Body mass index and albumin levels are prognostic factors for long-term survival in elders with limited performance status

Elderly long-term care facility residents typically have musculoskeletal conditions that may lead to long-term disability and increased mortality. Our main objective was to explore the relationship between body mass index (BMI), albumin levels, and mortality in elderly individuals with limited performance status. Among 182 participants (mean age, 78.8 years; 57% women), 11%, 64%, and 25% had serum albumin levels of <2.8, 2.8-3.5, and >3.5 g/dL, respectively. After multivariate adjustments, diastolic blood pressure >90 mmHg was associated with all-cause mortality [hazard ratio (HR) = 2.08, 95% confidence interval (CI) = 1.13-3.82; P = 0.018]. In addition, BMI <18.5 kg/m2 and albumin level <2.8 g/dL associated with higher mortality than BMI = 18.5-24 kg/m2 and albumin level > 3.5 g/dL (HR = 1.80, 95% CI = 1.11-2.94 and HR = 2.54, 95% CI 1.22-5.30, respectively; P = 0.018 and 0.013, respectively). Highest mortality was noted in participants with albumin levels <2.8 g/dL and BMIs <18.5 kg/m2 (HR = 6.12, 95% CI = 1.85-20.21, P = 0.003). Combined hypoalbuminemia (albumin level < 2.8 g/dL) and low BMI (<18.5 kg/m2) may be a useful prognostic indicator of high mortality risk in elderly individuals with limited performance status.


INTRODUCTION
Body mass index (BMI) is a simple anthropometric measure of nutritional status, but it is also an important mortality indicator among hospitalized patients [1]. In the elderly population, low BMI, which was defined as a BMI <18.5 kg/m 2 , is more related to increased mortality than is being overweight [2,3].
The serum albumin level is another important variable for assessing the nutritional status of patients with acute or chronic illness [4]. Hypoalbuminemia, defined as serum albumin levels <3.5 g/dL, is common among hospitalized patients [5] and is associated with mortality and increased risk for various diseases [6], including heart failure [7], cirrhosis [8], and nephrotic syndrome [9]. In addition, poor functional recovery from stroke was noted in a population of patients with low BMIs and serum albumin levels [10]. This suggests, the combination of BMI and serum or urine albumin levels may be predictive of allcause mortality in the general population [11].
Aging is a major risk factor adversely affecting life expectancy among the elderly [12,13]. Many of these individuals have a limited performance status (Eastern Cooperative Oncology Group (ECOG) [14] score >2) and use a wheelchair or are bedridden in long-term facilities. Indexes such as the Minimum Data Set (including: frailty index, nutrition, physical function and cognitive function) have been used to predict mortality in nursing homes [15][16][17]. However, most of those studies have focused on short-term mortality and used qualitative prognostic information instead of quantitative data [18]. We evaluated the utility of the combination of serum albumin and BMI for predicting mortality in elderly residents with limited performance status in long-term care facilities (ECOG score >2) and assessed whether a cut-off value could assist physicians in clinical settings.

RESULTS
In total, 228 individuals (42% men) living in eight different long-term care facilities met our inclusion criteria for BMI and serum albumin at the end of a 6year follow-up period. Among those, 46 (20%) were excluded due to ECOG scores <2 (Figure 1). Of the remaining 182 participants, 20 (11%) had serum albumin levels <2.8 g/dL, while 117 (64%) had serum albumin levels of 2.8-3.5 g/dL and 45 (25%) had levels >3.5 g/dL. Table 1 lists the basic characteristics of the participants segregated based on their albumin levels. The total number of deaths during follow-up was 139 (76%). The mean age of the participants with an ECOG score ≥2 was 78.8 years (57% women). Compared with the other two groups, participants with albumin levels <2.8 g/dL were older and had a significantly lower waist circumference (WC) and BMI as well as lower hemoglobin, albumin, total cholesterol, and triglyceride (TG) levels.
From year 1 of the follow-up period onward, participants with both BMIs <18.5 kg/m 2 and albumin levels <2.8 mg/dL had a lower survival rate (Figure 2A and 2B). On the other hand, the survival rate did not differ significantly between participants with albumin levels of 2.8-3.5 mg/dL or >3.5 mg/dL. The highest   Table 3).

DISCUSSION
Low serum albumin levels not only increase the risk of postoperative organ dysfunction, they are also predictive of mortality in the healthy elderly [19,20]. BMIs are inversely related to mortality in the elderly, and the combination of albumin level and BMI is used to predict risk of mortality in cardiac surgery and dialysis patients [21][22][23]. Our cohort study demonstrated that among elderly participants with limited performance status (ECOG ≥2), the mortality rate was 6-fold when the serum albumin level was <2.8 mg/dL and BMI was <18.5 kg/m 2 .

BMI
Low BMI (<18.5 kg/m 2 ) is predictive of mortality in the elder population. Large cohort studies of elders in Australia [24], Europe [25], and Taiwan [26] showed that being underweight associates with higher all-cause mortality risk. Similarly, we found an association between lower BMI and increased mortality risk in elders with limited musculoskeletal performance status ( Figure 2B). We also found that mortality among participants with BMIs <18.5 kg/m 2 and serum albumin levels <2.8 g/dL is higher than among those with BMIs <18.5 kg/m 2 and albumin levels >2.8 g/dL (Table 3A). This is consistent with the finding of Engelman et al. [22], who reported that elders with lower BMIs and serum albumin levels (<2.8 g/dL) showed greater mortality than those with lower BMIs but normal serum albumin levels. This suggests that low BMI alone does not increase the risk of all-cause mortality.

