790
LETTERS TO THE EDITOR
colleagues1 appropriately focused on an important investigation of cognitive decline, specifically, testing all particular cognitive domains rather than the extreme of dementia. Steenland’s article concludes that elderly adults with normal cognition at baseline who used statins had a slower rate of annual worsening on the Clinical Dementia Rating Scale Sum of Boxes (CDR-SOB) than nonusers. Arguably, this research illustrates that the rate of decline in subtle cognitive domains may be clinically relevant in monitoring statin response. The authors studied individuals with an average age of 73, the majority of whom were white women, although community-dwelling older adults have various ethnic backgrounds, especially in urban settings. The authors argue that the subjects were matched for underlying conditions, but it would be important to note what other comorbid conditions existed (e.g., chronic obstructive pulmonary disease, congestive heart failure, myocardial infarction, baseline lipid profiles, ratios of high- to low-density lipoprotein cholesterol to correlate dyslipidemia status in the two cohorts, how the authors defined heart disease, and whether participants were taking aspirin or other antiplatelet medications that could have confounded the results). The article did not mention about loss to follow-up and possible explanations for such. It also did not discuss any additional reasons for new cognitive impairment throughout the study (e.g., development of stroke). As emerging evidence connects functional decline with early cognitive decline,2 it would be helpful to have the mean or median baseline CDR-SOB score available in Table 1 to help readers appreciate the changes mentioned in Table 2. Given the unreliable nature of self-reported information, the question arises as to how the authors verified that the subjects were taking their statins. Were these “professional patients,” who would be much more cognitively and physically independent than the average older adult? If so, could the proposed relationship be generalized to the older population as a whole? A noteworthy problem is the presence of recall bias as a potential threat to internal validity given that the outcome of the study is a measure of cognitive decline, which in itself could contribute to recall bias. The spread between different statin medications appears to be consistent with the general participant population. Discussion regarding lipophilicity of different statins would have been of interest, because statins that are lipophilic are thought to provide more of a benefit in Alzheimer’s disease.3 Additionally, a vital question is left unanswered—how long have individuals in the subset been treated with statins, and what is the anticipated time to benefit? Given the small degree of correlation between cognitive decline and statin use, a concern arises over underlying confounders that may account for the outcomes seen in statin users who had normal cognition at baseline. Rather than highlight an association, the article could have elaborated on the clinical significance, or lack thereof, of the small relational differences between the two cognitive groups. The short follow-up of 2.8 and 3.4 years, respectively, may reflect an insufficient time to detect major differences. Future research should be aimed at establishing a long-term correlation between statin use and
APRIL 2014–VOL. 62, NO. 4
JAGS
cognitive decline, with potential implications for clinical practice. Steven Buslovich, MD, MSHCPM Brookdale Department of Geriatrics and Palliative Medicine Icahn School of Medicine at Mount Sinai, New York, New York, Rebecca Salbu Touro College of Pharmacy, Pharmacy and Health Outcomes, New York, New York Ravishankar Ramaswamy, MD, MS Brookdale Department of Geriatrics and Palliative Medicine Icahn School of Medicine at Mount Sinai, New York, New York,
ACKNOWLEDGMENTS Conflict of Interest: The editor in chief has reviewed the conflict of interest checklist provided by the authors and has determined that the authors have no financial or any other kind of personal conflicts with this paper. Author Contributions: SB: major contributor. RR: body and editing contributor. RS: pharmaceutical perspective contribution. Sponsor’s Role: None.
REFERENCES 1. Steenland K, Zhao L, Goldstein FC et al. Statins and cognitive decline in older adults with normal cognition or mild cognitive impairment. J Am Geriatr Soc 2013;61:1449–1455. 2. Rosso AL, Studenski SA, Chen WG et al. Aging, the central nervous system, and mobility. J Gerontol A Biol Sci Med Sci 2013;68A:1379–1386. 3. Shepardson NE, Shankar GM, Selkoe DJ. Cholesterol level and statin use in Alzheimer disease: I. Review of epidemiological and preclinical studies. Arch Neurol 2011;68:1239–1244.
