Risk for Health Events After Deprescribing Acetylcholinesterase Inhibitors in Nursing Home Residents With Severe Dementia

Abstract

BACKGROUND/OBJECTIVE:

Reevaluation of the appropriateness of acetylcholinesterase inhibitors (AChEIs) is recommended in older adults with severe dementia, given the lack of strong evidence to support their continued effectiveness and risk for medication-induced adverse events. We sought to evaluate the impact of deprescribing AChEIs on risk of all-cause events (hospitalizations, emergency department visits, and mortality) and serious falls or fractures in older nursing home (NH) residents with severe dementia.

DESIGN:

Analysis of 2015 to 2016 data from Medicare claims, Part D prescriptions, Minimum Data Set (MDS) version 3.0, Area Health Resource File, and Nursing Home Compare. Marginal structural models with inverse probability of treatment weights were used to evaluate the association of deprescribing AChEIs and all-cause negative events as well as serious falls or fractures.

SETTING:

US Medicare certified NHs.

PARTICIPANTS:

Nonskilled NH residents, aged 65 years and older, with severe dementia receiving AChEIs within the first 14 days of an MDS assessment in 2016 (n = 37 106).

RESULTS:

The sample was primarily white (78.7%), female (75.5%), and aged 80 years or older (77.4%). Deprescribing AChEIs was associated with an increased likelihood of all-cause negative events in unadjusted models (odds ratio [OR] = 1.17; 95% confidence interval [CI] = 1.11-1.23; P < .01), but not in fully adjusted models (adjusted OR [aOR] = 1.00; 95% CI = 0.94-1.06; P = .94). By contrast, deprescribing was associated with a reduced likelihood of serious falls or fractures in unadjusted models (OR = 0.59; 95% CI = 0.52-0.66; P < .001) and remained significant in adjusted models (aOR = 0.64; 95% CI = 0.56-0.73; P < .001).

CONCLUSION:

Deprescribing AChEIs was not associated with a significant increase in the likelihood for all-cause negative events and was associated with a reduced likelihood of falls and fractures in older NH residents with dementia. Our findings suggest that deprescribing AChEIs is a reasonable approach to reduce the risk of serious falls or fractures without increasing the risk for all-cause events. J Am Geriatr Soc 68:699-707, 2020.

Deprescribing acetylcholinesterase inhibitors (AChEIs) may be a reasonable strategy to reduce medication burden and the potential for adverse events in patients with severe dementia, given the lack of strong evidence for their long-term effectiveness in this population.¹ Although one of the main barriers to deprescribing AChEIs is the potential worsening of behavioral symptoms,^2–4 studies suggest that, in patients with severe dementia, any worsening of these symptoms that may be observed is likely not clinically significant.^5–7 Equally important clinical end points in the decision to discontinue AChEIs may be the occurrence of all-cause events (ie, hospitalizations, emergency department [ED] visits, and mortality) that result from deprescribing vs the occurrence of adverse events that may be induced by AChEIs (ie, medication-related adverse events).

Although randomized trials report a consistent and positive effect of AChEIs on cognitive function compared to placebo, nearly all have been conducted in individuals with mild or moderate dementia.¹ The limited number of studies that have included patients with severe dementia report findings that had minor clinical significance or were inconclusive, putting into question the cognitive benefits of these agents in patients with more advanced disease.¹ There is some evidence from clinical trials^8,9 and observational studies^10,11 to suggest that use of AChEIs may be associated with decreased mortality risk; however, it is uncertain by what mechanism this is achieved. It is also uncertain to what degree these benefits may apply to patients with severe dementia. For example, a recently published retrospective observational study¹² found that community-dwelling patients receiving AChEIs had significantly reduced mortality rates compared to patients not receiving treatment. However, when analyses were stratified by baseline mortality risk, this benefit was not seen for patients with a high risk for mortality, a designation that would likely apply to individuals with severe dementia.

