NPC Archive Item: Risks of anticholinergic drugs on cognitive decline and mortality examined

NOTE – This is an archive post from the NPC and has not been updated since first publication. Therefore, some hyperlinks may no longer be working.
MeReC Rapid Review NPC Logo
14 July 2011
A large observational study found that medications with anticholinergic activity were associated with greater cognitive decline in some groups of patients and increased mortality, compared with drugs without these effects. However, the study did not link the causes of death to anticholinergic activity and significant shortcomings limit its interpretation and application to practice.
Level of evidence:
Level 2 (limited quality patient-oriented evidence) according to the SORT criteria.
Patients taking anticholinergic medications may be concerned by the recent media coverage on risks of anticholinergics that were generated by this study. They should be advised not to stop any anticholinergic medication without a careful review of their treatment with their prescriber or other health professional. A number of medicines have anticholinergic effects: in elderly patients, or those at increased risk of cognitive decline who are using multiple medications which have anticholinergic activity, their continued use should be balanced against their risks.
What is the background to this?
A systematic review published in 2009 found a strong association between drugs with anticholinergic effects and acute cognitive impairment. That review did not identify sufficient studies that examined the long-term cognitive effects of anticholinergic drugs. Therefore, the authors of the current study retrospectively analysed data from the ongoing Medical Research Council Cognitive Function and Ageing Study (MRC CFAS). This involved examining MRC CFAS data from 1991 (baseline) and then comparing it to data from 1993.
For the purposes of this study, each participant’s anticholinergic burden was calculated using the Anticholinergic Burden Scale (ACB). This classifies all drugs according to their documented anticholinergic activity, 0 being none and 3 being high. Based on the ACB, medications were identified to have:
  • absent or no anticholinergic effects: ACB score 0
  • possible anticholinergic effects (drugs with serum anticholinergic activity but no clinically relevant negative cognitive effects): ACB score 1 e.g. atenolol, furosemide
  • definite anticholinergic effects (drugs with established and clinically relevant cognitive anticholinergic effects): moderate effect (ACB score 2, e.g. carbamazepine) and severe effect (ACB score 3, e.g. amitriptyline)
The researchers adjusted the data for confounders (e.g. age, sex, educational level, social class, number of non-anticholinergic medications, cognitive performance at baseline, and number of co-morbid health conditions from a defined list), and carried out a number of analyses.
Considering baseline data (n=12,250), the researchers compared the mean baseline Mini-Mental State Examination (MMSE) scores in the group of participants taking (n=5,709) or not taking (n=6,541) drugs with possible anticholinergic effects (ACB score 1) ; and the mean baseline MMSEs in the group of participants taking (n=508) or not taking (n=11,742) drugs with definite anticholinergic effects (ACB score 2 or 3). They also added up the ACB score for each anticholinergic medication a participant was taking to obtain a total ACB score for that person, and compared the MMSE scores of participants taking drugs with any anticholinergic effects to those of participants taking none.
After a two year follow-up period (n=8,334) the researchers assessed the effect of use of drugs with anticholinergic effects on MMSE score, to assess if there had been any cognitive decline at two years. They also attempted to assess if anticholinergic medications had any effect on mortality at the two-year follow up point.
The MMSE has a 30 point scale: the lower the score, the greater the cognitive impairment. NICE TA 217 states that the MMSE is frequently used to define the severity of dementia illnesses such as Alzheimer’s Disease, as follows:
MMSE 21–26 – mild Alzheimer’s disease
MMSE 10–20 – moderate Alzheimer’s disease
MMSE 10–14 – moderately severe Alzheimer’s disease
MMSE less than 10 – severe Alzheimer’s disease.
What does this study claim?
At baseline participants who had an ACB score of five or more had a mean (+Standard Deviation [SD]) MMSE score of 25.0+3.7, compared with 26.1+3.5 in those taking no anticholinergic medications (P=0.001). Taking any definite anticholinergic medication was associated with an MMSE score 0.82 points lower (95% confidence interval [CI] 0.55 to 1.10) than for no anticholinergic medications, although the median MMSE scores were the same. The researchers found no statistically significant association between taking possible anticholinergic medications and baseline MMSE score.
At two years, the average decline in MMSE was 0.8+2.9 points. Participants taking medication with definite anticholinergic effects had a 0.33-point greater decline in MMSE score (95%CI 0.03 to 0.64; P=0.03) than those not taking these drugs overall. When stratified by MMSE at baseline this was statistically significant only in participants who had an MMSE score of 26 to 30 at baseline (0.52 point decline, 95%CI 0.83 to 0.21, P=0.001). Participants with an ACB score of 4 or more had a 0.34 (95%CI 0.01 to 0.67) greater decrease in MMSE than those not taking anticholinergics, but again, when stratified by MMSE at baseline this was statistically significant only in participants who had had an MMSE score of 26 to 30 at baseline. The use of anticholinergics with possible anticholinergic effects at baseline (ACB score 1) was not associated with statistically significant differences in cognitive decline overall or when stratified by baseline MMSE.
The odds of death, at two years, for people on medications with definite anticholinergic effects (ACB score 2 or 3) or possible anticholinergic effects were higher than those not taking drugs with anticholinergic effects (odds ratio [OR] 1.68, 95%CI 1.30 to 2.16; P<0.001, and 1.56, 95%CI 1.36 to 1.79; P<0.001, respectively).
