Oral Nutritional Supplements and 30-Day Readmission Rate in Hypoalbuminemic Maintenance Hemodialysis Patients

Abstract

Introduction:

Hemodialysis patients are hospitalized nearly twice yearly, and 35% of these patients are re-hospitalized within 30-days post-discharge. We hypothesized that monitored oral nutritional supplementation (ONS) during hemodialysis treatment may decrease readmissions.

Methods:

A cohort of maintenance hemodialysis patients, treated at a large dialysis organization, who were hospitalized with a post-discharge albumin of ≤ 3.5 g/dL, without documented ONS use 90 days prior to the index hospitalization were identified. Individuals who received monitored intradialytic ONS post-discharge were compared to those without receipt of ONS. The outcome of interest was 30-day hospital readmissions. Logistic regression was used to assess the association between ONS receipt and 30-day readmission events, with adjustment for case-mix and laboratory variables.

Findings:

Of 5,479 eligible patients, ONS was prescribed to 1,420 individuals. Mean age was 64.6 +/− 14.1 (SD) years; median dialysis vintage was 3.9 years. There were 274 (19%) readmissions among ONS recipients versus 1571 (38.7%) among controls during the 30-day follow-up period. Individuals who did not receive ONS had increased odds of readmission [OR 2.26 (95% CI 1.02, 2.53)] in 30 days, as compared to those who did receive ONS post-discharge. In sensitivity analyses using a propensity score matched cohort, the odds ratio of readmissions within 30 days post-discharge was 1.71 (95% CI: 1.42, 2.07) for individuals who did not receive ONS as compared to those who received ONS.

Discussion:

Consumption of ONS during hemodialysis sessions is associated with reduced hospital readmission rates among in-center maintenance hemodialysis with severe hypoalbuminemia at 30 days post-hospital discharge.

Introduction

Hemodialysis patients are hospitalized frequently, averaging 2 hospitalizations per year.¹ Thirty-day readmission rates are exceptionally high, at 35%, with an additional 5% of dialysis patients dying within 30 days of an index hospitalization. This compares to 15% rehospitalization among Medicare beneficiaries age 65 or older in the general population.^1,2 Low serum albumin, which reflects both nutritional status and chronic inflammation, is associated with both mortality and hospitalization in hemodialysis (HD) patients.³

Many hemodialysis patients have poor nutritional status, with as many as 75% of patients requiring kidney replacement therapy showing evidence of protein-energy wasting (PEW).⁴ The process of hemodialysis itself induces a catabolic state; of note, this catabolic state can potentially be ameliorated with the use of oral intradialytic protein supplements.^5,6 Prior observational studies of in-center maintenance hemodialysis patients with a serum albumin of less than 3.5 mg/dL showed a reduction in mortality risk ranging from 9–35% in patients treated with intradialytic oral nutritional supplements (ONS).^7,8 Two more studies have indicated lower hospitalization rates in HD patients treated with monitored thrice weekly ONS during treatments.^9,10

Following acute illness requiring hospitalization, dialysis patients may be particularly vulnerable given catabolism associated with acute illness and likely reductions in food intake during hospitalization. Little is known about the use of ONS as an intervention to address the high rate of hospital readmission among dialysis patients. Accordingly, using data from a natural experiment in which oral nutritional supplements were introduced in a subset of dialysis facilities, we evaluated whether ONS impacted readmission in in-center hemodialysis patients with low baseline serum albumin levels in comparison to contemporaneous matched patients receiving hemodialysis in facilities that did not routinely utilize oral nutritional supplements.