Despite the uncertain impact of discontinuing AChEIs on all-cause events, one potential benefit is a reduction in the occurrence of serious medication-related adverse events that may result in hospitalization,¹³ specifically syncope.¹⁴ AChEI induced syncopal events or related symptoms may lead to avoidable hospitalizations for cardiac procedures, as well as an increased risk for falls and fractures, as demonstrated in a large population based study by Gill et al.¹⁵ However, no studies to date have examined whether discontinuing AChEIs actually results in reduced rates of serious fall or fracture.

The objective of this study was to evaluate the impact of deprescribing AChEIs on the occurrence of all-cause negative events (ie, hospitalizations, ED visits, and mortality), as well as hospitalizations or ED visits for serious falls and fractures in nursing home (NH) residents with severe dementia. The findings presented will address a critical knowledge gap with regard to the potential risks and benefits that may be associated with deprescribing AChEIs in this population, in which deprescribing is most likely to occur.

METHODS

Design and Data Sources

This study was a retrospective, longitudinal analysis of Medicare Part A and B claims, Master Beneficiary Summary File (MBSF), Part D prescription drug event data, the Minimum Data Set (MDS) version 3.0, the Area Health Resource File (AHRF), and Nursing Home Compare (NHC) for 2015 to 2016. The University of Pittsburgh Institutional Review Board deemed this study exempt.

Data originated from a random sample of 1 million Medicare beneficiaries, aged 65 years and older, with continuous enrollment in Medicare Parts A, B, and D in 2015 and a dementia diagnosis prior to 2016 based on the Chronic Conditions Warehouse algorithm for identifying Alzheimer disease or related disorders with International Classification of Diseases (ICD) codes.¹⁶ The MDS, a comprehensive health assessment tool administered to residents of Centers for Medicare and Medicaid Services–certified NHs at admission and at least every 90 days thereafter, served as the primary source of variables. The Medicare MBSF and Part A/B claims were used to identify comorbidities, inpatient and outpatient healthcare utilization in the year prior, and death date. Medicare Part D data provided information on prescriptions dispensed in outpatient and long-term care settings, including drug name, National Drug Code, date filled, dose, strength, quantity, days’ supply, and prescriber characteristics. The NHC and AHRF provided facility characteristics.^17,18

Sample

The final cohort consisted of nonskilled nursing stays for patients with severe dementia receiving AChEIs at index (Figure 1). Skilled NH stays require more advanced care, including intravenous medications and physical therapy, and are covered by Medicare Part A and, thus, medication data are not available. We used the MDS reason for assessment fields (A0310A, A0310B) to identify all MDS assessments for nonskilled NH stays,^19,20 beginning in 2016. NH episodes (n = 335 487) were constructed by matching the first assessment in 2016 to the closest discharge form or assigning the end of the study period (December 31, 2016) as the end date. The first day in which AChEI supply was observed was assigned as the AChEI index date. We required that residents had continuous enrollment in Medicare Parts A, B, and D for the duration of all episodes and the year prior (n = 14 388; 4.3% excluded). Episodes in which the resident had any Medicare skilled nursing facility claims overlapping the index date to episode end date were excluded (n = 52 390; 16.3% excluded). Episodes in which residents had severe dementia were identified using cognitive assessments in the MDS (n = 149 727; 55.6%). Specifically, the Brief Interview for Mental Status (BIMS)²¹ was used, or if unable to complete the BIMS, the Cognitive Performance Scale (CPS) was used.²² We used a BIMS score of 7 or less or a CPS score of 4 or greater to identify severe dementia, which have demonstrated acceptable sensitivity and specificity for severe cognitive impairment when compared to the modified Mini-Mental State Examination.²³ We then limited to residents receiving AChEIs at index by searching Part D records for generic drug names (donepezil, rivastigmine, and galantamine). Residents were considered treated if there was a prescription for an AChEI with an estimated days’ supply overlapping one of the initial 14 days of the episode (n = 43 996; 29.4%). We excluded episodes with 30 days or less of follow-up to allow time to observe potential discontinuation (n = 4158; 9.5% excluded). Finally, if residents had more than one episode meeting criteria, only the first was included (n = 2729; 6.9% excluded).

Figure 1.

