PPIs and Functional Decline in Older Adults
The methods of the PharmacosurVeillance in the elderly Care Study have been previously described. Briefly, all individuals aged 65 and older consecutively admitted to participating wards (11 acute care medical wards, 3 long-term care and rehabilitation units) from April 1 to June 30, 2007, were asked to participate in the study. Overall, 762 individuals were screened in the survey period, but 72 (9.4%) refused to participate. To avoid setting-dependent heterogeneity, individuals primarily admitted to long-term rehabilitation (3 of the 14 centers participating in the PharmacosurVeillance in the elderly care studies) (N = 159), but not individuals discharged from the acute care hospital to intermediate units or seeking intermediate care after being discharged at home, were excluded from the present study. Individuals who died during their hospital stay (n = 25), those receiving a histamine-2 receptor antagonist prescription at discharge (n = 15), those completely dependent at the time of discharge (n = 54), and those who died during follow-up (n = 36) were also excluded, leaving a final sample of 401 participants with complete baseline and follow-up data for study analyses.
All participants gave written informed consent. Information was collected on demographic, socioeconomic, and clinical characteristics, with a special emphasis on pharmacological therapy and comprehensive geriatric assessment. Once discharged, participants underwent follow-up visits every 3 months for 1 year. All participants or a relative or caregiver were contacted by telephone to plan the follow-up visit. At each follow-up visit, information regarding vital status, ADLs, adverse drug reactions (ADRs), and changes in drug prescriptions (date of start or withdrawal for each drug) were collected. The ethical committee of the Italian National Research Center on Aging, Ancona, Italy, approved the study protocol.
Because older adults discharged from the hospital may have dramatically greater care needs during the year after hospitalization than before, the outcome of the present study was the loss of independence in at least one ADL from the discharge examination to the 12-month follow-up visit. ADL scale items included transferring, dressing, feeding, toileting, and bathing. Performance on individual ADLs was scored according to five levels of difficulty: without difficulty, with difficulty but without help, help for only part of the activity, help for total activity, and not able to perform. Participants needing help for total activity or not able to perform were considered dependent. Information about functional status was collected in interviews with participants and caregivers during the visits.
Drugs were coded using the Anatomical and Therapeutic Classification (ATC). For each prescription drug, dates of start and withdrawal, type, and dosage were recorded. Participants taking PPIs were identified according to ATC code A02BC. Participants receiving a PPI prescription at the time of discharge were defined as users of PPIs. An analytical variable was created to investigate the association between duration of PPI therapy and study outcome as no use, less than 3 months, or 3 months or more. An analytical variable was also created to investigate the relationship between PPI dosage and study outcome. Participants receiving omeprazole 10 to 20 mg/d, pantoprazole 10 to 20 mg/d, lansoprazole 15 mg/d, rabeprazole 10 mg/d, or esomeprazole 20 mg/d were classified as receiving low-dose PPIs; participants receiving omeprazole 40 mg/d, pantoprazole 40 mg/d, lansoprazole 30 mg/d, rabeprazole 20 mg/d, or esomeprazole 40 mg/d were classified as receiving high-dose PPIs.
Variables considered in the analyses included factors known to affect prognosis in elderly populations: age, sex, cognitive impairment (age- and education-adjusted Mini-Mental State Examination score), depression (Geriatric Depression Scale score), dependency in ADLs, continence, nutritional status (body mass index, serum albumin), and overall comorbidity (Cumulative Illness Rating Scale (CIRS) comorbidity score).
Information on the number of drugs prescribed at discharge and discharge prescription of antithrombotics and nonsteroidal anti-inflammatory drugs was also included in the analysis. Finally, cardiovascular diseases (including coronary artery disease, carotid and peripheral arterial disease, atrial fibrillation, venous thrombosis, pulmonary embolism), heart failure, prior stroke or transient ischemic attack, GERD, peptic ulcer, diarrhea, infections, fractures, chronic obstructive pulmonary disease, chronic kidney disease (CKD), and anemia were also considered as potential confounders.