RESPONSE TO BUSLOVICH AND COLLEAGUES To the Editor: We thank Buslovich and colleagues for their letter1 regarding our article on statins.2 Many of the questions that they raise refer to problems inherent to observational studies (vs randomized trials). Several trials of statins have been conducted that have considered cognition as a secondary endpoint but have had limitations, including restriction to individuals with cardiovascular disease, no separation between individuals with normal cognition and those with mild cognitive impairment at baseline, and lack of detailed longitudinal cognitive data. The current study took advantage of an observational data set with detailed longitudinal cognitive data and found protective effects of statins against cognitive decline for those with baseline normal cognition. As in any observational study, the findings could have resulted from uncontrolled confounding, but a number of possible confounders, including demographic characteristics, vascular comorbidities, and measured blood pressure were adjusted for. As Buslovich and colleagues noted, the comorbidity data were self-reported, a
JAGS
APRIL 2014–VOL. 62, NO. 4
limitation. Lipid medication data were also self-reported, but restriction of the study to those who continuously or never took statins should have reduced misclassification, and reporting errors are less likely in individuals with normal cognition. Buslovich and colleagues also commented on several other things. The analyses were not conducted according to subtype of statins largely because there were insufficient numbers in different categories for analyses with good statistical power. Data on aspirin use were not included in analyses, because these were considered less accurate than data on prescription medications, and aspirin was not a priori considered to be a likely confounder. Other data, such as ratios of high-density lipoprotein cholesterol to low-density lipoprotein cholesterol, were not available. As shown in Table 1 and Figure 2, most people with normal cognition had a Clinical Dementia Rating sum of boxes score of 0 at baseline. Buslovich and colleagues comment that we did not know the reasons for loss to follow-up, a limitation. Indeed, given that the cohort was a dynamic one with new people entering over time, it is difficult to determine the difference between loss to follow-up and simply missing a visit. We agree with Buslovich and colleagues that the protective effects of statins were modest and that the population from Alzheimer’s disease centers was a convenience sample, limiting the generalizability of the findings. We also agree that other important limitations were lack of data on lifetime statin use and short follow-up, but given the recent Food and Drug Administration decision calling attention to instances of memory loss in statin users, we believed it important to note that, in a large population of elderly adults with normal cognition, statin use was modestly protective against cognitive decline. Kyle Steenland, PhD Liping Zhao, MS Department of Environmental Health, Rollins School of Public Health, Emory University, Atlanta, Georgia Felicia C. Goldstein, PhD Allan I. Levey, MD, PhD Department of Neurology, School of Medicine, Emory University, Atlanta, Georgia
ACKNOWLEDGMENTS Conflict of Interest: The editor in chief has reviewed the conflict of interest checklist provided by the authors and has determined that the authors have no financial or any other kind of personal conflicts with this paper. Author Contributions: All authors contributed to this paper. Sponsor’s Role: None.