Sample construction. AChEI indicates acetylcholinesterase inhibitor; CY, Calendar Year; MDS, Minimum Data Set; NH, nursing home; SNF, skilled nursing facility.

A longitudinal data set was created in which each resident could have multiple MDS assessments from episode start until deprescribing or censoring. The final cohort included 37 106 residents with 109 240 assessments for all-cause events and 114 648 assessments for falls and fractures.

Dependent Variables

The dependent variables in this analysis were all-cause negative events and medication-related events of serious falls or fractures. All-cause negative events included ED visits, hospitalizations, and death for any reason. Hospitalizations and ED visits were identified by searching for Medicare claims occurring since AChEI index date until the earliest of the following: NH discharge date, death date, or end of study (December 31, 2016). Death dates were extracted from the Medicare MBSF. Serious falls and fractures were identified in a similar fashion, but using claims for hospitalizations or ED visits associated with specific ICD-9 or ICD-10 codes.²⁴

Independent Variables

The primary independent variable in this analysis was whether AChEIs were deprescribed. Deprescribing was defined as discontinuation, signaled by a subsequent gap in therapy of at least 30 days based on prescription fill dates and last day of supply, with the 31st gap day in the period serving as the discontinuation date. The discontinuation date was defined as such to avoid the potential for immortal time bias, in which a period of immortal time, usually required for the observation of a specific event, may be incorrectly attributed to the exposed group.^25,26 Deprescribing was treated as a time-varying exposure in our data set and was coded as positive if the 31st gap day in medication supply occurred on or after the assessment start date and before either the assessment stop date or the event date.

Covariates were extracted from the MDS, Medicare Part D Prescription Drug Event records, Medicare Part A and B claims, NHC, and the AHRF. We previously identified that a number of clinical factors corresponding to limited prognosis or deteriorating clinical status were associated with increased likelihood for deprescribing, while system-level factors may act as barriers to deprescribing.²⁷ The full list of covariates included: demographics (age, sex, race/ethnicity, and marital status), clinical assessment factors (MDS assessment form type, resident ability to be understood, poor appetite, urinary incontinence, swallowing disorder, parenteral nutrition or tube feeds, mechanically altered diet, recent weight loss, shortness of breath, dehydration, cancer, end-stage renal disease, heart failure, activities of daily living, limited prognosis or hospice utilization, antidepressant use, antipsychotic use, benzodiazepine use, strong anticholinergic use, AChEI type, memantine use, total number of medications, Charlson Comorbidity Index, all-cause and cause-specific hospitalizations in 90 days prior to index date, and location prior to NH residence), environment of care (NH geographic region, facility size, and rurality), and provider specialty (AChEI prescriber specialty: primary care, geriatrics, or other).

Demographic, environment of care, and provider specialty variables were treated as time invariant and were measured at the time of the index MDS assessment, while clinical assessment factors were created as time-varying and were measured at the time of each MDS assessment.

Statistical Analyses

Analyses were performed using SAS, version 9.4 (SAS Institute Inc) and STATA 14 (StataCorp). Missing observations (<5% total) were addressed with single imputation using chained equations, including all covariates in the imputation.²⁸

We used marginal structural models with inverse probability of treatment weights (IPTW) to address the potential for time dependent confounding.^29–34 IPTWs were used to model each subject’s propensity for being deprescribed, considering the subject’s history of covariates.^30,35 We also addressed the potential for loss to follow-up by using inverse probability of censoring weights (IPCW), calculated as the probability of remaining uncensored at the time of each MDS assessment. The final stabilized inverse propensity weight for each assessment was the product of IPTW and IPCW. Additional information on the derivation of IPTWs and IPCWs and the implementation of marginal structural models can be found in supplementary materials.

Each record ended with the event of interest, censoring due to death or discharge, or the day before the next MDS assessment start date (if no event or censoring). Sample characteristics were calculated on both the patient level and the assessment level (ie, at the time of each MDS assessment). We evaluated the balance of covariates across deprescribing status after weighting using standardized differences in each sample.³⁶

Analyses were conducted using an intent-to-treat approach for deprescribing, where residents remained in this category until censoring or end of follow-up. The primary analysis of the association of deprescribing and each outcome used a pooled logistic regression model for discrete events, analogous to a Cox proportional hazards model, and was weighted by the product of IPTWs and IPCWs. We used robust SEs to account for the correlation between observations from the same individual.