To minimize potential confounding due to the observational design of the study, a propensity score matching analysis was performed. A propensity score was estimated for the likelihood of receiving PPIs at discharge using multivariable logistic regression analysis conditional on all above-listed covariates at baseline. Each case (PPI user) was matched to a control on the basis of the propensity score. All analyses were performed before and after propensity score matching as follows.
First, users of PPIs were compared with nonusers with regard to study variables. Therefore, participants losing independence in at least one ADL were compared with those not experiencing functional decline during follow-up. The chi-square test or one way analysis of variance was used when appropriate. Finally, logistic regression models adjusted for age, sex, and confounders known to increase the risk of functional decline in older populations, including hypoalbuminemia, cognitive impairment, number of lost ADLs at discharge, incontinence, and CIRS comorbidity, were constructed. Cardiovascular disease, heart failure, prior stroke or transient ischemic attack were also included in multivariable models because they may increase the risk of functional decline or the use of PPIs in individuals taking antithrombotics. CKD was also considered to be a potential confounder because of its effect on functional status in older adults.
The analysis was repeated after excluding individuals with established indications for PPI prescription (peptic ulcer, GERD, long-term use of nonsteroidal anti-inflammatory drugs, and combined treatment with antithrombotics and systemic steroids). The analysis was also repeated after including individuals who died during follow-up (n = 36), on the assumption that death equates with complete loss of independency. Finally, a sensitivity analysis was performed using control drug classes. Two drug classes were chosen (beta-blockers and antipsychotics) to be included in the logistic regression models instead of PPIs. It was reasoned that beta-blockers are more frequently prescribed in individuals with moderate to severe heart failure, a condition associated with a high risk of functional decline. Antipsychotics were chosen because of their known adverse effects on independency. All analyses were performed using SPSS statistical software for Windows (SPSS, Inc., Chicago, IL).
Methods
The methods of the PharmacosurVeillance in the elderly Care Study have been previously described. Briefly, all individuals aged 65 and older consecutively admitted to participating wards (11 acute care medical wards, 3 long-term care and rehabilitation units) from April 1 to June 30, 2007, were asked to participate in the study. Overall, 762 individuals were screened in the survey period, but 72 (9.4%) refused to participate. To avoid setting-dependent heterogeneity, individuals primarily admitted to long-term rehabilitation (3 of the 14 centers participating in the PharmacosurVeillance in the elderly care studies) (N = 159), but not individuals discharged from the acute care hospital to intermediate units or seeking intermediate care after being discharged at home, were excluded from the present study. Individuals who died during their hospital stay (n = 25), those receiving a histamine-2 receptor antagonist prescription at discharge (n = 15), those completely dependent at the time of discharge (n = 54), and those who died during follow-up (n = 36) were also excluded, leaving a final sample of 401 participants with complete baseline and follow-up data for study analyses.
All participants gave written informed consent. Information was collected on demographic, socioeconomic, and clinical characteristics, with a special emphasis on pharmacological therapy and comprehensive geriatric assessment. Once discharged, participants underwent follow-up visits every 3 months for 1 year. All participants or a relative or caregiver were contacted by telephone to plan the follow-up visit. At each follow-up visit, information regarding vital status, ADLs, adverse drug reactions (ADRs), and changes in drug prescriptions (date of start or withdrawal for each drug) were collected. The ethical committee of the Italian National Research Center on Aging, Ancona, Italy, approved the study protocol.
Because older adults discharged from the hospital may have dramatically greater care needs during the year after hospitalization than before, the outcome of the present study was the loss of independence in at least one ADL from the discharge examination to the 12-month follow-up visit. ADL scale items included transferring, dressing, feeding, toileting, and bathing. Performance on individual ADLs was scored according to five levels of difficulty: without difficulty, with difficulty but without help, help for only part of the activity, help for total activity, and not able to perform. Participants needing help for total activity or not able to perform were considered dependent. Information about functional status was collected in interviews with participants and caregivers during the visits.