Materials & Methods

Study Population

A large dialysis organization (LDO) that operates more than 2,150 outpatient dialysis facilities initiated a pilot program in 2007 that encouraged the use of monitored intradialytic ONS for individuals with a serum albumin level ≤ 3.5 g/dL who were able to consume oral supplements. This program was adopted at the discretion of the treating nephrologist, although participation was not mandatory.^7,10. For the present study, in-center hemodialysis patients treated at the LDO who were hospitalized for any reason from January 1, 2011 to December 31, 2011 and had a post-discharge albumin of ≤ 3.5 g/dL without receipt of ONS during the 90-day period prior to their admission were included. This group was divided based on whether they received or did not receive ONS following discharge ONS. Individuals were excluded if they died within 30 days post-discharge, baseline serum albumin was ≥3.5 mg/dL within 30 days of the index hospitalization, or if albumin was not measured either prior to hospitalization or within 30 days post-discharge (Figure 1A). Serum albumin concentration was measured each month using the bromocresol green method by a commercial laboratory (Spectra Laboratories, www.spectra-labs.com).

Although the practice of providing ONS to qualified in-center patients was encouraged at that time, penetration was incomplete, reflecting nephrologist and medical director preference or patient deferment. In general, the medical director would promote ONS use within a facility and the treating nephrologist would prescribe the ONS protocol. For patients prescribed an ONS, any one of six intradialytic ONS, as outlined in Table 1 were available, based on patient preference and individual facility availability. Patients received their option of one dose of ONS per treatment, three times per week. No preferred ONS was prescribed, and patients had options to choose with each treatment to avoid taste fatigue. Documentation of ONS receipt and consumption was performed at each hemodialysis treatment.

Table 1:

Oral nutritional supplement product choices

Nutrition Product	Protein (grams)	Calories
Nepro®Carb Steady (8 fluid ounces) Abbott Nutrition, Columbus, Ohio	19	425
ProStat® RC (1 fluid ounce) Medical Nutrition, Englewood, New Jersey	15	60
ZonePerfect™ (50 gm bar) Abbott Nutrition, Columbus, Ohio	14	210
VitalProteinRX™ (60 gm bar) VitalRemedyMD, Inc., Boca Raton, Florida	20	210
Liquacel™ (1 ounce) Global Health Products, Inc. Rochester, New York	16	90
BodyQuest Protein Ice Cream (4 ounces) Body Quest Ice Cream, LLC, Atlanta, Georgia	16	160

Covariates

Patient-level data was extracted from the LDO data warehouse.¹¹ Age, sex, race, body mass index (BMI), diabetes status, dialysis vintage, vascular access type, and hospitalizations in the quarter preceding protocol implementation (October 1 to December 31, 2010) were determined. Baseline laboratory values, including serum albumin, hemoglobin, phosphorus, and eKt/V during the fourth quarter of 2010, were also used for multivariable adjustment in models assessing the association between ONS use and re-hospitalization. This study was exempt from research ethics committee approval because data was deidentified and not linked to patient-specific characteristics. As such, informed consent was waived due to de-identified information.

Outcomes

The primary study outcome was readmission to a hospital within 30 days after discharge that occurred between January 1, 2011 to December 31, 2011. Hospitalization data was captured from the LDO data warehouse, which stores admission and discharge dates for hospitalizations and lists ICD-9 coding associated with each hospitalization. Readmission for any medical problem was measured as an event. Secondary outcomes included extended periods of readmission within 60 days and 90 days of an index hospital discharge

Statistical Analysis

Baseline patient characteristics were compared using t tests or χ² tests for continuous and categorical variables, respectively. Logistic regression was used to assess the association between receipt of oral nutritional supplement and 30-day readmission events. Individual data was adjusted for case-mix (age, sex, race, diabetes mellitus, dialysis vintage, vascular access, BMI, and hospitalization in Q4, 2010) as well as case-mix plus laboratory variables (3-month mean values for albumin, hemoglobin, phosphorus, and eKt/V for October - December, 2010).

The following sensitivity analyses were performed: (1) propensity score matching and (2) mixed effect models. Propensity scores were developed based on the predicted probability of receipt of ONS and included case-mix variables from patient-level data: age, sex, race, diabetes mellitus, dialysis vintage, vascular access, BMI, and hospitalization in Q4, 2010. Individuals who received ONS were matched with patients who did not receive ONS in a 1:2 ratio, based on propensity scores plus two markers to ensure equal post-discharge (pre-study follow-up) survival time to avoid immortal time bias. These markers indicating active treatment included survival time from discharge to initiation of ONS and survival time from post-hospitalization serum albumin assessment to initiation of ONS (Fig S1A and S1B). Mixed effect model analysis was done using the dialysis facility as a random effect first, and a second model adding a variable indicating the proportion of patients within a facility that receive ONS therapy. All analyses were performed in SAS, version 9.3 (SAS Institute Inc).