We conducted several sensitivity analyses. We first evaluated the influence of extreme weights in our sample. In the second sensitivity analysis, we reconducted our analyses in a per-protocol fashion, where residents who were deprescribed but then later filled a new AChEI prescription (ie, “restarters”) were censored in the period following the new AChEI prescription. Finally, we examined the effect of deprescribing AChEIs on both outcomes in subgroups stratified by whether residents were receiving memantine at baseline.

RESULTS

Sample Characteristics

Time-invariant sample characteristics, including demographics, environment of care, and provider specialty, are presented at baseline in Table 1. Time-varying clinical characteristics are presented at baseline on the resident level and during follow-up at the assessment level in Table 2. At the index date, the sample was primarily white (78.7%), aged 80 years or older (77.4%), and female (75.5%). Most residents in the sample had already been residing in the NH at their index date (87.7%) as opposed to newly admitted (12.3%). Weighted sample characteristics are presented in Supplementary Tables S1 and S2. All sample characteristics showed adequate balance after applying IPTWs, with the exception of MDS assessment type, hospitalizations, memantine use, total number of medications, geographic region, and rurality.

Table 1.

Baseline Time Invariant Sample Characteristics (N = 37 106 Residents)

Characteristics	No. (%) of Residents
Demographics
Age, y
65–69	901 (2.4)
70–79	7496 (20.2)
80–89	17 922 (48.3)
≥90	10 787 (29.1)
Sex
Male	9092 (24.5)
Female	28 014 (75.5)
Race/ethnicity
White	29 210 (78.7)
Black	4405 (11.9)
Hispanic	2049 (5.5)
Other	1442 (3.9)
Current marital status
Married	8111 (22.1)
Not Married	28 918 (77.9)
Environment of Care
Geographic region
Midwest	10 131 (27.3)
Northeast	6631 (17.9)
South	17 279 (46.6)
West	3065 (8.3)
Certified beds
<50	1904 (5.1)
50–99	10 613 (28.6)
100–199	20 335 (54.8)
≥200	4254 (11.5)
Rural/urban continuum
Urban	25 715 (69.3)
Rural	9925 (26.7)
Highly rural	1466 (4.0)
Provider Specialty
Prescriber specialty
Geriatrics	3167 (8.5)
Primary care	3963 (10.7)
Other	29 976 (80.8)

Table 2.