Drugs were coded using the Anatomical and Therapeutic Classification (ATC). For each prescription drug, dates of start and withdrawal, type, and dosage were recorded. Participants taking PPIs were identified according to ATC code A02BC. Participants receiving a PPI prescription at the time of discharge were defined as users of PPIs. An analytical variable was created to investigate the association between duration of PPI therapy and study outcome as no use, less than 3 months, or 3 months or more. An analytical variable was also created to investigate the relationship between PPI dosage and study outcome. Participants receiving omeprazole 10 to 20 mg/d, pantoprazole 10 to 20 mg/d, lansoprazole 15 mg/d, rabeprazole 10 mg/d, or esomeprazole 20 mg/d were classified as receiving low-dose PPIs; participants receiving omeprazole 40 mg/d, pantoprazole 40 mg/d, lansoprazole 30 mg/d, rabeprazole 20 mg/d, or esomeprazole 40 mg/d were classified as receiving high-dose PPIs.
Variables considered in the analyses included factors known to affect prognosis in elderly populations: age, sex, cognitive impairment (age- and education-adjusted Mini-Mental State Examination score), depression (Geriatric Depression Scale score), dependency in ADLs, continence, nutritional status (body mass index, serum albumin), and overall comorbidity (Cumulative Illness Rating Scale (CIRS) comorbidity score).
Information on the number of drugs prescribed at discharge and discharge prescription of antithrombotics and nonsteroidal anti-inflammatory drugs was also included in the analysis. Finally, cardiovascular diseases (including coronary artery disease, carotid and peripheral arterial disease, atrial fibrillation, venous thrombosis, pulmonary embolism), heart failure, prior stroke or transient ischemic attack, GERD, peptic ulcer, diarrhea, infections, fractures, chronic obstructive pulmonary disease, chronic kidney disease (CKD), and anemia were also considered as potential confounders.
Analytical Approach
To minimize potential confounding due to the observational design of the study, a propensity score matching analysis was performed. A propensity score was estimated for the likelihood of receiving PPIs at discharge using multivariable logistic regression analysis conditional on all above-listed covariates at baseline. Each case (PPI user) was matched to a control on the basis of the propensity score. All analyses were performed before and after propensity score matching as follows.
First, users of PPIs were compared with nonusers with regard to study variables. Therefore, participants losing independence in at least one ADL were compared with those not experiencing functional decline during follow-up. The chi-square test or one way analysis of variance was used when appropriate. Finally, logistic regression models adjusted for age, sex, and confounders known to increase the risk of functional decline in older populations, including hypoalbuminemia, cognitive impairment, number of lost ADLs at discharge, incontinence, and CIRS comorbidity, were constructed. Cardiovascular disease, heart failure, prior stroke or transient ischemic attack were also included in multivariable models because they may increase the risk of functional decline or the use of PPIs in individuals taking antithrombotics. CKD was also considered to be a potential confounder because of its effect on functional status in older adults.
The analysis was repeated after excluding individuals with established indications for PPI prescription (peptic ulcer, GERD, long-term use of nonsteroidal anti-inflammatory drugs, and combined treatment with antithrombotics and systemic steroids). The analysis was also repeated after including individuals who died during follow-up (n = 36), on the assumption that death equates with complete loss of independency. Finally, a sensitivity analysis was performed using control drug classes. Two drug classes were chosen (beta-blockers and antipsychotics) to be included in the logistic regression models instead of PPIs. It was reasoned that beta-blockers are more frequently prescribed in individuals with moderate to severe heart failure, a condition associated with a high risk of functional decline. Antipsychotics were chosen because of their known adverse effects on independency. All analyses were performed using SPSS statistical software for Windows (SPSS, Inc., Chicago, IL).