Results

Baseline Characteristics

Baseline characteristics for the patients are shown in Table 2. There were 5,479 eligible patients with index discharges from 1,503 LDO facilities, 25.9% of whom received ONS. In the non-matched cohort, individuals who received ONS were more likely to be men, white race, and have an arteriovenous fistula for dialysis access. Individuals who received ONS had a higher percentage of hospitalizations in the fourth quarter of 2010, just prior to the study (35.3% vs. 24.8%, p<0.001). Dialysis vintage was similar between groups.

Table 2:

Baseline characteristics of study participants

Baseline Characteristics	All Patients	ONS	Controls
N	5,479	1,420	4059
Age, years	64.6 (14.1)	64.6 (13.6)	64.6 (14.2)
Male, %	52.7	56.9*	51.3*
Race
White	56.4	58.5*	55.7*
Black	38.7	35.8	39.7
Other	4.9	5.8	4.8
Diabetes mellitus, %	64.9	63.8	65.3
Vintage, years	3.9 (3.9)	3.9 (3.7)	3.9 (3.9)
Access Type
Fistula	46.5	54.0	44.2
Graft	23.1	23.7	22.9
Catheter	29.4	22.8	31.7
Unknown	1.0	0.5	1.2
BMI, kg/m2	28.2 (7.9)	28.6 (7.5)	28.1 (8.0)
Hospital in Q4, 2010, %	32.5	35.3**	24.8**

Continuous variables reported as mean (standard deviation). Abbreviations: n = number of individuals; BMI = body mass index; Hospital in Q4, 2010 = hospitalization in the fourth quarter, prior to study.

^*p<0.05

^**p <0.0001

Readmission Outcome

Among all individuals who received ONS prescription, 19.3% (n=274) were readmitted, as compared to the control group readmission outcome of 38.7% (n=1571). The odds ratios for 30-day readmission rates observed in patients not treated by ONS, after adjustment for case-mix and laboratory variables are presented in Table 3. In unadjusted and adjusted analyses, individuals who did not receive ONS had increased odds of readmission at 30 days, as compared to individuals who did receive ONS [unadjusted OR 2.64 (2.28, 3.06) and case-mix and lab-adjusted OR 2.26 (1.92, 2.66)]. This significant association persisted during extended follow-up at 60 days [adjusted OR 1.94 (1.69, 2.23)] and at 90 days [adjusted OR 1.89 (1.65, 2.17)] (Table 3).

Table 3:

Association of treatment with ONS after discharge with readmission at 30 days, 60 days, and 90 days.

		Readmission in 30 Days	Readmission in 60 Days	Readmission in 90 Days
Receipt of ONS after discharge, n (%)	ONS	274 (19.3)	513 (36.1)	606 (42.7)
	Controls	1571 (38.7)	2265 (55.8)	2503 (61.7)
Odds Ratio*	Unadjusted†	2.64 (2.28, 3.06)	2.23 (1.97, 2.53)	2.16 (1.91, 2.44)
	Adjusted‡	2.26 (1.92, 2.53)	1.94 (1.69, 2.23)	1.89 (1.65, 2.17)

Abbreviations: ONS, oral nutritional supplement; OR, Odds Ratio. Odds ratios are presented as OR (95% confidence intervals).