Time-Varying Clinical Characteristics

Variable	Full Sample At Index Date (N = 37 106 Residents)	All-Cause Events Analysis, Assessment Level (N = 109 240 Assessments)	Falls/Fractures Analysis, Assessment Level (N = 114 647 Assessments)
Entered from
Community	7575 (20.4)	22 018 (20.2)	22 850 (19.9)
Hospital	25 275 (68.1)	74 676 (68.4)	78 630 (68.6)
NH or other LTC facility	4256 (11.5)	12 546 (11.5)	13 168 (11.5)
MDS assessment type
Admission	4576 (12.3)	4576 (4.2)	4576 (4.0)
Quarterly	24 434 (65.8)	78 051 (71.4)	81 972 (71.5)
Annual	6001 (16.2)	19 598 (17.9)	20 575 (17.9)
Significant change in status	2095 (5.6)	7015 (6.4)	7525 (6.6)
Charlson Comorbidity Index
0–1	6054 (16.3)	17 095 (15.6)	17 639 (15.4)
2–3	9162 (24.7)	28 203 (25.8)	29 371 (25.6)
4–5	8702 (23.5)	26 667 (24.4)	27 994 (24.4)
≥6	13 188 (35.5)	37 275 (34.1)	39 644 (34.6)
Makes self understood
Understood	14 721 (39.7)	41 022 (37.6)	42 987 (37.5)
Usually understood	9961 (26.8)	29 437 (26.9)	30 859 (26.9)
Sometimes understood	8084 (21.8)	24 000 (22.0)	25 148 (21.9)
Rarely/never understood	4340 (11.7)	14 781 (13.5)	15 654 (13.7)
PHQ 9 score, mean (SD)	2.22 (3.40)	2.16 (3.39)	2.16 (3.40)
Aggressive behavior scale score, mean (SD)	0.56 (1.34)	0.53 (1.29)	0.53 (1.29)
Activities of daily living score
1–7	2849 (7.7)	7968 (7.3)	8359 (7.3)
8–14	6033 (16.3)	16 502 (15.1)	17 268 (15.1)
15–21	19 426 (52.4)	56 793 (52.0)	59 437 (51.8)
22–28	8798 (23.7)	27 977 (25.6)	29 584 (25.8)
Urinary incontinence
Continent	3954 (10.7)	10 220 (9.4)	10 720 (9.4)
Occasionally incontinent	5439 (14.7)	14 420 (13.2)	15 114 (13.2)
Frequently incontinent	12 281 (33.1)	35 465 (32.5)	37 093 (32.4)
Always incontinent	14 644 (39.5)	47 338 (43.3)	49 734 (43.4)
Indwelling catheter	788 (2.1)	1797 (1.6)	1987 (1.7)
Cancer	1555 (4.2)	4462 (4.1)	4677 (4.1)
Heart failure	5809 (15.7)	16 697 (15.3)	17 584 (15.3)
End-stage renal disease	3419 (9.2)	9960 (9.1)	10 537 (9.2)
Short of breath	2469 (6.7)	6922 (6.3)	7353 (6.4)
Poor appetite	4946 (13.3)	14 490 (13.3)	15 276 (13.3)
Weight loss	2389 (6.4)	6952 (6.4)	7416 (6.5)
Swallowing difficulty	1247 (3.4)	3790 (3.5)	3999 (3.5)
Mechanically altered diet	20 311 (54.7)	56 341 (51.6)	59 107 (51.6)
IV/parenteral nutrition or feeding tube	1262 (3.4)	2961 (2.7)	3463 (3.0)
Hospice or limited prognosis	1875 (5.1)	6777 (6.2)	7278 (6.3)
Hospitalizations/ED visits (90 d prior to index)
None	27 980 (75.4)
Cause specific (fall, fracture, or syncope)	3150 (8.5)
Other cause	5976 (16.1)
AChEI at index date
Donepezil	28 877 (77.8)	84 455 (77.3)	88 587 (77.3)
Donepezil/memantine	832 (2.2)	2485 (2.3)	2564 (2.2)
Galantamine	483 (1.3)	1927 (1.8)	2044 (1.8)
Rivastigmine (oral)	5403 (14.6)	15 825 (14.5)	16 653 (14.5)
Rivastigmine (transdermal)	1511 (4.1)	4548 (4.2)	4800 (4.2)
Memantine use	15 199 (41.0)	44 342 (40.6)	46 535 (40.6)
Benzodiazepine and/or Z drug	5442 (14.7)	14 846 (13.6)	15 706 (13.7)
Antipsychotic use	9003 (24.3)	23 999 (22.0)	25 226 (22.0)
Antidepressant use	21 071 (56.8)	60 832 (55.7)	63 791 (55.6)
Highly anticholinergic drugs (AGS Beers Criteria®)	5519 (14.9)	14 776 (13.5)	15 524 (13.5)
Total no. of medications, mean (SD)	5.9 (3.1)	5.7 (3.1)	5.7 (3.1)

Note: Data are given as number (percentage), unless otherwise indicated.

Abbreviations: AChEI, acetylcholinesterase inhibitor; ED, emergency department; IV, intravenous; LTC, long-term care; MDS, Minimum Data Set; NH, nursing home; PHQ, Patient Health Questionnaire.