REFERENCES 1. Buslovich S, Salbu R, Ramaswamy R. Statins and the trajectory of cognitive decline. J Am Geriatr Soc 2014;62:789–790. 2. Steenland K, Zhao L, Goldstein FC et al. Statins and cognitive decline in older adults with normal cognition or mild cognitive impairment. J Am Geriatr Soc 2013;61:1449–1455
LETTERS TO THE EDITOR
791
DRUG BURDEN INDEX FOR INTERNATIONAL ASSESSMENT OF THE FUNCTIONAL BURDEN OF MEDICATIONS IN OLDER PEOPLE To the Editor: We thank Faure and colleagues for their interest in use of the Drug Burden Index (DBI). We share their desire to generalize this tool to allow it to be used more easily internationally. We wish to discuss their approach as outlined in the research letter, “A Standard International Version of the Drug Burden Index for CrossNational Comparison of the Functional Burden of Medications in Older People,”1 and to consider the implications not only for the proposed utility of comparing DBI between countries with existing formularies, but also for calculation of DBI in countries with limited national formularies, in which the tools to readily calculate a national DBI are not available. For countries with well-developed drug regulatory authorities and an advanced drug formulary structure, DBI as originally defined2 can be calculated from the formulary to measure exposure to medications with clinical anticholinergic or sedative effects. DBI has been calculated and validated against a range of clinical outcomes in different continents, countries, and settings, including the United States,2 Australia,3 the United Kingdom,4 and Finland.5 The DBI uses d, the minimum registered or licensed dose on the national formulary, as an estimate of the dose required to provide 50% of the maximal effect (DR50). This is based on the pharmacological principle that a registered dose must have some efficacy and that the minimum dose is likely to have less than maximal efficacy. The actual DR50 for most therapeutic drugs in older adults is not known. There are some differences between countries in the minimum registered or licensed doses of medicines used in the calculations, which may be due to the effect of ethnicity on drug response6 or to other issues influencing the regulatory decisions. For example, diazepam has a minimum registered or licensed dose of 4 mg in the United States and 5 mg in Australia. In studies comparing DBI exposure between countries with well-developed national formularies, we agree that a consistent estimate of DR50 is required when differences between countries in minimum registered or licensed dose are unlikely to be related to the effect of ethnicity on pharmacological exposure and response. Using the median minimum registered or licensed dose for the countries studied should give a better estimate of DR50 than the World Health Organization (WHO) Defined Daily Dose (DDD)7 proposed in DBI-WHO1 for the reasons outlined below. For countries in which the WHO list of essential medications and WHO model formulary are used as primary resources to support and improve medication prescribing, we believe that calculation of DBI may best be performed using the WHO model formulary recommended adult starting dosage.8 Review of the WHO model formulary adult starting doses indicates that they rather closely reflect the minimum registered or licensed doses in the countries in which DBI has been validated against functional status and physical performance. This dose is, in most cases, substantially lower than the DDD7 substituted for d in calculation of DBI-WHO.1 DDD is the average maintenance
LETTERS TO THE EDITOR
colleagues1 appropriately focused on an important investigation of cognitive decline, specifically, testing all particular cognitive domains rather than the extreme of dementia. Steenland’s article concludes that elderly adults with normal cognition at baseline who used statins had a slower rate of annual worsening on the Clinical Dementia Rating Scale Sum of Boxes (CDR-SOB) than nonusers. Arguably, this research illustrates that the rate of decline in subtle cognitive domains may be clinically relevant in monitoring statin response. The authors studied individuals with an average age of 73, the majority of whom were white women, although community-dwelling older adults have various ethnic backgrounds, especially in urban settings. The authors argue that the subjects were matched for underlying conditions, but it would be important to note what other comorbid conditions existed (e.g., chronic obstructive pulmonary disease, congestive heart failure, myocardial infarction, baseline lipid profiles, ratios of high- to low-density lipoprotein cholesterol to correlate dyslipidemia status in the two cohorts, how the authors defined heart disease, and whether participants were taking aspirin or other antiplatelet medications that could have confounded the results). The article did not mention about loss to follow-up and possible explanations for such. It also did not discuss any additional reasons for new cognitive impairment throughout the study (e.g., development of stroke). As emerging evidence connects functional decline with early cognitive decline,2 it would be helpful to have the mean or median baseline CDR-SOB score available in Table 1 to help readers appreciate the changes mentioned in Table 2. Given the unreliable nature of self-reported information, the question arises as to how the authors verified that the subjects were taking their statins. Were these “professional patients,” who would be much more cognitively and physically independent than the average older adult? If so, could the proposed relationship be generalized to the older population as a whole? A noteworthy problem is the presence of recall bias as a potential threat to internal validity given that the outcome of the study is a measure of cognitive decline, which in itself could contribute to recall bias. The spread between different statin medications appears to be consistent with the general