 |
Periodicals Home | Search | User Pref |
Help | Logout |
| AJKD Home | Table
of Contents |
All Issues | Order | About this Journal |
<< Issue |
>> Issue |
|
Periodicals Home | Search | User Pref |
Help | Logout |
| AJKD Home | Table
of Contents |
All Issues | Order | About this Journal |
<< Issue |
>> Issue |
 |
Kamyar Kalantar-Zadeh, MD [MEDLINE LOOKUP]Sections |
|
| Abstract | TOP |
Malnutrition inflammation complex syndrome (MICS) occurs commonly in maintenance hemodialysis (MHD) patients and may correlate with increased morbidity and mortality. An optimal, comprehensive, quantitative system that assesses MICS could be a useful measure of clinical status and may be a predictor of outcome in MHD patients. We therefore attempted to develop and validate such an instrument, comparing it with conventional measures of nutrition and inflammation, as well as prospective hospitalization and mortality. Using components of the conventional Subjective Global Assessment (SGA), a semiquantitative scale with three severity levels, the Dialysis Malnutrition Score (DMS), a fully quantitative scoring system consisting of 7 SGA components, with total score ranging between 7 (normal) and 35 (severely malnourished), was recently developed. To improve the DMS, we added three new elements to the 7 DMS components: body mass index, serum albumin level, and total iron-binding capacity to represent serum transferrin level. This new comprehensive Malnutrition-Inflammation Score (MIS) has 10 components, each with four levels of severity, from 0 (normal) to 3 (very severe). The sum of all 10 MIS components ranges from 0 to 30, denoting increasing degree of severity. These scores were compared with anthropometric measurements, near-infrared-measured body fat percentage, laboratory measures that included serum C-reactive protein (CRP), and 12-month prospective hospitalization and mortality rates. Eighty-three outpatients (44 men, 39 women; age, 59 ± 15 years) on MHD therapy for at least 3 months (43 ± 33 months) were evaluated at the beginning of this study and followed up for 1 year. The SGA, DMS, and MIS were assessed simultaneously on all patients by a trained physician. Case-mix-adjusted correlation coefficients for the MIS were significant for hospitalization days (r = 0.45; P < 0.001) and frequency of hospitalization (r = 0.46; P < 0.001). Compared with the SGA and DMS, most pertinent correlation coefficients were stronger with the MIS. The MIS, but not the SGA or DMS, correlated significantly with creatinine level, hematocrit, and CRP level. During the 12-month follow-up, 9 patients died and 6 patients left the cohort. The Cox proportional hazard-calculated relative risk for death for each 10-unit increase in the MIS was 10.43 (95% confidence interval, 2.28 to 47.64; P = 0.002). The MIS was superior to its components or different subversions for predicting mortality. The MIS appears to be a comprehensive scoring system with significant associations with prospective hospitalization and mortality, as well as measures of nutrition, inflammation, and anemia in MHD patients. The MIS may be superior to the conventional SGA and the DMS, as well as to individual laboratory values, as a predictor of dialysis outcome and an indicator of MICS. © 2001 by the National Kidney Foundation, Inc.
| (Click on a term to search this journal for other articles containing that term.) |
| Index words: Dialysis,
malnutrition,
inflammation,
nutritional
assessment, hospitalization,
mortality,
near
infrared, albumin,
transferrin,
C-reactive
protein (CRP) |
| Methods | TOP |
Patients
The outpatient chronic
dialysis program at San Francisco General Hospital (San Francisco, CA) treated
91 adult MHD patients at the time of the study. Inclusion criteria were patients
undergoing MHD for at least 3 months and aged 18 years or older. Three patients
did not meet these criteria. Of 88 eligible MHD patients, 2 patients were
hospitalized in other centers at the time of the study and 3 patients did not
agree to participate. Therefore, 83 individuals (44 men, 39 women) agreed to
enroll onto the study. The study was approved by the institutional review board,
and written informed consent was obtained from all participants.
Conventional Subjective Global
Assessment
The Subjective Global Assessment (SGA) of
nutritional states, as it is commonly used in nephrology, is a semiquantitative
scoring system based on history and physical examination.9,10
The history consists of five components: weight loss during the preceding 6
months, gastrointestinal (GI) symptoms, food intake, functional capacity, and
comorbidities. Each of these features is scored separately as A, B, or C,
reflecting well-nourished to severely malnourished categories. The physical
examination includes two components: loss of subcutaneous fat and muscle
wasting. (The presence of edema or ascites is the third component of the
original SGA physical examination, usually not used for dialysis patients.)
These two components are classified from 0 to 3, representing normal to severely
abnormal. Data are scored subjectively, and patients are classified in terms of
the three major SGA scores: A, well nourished; B, mild to moderate malnutrition;
and C, severe malnutrition. Details on methods for SGA evaluation in dialysis
patients are available as an appendix in a previously published article,10
available on the web site of the American Journal of Kidney Diseases (http://www.ajkd.org).