So what?
Although this study’s findings broadly support the existing work evaluating anticholinergic exposure and cognitive impairment, the study has significant shortcomings which limit its interpretation and application to clinical practice.
Observational studies, such as this one, can prove only association not causation and are prone to confounding. Unlike in the setting of a randomised controlled trial (RCT), in ‘real life’, treatment plans are chosen, changed, or actively not chosen in the light of individual patients’ risk factors, preferences and tolerability or response to other drugs. Thus, observed differences in outcomes may be due to differences among the patients, not only the different treatments. Importantly, in this study, the data were adjusted only for the number of self-reported co-existing morbidities in a specified list, not their nature or severity.
The MRC CFAS study attempted to consider all the medications participants were taking, including those purchased over the counter (by obtaining these data through interviewing and/or direct observation). This is a potential strength, however, the analysis discussed here looked only at the medications the participants were taking at baseline and at the two year follow-up point. During the intervening period, any changes to anticholinergic burden due to dosage fluctuations, interrupted use, or compliance issues were not considered. Moreover, 21% of those not taking anticholinergics at baseline were doing so two years later, and for those taking anticholinergics at baseline, 17% were no longer users by the follow-up.
Calculating the cumulative anticholinergic burden per participant was done by adding up the ACB score for each anticholinergic drug the person was taking. This does not account for the type (‘possible’ or ‘definite’ anticholinergic activity) or dose of the different anticholinergics being used by the individual, and therefore it is questionable if this is an accurate way to measure of a person’s total anticholinergic burden.
More fundamentally, there are significant question over some aspects of the analysis. The text suggests that the baseline analysis included a comparison of the MMSE of users of definite anticholinergics with that of people not taking anticholinergics; and a similar comparison between the MMSE of users of possible anticholinergics and people not taking anticholinergics. However, the numbers reported in the table suggest that the comparisons were between users of definite anticholinergics and all other participants in the cohort (i.e. including users of possible anticholinergics); and between users of possible anticholinergics and all other participants in the cohort (i.e. including users of definite anticholinergics). If this is truly what was done, this is a significant flaw which substantially limits the confidence which can be placed in the conclusions the authors draw. Insufficient data are presented for the results at follow up, including mortality, to ascertain whether similar concerns apply to these results.
Notwithstanding the limitations expressed above, the practical applications of this study is limited in that many of the drugs scored as having ACB score 1 (possible) or 2 (definite) anticholinergic effects would perhaps not usually be considered as being ‘anticholinergic drugs’ by many practitioners. In addition, one must question the clinical significance of a mean 0.5 point greater decline in MMSE over two years, in people with a high MMSE score at baseline, set in the context of an average 0.8 point decline with substantial variation: the standard deviation of the mean decline was +2.9 points, suggesting a 95% confidence interval of +5.7 points (noting also that, clearly, it is not possible for an individual to score fractions of a point on the MMSE).
Importantly, the study failed to link cause of death to anticholinergic activity. The researchers highlight that medications with anticholinergic activity are used for many diseases, such as hypertension and congestive heart failure, which were not adjusted for in this analysis and so the mortality findings may simply reflect the prevalence of anticholinergic prescribing in disease states with significant mortality.
Due to the significant limitations of this study, further observational research on a more contemporary cohort is needed, using a matched case-control approach, for example, and more comprehensive measures of measuring medication utilisation for assessing long-term exposure to anticholinergic medications.
Cohort study in participants enrolled in the Medical Research Council Cognitive Function and Ageing Study (MRC CFAS) between 1991 and 1993.
13.004 participants aged 65years and older. At baseline, the mean age of the participants was 75 years, and 60% were women. The mean Mini-Mental State Examination (MMSE) score at baseline was 25.9. Of the 12,250 participants with complete medication data and MMSE scores at baseline, by two-year follow up, 1,223 (10%) had died, 2,493 (20%) had dropped out, and 8,534 had completed the two-year follow-up survey (8,334 with complete MMSE). At baseline, 47% (5,709) of participants used a medication with possible anticholinergic properties, and 4% (508) of participants used a drug with definite anticholinergic properties.
Intervention and comparison
Initially, baseline use of possible or definite anticholinergics was identified according to the ACB scale and cognition determined using the MMSE. The main outcome measure was decline in the MMSE score at a two-year follow-up point.
Outcomes and results
Table 1. 1991: Association between anticholinergic medication use and MMSE score at baseline (N=12,250)
MMSE Adjusted MMSE difference*
ACB measure n % Mean + SD Median (interquartile range) Estimate (95% CI) P-value
Model 1
Sum of ACB score
0 6,327 52 26.1+3.5 27 (24–29) 0.00 0.001
1 3,273 27 25.8+3.4 27 (24–28) -0.003