participant population. Discussion regarding lipophilicity of different statins would have been of interest, because statins that are lipophilic are thought to provide more of a benefit in Alzheimer’s disease.3 Additionally, a vital question is left unanswered—how long have individuals in the subset been treated with statins, and what is the anticipated time to benefit? Given the small degree of correlation between cognitive decline and statin use, a concern arises over underlying confounders that may account for the outcomes seen in statin users who had normal cognition at baseline. Rather than highlight an association, the article could have elaborated on the clinical significance, or lack thereof, of the small relational differences between the two cognitive groups. The short follow-up of 2.8 and 3.4 years, respectively, may reflect an insufficient time to detect major differences. Future research should be aimed at establishing a long-term correlation between statin use and
APRIL 2014–VOL. 62, NO. 4
JAGS
cognitive decline, with potential implications for clinical practice. Steven Buslovich, MD, MSHCPM Brookdale Department of Geriatrics and Palliative Medicine Icahn School of Medicine at Mount Sinai, New York, New York, Rebecca Salbu Touro College of Pharmacy, Pharmacy and Health Outcomes, New York, New York Ravishankar Ramaswamy, MD, MS Brookdale Department of Geriatrics and Palliative Medicine Icahn School of Medicine at Mount Sinai, New York, New York,
ACKNOWLEDGMENTS Conflict of Interest: The editor in chief has reviewed the conflict of interest checklist provided by the authors and has determined that the authors have no financial or any other kind of personal conflicts with this paper. Author Contributions: SB: major contributor. RR: body and editing contributor. RS: pharmaceutical perspective contribution. Sponsor’s Role: None.
REFERENCES 1. Steenland K, Zhao L, Goldstein FC et al. Statins and cognitive decline in older adults with normal cognition or mild cognitive impairment. J Am Geriatr Soc 2013;61:1449–1455. 2. Rosso AL, Studenski SA, Chen WG et al. Aging, the central nervous system, and mobility. J Gerontol A Biol Sci Med Sci 2013;68A:1379–1386. 3. Shepardson NE, Shankar GM, Selkoe DJ. Cholesterol level and statin use in Alzheimer disease: I. Review of epidemiological and preclinical studies. Arch Neurol 2011;68:1239–1244.
RESPONSE TO BUSLOVICH AND COLLEAGUES To the Editor: We thank Buslovich and colleagues for their letter1 regarding our article on statins.2 Many of the questions that they raise refer to problems inherent to observational studies (vs randomized trials). Several trials of statins have been conducted that have considered cognition as a secondary endpoint but have had limitations, including restriction to individuals with cardiovascular disease, no separation between individuals with normal cognition and those with mild cognitive impairment at baseline, and lack of detailed longitudinal cognitive data. The current study took advantage of an observational data set with detailed longitudinal cognitive data and found protective effects of statins against cognitive decline for those with baseline normal cognition. As in any observational study, the findings could have resulted from uncontrolled confounding, but a number of possible confounders, including demographic characteristics, vascular comorbidities, and measured blood pressure were adjusted for. As Buslovich and colleagues noted, the comorbidity data were self-reported, a
JAGS
APRIL 2014–VOL. 62, NO. 4
limitation. Lipid medication data were also self-reported, but restriction of the study to those who continuously or never took statins should have reduced misclassification, and reporting errors are less likely in individuals with normal cognition. Buslovich and colleagues also commented on several other things. The analyses were not conducted according to subtype of statins largely because there were insufficient numbers in different categories for analyses with good statistical power. Data on aspirin use were not included in analyses, because these were considered less accurate than data on prescription medications, and aspirin was not a priori considered to be a likely confounder. Other data, such as ratios of high-density lipoprotein cholesterol to low-density lipoprotein cholesterol, were not available. As shown in Table 1 and Figure 2, most people with normal cognition had a Clinical Dementia Rating sum of boxes score of 0 at baseline. Buslovich and colleagues comment that we did not know the reasons for loss to follow-up, a limitation. Indeed, given that the cohort was a dynamic one with new people entering over time, it is difficult to determine the difference between loss to follow-up and simply missing a visit. We agree with Buslovich and colleagues that the protective effects of statins were modest and that the population from Alzheimer’s disease centers was a convenience sample, limiting the generalizability of the findings. We also agree that other important limitations were lack of data on lifetime statin use and short follow-up, but given the recent Food and Drug Administration decision calling attention to instances of memory loss in statin users, we believed it important to note that, in a large population of elderly adults with normal cognition, statin use was modestly protective against cognitive decline. Kyle Steenland, PhD Liping Zhao, MS Department of Environmental Health, Rollins School of Public Health, Emory University, Atlanta, Georgia Felicia C. Goldstein, PhD Allan I. Levey, MD, PhD Department of Neurology, School of Medicine, Emory University, Atlanta, Georgia
ACKNOWLEDGMENTS Conflict of Interest: The editor in chief has reviewed the conflict of interest checklist provided by the authors and has determined that the authors have no financial or any other kind of personal conflicts with this paper. Author Contributions: All authors contributed to this paper. Sponsor’s Role: None.