Primary Analyses

All-cause negative events occurred in 31.6% of residents, and hospitalizations for falls or fractures occurred in 9.2% of residents, over a median follow-up time of 226 days (interquartile range [IQR] = 92–312 days) and 259 days (IQR = 110–318 days), respectively. This equated to event rates of 0.56 all-cause negative events and 0.15 falls or fractures per person-year. ED visits were the most common all-cause event, occurring in 16.1% of residents, followed by hospitalizations (8.4%) and death (7.1%). All-cause events occurred at a higher rate in residents who had AChEIs deprescribed vs those who did not (0.62 vs 0.55 events per person-year). Falls or fractures had a lower event rate in residents who had AChEIs deprescribed vs those who did not (0.09 vs 0.16 events per person-year).

Model results are presented in Figure 2. In unadjusted analyses, deprescribing AChEIs was associated with an increased likelihood for all-cause negative events and a reduced likelihood for falls or fractures. In adjusted analyses, deprescribing AChEIs was no longer significantly associated with all-cause negative events, but remained significantly associated with a reduced likelihood of falls or fractures.

Figure 2.

Association of deprescribing acetylcholinesterase inhibitors (AChEIs) with all-cause negative events and falls or fractures. aOR indicates adjusted OR; CI, confidence interval; OR, odds ratio.

Sensitivity Analyses

Excluding observations with IPTWs outside of the 1st and 99th percentiles and top-coding values at the 1st and 99th percentiles resulted in no substantive changes to our findings (Supplementary Tables S3 and S4). Applying a per-protocol approach resulted in less than 2% of MDS assessments being excluded from analyses due to potential medication restarts, but our findings remained essentially unchanged (all-cause events: adjusted odds ratio [aOR] = 1.03 [95% confidence interval [CI] = 0.97-1.10] [P = .34]; falls or fractures: aOR = 0.66 [95% CI = 0.57-0.75] [P < .001]).

Analyses stratified by memantine use at baseline are presented in Supplementary Table S5. Results remained consistent regardless of baseline memantine use. Deprescribing AChEIs was not significantly associated with all-cause events (memantine at baseline: aOR = 1.04 [95% CI = 0.95-1.15] [P = .40]; no memantine at baseline: aOR = 0.99 [95% CI = 0.92-1.07] [P = .89]), and remained associated with a reduced likelihood of falls or fractures (memantine at baseline: aOR = 0.70 [95% CI = 0.57-0.87] [P < .001]; no memantine at baseline: aOR = 0.60 [95% CI = 0.51-0.70] [P < .001]).

DISCUSSION

In a large national sample of NH residents with severe dementia, we found that deprescribing AChEIs was not associated with the occurrence of all-cause negative events and was associated with a reduced likelihood of hospitalization due to falls or fractures. The findings from this analysis address a major gap in the literature related to the tolerability and safety of deprescribing AChEIs in NH residents and suggest that the discontinuation is reasonable in individuals with severe dementia given that it does not increase the likelihood of negative events.

To our knowledge, no large observational studies have examined the impact of discontinuing AChEIs on the occurrence of negative events. Smaller studies that have examined the impact of discontinuing AChEIs on all-cause negative events present conflicting results. Two observational studies^37,38 reported higher mortality rates among patients who experienced gaps in AChEI therapy. By contrast, a recent meta-analysis of five randomized controlled trials reported no significant differences in dropout rates due to deaths or adverse events, such as falls and gastrointestinal symptoms, between individuals who discontinued AChEIs and those who continued therapy.³⁹ Only one of the above studies included any patients with severe dementia,⁴⁰ despite the fact that these patients may be most likely to discontinue these agents due to lack of perceived benefit relative to risks.^21,22,41,42

The strengths of the present study overcome the limitations of prior investigations and enhance the clinical significance of our findings, while contributing new information to the literature on deprescribing. We used a large, nationally representative sample of NH residents and focused on those with severe dementia, who are arguably the most clinically relevant population for deprescribing AChEIs.⁴³ In addition to all-cause negative events, we also evaluated the effect of deprescribing AChEIs on medication-related adverse events (ie, falls and fractures) in an effort to address the potential benefits of deprescribing, as well as the potential risks. To our knowledge, no prior studies have examined the effect of deprescribing on falls and fractures.