Albumin
A large cross-sectional longitudinal cohort study of a general elderly population (mean age = 73 years) found an association between albumin levels <3.6 g/dL and higher mortality rates (HR = 2.8, 95% CI = 2.5-3.3) [26]. We found a similar association among limited-performance elders (mean age = 78.8 ± 7.5) when serum albumin levels were <3.5 mg/dL (Figure 2A). In addition, a low serum albumin level is a prognostic factor in cardiovascular disease [27], malnutrition [28], inflammatory reactions [29] and nephrotic syndrome [30] in elders. Similarly, we found that lower serum albumin levels associate with an increased risk of death (HR = 2.54, 95% CI = 1.22-5.30, P = 0.013) after adjusting for several potential confounders, including age, renal function, and BP (Table 2). Elderly patients undergoing surgical intervention are at higher risk than younger patients for such postoperative morbidities as pneumonia (OR = 1.76, 95% CI = 1.34-2.33, P <0.001) and sepsis (OR = 2.29, 95% CI = 1.22-4.30, P = 0.010) [31]. Consequently, a lower serum albumin may reflect physiological impairments such as inflammation, malnutrition or malabsorption in older adults [32]. Importantly, the survival rate among older patients depends not only on the aforementioned physiological impairments, but also on such preexisting conditions as limited performance status, cognitive impairment, and low BMI [33].

Albumin and BMI
Combined low albumin levels and BMIs in the elderly associate with poor prognoses in several diseases [10,[34][35][36][37]. Mechanisms to explain lower serum albumin [35,38] and BMI [39] may involve chronic inflammation, which appears to be a common factor in various diseases affecting the elderly, including cancer, cardiovascular disease, diabetes, cognitive impairment, frailty, sarcopenia [40,41]. Compared to individuals without inflammation or obesity, patients exhibiting chronic inflammation have a greater risk of mortality [HR = 2.68, 95%CI = 2.14-3.35] when BMI is <21.5 kg/m 2 [42]. Moreover, inflammation may be the main cause of reduced serum albumin levels [43], and increasing the BMI can decrease the negative consequences of inflammation [42]. Our finding that mortality was highest when serum albumin was <2.8 g/dL and BMI was <18.5 kg/m 2 is consistent with those earlier findings (Table 3A).
We also found that the risk of mortality increased about 2-fold in all categories of BMI and albumin when DBP was >90 mmHg (Table 3B). Likewise, Athanase et al. reported that all-cause mortality in the elderly is increased when DBP is >80 mmHg [44]. Thus, hypertension alone can increase the risk of all-cause mortality. However, a recent large prospective study of the general population reported that the combination of low-grade albuminuria and BMI predict mortality more accurately than low-grade albuminuria or BMI alone [11]. Our findings in elders with limited performance   [45]. Albuminuria can be affected by both physiological and abnormal stressors, such as exercise [46], posture [47], urinary tract infection [48], and myocardial ischemia [49]. Serum albumin is less affected by these conditions. For longterm medical care, therefore, combining BMI, albumin levels, and DBP as health condition indicators may be suitable for elderly individuals with limited performance status

Strengths and limitations
A key strength of our study is the relatively long followup period for elderly individuals with limited performance status. However, there were also several limitations. First, the number of participants was small. Second, the study did not use recent biological data or anthropometric indices. Therefore, changes in those factors could influence the results. Third, the height of some patients could not be measured, so arm length was used to estimate their BMI, which may have introduced bias into our results. Finally, because all of our participants were Taiwanese, our findings may not be applicable to other ethnic groups.

CONCLUSION
Combined hypoalbuminemia (<2.8 g/dL) and low BMI (<18.5 kg/m 2 ) may be a useful prognostic indicator for predicting high risk of mortality in elderly individuals with limited performance status (ECOG score ≥2). These findings could aid in early identification of relevant vulnerable populations.

Participants
The nutritional status of elderly individuals in long-term care facilities (NSELCF) study is an ongoing longitudinal cohort study, which recruited 374 >65year-old residents in eight long-term care facilities in Taiwan [50]. By the end of the follow-up on December 31, 2009, only 265 participants (men, 43.0%) remained. Of those, 37 were excluded because they had missing anthropometric or laboratory data, and 46 were excluded because they had ECOG scores <2 (Figure 1). Ultimately, 182 individuals were analyzed. Medical ethics approval for participant recruitment and data analyses was obtained from the Institutional Review Board of the China Medical University Hospital. All participants provided written informed consent.

Measurements
The anthropometric indices included age, sex, weight (to the nearest 0.1 kg), height (to the nearest 0.1 cm), BMI (calculated as weight (kg) / estimated height squared (m 2 )), WC (to the nearest 0.1 cm), blood pressure (in mmHg), and ECOG score. The participants were divided into five categories from 0 (completely active) to 4 (completely disabled). For participants whose height measurement could not be directly measured, height was estimated based on arm span (length from the fingertips of one hand to those of the other hand [51]).
Blood was drawn while patients were in a lying or seated position. In addition to blood counts, the chemistry profile included creatinine, BUN, albumin, fasting glucose, cholesterol, TG, high-density lipoprotein cholesterol (HDL-C), low-density lipoprotein cholesterol (LDL-C), and uric acid levels.

Mortality
Mortality data were obtained from the Department of Health and Welfare in Taiwan from January 2003 to December 2009.