Anticholinergics and Risk of Pneumonia in Elderly Adults
One thousand thirty-nine cases and 2,022 matched controls were included in this analysis (Table 1); these participants filled 14,897 prescriptions for anticholinergic medications during the year before the index date (Figure 2). For those with any anticholinergic prescription fills in the prior year (64% of participants), the number of fills varied from 1 to 97, with a median of 4 fills per participant. Of all anticholinergic prescriptions filled in the year before the index date, the most common classes of medications were pain medications (20.7%), gastrointestinal medications (20.7%), antidepressants (19.6%), benzodiazepines (6.2%), and urinary incontinence medications (3.7%). The most common individual medications filled were ranitidine (14.7%), digoxin (10.0%), and acetaminophen with hydrocodone (9.6%). Appendix 1 provides a complete list of anticholinergic medications filled for study participants in the year before the index date. Approximately 43% of participants had filled at least one anticholinergic prescription within 90 days before the index date (acute use). For those with any acute fills, the number of prescription fills within 90 days of the index date varied from 1 to 26, with a median of 2 fills. A similar proportion of participants (42%) had filled at least three prescriptions for anticholinergic medications within the year before the index date (defined as chronic use). There was significant overlap between acute users and chronic users; of participants who had filled at least one anticholinergic prescription within 90 days of the index date, 82% had also filled three or more prescriptions in the prior year. Only 21% of participants were classified as past users, having filled one or more prescriptions in the past year but none within the past 90 days.
(Enlarge Image)
Figure 2.
Frequency of anticholinergic prescription fills for study participants in the year before the index date, according to medication class.
A strong association was found between acute use of any anticholinergic medication and risk of community-acquired pneumonia, even after adjustment for confounding variables (Table 2, aOR = 2.55, 95% CI = 2.08–3.13), but past use of anticholinergic medications was not associated with risk of pneumonia after adjustment for all confounders (aOR = 1.19, 95% CI = 0.92–1.53). There was a significant association between chronic use of anticholinergics and risk of pneumonia (aOR = 2.07, 95% CI = 1.68–2.54), but this estimate was not significantly different from that of nonchronic use (1–2 fills in the prior year) (aOR = 1.83, 95% CI = 1.44–2.32).
When analyses were stratified according to anticholinergic potency (Table 2, Models 1a and 2a), risk estimates for use of high-potency anticholinergics were not substantively different from those for low potency anticholinergics for acute use (high potency, aOR = 2.63, 95% CI = 1.99–3.48; low potency, aOR = 2.54, 95% CI = 2.04–3.16) or chronic use (high-potency, aOR = 2.33, 95% CI = 1.78–3.13; low potency, aOR = 1.90, 95% CI = 1.51–2.40).
Because there was so much overlap between acute and chronic users (82% of acute users were also chronic users), post hoc analyses were conducted to try to further delineate whether the greater risk observed with acute use was primarily limited to recent initiators of anticholinergic medications. In this exploratory model, new users (with ≥1 fills in the 90 days before the index date but none in the remainder of the prior year) had a risk of pneumonia that was nearly five times as great as that of nonusers (aOR = 4.96, 95% CI = 3.47–6.19). New users made up a small proportion of acute users (n = 199, 15% of all acute users), and in this model acute users who were not new initiators still had a significantly greater risk of pneumonia (aOR = 2.23, 95% 1.81–2.76). When new users were restricted to those who had a fill within 30 days of the index date but no other fills in the prior year (n = 102), the association was yet stronger (aOR = 8.33, 95% CI = 5.03–13.8).
The exploratory dose-response analysis showed that the risk of CAP was similarly high for participants who filled one, two, three, four, or five or more anticholinergic prescriptions in the 90 days before the index date but that there was no clear pattern of greater risk with additional fills (Table 3).
In the analysis excluding all anticholinergic opiates (Table 3), acute use of any anticholinergic medication was still associated with risk of pneumonia, although the risk was attenuated (aOR = 1.56, 95% CI = 1.29–1.89). Further adjustment for current or past use of any opiate, including nonanticholinergic opiates, did not significantly change the risk estimate (aOR = 1.51, 1.24–1.83). In the analysis excluding prescription fills for cough suppressants in the 5 days before the index date, the association between acute anticholinergic use and pneumonia risk persisted, although it was slightly attenuated (Table 3, aOR = 1.96, 95% CI = 1.58–2.44).
