| JOURNAL HOME | CME HOME | THIS MONTH | PAST ISSUES | ETOC | COLLECTIONS |
| AUTHORS | REVIEWERS | EDITORIAL BOARD | FEEDBACK | RSS | HELP |
| ||||||||||||||
| ||||||||||||||
|
|
|
|
||||||||||||
|
||||||||||||||
Departments of *Anesthesia and Perioperative Care, and
Epidemiology and Biostatistics, University of California, San Francisco, California
Address correspondence and reprint requests to Dr. Leung, University of California-San Francisco, Mount Zion Medical Center, Department of Anesthesia and Perioperative Care, 1600 Divisadero St., Rm. C-355, San Francisco, CA 94115. Address e-mail to jmleung{at}itsa.ucsf.edu
 |
Abstract |
|---|
 Top Abstract IntroductionMethodsResultsDiscussionReferences |
|---|
70 yr of age undergoing noncardiac surgery. Potential clinical risk factors were measured and evaluated for their association with the occurrence of long-term postoperative mortality. Long-term survival was determined by using the Kaplan-Meier method. Multivariate correlates of survival were analyzed with the Cox proportional hazards model. The survival of the study group was also compared with the age- and gender-matched general United States population. Five hundred seventeen patients who survived the initial hospitalization were studied. The mean follow-up duration was 28.6 ± 12.8 mo. One hundred sixty-four of 517 patients (31.7%) were deceased at the time of follow-up. A history of cancer (hazard ratio [HR] 2.44, 95% confidence interval [CI] 1.78–3.38, P < 0.0001), ASA physical status >II (HR 2.27, 95% CI 1.61–3.21, P < 0.0001), neurologic disease (HR 1.59, 95% CI 1.13–2.24, P = 0.008), age (HR 1.42 per decade, 95% CI 1.11–1.81, P = 0.005), postoperative pulmonary complications (HR 2.41, 95% CI 1.30–4.48, P = 0.005), and renal complications (HR 6.07, 95% CI 2.23–16.52, P < 0.0001) were significant independent predictors of decreased long-term survival. Compared with the United States population, patients with complications had a greater increase in mortality risk in the first 3 mo after surgery (HR 7.3 versus general population) than those without complications (HR 2.9, P = 0.023). An effort to improve perioperative care delivery to elderly surgical patients must include measures to minimize in-hospital postoperative complications, particularly those involving the pulmonary and renal systems. IMPLICATIONS: Elderly surgical patients who have had an in-hospital postoperative complication have increased mortality, particularly in the first 3 mo after surgery. Efforts to improve perioperative care delivery to elderly surgical patients must include measures to minimize in-hospital postoperative complications, particularly those involving the pulmonary and renal systems.
|
Introduction |
|---|
|
TopAbstractIntroductionMethodsResultsDiscussionReferences |
|---|
65 yr of age are the group with the fastest growth in the United States (US) population. This aging population has tremendous implications for the practice of surgery and anesthesia. Aging increases the likelihood that a patient will require an operative procedure. Whereas approximately 12% of those aged 45–60 yr are operated on each year, this number increases to >21% in those aged 65 yr (1). With the expected increase in life expectancy, more elderly patients are expected to undergo surgery and anesthesia.
Although operative mortality has decreased in those aged 65 yr undergoing surgery, possibly because of advances in surgical, anesthetic, and monitoring techniques (2–4), perioperative morbidity continues to be more frequent in the elderly. Recent work by our group in elderly surgical patients undergoing noncardiac surgery demonstrated that 21% developed 1 or more in-hospital postoperative complications, involving primarily the cardiovascular, neurologic, and pulmonary systems (3). Whether the occurrence of nonfatal in-hospital postoperative complications decreases long-term survival is unknown. The primary objective of this study was to evaluate whether patients who have had such complications have shortened long-term survival, controlling for other clinical predictors.
|
Methods |
|---|
|
TopAbstractIntroductionMethodsResultsDiscussionReferences |
|---|
70 yr of age undergoing elective or emergent noncardiac surgery requiring general, regional, or general/regional combined anesthesia were prospectively studied. The study was conducted during a 1-yr period, from March 1997 to February 1998, at the University of California, San Francisco-Mount Zion Medical Center. This hospital is typical of a community medical center providing health care to a heterogeneous group of elderly patients living in the San Francisco neighborhood. Patients requiring local anesthesia, with or without IV sedation, and those undergoing cardiac surgery were excluded. The institutional committee on human research approved the inpatient part of the study to be exempted from the requirement of obtaining informed consent from patients because it involved the collection of existing records and/or diagnostic data that were publicly available. Informed consent from patients for the long-term follow-up study was obtained through subject contact letters that were sent to patients after hospital discharge at long-term follow-up.
