|Home | About | Journals | Submit | Contact Us | Français|
To determine the prevalence and factors associated with use of potentially inappropriate medications (PIMs) in elderly surgical patients.
Retrospective cohort study.
379 acute care hospitals participating in the nationally representative Perspective database (2006-2008).
Individuals age 65 and older undergoing major inpatient gastrointestinal, gynecologic, urologic, and orthopedic surgery (n=272,351).
Medications were classified as PIMs using previously published criteria. Information about patient and provider characteristics and administration of PIMs was obtained from hospital discharge file data. We used logistic regression techniques to examine factors associated with use of PIMsin the perioperative period.
272,351 patients aged 65 or older underwent one of our target surgical procedures, and one quarter of patients received at least one PIM during their surgical admission. Meperidine was the most frequently prescribed PIM (37,855, 14% of patients). In adjusted analysis, PIM use was less likely as age advanced (adjusted OR 0.98 per year of age, 95%CI 0.97-0.98) and among men (adjusted OR 0.83, 95%CI 0.81-0.85). PIMs were most likely to be prescribed to patients cared for by an orthopedic surgeon (adjusted OR 1.22, 95%CI 1.08-1.40 v. general surgeons). Compared to patients undergoing surgery in the Northeast, patients in the West and South were more likely to receive a PIM (adjusted OR 1.79, 95%CI 1.02-3.16 and adjusted OR 2.24, 95%CI 1.38-3.64, respectively).
Receipt of PIMs in elderly surgical patients is common and varies widely across providers, geographic regions, and by patient characteristics. Interventions aimed at reducing the use of PIMs in the perioperative period should be considered in quality improvement efforts.
As the population ages, an increasing number of elderly patients will be candidates for major surgery. Currently, over 50% of orthopedic, urologic, and general surgery procedures are performed in patients age 65 and older.1 Numerous studies have documented that morbidity and mortality after major surgery increases dramatically with age.2-4 For this reason, there is increasing interest in identifying potential processes of care that can improve surgical care in the growing population of elderly patients.5, 6
Although there is great enthusiasm for improving surgical care in the elderly, there is little agreement about where to start. A recent review of potential quality indicators for elderly surgical patients identified over ninety candidate measures.5 One potential quality indicator – the avoidance of high risk medications – is easily measured and an attractive target for quality improvement. An explicit list of potentially inappropriate medications (PIMs) in the elderly was initially proposed by Beers et al (ref) and subsequently refined by Zhan et al.7. Zhan criteria medications include 33 agents considered inappropriate for use in individuals age 65 and older. PIM use has been associated with adverse drug reactions in elders such as delirium, falls, and urinary retention.8 This list contains several types of medications – including sedatives and pain medications – that are commonly used in surgical patients. Although PIM use has been studied in elder patients hospitalized with medical diagnoses,8-11 it has not been studied in the elder surgical population.
Currently, the prevalence and risk factors for perioperative PIM use in elder surgical patients is unknown. For this reason, we used data from a nationally representative sample of US hospitals (Perspective database) to study PIM use among patients age 65 and older undergoing major gastrointestinal, urologic, gynecologic, and orthopedic surgery. To better understand risk factors for receipt of a PIM in the perioperative period, we examined patient and provider characteristics associated the PIM use.
We used the 2006-2008 Perspective database (Premier Inc., Charlotte, North Carolina) to identify a retrospective cohort of surgical patients age 65 and older. The Perspective database was developed by Premier Healthcare Informatics for measuring quality and health care utilization. Premier hospitals are located in all regions of the United States, are predominantly small to mid-size non-teaching facilities, and serve a largely urban patient population. Using ICD-9-CM procedure codes defined by the Healthcare Cost and Utilization Project Clinical Classification System, we identified patients age 65 and older undergoing major inpatient gastrointestinal, urologic, gynecologic, and orthopedic surgery.(http://www.hcupus.ahrq.gov/toolssoftware/ccs/ccs.jsp). The Healthcare Cost and Utilization Project Clinical Classification System is a tool for clustering ICD-9-CM procedure codes into clinically meaningful categories. To capture patients undergoing elective inpatient surgery, we excluded patients admitted through the emergency department, transferred from an acute care hospital or had a length of stay shorter than 3 days. Our final cohort included 272,351 who underwent elective inpatient surgery.