^*Odds ratios are presented as OR (95% confidence intervals)

^†Unadjusted logistic regression

^‡Adjusted for age, gender, race, DM, dialysis vintage, vascular access, BMI, hospitalizations in Q4 of 2010, 3-month mean albumin, hemoglobin, phosphorus, and eKt/V

Sensitivity Analysis

In a sensitivity analysis using 1:2 propensity score matching, there was no difference between controls and those receiving ONS with respect to access type and Q4 hospitalizations prior to the enrollment date (Table S1). There was a small difference in dialysis vintage (4.5±3.9 vs. 4.0±3.9 years, p<0.05) among the matched group, with slightly longer vintage in the ONS treated group. In a sensitivity analysis using propensity score matched, fully adjusted model, there was slight attenuation in the association between receipt of ONS and readmission risk, although the odds of readmission remained significant at 30 days, 60 days, and 90 days, with odds ratio of 1.71 (1.42, 2.07), 1.45 (1.23, 1.71), and 1.39 (1.18, 1.63) respectively. In an additional anlysis using fully adjusted models for case mix and laboratory data, accounting for effects of facility-based practices, individuals who did not receive ONS remained at increased odds for 30-day readmission hospitalizations [OR 2.26 (1.92, 2.66) for all patients and OR 1.71 (1.42, 2.07) for matched cohort] (Table S2). There was no substantial impact on the main findings and no statistically significant interaction between the readmission by ONS group and the proportion of facility patients receiving ONS.

Discussion

The use of intradialytic ONS in individuals with low serum albumin treated with maintenance hemodialysis is associated with reduced odds of readmission within 30 days, 60 days and 90 days. These associations remained significant after adjustment for case-mix, baseline laboratory data, and use of propensity score matching. Additional mixed-effects model analyses using factors that account for dialysis facility adoption of ONS and facility effect showed similar results. These findings support the hypothesis that a simple intervention driven by care providers may impact readmissions for a population with known high readmission rates.

The present study expands on the previously published reports of intradialytic oral nutritional supplementation in a hemodialysis population that have shown that use of nutritional supplement protocols during maintenance hemodialysis treatments is associated with a reduction in mortality between 9–34%, even after accounting for facility and individual patient characteristics including serum albumin levels.^7,8 In addition to mortality benefits seen in ONS-treated individuals, the impact of ONS on hospital admissions has also been investigated among non-hemodialysis and hemodialysis patients.^9,10,12 In a systematic review and meta-analysis of randomized clinical trials with primarily non-hemodialysis adult patients of any nutritional status, ONS use was associated with a 41% reduction in odds of hospitalization or readmission [OR 0.591; 95% confidence limits: 0.43, 0.80].¹² No patients treated with hemodialysis were reported in this study, although several subgroup analyses showed the benefit to be stronger in an advanced age population. Additional evidence of improvement in hospitalization length of stay and readmissions was reported with the use of inpatient ONS from a general adult population cohort within a hospitalization billing database containing 44.0 million adult inpatient episodes at 460 sites during the years 2000 to 2010.¹³ A reduction in length of stay by 2.3 days [21% decline; 95% confidence limits: 2.16, 2.42 days] was associated with $4,734 [95% confidence limits: $4,714, $4,754] lower hospitalization cost in those treated with ONS. In the matched analysis for the subset of 862,960 patient readmissions within 30 days, the probability of 30-day readmissions declined by 6.7% [absolute decline of 2.3%, 95% confidence limits: 1.9%, 2.7%]. Among a maintenance hemodialysis population, hospitalization rates were lower in individuals who were treated with ONS as compared to those who were not, after adjustment for key confounders including age, dialysis vintage, and vascular access.^9,10 Lacson et al further adjusted for key laboratory confounders including albumin and facility level standardized mortality ratio. Despite all of these adjustments, hospitalization rates remained lower [OR: 0.90 (0.86, 0.95), favoring ONS use] among those treated with ONS in the matched cohort that required active hemodialysis treatment for at least 30 days to allow for the opportunity to be offered ONS. The present study adds to the dialysis literature that readmission rates also are impacted by ONS treatment, with a reduction in 30-day, 60-day, and 90-day readmission rates with provision of ONS.