In-Hospital Follow-Up
The preoperative demographic, medical history, intraoperative events, and predefined in-hospital postoperative complications were recorded for each patient who met the study entry criteria. The in-hospital course was followed daily by one of the investigators until discharge for the new occurrence of postoperative outcomes: cardiac complications included ischemic cardiac complications (new occurrence of chest pain, electrocardiographic changes, or cardiac enzyme changes), clinically diagnosed myocardial infarction, dysrhythmias, or heart failure. Respiratory complications included clinically diagnosed pulmonary edema, tracheal reintubation, pulmonary consolidation on chest radiograph, pneumothorax or clinically significant pleural effusion. Renal insufficiency indicators, chosen for their clinical significance, were defined as a new requirement for dialysis postoperatively, or increase of serum creatinine (preoperative value <1.5 mg/dL, increasing to a postoperative value >1.8 mg/dL or preoperative value >1.5 mg/dL which increased by 30% postoperatively above the preoperative value). Neurologic complications were defined as the new occurrence of transient ischemic attack or stroke, delirium, or confusion. Infection required documentation of a positive culture. Gastrointestinal complications were defined as bowel ischemia, perforation, bleeding, cholecystitis, pancreatitis, or clinical diagnosis of postoperative hepatic insufficiency. Thromboembolic events were defined as deep venous thrombosis or pulmonary embolism. Other postoperative outcomes measured included death, surgical complications, and reoperation during the same hospitalization. Two investigators verified all outcomes. In addition, the risk of surgery was classified into low, intermediate, and high risk (5).
Long-Term Follow-Up
A self-administered questionnaire was mailed in 2000, 2–3 yr after the initial operations, to those patients who survived the initial hospitalization. Survival status of the nonresponders was tracked by searching the hospital medical records and the National Death Index. The main study end-point was all-cause mortality up to 3 yr after the initial surgical procedure. The date of death was recorded for those who were deceased at long-term follow-up.
Data were analyzed by using Stata 5.0 (College Station, TX) and SAS software, PC computer release version 8 (SAS Institute, Inc., Cary, NC). Categorical variables with 2 levels such as a history of hypertension, diabetes mellitus, heart disease, or other co-morbid diseases, etc., were coded as 0 = absent and 1 = present and unknown (missing) if information could not be determined from medical record review. Nonordered categorical data with more than two levels (types of surgery or anesthesia, etc.) were entered as K-1 dummy variables (indicator variables).
Cox proportional hazards models were used to examine covariate effects on mortality while accounting for possible confounding and effect modification. The covariates considered in the models included nonfatal in-hospital postoperative complications, age, sex, race, body mass index, American Society of Anesthesiologists (ASA) classification, surgical risk, postoperative hematocrit, history of: cancer, cardiac diseases (including angina, myocardial infarction, congestive heart failure, valvular disease, or dysrhythmias), diabetes mellitus, hypertension, neurologic disease, renal disease, pulmonary disease, vascular disease, and smoking. The covariates with a univariate P < 0.10 were entered into the Cox proportional hazard model. Hazard ratios (HR) with 95% confidence interval (CI) were presented. P value < 0.05 was considered significant. The proportional hazard assumption for a given predictor of interest was assessed by the goodness-of-fit testing approach using the 
2 statistic.
