There is tremendous controversy regarding the adequacy/effectiveness of the nutritional care provided in VA Transitional Care Units (TCUs) - now frequently called Geriatric Evaluation and Management Units (GEMs). The interrelationship between concurrent inflammatory disease, the adequacy of a resident's nutrient consumption, the development or resolution of putative nutritional deficits, and clinical outcomes is not established. Given the known obstacles to increasing nutrient intake (e.g. cost, resident acceptance, associated morbidity) and the lack of proven effectiveness of all forms of nutrition support and nutritional supplementation to improve clinical outcomes, a better understanding of these interrelationships is needed and was the focus of this study.
The primary objective of this study was to determine how best to define nutritional risk (i.e., risk for adverse clinical outcomes due to inadequate nutrient intake) among older VA TCU residents. As part of this objective, we wanted to develop a better understanding of the interrelationship between nutrient intake, weight change, serum concentration of albumins, health status/illness severity, and mortality. The secondary objective was to develop a prediction model for identifying which TCU residents are likely to have ongoing problems with low nutrient intake.
To meet these objectives, older, non-terminally ill veterans admitted to a VA TCU for recuperative and rehabilitative care were prospectively studied using previously validated methods. After completing a comprehensive admission assessment, each subject was monitored closely throughout his/her TCU stay with serial nutrient intake and recurring metabolic, functional, neuropsychological, cognitive, nutritional, and medical assessments. For one year post discharge, each subject was monitored by CPRS, phone, and clinic visit to determine survival and days of institutional care. Strengths of associations were assessed using univariable and multivariable analytic techniques including logistic and Cox Proportional-Hazards analyses.
Of the 540 study eligible patients admitted to the TCU between March 2006 and March 2011, 446 (83%) entered the study. The subjects (mean age 78.6+7.5 yrs) were predominantly white (88%) males (98%) admitted from either an acute care hospital (74%) or home (26%). All had multiple stable (mean 16+4) and uncontrolled (mean 4+2) comorbid health problems, were taking multiple medications (mean 14+5), and had evidence of ongoing inflammation (e.g., median hs-CRP of 12 mg/dL, IQR 5-42 mg/dL). After a median LOS on the GEM of 21 days (IQR 14-40 days), the subjects were discharged to home (76%), death (1%), nursing home (9%), or an acute care hospital (9%). In the year following GEM discharge, 50% had a least one hospital readmission and 21% died.
1. Despite the unit's strong focus on nutrition, 26% of subjects maintained an average daily energy intake <70% of their predicted needs while 49% consumed <90% of needs. A majority (62%) improved their nutrient intakes during their TCU stays; however, there was a large variance in baseline nutrient intake and its change during the TCU stay.
2. Traditional measures of nutritional assessment were found to lack sensitivity and specificity as indicators of either nutritional status or adequacy of nutrient intake. Both prealbumin and its change during the hospitalization were positively correlated with protein and energy intake but inversely correlated with markers of inflammation, particularly hs-CRP and interleukin-6 (IL-6). In multivariable models, markers of inflammation accounted for 56% of the variance in prealbumin change; whereas, protein intake accounted for 6%. Similar results were found for albumin. The nutritional significance of body weight and its change often could not be assessed due to fluctuations in total body water. Between admission and discharge, a majority of the subjects (53%) experienced a weight change of greater than 2.5%. The amount of weight change was strongly and positively correlated with multiple indicators of both nutrient intake and change in the amount of body water, such as peripheral edema. By multivariable analysis, the strongest predictor of weight change was change in maximal calf circumference change (partial R2=0.35, p<0.0001), followed by average daily energy intake (partial R2=0.14, p<0.0001. Given the potential confounding effects in this population of inflammation and fluctuations in total body water, neither albumin, prealbumin, body weight, or their change is an adequate substitute for a more in-depth nutritional assessment including detailed calorie counts.
3. Change in physical function between admission and discharge was positively and independently correlated with both change in nutrient intake and change in markers of inflammation. The adjusted odds of experiencing a clinically significant improvement in functional status were 2-3 times greater for subjects in the upper quartile of change (i.e., greatest improvement) compared to the lowest for both protein intake and serum hs-CRP levels. Future studies are needed to determine whether combined interventions that target these factors improve recovery during hospitalization for this population.
4. The subjects' energy requirements were estimated based on the use of standard formulas that were validated in this population against resting metabolic rate measured by indirect calorimetry (mRMR). For all 10 published formulas evaluated, estimated RMR (eRMR) was highly correlated with mRMR (r = 0.70 to 0.74, p<0.0001 in all cases). However, all but one formula overestimated RMR by an average of 5% to 37%. The Harris-Benedict formula was one of the most accurate differing from mRMR by an average of 5+15%. A new formula derived using least-squared regression and including an estimate of GFR calculated using the Cockcroft-Gault formula (RMRe = 7.51 x wt(kg) - 4.55 x Age + 2.78 x GFRcg + 1060.50) produced results that differed from mRMR by an average of 1+12%. This new formula needs to be validated in additional studies.
5. Based on analysis of all admission and discharge variables for the first 235 patients to complete the study, the strongest predictor of post-discharge mortality by Cox Regression was the change in TNFa. When TNFa increased, the subsequent risk of death over 7 months was 29%; for the remaining subjects, death risk was 5% (p=0.045, Chi Square=4.0).
6. The kidneys play a significant role in the clearance of TNFa and its soluble receptors (sTNFrI and II). In this study, sTNFrI was found to be a stronger predictor of post-discharge mortality than were indicators of renal function (cystatin C, creatinine, BUN, and calculated GRF) indicating that persistent inflammation may be the mechanism for the higher mortality seen in TCU patients with renal impairment.