REFERENCES 1. Buslovich S, Salbu R, Ramaswamy R. Statins and the trajectory of cognitive decline. J Am Geriatr Soc 2014;62:789–790. 2. Steenland K, Zhao L, Goldstein FC et al. Statins and cognitive decline in older adults with normal cognition or mild cognitive impairment. J Am Geriatr Soc 2013;61:1449–1455
LETTERS TO THE EDITOR
791
DRUG BURDEN INDEX FOR INTERNATIONAL ASSESSMENT OF THE FUNCTIONAL BURDEN OF MEDICATIONS IN OLDER PEOPLE To the Editor: We thank Faure and colleagues for their interest in use of the Drug Burden Index (DBI). We share their desire to generalize this tool to allow it to be used more easily internationally. We wish to discuss their approach as outlined in the research letter, “A Standard International Version of the Drug Burden Index for CrossNational Comparison of the Functional Burden of Medications in Older People,”1 and to consider the implications not only for the proposed utility of comparing DBI between countries with existing formularies, but also for calculation of DBI in countries with limited national formularies, in which the tools to readily calculate a national DBI are not available. For countries with well-developed drug regulatory authorities and an advanced drug formulary structure, DBI as originally defined2 can be calculated from the formulary to measure exposure to medications with clinical anticholinergic or sedative effects. DBI has been calculated and validated against a range of clinical outcomes in different continents, countries, and settings, including the United States,2 Australia,3 the United Kingdom,4 and Finland.5 The DBI uses d, the minimum registered or licensed dose on the national formulary, as an estimate of the dose required to provide 50% of the maximal effect (DR50). This is based on the pharmacological principle that a registered dose must have some efficacy and that the minimum dose is likely to have less than maximal efficacy. The actual DR50 for most therapeutic drugs in older adults is not known. There are some differences between countries in the minimum registered or licensed doses of medicines used in the calculations, which may be due to the effect of ethnicity on drug response6 or to other issues influencing the regulatory decisions. For example, diazepam has a minimum registered or licensed dose of 4 mg in the United States and 5 mg in Australia. In studies comparing DBI exposure between countries with well-developed national formularies, we agree that a consistent estimate of DR50 is required when differences between countries in minimum registered or licensed dose are unlikely to be related to the effect of ethnicity on pharmacological exposure and response. Using the median minimum registered or licensed dose for the countries studied should give a better estimate of DR50 than the World Health Organization (WHO) Defined Daily Dose (DDD)7 proposed in DBI-WHO1 for the reasons outlined below. For countries in which the WHO list of essential medications and WHO model formulary are used as primary resources to support and improve medication prescribing, we believe that calculation of DBI may best be performed using the WHO model formulary recommended adult starting dosage.8 Review of the WHO model formulary adult starting doses indicates that they rather closely reflect the minimum registered or licensed doses in the countries in which DBI has been validated against functional status and physical performance. This dose is, in most cases, substantially lower than the DDD7 substituted for d in calculation of DBI-WHO.1 DDD is the average maintenance