Patient demographic data were obtained from the hospital discharge file data. We used ICD 9 diagnosis codes to assess patient comorbidities using methods described by Elixhauser et al.12 Comorbidites with prevalence greater than 1% were included in the analysis. Hospital characteristic including teaching status, location type (rural v. urban), and geographic region were obtained from the Premier database. Information about the specialty of the attending physician and inpatient consultation by a non-surgeon (internist, hospitalist, family practitioner, or geriatrician) was obtained from discharge abstracts.
Perspective contains a date-stamped log of all medications charged for during each hospitalization. A charge for a medication any time between the day of surgery and the day of discharge was considered postoperative receipt of a drug. We defined potentially inappropriate medication use as receipt of one or more of the 33 drugs classified as potentially inappropriate in individuals age 65 and older as defined by Zhan et al.7 In a review by expert panel, PIMs were classified in 3 categories – ‘always avoid’, ‘rarely appropriate’, and ‘some indications but often misused. Because administrative data are limited in their ability to capture all clinical conditions, we were not able to determine the specific indication for each medication administered. For this reason, we did not include medications classified as having ‘some indications’ in the elderly in our definition of PIM exposure used in our risk factor analysis. Summary statistics of patient and provider variables presented as proportions for categorical variables and means and standard deviations for continuous variables. Frequency of PIM use was also measured at the individual hospital level among hospitals with at least 50 surgical patients during the study period.
We first described study patients and providers using univariable methods. Multivariable GEE models (Proc Genmod in SAS) were used to adjust for clustering of patients within hospitals and calculate adjusted odds ratios and 95% confidence intervals for receipt of PIMs. Models were constructed using manual variable selection methods. Covariates (confounding factors) were selected for inclusion if they were associated with the outcome at p<0.01. All analyses were carried out using SAS version 9.2 (SAS Institute, Inc. Cary, NC). The institutional review board at UCSF approved our study.
272,351 patients age 65 and older underwent one of the nine procedures examined. The most frequently performed procedures were arthroplasty (n=139,958), hip replacement (n=67,401) and colorectal resection (n=40,400). The mean age was 75 and 35% were male. Seventy five percent were white. The most common comorbidities were hypertension (70%), diabetes (21%), and iron deficiency anemia (19%).
The majority of patients (71%) were under the care of an orthopedic surgeon during their surgical hospitalization. General surgeons were the attending physicians for over 40,000 patients. Overall, 26% of surgical patients received a medical consultation after surgery. Hospitals were most frequently located in the southern United States (43%) and were in an urban setting (88%).
Overall, over half of patients undergoing surgery received a PIM during their surgical hospitalization. Twenty five percent received a PIM categorized as ‘always avoid’ or ‘rarely appropriate’. Among patients receiving an ‘always avoid’ medication, meperidine was the most frequently prescribed medication with 37,855 (14%) patients receiving the drug after surgery. In the ‘rarely appropriate’ category, the most commonly prescribed medications were sedatives and muscle relaxants – diazepam (3%) and cyclobenzaprine (3%). Two medications commonly used in surgical patients – diphenhydramine and promethazine – were given to a large number of patients (59,967 (22%) and 52,027 (19%), respectively). There was substantial variation in frequency of ‘always avoid’ or ‘rarely appropriate’ PIM use across individual hospitals (Figure 1). Over 40% of all hospitals administered a PIM to fewer than 20% of elderly patients. Very high frequency PIM use (>80%) was observed in only 4 hospitals. Variation in meperidine use at the hospital level was pronounced. In over a quarter (26%) of all hospitals, fewer than 5% of patients received meperidine during their surgical hospitalization. In nearly half of all hospitals (47%), fewer than 10% of patients received meperidine.