Use of intradialytic ONS may mitigate the catabolic impact of hemodialysis by providing substrate to minimize protein-energy wasting (PEW). The concept of PEW syndrome in ESRD is a multifactorial process that occurs through dysregulation of catabolic and nutritional deficiencies in those with advanced kidney disease.¹⁴ The procedure of dialysis contributes to PEW by increased resting energy expenditure during the HD procedure through inflammation as well as protein loss through the dialysis itself, leading to whole-body and skeletal muscle protein homeostasis alterations. During acute illnesses and hospitalizations, increased catabolism is stimulated through inflammation, metabolic derangements and underlying comorbid conditions, which in turn leads to further dysregulation of the protein-energy wasting spectrum. Use of intradialytic ONS has been shown to improve whole-body energy balance in a sustainable manner, as compared with neutral or negative balance in those not receiving ONS.¹⁵ Furthermore, the impact of interrupting the catabolic cycle during dialysis has been shown to extend beyond the duration of dialysis treatment.⁵ It is possible that cellular pathways leading to muscle wasting are activated by the catabolic stimulus occurring during hemodialysis and that intradialytic ONS plays a role in interrupting the catabolic cycle.^14,16

In general, chronic disease conditions such as ESRD are more prone to hospital readmissions. In 2012, a study by Jencks et al evaluated the readmission rates among Medicare beneficiaries, showing ESRD as a significant predictor for 30-day readmission, with a hazard ratio (95% CI) of 1.417 (1.409, 1.425), second only to the number of prior hospitalizations.² This outcome will become more relevant in the present environment due to anticipated changes in patterns of reimbursement, managed by CMS/Medicare’s Affordable Care Act, which articulated a major focus on quality of care by implementing activities to prevent hospital readmissions. In particular, the ESRD Quality Incentive Program (QIP) has embraced this shift to value-based purchasing, in which improved outcomes and lower cost are rewarded, with the goal of promoting effective prevention and treatment practices for leading causes of morbidity and mortality through quality care delivery, coordination, and multilevel engagement.¹⁷ A core measure of standardized readmission ratio (SRR) was recently approved by CMS as part of the Final Rule for value-based purchasing, where this risk-adjusted SRR (defined as # of observed unplanned readmissions relative to the # of expected unplanned readmissions) is now incorporated into clinical measurement as a benchmark to assess care coordination under the QIP.¹⁷

A major barrier towards improvement in this quality measure of readmissions is the complexity of care necessary to manage an individual with ESRD. Even in a less medically complicated population, multi-component, multi-disciplinary interventions have been shown to be most effective to prevent 30-day hospital readmissions.¹⁸ The most successful care plans focus on augmenting the patient’s capacity for self-care and ability to mobilize supportive care structures in order to improve outcomes. These interventions also require huge investments and high resource utilization. In addition to the complications posed by heavy burden of chronic illness, patients with ESRD have the additional complexity posed by transition of care from the hospital to the outpatient nephrology center. This concept of transition of care is defined by a set of actions designed to ensure coordination and continuity of health care during transitions between different healthcare settings of levels of care.¹⁹ This vulnerable time period incorporates risks related to systems practices, from flow of information from hospital to the dialysis unit, timely evaluation by a nephrologist after discharge including dry weight adjustments and medication reconciliation. Implementing effective interventions during care transition is often resource intensive, thus requiring efficient and preferably, careful parsimonious evidence based selections.

There are few modifiable risk factors that have been identified in the dialysis population, although anemia and subsequent treatment with erythropoietin stimulating agents has been associated with a reduction in risk of readmission.^20,21 Arguably, the postulated patho-biological pathways for effecting this association were unclear and over time, the anticipated benefit of these interventions diminished, particularly in the cohort treated with vitamin D administration.²¹ Hypoalbuminemia, defined as a serum albumin less than 3.5 g/dL, was associated with an increased risk of index hospitalization and subsequent hospitalizations in individuals with ESRD, and in individuals who achieved a serum albumin greater than 3.5 g/dL, re-admission rates were found to be lower.^22,23 Focusing on the modifiable root cause underlying the observed hypoalbuminemia, to include dialysis-associated catabolism, may have a more durable impact than efforts directed at simply modifying ESA dose. Recognition that dialysis facilities and providers, through simple interventions such as ONS supplementation, have another viable, less complex therapeutic option to improve transition from the hospital to the home and potentially aid in preventing readmissions, is thus of great value. Of note, all post-hospitalization interventions do not preclude post-discharge care coordination, including contact with healthcare personnel such as nephrologists and other allied healthcare providers, that may potentiate the positive impact of any intervention.²⁴