In unadjusted analyses, deprescribing AChEIs was associated with an increased likelihood for all-cause negative events. Interestingly, this association was no longer significant in adjusted models, suggesting that there is a substantial degree of confounding by indication, where those who are most likely to be deprescribed are those also at high risk for hospitalizations, ED visits, and/or death. This is not surprising given that severe dementia itself is considered to be a life limiting condition,⁴⁴ and in our prior work, we identified that patients with clinical characteristics that signal a decline in health status were more likely to discontinue AChEIs.^27,45,46 Thus, the observed increase in all-cause negative events in unadjusted analyses was likely attributable to a higher baseline mortality risk due to severe dementia rather than the effects of AChEI discontinuation.

Deprescribing AChEIs may confer some benefits in NH residents with severe dementia, as evidenced by the reduction in the likelihood of hospitalizations for serious falls or fractures that was observed in both adjusted and unadjusted analyses. Previous investigations have identified bradycardia and syncope as significant adverse effects of AChEIs that can precipitate the occurrence of falls and fractures.^14,15 This is also supported by our prior work, in which we identified that discontinuing AChEIs was more likely in NH residents with a hospitalization for bradycardia, fall, or fracture in the prior 90 days,²⁷ potentially in response to one of these negative events. We do acknowledge that while our outcome definition for falls and fractures was sensitive, it was not specific to events due to bradycardia or syncope and we did not have measures of blood pressure at the time of event available in claims to adjudicate these events. As a result, it is possible that some of this effect is not directly attributable to a reduction in adverse events achieved through deprescribing AChEIs.

There are several limitations that should be acknowledged. Although it is unlikely that a gap in days supply of greater than 30 days would occur unintentionally in a NH population, we were not able to confirm intentional vs unintentional discontinuation based on refill records. We also acknowledge that the rate of falls or fractures in our study may have been underestimated due to goals of care or advance care planning that may prevent some residents from being transferred to an acute setting. We also may have missed early adverse events by imposing a minimum length of stay requirement of 30 days as well other less serious adverse events not captured by claims. We only included the first qualifying episode for residents who had multiple stays during follow-up. Thus, there may have been differences in terms of overall health status and, thus, vulnerability to adverse events between their first and subsequent stays. However, the proportion of residents with multiple stays was small, so it is unlikely that this would have influenced our results substantially. Finally, although we used marginal structural models to address the potential for time-varying confounding, we cannot rule out the possibility of residual confounding or the potential unmeasured confounders.

Future studies should continue to address the impact of deprescribing AChEIs on other medication-specific adverse events, such as gastrointestinal effects, weight loss, and urinary symptoms, which may significantly affect resident quality of life.

CONCLUSIONS

This study found that deprescribing AChEIs was not associated with a significant increase in the likelihood of all-cause negative events, but was associated with a decrease in the likelihood of hospitalizations due to serious falls and fractures in older NH residents with severe dementia. Our findings suggest that deprescribing AChEIs may be an effective strategy to reduce the occurrence of serious medication-related adverse events without substantial risk.

Editor’s Note.

Our Journal and many others have published papers like this one that call into question the benefit/risk ratio of acetylcholinesterase inhibitors (AChEIs), especially as cognitive impairment progresses. The initial decision about whether to start an AChEI is a person-centered one – like many of the decisions we make in treating older patients. The best evidence for the effectiveness of these drugs is in delaying the need for institutionalization. But, that is a very complicated outcome because it is dependent on so many factors other than any slight stabilization or improvement in cognitive function these drugs might provide.

There is more convincing evidence emerging about the potential harms of these drugs. Many students and residents do not think about the potential cardiac effects of AChEIs through causing bradycardia that can result in hypotension, cerebral hypo-perfusion, near syncope, falls and injuries. While making hospital rounds I have seen several older patients who have had pacemakers inserted while on an AChEI. These drugs have in fact been shown to be associated with syncope, pacemaker placement, and injurious falls, and their initiation is associated with emergency department visits and hospitalizations, as demonstrated by a paper recently published in this Journal.

The current paper by Niznik and colleagues adds further evidence to these associations, by demonstrating that deprescribing of AChEIs is associated with fewer falls and fractures among people with dementia residing in nursing homes.

-Joseph G. Ouslander, MD