Results
One thousand thirty-nine cases and 2,022 matched controls were included in this analysis (Table 1); these participants filled 14,897 prescriptions for anticholinergic medications during the year before the index date (Figure 2). For those with any anticholinergic prescription fills in the prior year (64% of participants), the number of fills varied from 1 to 97, with a median of 4 fills per participant. Of all anticholinergic prescriptions filled in the year before the index date, the most common classes of medications were pain medications (20.7%), gastrointestinal medications (20.7%), antidepressants (19.6%), benzodiazepines (6.2%), and urinary incontinence medications (3.7%). The most common individual medications filled were ranitidine (14.7%), digoxin (10.0%), and acetaminophen with hydrocodone (9.6%). Appendix 1 provides a complete list of anticholinergic medications filled for study participants in the year before the index date. Approximately 43% of participants had filled at least one anticholinergic prescription within 90 days before the index date (acute use). For those with any acute fills, the number of prescription fills within 90 days of the index date varied from 1 to 26, with a median of 2 fills. A similar proportion of participants (42%) had filled at least three prescriptions for anticholinergic medications within the year before the index date (defined as chronic use). There was significant overlap between acute users and chronic users; of participants who had filled at least one anticholinergic prescription within 90 days of the index date, 82% had also filled three or more prescriptions in the prior year. Only 21% of participants were classified as past users, having filled one or more prescriptions in the past year but none within the past 90 days.
(Enlarge Image)
Figure 2.
Frequency of anticholinergic prescription fills for study participants in the year before the index date, according to medication class.
A strong association was found between acute use of any anticholinergic medication and risk of community-acquired pneumonia, even after adjustment for confounding variables (Table 2, aOR = 2.55, 95% CI = 2.08–3.13), but past use of anticholinergic medications was not associated with risk of pneumonia after adjustment for all confounders (aOR = 1.19, 95% CI = 0.92–1.53). There was a significant association between chronic use of anticholinergics and risk of pneumonia (aOR = 2.07, 95% CI = 1.68–2.54), but this estimate was not significantly different from that of nonchronic use (1–2 fills in the prior year) (aOR = 1.83, 95% CI = 1.44–2.32).
When analyses were stratified according to anticholinergic potency (Table 2, Models 1a and 2a), risk estimates for use of high-potency anticholinergics were not substantively different from those for low potency anticholinergics for acute use (high potency, aOR = 2.63, 95% CI = 1.99–3.48; low potency, aOR = 2.54, 95% CI = 2.04–3.16) or chronic use (high-potency, aOR = 2.33, 95% CI = 1.78–3.13; low potency, aOR = 1.90, 95% CI = 1.51–2.40).
Because there was so much overlap between acute and chronic users (82% of acute users were also chronic users), post hoc analyses were conducted to try to further delineate whether the greater risk observed with acute use was primarily limited to recent initiators of anticholinergic medications. In this exploratory model, new users (with ≥1 fills in the 90 days before the index date but none in the remainder of the prior year) had a risk of pneumonia that was nearly five times as great as that of nonusers (aOR = 4.96, 95% CI = 3.47–6.19). New users made up a small proportion of acute users (n = 199, 15% of all acute users), and in this model acute users who were not new initiators still had a significantly greater risk of pneumonia (aOR = 2.23, 95% 1.81–2.76). When new users were restricted to those who had a fill within 30 days of the index date but no other fills in the prior year (n = 102), the association was yet stronger (aOR = 8.33, 95% CI = 5.03–13.8).
The exploratory dose-response analysis showed that the risk of CAP was similarly high for participants who filled one, two, three, four, or five or more anticholinergic prescriptions in the 90 days before the index date but that there was no clear pattern of greater risk with additional fills (Table 3).
Sensitivity Analyses
In the analysis excluding all anticholinergic opiates (Table 3), acute use of any anticholinergic medication was still associated with risk of pneumonia, although the risk was attenuated (aOR = 1.56, 95% CI = 1.29–1.89). Further adjustment for current or past use of any opiate, including nonanticholinergic opiates, did not significantly change the risk estimate (aOR = 1.51, 1.24–1.83). In the analysis excluding prescription fills for cough suppressants in the 5 days before the index date, the association between acute anticholinergic use and pneumonia risk persisted, although it was slightly attenuated (Table 3, aOR = 1.96, 95% CI = 1.58–2.44).