This study has several limitations. First, only directly observed therapy with receipt of ONS was captured, and individuals whose post-discharge care may have been altered by unmonitored ONS use were not accounted for in this study. However, such instances would have likely led to a bias towards the null hypothesis by dilution of the effect size between those treated with ONS and those without. Second, a contributing factor as part of the ONS intervention (if not an alternative explanation for finding a beneficial association in the ONS-treated group) could be the unappreciated increase in interaction associated with direct observation and documentation of ONS by dialysis staff, which could have led to more prompt attention to subtle symptoms or findings of decline in a patient. Third, a competing risk of death was not taken into account, which may bias these results. In a prior study, it has been shown that ONS use is associated with improved survival even within 30 days, after matching and adjustment, thus the contribution of death as a competing risk within the 30-day readmission period may be less.⁷ Fourth, by nature of the observational design, we were unable to account for additional practice pattern variations between facilities. Thus, differences in nephrologist follow-up time or differences in facility average treatment times could have influenced outcomes of readmissions. However, statistical analysis using mixed effects models were performed to mitigate facility-level confounding influence, which did not substantially change the results. In addition, without randomization, there were differences between the individual characteristics of the ONS treated and untreated groups, specifically access type, sex, and race. Using propensity score matching technique, however, the matched cohort was balanced in terms of baseline characteristics, and the relationship between ONS and readmissions remained significant. Finally, we did not capture the admission diagnosis for the index hospitalization and for the subsequent readmissions. There may in fact be an interaction in which certain hospitalizations (such as heart failure) may have different risks of readmissions, with a variance in distribution of reasons for hospitalization that could bias our results.

This study also has several strengths. First, the cohort study is comprised of prevalent hemodialysis patients from a large, nationwide dialysis provider, which captures representative patients from a variety of locations. Thus, regional practice differences would be less confounding with this population mix, likely increasing generalizability. Second, the use of propensity score matching in sensitivity analysis allows for balancing measured confounders, including markers for severity of illness, baseline nutritional status and baseline functional status. Third, we used multiple time points for monitoring readmission rates, extending the durability of outcomes for up to 90 days post-discharge. Finally, by eliminating those who did not have the “active on treatment” variables, this study reduced the possible confounding association between the absence of an opportunity to receive ONS [e.g. by early death or readmission] and outcome.

In summary, provision of monitored intra-dialytic oral nutritional supplements in a population of recently discharged maintenance hemodialysis patients with a serum albumin of less than 3.5 g/dL was associated with a reduction in the odds of readmission at 30-day, 60-day, and 90-day time points. Given the vulnerability of readmissions in the maintenance hemodialysis population, information derived from this study has the potential to significantly impact healthcare delivery by providing a simple, easily implementable treatment option for a modifiable readmission risk factor within this population. The authors propose that monitored intradialytic ONS provided to hypoalbuminemic hemodialysis patients after hospital discharge is a minimal-risk, relatively low cost, simple intervention that can be done to reduce readmission rates among a population at high-risk for hospitalization and readmissions. We await findings from a prospective cluster randomized trial that will offer monitored intradialytic ONS for all hemodialysis patients, with particular interest in those receiving the intervention post-hospital discharge.²⁵

Supplementary Material

Supplementary Tables 1 and 2

Table S1: Baseline characteristics of study participants, using propensity score matched groups

Table S2: Association of ONS treatment and 30-day readmissions using fixed effect vs mixed effect models

Supplemental Figure 1a

Figure S1A: Patients were screened from hospital discharge and included if baseline albumin was measured and was less than 3.5 mg/dL. Abbreviations: ONS, oral nutritional supplement.

Supplemental Figure 1b

Figure S1B: Study design of individuals who were discharged from the hospital with serum albumin ≤ 3.5 g/dl, matched according to ONS use or non-use, followed for 30 days.