(-0.11 to 0.16)
2 1,441 12 25.6+3.6 26 (24–28) 0.09

(-0.27 to 0.10)
3 719 6 25.6+3.6 26 (24–28) 0.10

(-0.35 to 0.14)
4 306 2 25.1+3.9 26 (23–28) 0.55

(-0.91 to -0.19)
>5 184 2 25.0+3.7 26 (23–28) 0.70

(-1.16 to -0.25)
Model 2
Sum of ACB (per point) 12,250 100 25.9+3.5 27 (24–28) -0.09

(-1.14 to -0.04)
Model 3
Any anticholinergic drugs
No 6,541 53 26.1+3.5 27(24-28) 0.00 0.71
Yes 5,709 47 25.7+3.5 27(24-28) 0.02

(-0.10 to 0.14)
No 11,742 96 25.9+3.5 27(24-28) 0.00 <0.001
Yes 508 4 24.9+3.9 27(23-28) -0.82

(-1.10 to -0.55)
*Adjusted for age, sex, education, social class, number of nonanticholinergic medications, and number of comorbidities.
Table 2. 1993: Association* between anticholinergic medication use and additional decline on the MMSE at 2 years stratified according to MMSE score at baseline (N=8,334)
Baseline MMSE
26–30 (n=6,041)
22–25 (n=1,769)
0–21 (n=524)
ACB Measure Estimate (95%CI) P-value Estimate (95%CI) P-value Estimate (95%CI) P- value Estimate (95%CI) P-value
Model 1
Sum of ACB score
0 0.00 0.02 0.00 0.03 0.00 0.11 0.00 0.09
1 0.05

(-0.08 to 0.19)

(-0.45 to 0.32)

(-0.60 to 1.27)

(-0.11 to 0.17)
2 0.01

(-0.19 to 0.21)

(-0.53 to 0.54)

(-1.31 to 1.14)

(-0.20 to 0.21)
3 -0.21

(-0.47 to 0.05)

(-1.00 to 0.45)

(-0.70 to 2.65)

(-0.42 to 0.12)
4+ -0.51

(-0.84 to -0.18)

(-1.24 to 0.36)

(-0.18 to 3.87)

(-0.67 to -0.01)
Model 2
Sum of ACB score
Per point -0.07

(-0.13 to -0.02)
0.01 -0.07

(-0.21 to 0.07)
0.32 0.32

(-0.02 to 0.65)
0.07 -0.05

(-0.10 to 0.01)
Model 3
Any anticholinergic drugs
Possible (1) 0.02

(-0.10 to 0.14)
0.75 -0.05

(-0.39 to 0.29)
0.77 0.38

(-0.43 to 1.20)
0.35 0.02

(-0.11 to 0.14)
Definite (2–3) -0.52

(-0.83 to
<0.001 -0.32

(-1.10 to 0.46)
0.42 1.10

(-0.61 to 2.82)
0.21 -0.33

(-0.64 to -0.03)
*Models were adjusted for age, sex, education, social class, number of nonanticholinergic medications, and number of comorbidities.
Association* between anticholinergic use and mortality and attrition
After adjusting for age, sex, baseline MMSE score, education, social class, number of nonanticholinergic medications, and number of health conditions, taking anticholinergic medications was associated with death at two years, with a dose-response effect on ACB score. For every additional ACB point scored, the odds of death increased by 26% (OR 1.26, 95%CI 1.20 to 1.32). The effect was present for the use of definite anticholinergics and possible anticholinergics (OR1.68, 95%CI 1.30 to 2.16; P<0.001, and 1.56, 95%CI 1.36 to 1.79; P<0.001, respectively).
CFAS is funded by the Medical Research Council, UK (Grant G9901400)
Further information on safe prescribing is available in the NPC’s 10 Top tips for GPs, guide to medication review and e-learning materials on evidence-informed decision making.
Further information on prescribing and safety information for individual drugs can be found on NHS Evidence
Please comment on this rapid review in the NPC discussion rooms, or using our feedback form.
Make sure you are signed up to NPC Email updates — the free email alerting system that keeps you up to date with the NPC news and outputs relevant to you