Overall survival rates were summarized by using a Kaplan-Meier curve. For comparison, an age- and gender-matched curve for the US general population was calculated from published vital statistics (6). For each patient, the expected survival curve was calculated from the published death rates for a person of the same sex starting at the same age, and these were then averaged across all patients to obtain the overall comparison curve. Formal comparisons of patients, and subsets of patients, with the general population were performed by proportional hazards regression with age as the time scale, but with death rates in the general population treated as a known baseline hazard. This was accomplished by using the Statistical Analysis System’s LIFEREG procedure (SAS Institute) to fit exponential regressions, with the US population rates represented by 10,000 persons of each sex at each year of age from 70 to 99 yr (a total of 600,000 observations, enough to ensure that the model accurately reproduced the population baseline hazard despite the addition of the patients). Terms were included in the models to estimate how the risk of patients or subgroups of patients differed from the population-defined baseline for the same sex and age. Because the age of patients changed over the multiyear follow-up in this study, a single patient’s observation was split at each birthday into multiple, shorter observations to maintain the accuracy of the age covariate. Because surgical patients may become more similar to the general population as they survive longer past surgery, observations were also split into the first 3 mo after surgery, 3–12 mo after surgery, and >12 mo after surgery. Such splitting into multiple observations is the standard method for modeling time-dependent covariates in exponential regression. Subgroups were defined by in-hospital complication versus no in-hospital complication, and by age 70–79 yr at surgery versus age 80 yr at surgery. The risk increases in different subgroups were compared by using likelihood ratio tests comparing models with the subgroups assumed to have the same risk increases versus models allowing the risk increases to differ in the different subgroups. Note that current year of age is already controlled for in the models; the subgrouping by age at surgery was designed to investigate whether older patients were more highly selected, e.g., patients 80 yr of age more often forgo surgery unless their prognoses for survival are better than average. In all analyses, patients were not counted as being at risk of death until after discharge, because survival to discharge was required for inclusion in this study.
|
Results |
|---|
|
TopAbstractIntroductionMethodsResultsDiscussionReferences |
|---|
|
Univariate predictors of long-term mortality are shown in Table 2. In the Cox proportional hazards model, a history of cancer (HR 2.44, 95% CI 1.78–3.38, P < 0.0001), ASA classification >II (HR 2.27, 95% CI 1.61–3.21, P < 0.0001), a history of neurologic disease (HR 1.59, 95% CI 1.13–2.24, P = 0.008), and age (HR 1.42, 95% CI 1.11–1.81, P = 0.005) were significant independent predictors of decreased long-term survival (Table 3). The age effect was not different between men and women. The goodness-of-fit testing demonstrated that the data overall conformed to the proportional hazard assumption (P = 0.19). However, a test specifically for age did suggest a departure from the proportional hazard assumption (P = 0.02). This suggests that the effect of age may change with time since surgery, a possibility that we explore in more detail in the comparison to the general population below.
|
|
Table 4 shows the increased estimated risks of death for surgical patients compared with what would be expected for the general population of the same age and sex. As expected, the excess risk declined with time since surgery, both overall and for all subgroups examined. The risk for those with complications was greater than for those without complications in the first 3 mo postsurgery (relative risk [RR] 7.3 versus 2.9, P = 0.023) (Fig. 1), similar for those with and without complications from 3 to 12 mo (RR 2.4 versus 2.3, P = 0.95), and possibly increased for those with complications compared with those without beyond 12 mo (RR 1.9 versus 1.32, P = 0.16). Further subdividing by age at surgery showed that those aged 80 yr who had no complication had less risk than those aged 70–79 yr (P = 0.0074, likelihood ratio test over all 3 time periods after surgery), but age at surgery did have an effect among those with a complication (P = 0.57). The presence of a complication minimally affected the 70–79 age group (P = 0.59 overall, with actually decreased risk beyond 3 mo), but the impact was substantial in those aged 80 yr (P = 0.031).
|
|
|
Discussion |
|---|
|
TopAbstractIntroductionMethodsResultsDiscussionReferences |
|---|
Age Effect
Previous studies investigating long-term survival after surgery in elderly patients had focused on specific types of surgical interventions such as cardiac (7–9), orthopedic (10), or cancer surgery (11,12). Many of these studies, which reported comparable long-term results in geriatric patients as compared with their younger counterparts, included elderly patients who were highly selected for the specific surgery (13–15). In addition, many of these studies had small sample sizes precluding the determination of the additional effect of coexisting chronic medical conditions on long-term outcomes (16,17). A study using Medicare data from 1987 to 1990 of 24,461 patients aged 80 years who had undergone coronary artery bypass graft surgery, reported more frequent mortality rates for elderly patients during both the in-hospital and 30-days follow-up (18). In that study, although the initial surgical risk was high, octogenarians who underwent bypass surgery had long-term survival rate similar to that of the general US octogenarian population. However, that study only considered age as a predictor and did not include other clinical data such as the impact of postoperative complications on survival.