In multivariate analysis, several patient factors were associated with receipt of a “always avoid’ or ‘rarely appropriate’ PIM. Men were less likely to receive a PIM than women (adjusted odds ratio (AOR) 0.83, 95%CI 0.81-0.85). Compared to younger patients (age 65-69), patient in all older age groups were less likely to receive a PIM – 30% v. 26% for age 70-74 (AOR 0.89, 95%CI 0.87-0.91), 24% for age 75-79 (AOR 0.82, 95%CI 0.79-0.85), 22% for age 80+ (AOR 0.71, 95%CI 0.68-0.74). Both black and Hispanic elders were less likely to receive a PIM than white elders (AOR 0.80, 95%CI 0.76-0.85 and AOR 0.88, 95%CI 0.81-0.96, respectively). Several individual patient comorbidities were associated with small increases in the risk of receiving a PIM. Patients diagnosed with COPD (AOR 1.09, 95%CI 1.07-1.12), congestive heart failure (AOR 1.06, 95%CI 1.02-1.10), hypothyroidism (AOR 1.05, 95%CI 1.03-1.07), depression (AOR 1.10, 95%CI 1.07-1.14), psychoses (AOR 1.17, 1.09-1.1.25), anemia (AOR 1.04, 95%CI 1.01-1.07), weight loss (AOR 1.24, 95%CI 1.15-1.34), and rheumatoid/collagen disorders (AOR 1.08, 95%CI 1.03-1.14) were more likely to receive a PIM than patients who were not diagnosed with these conditions. Compared to patients without diabetes, diabetics were less likely to receive a PIM (AOR 0.94, 95%CI 0.92-0.96).
There was substantial variation in PIM use across physician and hospital type. PIM was common among patients under the care of orthopedic surgeons (26%) 22% general surgeons (22%), gynecologic surgeons (23%) and urologic surgeons (19%). Compared with general surgery specialty, orthopedic surgery specialty was the single surgical specialty significantly associated with PIM administration after adjustment for other patient and provider characteristics (AOR 1.22, 95%CI 1.08-1.40). Rates of PIM use were highest among patients under the care of non-surgeon physician (32%, AOR 1.40, 95%CI 1.26-1.55). The presence of a medical consultation was associated with increased likelihood of receiving a PIM (AOR 1.09, 95%CI 1.04-1.14). Among all hospital characteristics examined, only geographic region was independently associated with increased use of PIM in the elderly. Rates of PIM use varied across regions -- 14% in the Northeast, 23% in the Midwest, 26% in the West, and 30% in the South. Compared to hospitals in the Northeast, hospitals in the West and South and were more likely to prescribe a PIM during the postoperative period (AOR 1.79, 95%CI 1.02-3.16 and AOR 2.24, 95%CI 1.38-3.64, respectively).
In this retrospective cohort of elderly patients undergoing elective major surgery, we found a very high prevalence of PIM use. In addition, we found substantial variation across patients and providers. Potentially inappropriate medication was associated with younger age, female gender, white race, and several individual comorbid conditions. Two provider characteristics – orthopedic specialty and medicine consultation – were associated with increased PIM prescribing. Elderly patients who underwent surgery in the South were more likely to receive a PIM compared to patients who underwent surgery in other geographic regions.