7. Depression was measured by the Geriatric Depression Scale-15 items (GDS), apathy was derived from 3 items on the GDS related to motivation (GDS-apathy), and executive function was measured by verbal fluency, digit span and Florida praxis battery. Baseline measures of apathy, depression, executive function did not predict either energy or protein intake. However baseline GDS and GDS- apathy were negatively correlated to change in walking and walking endurance even after controlling for change in inflammatory markers.
The study indicates that persistent anorexia and ongoing dysregulated subclinical inflammation are highly prevalent among older recuperative care and rehabilitation patients and are strongly associated with functional recovery and subsequent mortality. The study also indicates a need to reassess current in-hospital nutrition assessment practices given the lack of sensitivity and specificity of commonly employed nutritional assessment parameters to identify older patients at high nutritional risk. Multimodality interventions that target both the ongoing inflammation and the nutritional deficits need to be developed and tested to assess their impact on improving clinical outcomes in this population.
- Sullivan DH, Johnson LE, Dennis RA, Roberson PK, Garner KK, Padala PR, Padala KP, Bopp MM. Nutrient intake, peripheral edema, and weight change in elderly recuperative care patients. The journals of gerontology. Series A, Biological sciences and medical sciences. 2013 Jun 1; 68(6):712-8.
- Dennis RA, Johnson LE, Roberson PK, Heif M, Bopp MM, Garner KK, Padala KP, Padala PR, Dubbert PM, Sullivan DH. Changes in activities of daily living, nutrient intake, and systemic inflammation in elderly adults receiving recuperative care. Journal of the American Geriatrics Society. 2012 Dec 1; 60(12):2246-53.
- Sullivan DH, Johnson LE, Dennis RA, Roberson PK, Heif M, Garner KK, Bopp MM. The Interrelationships among albumin, nutrient intake, and inflammation in elderly recuperative care patients. The Journal of Nutrition, Health & Aging. 2011 Jan 1; 15(4):311-5.
- Dennis RA, Johnson LE, Roberson PK, Heif M, Bopp MM, Cook J, Sullivan DH. Changes in prealbumin, nutrient intake, and systemic inflammation in elderly recuperative care patients. Journal of the American Geriatrics Society. 2008 Jul 1; 56(7):1270-5.
- Sullivan DH, Roberson PK, Johnson LE, Mendiratta P, Bopp MM, Bishara O. Association between inflammation-associated cytokines, serum albumins, and mortality in the elderly. Journal of The American Medical Directors Association. 2007 Sep 1; 8(7):458-63.
- Sullivan D. Vitamin D: An Urgent Need for More Research. [Cyberseminar]. 2011 Jun 29.
- Padala PR, Dubbert PM, Padala KP, Garner K, Dennis RA, Bopp M, Sullivan D. Mood and Neurocognitive Measures to Predict Intake and Function in Elderly in a Recuperative Care Unit. Poster session presented at: American Geriatrics Society Annual Meeting; 2012 May 3; Seattle, WA.
- Sullivan D, Padala KP, Garner KK, Dennis RA, Bopp M, Padala PR. Accuracy of Formulas to Predict Resting Metabolic Rate in Elderly Admitted to a Recuperative Care Unit. Poster session presented at: American Geriatrics Society Annual Meeting; 2012 May 3; Seattle, WA.
- Sullivan D. Nutritional Issues in Older Adults. Paper presented at: American College of Physicians Internal Medicine Annual Meeting; 2011 Apr 7; San Diego, CA.
- Sullivan D. Changes in Activities of Daily Living, Nutrient Intake and Systemic Inflammation in Elderly Patients. Paper presented at: Gerontological Society of America Annual Scientific Meeting; 2010 Nov 22; New Orleans, LA.
- Sullivan D. Human Subjects Protection: Special considerations in studies of vulnerable elderly populations. Paper presented at: VA VISN 16 Research Conference; 2010 Oct 14; Jackson, MS.
- Sullivan D. Disease Models of Inflammation: What can be learned. Paper presented at: American Geriatrics Society Annual Meeting; 2010 May 15; Orlando, FL.
- Sullivan D. Inflammation and Nutrition: Where We Are and Where We Need to Be. Paper presented at: American Society for Parenteral and Enteral Nutrition's Clinical Nutrition Week; 2010 Feb 12; Las Vegas, NV.
- Sullivan D. Clinical Markers of Inflammatory Processes. Paper presented at: American Dietetic Association Food and Nutrition Conference and Expo; 2009 Oct 19; Denver, CO.
- Heif M, Johnson L, Dennis RA, Roberson P, Bopp M, Cook J, Sullivan D. The Interrelationship among Renal Impairment, Inflammatory Markers and Mortality in Elderly Transitional Care Unit Patients. Paper presented at: Gerontological Society of America Annual Scientific Meeting; 2007 Nov 16; San Francisco, CA.
- Dennis RA, Johnson LE, Roberson PK, Heif M, Bopp MM, Cook J, Sullivan D. Associated changes in systemic inflammation, prealbumin, and nutrient intake in elderly recuperative care patients. Poster session presented at: American Geriatrics Society Annual Meeting; 2007 May 4; Seattle, WA.
- Johnson LE, Dennis RA, Roberson PK, Heif M, Bopp MM, Cook J, Sullivan D. The Interrelationship Among Inflammatory Markers, Serum Albumin, and Death in Elderly Transitional Care Patients. Poster session presented at: American Geriatrics Society Annual Meeting; 2007 May 4; Seattle, WA.
Aging, Older Veterans' Health and Care, Health Systems
Frail elderly, Frailty, Long-term care, Outcomes, Outcomes - Patient, Quality assessment