthanum Carbonate Group. Lanthanum carbonate abstract, the Results and Discussion sections, Tables 1, 2, 3, and effectively controls serum phosphate without affecting serum calcium levels in 4, the Figure, and the Supplement. patients undergoing hemodialysis. Ther Apher Dial. 2008;12(1):55-61. doi:10. 1111/j.1744-9987.2007.00541.x Brendan Smyth, MBBS

In Reply On behalf of my coauthors, I write to report errors in Author Affiliation: The George Institute for Global Health, University of New South Wales, Newtown, New South Wales, Australia. our article, “Representativeness of Randomized Clinical Trial Corresponding Author: Brendan Smyth, MBBS, The George Institute for Global Cohorts in End-stage Kidney Disease: A Meta-analysis,”1 that Health, University of New South Wales, 1 King St, Newtown, Australia 2042 was published online in July 8, 2019, in JAMA Internal ([email protected]). Medicine. In that article, we reported on a meta-analysis of 189 Conflict of Interest Disclosures: Dr Smyth reported being supported by an randomized clinical trials (RCTs) comprising 80 104 partici- Australian Government Research Training Program scholarship from the pants to determine whether participants in large, multi- University of Sydney during the conduct of the study and receiving nonfinancial support from Roche International outside the submitted work. center dialysis trials were similar to the general population 1. Smyth B, Haber A, Trongtrakul K, et al. Representativeness of randomized undergoing dialysis in terms of age, comorbidities, and mor- clinical trial cohorts in end-stage kidney disease: a meta-analysis [published tality rate. We found that “participants in large, multicenter online July 8, 2019]. JAMA Intern Med. RCTs of patients with end-stage kidney disease undergoing 2. Jiang M, Lin Y, Su S. Studies making use of the same randomized clinical trial dialysis are younger, have a different pattern of comorbidi- cohorts. JAMA Intern Med. doi:10.1001/jamainternmed.2019.4621 ties, and have a lower mortality rate than the general popula- 3. Kauric-Klein Z, Peters RM, Yarandi HN. Self-efficacy and blood pressure tion of patients undergoing dialysis.” self-care behaviors in patients on chronic hemodialysis. West J Nurs Res.2017; 39(7):886-905. doi:10.1177/0193945916661322 I thank Jiang et al2 for bringing to our attention their concerns about the inclusion of some of the trial cohorts in our 4. Kauric-Klein Z. Improving blood pressure control in end stage renal disease through a supportive educative nursing intervention. Nephrol Nurs J. 2012;39 meta-analysis and the possibility that some of these are (3):217-228. 3-15 duplicates. I have reviewed all of the articles in question. 5. Kauric-Klein Z. Blood pressure knowledge in hypertensive hemodialysis Of the 6 identified pairs, 4 are clearly distinct cohorts.7-15 How- patients. CANNT J. 2012;22(4):18-25. ever, 2 pairs represent duplicated cohorts, of which 1 cohort 6. Martin KJ; Epoetin Delta 3001 Study Group. Epoetin delta in the was published 3 times.3-7 management of renal anaemia: results of a 6-month study. Nephrol Dial Transplant. The cohorts reported in the 3 articles by Kauric-Klein et al 2007;22(10):3052-3054. doi:10.1093/ndt/gfm490 that were published in 20124,5 and 20173 appear to be the same. 7. Martin KJ. The first human cell line-derived erythropoietin, epoetin-delta 3 4,5 (Dynepo), in the management of anemia in patients with chronic kidney The later article does not reference the former articles and disease. Clin Nephrol. 2007;68(1):26-31. doi:10.5414/CNP68026 makes no mention of the cohort being published previously. 8. Chen N, Wu X, Ding X, et al. Sevelamer carbonate lowers serum phosphorus 6,7 The cohorts in the 2 articles published by Martin et al effectively in haemodialysis patients: a randomized, double-blind, also appear to be the same. One assesses treatment for 24 placebo-controlled, dose-titration study. Nephrol Dial Transplant. 2014;29(1): weeks,6 and the other reports on an additional 28 weeks of data 152-160. doi:10.1093/ndt/gft232 (including the participants in the first trial),7 but neither 9. Chen JB, Chiang SS, Chen HC, et al. Efficacy and safety of SBR759, a novel calcium-free, iron(III)-based phosphate binder, in Asian patients undergoing article references or mentions the publication of the same hemodialysis: A 12-week, randomized, open-label, dose-titration study versus data elsewhere. sevelamer hydrochloride. Nephrology (Carlton). 2011;16(8):743-750. doi:10.1111/ In our systematic review, we identified 5229 records; j.1440-1797.2011.01509.x after removal of duplicates by screening titles and abstracts, 10. Locatelli F, Dimkovic N, Spasovski G. Evaluation of colestilan in chronic we reviewed 545 full-text articles. Of these articles, 356 were kidney disease dialysis patients with hyperphosphataemia and dyslipidaemia: a randomized, placebo-controlled, multiple fixed-dose trial. Nephrol Dial excluded, leaving 189 trials for our meta-analysis. Following Transplant. 2013;28(7):1874-1888. doi:10.1093/ndt/gft064 our protocol, I reviewed all of the trials with 2 of my coau- 11. Locatelli F, Spasovski G, Dimkovic N, Wanner C, Dellanna F, Pontoriero G. thors (A.H. and K.T.) to identify and exclude reports of trials The effects of colestilan versus placebo and sevelamer in patients with CKD 5D that were secondary or accessory publications from the same and hyperphosphataemia: a 1-year prospective randomized study. Nephrol Dial Transplant. 2014;29(5):1061-1073. doi:10.1093/ndt/gft476 cohort. While we were effective in excluding articles that acknowledged the publication of data from the same cohort 12. Moustafa M, Lehrner L, Al-Saghir F, et al. A randomized, double-blind, placebo-controlled, dose-ranging study using Genz-644470 and sevelamer elsewhere, among the many articles reviewed, we failed to note carbonate in hyperphosphatemic chronic kidney disease patients on the similarities between these articles. hemodialysis. Int J Nephrol Renovasc Dis. 2014;7:141-152. doi:10.2147/IJNRD.S56217 Thus, I have reanalyzed all the data based on 186 trials and 13. Moustafa M, Aronoff GR, Chandran C, et al. Phase IIa study of the 79 104 participants—without the duplicated data4-6—and found immunogenicity and safety of the novel Staphylococcus aureus vaccine V710 in adults with end-stage renal disease receiving hemodialysis. Clin Vaccine Immunol. that the overall age and mortality rate remain the same; there 2012;19(9):1509-1516. doi:10.1128/CVI.00034-12 was a difference of 0.1 years in mean age when analysis was 14. Shigematsu T; Lanthanum Carbonate Group. Multicenter prospective restricted to US-only studies. The findings of the new analy- randomized, double-blind comparative study between lanthanum carbonate sis without these studies do not affect the overall results, con- and calcium carbonate as phosphate binders in Japanese hemodialysis patients clusions, or interpretations of our meta-analysis. I confirm that with hyperphosphatemia. Clin Nephrol. 2008;70(5):404-410. doi:10.5414/ CNP70404 there are no other errors that need attention and apologize to 15. Shigematsu T; Lanthanum Carbonate Research Group. Lanthanum the readers and editors of JAMA Internal Medicine for any con- carbonate effectively controls serum phosphate without affecting serum fusion. I have requested that the article be corrected online. calcium levels in patients undergoing hemodialysis. Ther Apher Dial. 2008;12(1): The corrected analysis affects information in the original 55-61. doi:10.1111/j.1744-9987.2007.00541.x