Although our study examines PIM use among surgical patients, our findings are consistent with prior studies of older adults hospitalized with medical diagnoses.7-10 Zhan et al found that over 20% of community-dwelling elderly patients in a nationally representative survey had received at least one PIM.7 In a retrospective study of elders hospitalized with medical diagnoses, Rothberg et al found that nearly half of the patients received a PIM during their hospital stay.9 Similar to our findings, they found that men and patients age 85 and older were less likely to receive a PIM. They also found that the proportion patients receiving PIMs varied regionally and was highest in the Southern United States. Studies of PIM use in community dwelling elders have also found regional variation in prescribing patterns.9, 13 Variation in PIM use by physician specialty has also been documented among patients admitted with medical diagnoses where cardiologists and internists are more likely to prescribe PIMS than geriatricians.9 Smaller studies examining meperidine use in hospitalized patients age 65 and older found that surgical patients were 7 times as likely than medical patients to receive meperidine.14
Our study has several limitations. First, some PIMs prescribed in the inpatient setting may in fact be medications used chronically and their use may not be wholly attributable to the surgical hospitalization. Because the Perspective database does not contain information about outpatient care, we were unable determine if the prescribed PIMs were medications that patients were receiving prior to surgery or if they were medications initiated for the first time in the postoperative period. The most frequently prescribed PIMs, however, were medications often prescribed for postoperative treatment for pain, insomnia, and nausea and are unlikely to reflect continuation of outpatient medication regimens. Second, lists of medications considered inappropriate in the elderly were developed and studied in community dwelling elder and nursing home residents. While the Zhan criteria were developed with outpatients in mind, the drug use recommendations apply strongly to inpatient care. For example, a robust literature shows that meperidine has higher rates of adverse effects than other opioids15-17, its use is discouraged by the health policy organizations.18, 19 Consequently it receives the most stringent warning of “never appropriate” under the Zhan criteria. Other Zhan criteria drugs, also appropriate in selected settings, are very often misused in the hospital environment. For example, benzodiazepines, sedating antihistamines, and strongly anticholinergic tricyclic antidepressants, which are commonly used to help hospitalized patients sleep, are well-recognized risk factors for delirium and a host of other serious outcomes in surgical patients and should only be used rarely and with great caution in hospitalized elders.20 Finally, adverse drug reactions (i.e. delirium, falls) are the most common complications of to capture these complications,21 we are unable to measure direct harm from PIM use in this study.
Our study has important implications and suggests opportunities for quality improvement in surgical care in the elderly.Efforts to identify targets for quality improvement in geriatric perioperative care are essential to optimize outcomes in this vulnerable population.Although we found variation across patients and providers, PIM use was frequent in all subgoups of patients and provider types suggesting the need for broad reaching efforts to minimize use of potentially harmful medications in the perioperative period. For medications most frequently prescribed in elder surgical patients – meperidine, diphenhydramine, and promethazine – there are effective and safer alternative analgesics, sedatives, and anti-emetics available. Moreover, hospital-specific rates of meperidine use – the most frequently prescribed ‘always avoid’ PIM – are low in the majority of hospitals studied, suggesting that individual institutions have successfully reduced inappropriate prescribing. Interventions aimed at optimizing medication choice in the elderly after surgery – e.g. provider education, computerized medication order entry alerts, and formulary restriction– could be implemented.22 These interventions, in conjunction with other processes of care aimed at optimizing surgical care in the elderly, could be used to develop standardized care protocols that could optimize care in the growing population of older surgical patients.
Funding/Support : Work on this study was funded in part by Dr. Auerbach's National Heart, Blood, Lund Institute Award (K24HL098372). Dr. Finlayson was supported by a National Institute on Aging/Paul B. Beeson Clinical Scientist Development Award in Aging (K08AG028965). Dr Steinman was supported by the National Institute on Aging and the American Federation for Aging Research (K23AG030999).
Sponsor's Role: The funding sources had no role in the design, methods, subject recruitment, data collections, analysis and preparation of paper.
|Elements of Financial/Personal Conflicts||Emily Finlayson||Judith Maselli||Michael Steinman||Michael Rothberg||Peter Lindenauer||Andrew Auerbach|
|Employment or Affiliation||X||X||X||X||X||X|
*Authors can be listed by abbreviations of their names.
Study concept and design: Finlayson, Auerbach
Acquisition of subjects and/or data: Auerbach
Analysis and interpretation of data: Finlayson, Maselli, Steinman, Rothman, Lindenauer, Auerbach
Preparation/critical revision of manuscript: Finlayson, Steinman, Rothman, Lindenauer, Auerbach
Conflict of Interest: All authors report no conflict of interest related to any aspect of the study.