The proportion of patients who survived at three years in our study is comparable to that reported by studies that have followed patients of similar ages. For example, a study of patients with hip fractures (mean age 81 years) showed a survival of 65.5% during a mean follow-up period of 2.7 years (19). Another study of patients 75 years of age undergoing proximal femoral fracture treatment, showed a survival rate of 59% at 30-months follow-up (20). Other studies that followed patients who were highly selected for the planned surgery demonstrated higher survival rates (71%–93%) during a 2–3 year follow-up period (18,21,22).
Other Predictors of Long-Term Mortality
Our results are in agreement with a report in open-heart surgical patients demonstrating that survival was significantly impacted by the presence of co-morbid conditions (23). Specifically, a history of cancer was similarly found to predict long-term mortality by Yamamuro et al. (24) who studied a group of elderly patients undergoing repeat coronary artery bypass graft surgery. The relationship between co-morbid illness and survival has also been described in nonsurgical populations, including hospital-based studies (25,26) and patients followed in an outpatient geriatric clinic (27).
The number of investigations that included perioperative complications as potential risk factors for long-term mortality was limited, but two studies of elderly cardiac surgical patients (7,15) did report postoperative renal dysfunction to be an important predictor. One other study in octogenarians undergoing aortic valve replacement reported that perioperative complications decreased long-term survival but did not identify the types of postoperative complications that were associated with decreased long-term survival (28).
Timing of Highest Risk
The impact of postoperative complications on long-term survival was most evident in the first three months postoperatively. The predictive value of postoperative complications diminished with time, suggesting that other chronic conditions present in this age group may have replaced postoperative complications as more important causes of mortality as time elapsed. Patients aged 80 years do better than those aged 70–79 years if there are no complications. This finding may have been a result of the fact that those aged 80 years were more highly selected to undergo surgery than their younger counterpart, but they had worse survival if they developed a postoperative complication, erasing most or all of the selection advantage.
Clinical Implications
Our study demonstrates that, in addition to co-morbid conditions, in-hospital adverse outcomes shorten long-term survival in elderly patients who have undergone noncardiac surgery. Although our study demonstrated that postoperative pulmonary and renal complications increased the hazards of long-term mortality, we did not identify the responsible mechanisms. Further studies are necessary to elucidate the etiology of such complications and whether they may be modifiable. Our results also suggest that the medical goals for geriatric surgical patients should extend beyond the in-hospital period. Because the high-risk period seems to be the first three postoperative months, more intense surveillance of the patients who develop in-hospital postoperative complications is indicated.
Finally, our results suggest that old age alone should not be a limiting factor for surgery because long-term survival in those who have undergone surgery without complications is similar to their age- and gender-matched counterpart.
In conclusion, elderly surgical patients who have had an in-hospital postoperative complication have increased mortality, particularly in the first three months after surgery. Efforts to improve perioperative care delivery to elderly surgical patients must include measures to minimize in-hospital postoperative complications, particularly those involving the pulmonary and renal systems.
|
Acknowledgments |
|---|
The authors thank Jessica Watson and Alan Bostrom for assistance in statistical analysis.
|
References |
|---|
|
TopAbstractIntroductionMethodsResultsDiscussionReferences |
|---|
This article has been cited by other articles:
|
|
 |
|
  J. L. Daban, G. P. De Saint Maurice, E. Batjom, E. Falzone, and S. Ausset Postoperative myocardial damages are a key issue in patients' outcome after hip fracture Age Ageing, July 1, 2009; 38(4): 488 - 489. [Full Text] [PDF]
|
|
|
|
 |
|
 A. Qaseem, V. Snow, N. Fitterman, E. R. Hornbake, V. A. Lawrence, G. W. Smetana, K. Weiss, D. K. Owens, and for the Clinical Efficacy Assessment Subcommittee Risk assessment for and strategies to reduce perioperative pulmonary complications for patients undergoing noncardiothoracic surgery: a guideline from the American College of Physicians. Ann Intern Med, April 18, 2006; 144(8): 575 - 580. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
G. W. Smetana, V. A. Lawrence, and J. E. Cornell Preoperative pulmonary risk stratification for noncardiothoracic surgery: systematic review for the American College of Physicians. Ann Intern Med, April 18, 2006; 144(8): 581 - 595. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 M. Rusca, S. Proietti, P. Schnyder, P. Frascarolo, G. Hedenstierna, D. R. Spahn, and L. Magnusson Prevention of Atelectasis Formation During Induction of General Anesthesia Anesth. Analg., December 1, 2003; 97(6): 1835 - 1839. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
|