jamainternalmedicine.com (Reprinted) JAMA Internal Medicine October 2019 Volume 179, Number 10 1447

Downloaded From: https://jamanetwork.com/ on 09/27/2021 Letters

CORRECTION Error in Dates of Data Analysis in the Methods: The Research Letter titled,

“Prevalence of Inappropriateness of Parenteral Vitamin B12 Administration Pervasive Errors Due to Duplicate Trial Cohorts: With regard to the Original in Ontario, Canada,”1 published online July 15, 2019, contained an error in Investigation “Representativeness of Randomized Clinical Trial Cohorts in the Methods section regarding the dates of data analysis. The dates of data 1 End-stage Kidney Disease: A Meta-analysis,”by Smyth et al, published online July analysis were written to be from July 26, 2019, to November 22, 2019. The 2 8, 2019, Jiang et al raised concerns about the inclusion of some of the trial dates should be from July 26, 2018, to November 22, 2018. The dates have cohorts in the meta-analysis and the possibility that some of these cohorts might been corrected. be duplicates. Of the 6 pairs of cohorts identified by Jiang et al,2 4 are clearly 1. Silverstein WK, Lin Y, Dharma C, Croxford R, Earle CC, Cheung MC. Prevalence distinct cohorts. However, 2 pairs represent duplicated cohorts, of which 1 cohort of inappropriateness of parenteral vitamin B administration in Ontario, Canada was published 3 times. The authors of the meta-analysis reanalyzed all the data 12 [published online July 15, 2019]. JAMA Intern Med. 2019. doi:10.1001/ based on 186 trials and 79 104 participants (ie, without the duplicated data) and jamainternmed.2019.1859 found that the overall age, mortality rate, and study conclusions remain the same.3 The corrected analysis affects information in the original abstract, the Results and Discussion sections, Tables 1, 2, 3, and 4, the Figure, and the Supplement. This Incomplete Title: The Original Investigation by Budathoki et al1 published article was corrected online. online August 26, 2019, had an incomplete title. The correct full title is 1. Smyth B, Haber A, Trongtrakul K, et al. Representativeness of randomized “Association of Animal and Plant Protein Intake With All-Cause and clinical trial cohorts in end-stage kidney disease: a meta-analysis [published Cause-Specific Mortality in a Japanese Cohort.” This article was corrected online July 8, 2019]. JAMA Intern Med. online. 2. Jiang M, Lin Y, Su S. Studies making use of the same randomized clinical trial 1. Budathoki S, Sawada N, Iwasaki M, et al. Association of animal and plant cohorts. JAMA Intern Med. doi:10.1001/jamainternmed.2019.4621 protein intake with all-cause and cause-specific mortality in a Japanese cohort 3. Smyth B. Studies making use of the same randomized clinical trial [published online August 26, 2019]. JAMA Intern Med. doi:10.1001/ cohorts—reply. JAMA Intern Med. doi:10.1001/jamainternmed.2019.4624 jamainternmed.2019.2806

1448 JAMA Internal Medicine October 2019 Volume 179, Number 10 (Reprinted) jamainternalmedicine.com

Downloaded From: https://jamanetwork.com/ on 09/27/2021