Dialysis Malnutrition
Score
Using components of the conventional SGA, one of the
authors recently developed a quantitative scoring system, the Dialysis
Malnutrition Score (DMS),11
that consists of seven components of the conventional SGA: weight change,
dietary intake, GI symptoms, functional capacity, comorbidity, subcutaneous fat,
and signs of muscle wasting. Ascites and edema were deleted, and number of years
of dialysis therapy (vintage) was added to the comorbidity component. Each
component has a score ranging from 1 (normal) to 5 (severely abnormal). Thus,
the DMS, ie, the sum of all seven components, is a number from 7 (normal) to 35
(severely malnourished); a higher DMS represents a greater degree of
protein-energy malnutrition. In a recent preliminary report from a
cross-sectional study using a different pool of patients, the DMS correlated
significantly with anthropometric values and laboratory measures of nutritional
status in MHD patients.11
Malnutrition Inflammation
Score
To attempt to make the scoring system more comprehensive
and quantitative, evaluation criteria for the 7 DMS components were revised, and
three new items were added: body mass index (BMI), serum albumin level, and
total iron-binding capacity (TIBC). Moreover, the number of severity levels of
each component was reduced from five to four levels because in the previous
study, we noted that the fifth level of the DMS was almost never used.11
Thus, the Malnutrition Inflammation Score (MIS) has 10 components, each with
four levels of severity, from 0 (normal) to 3 (severely abnormal). The sum of
all 10 MIS components ranges from 0 (normal) to 30 (severely malnourished); a
higher score reflects a more severe degree of malnutrition and inflammation.
Figure 1 shows our scoring sheet, which consists of four sections
(nutritional history, physical examination, BMI, and laboratory values) and 10
components.
| Fig. 1. Components of the
comprehensive MIS. |
) |
|
|
The history section includes 5 components adopted from the original
SGA.9
Weight change is determined as the change in edema-free posthemodialysis body
weight in the past 6 months. The lowest score (0) is given if weight loss is
less than 0.5 kg or there is an increase in body weight. Score 1 indicates a
minor loss of at least 0.5 kg, but less than 1.0 kg. Score 2 is given for weight
loss of at least 1.0 kg, but less than 5% of body weight, and score 3 indicates
weight loss of 5% or greater.
Dietary intake is scored 0 if it is the
usual intake of solid foods, with no recent decrease in amount or quality of
meals. A score of 1 indicates a slightly suboptimal solid diet, 2 indicates a
full-liquid diet or moderate decrease in food intake, and 3 indicates a daily
nutrient intake that would be incompatible with life on a chronic basis.
GI symptoms are scored 0 if the patient has a good appetite and no GI
symptoms; 1, mildly decreased appetite or mild nausea; 2, occasional vomiting or
other moderate GI symptoms, such as abdominal pain; and 3, diarrhea, frequent
vomiting, or severe anorexia.
Functional capacity is scored 0 for normal
functional capacity or a considerable improvement in level of previous
functional impairment. A score of 1 indicates mild or occasional difficulty with
baseline ambulation or feeling tired frequently; 2, difficulty with independent
activities; and 3, restriction to light activity or a persistent bed- and/or
chairbound state.
As in the modified SGA version (DMS), comorbidity
includes vintage (number of years on dialysis therapy) because the element of
time may have a bearing on the degree of malnutrition and inflammation.10,12
Thus, comorbidity is scored 0 if there are no other medical illnesses and the
patient has undergone hemodialysis therapy for less than 1 year; 1, mild
comorbidity, excluding such major comorbid conditions (MCCs) as congestive heart
failure class III or IV, severe coronary artery diseases, clinically evident
acquired immunodeficiency syndrome, moderate to severe chronic obstructive
pulmonary disease, and metastatic malignancies, or dialysis therapy for 1 to 4
years; score 2, moderate comorbidity (including one of the diseases listed under
MCCs) or dialysis therapy for more than 4 years; and score 3, two or more MCCs.
The existence of diabetes per se is not accounted for if the previously
mentioned comorbidities do not exist. Instead, comorbidities that may be a risk
for poor outcomes in patients with diabetes are examined individually.
The physical examination section consists of two components. Body fat
stores are scored by assessing subcutaneous fat deposition in four body areas,
ie, below the eyes, triceps and biceps areas, and chest. Signs of muscle wasting
are obtained by briefly examining seven sites: the temple, clavicle, scapula,
ribs (intercostal spaces), quadriceps, knee, and interosseous muscles. For each
of these two components, a score of 0 through 3, representing normal to severe
changes, is assigned according to conventional SGA guidelines based on criteria
specified elsewhere.9,10
We added a body weight function adjusted for height to the scoring
system because it has predictive value for dialysis mortality.7
BMI, a ratio of end-dialysis weight (in kilograms) to height squared (in square
meters), was selected to represent height-standardized weight.13,14
BMI was graded in four levels, 0 through 3, representing BMI greater than 20, 18
to 19.99, 16 to 17.99, and less than 16 kg/m2, respectively.
The fourth MIS section includes two laboratory values. Serum albumin
level is frequently a strong predictor of mortality among patients with
ESRD,15,16
and hypoalbuminemia may represent a response to inflammation (acute-phase
reaction),17,18
as well as low protein intake. Serum TIBC reflects serum transferrin
concentration and correlates significantly with nutritional state in dialysis
patients,10
although it also changes with inflammation and iron store fluctuations.
Therefore, these two laboratory values now comprise 20% of the total MIS score
(see Fig 1
for details).
In this study, a trained physician
(K.K.-Z.) scored each patient within 5 to 15 minutes before anthropometric
measurements were performed. To evaluate the degree of reproducibility, the same
physician repeated the MIS assessment after 1 week on a subset of 15 patients
without reference to the first MIS evaluation. The correlation coefficient
(r) between the two MIS assessments was 0.91, denoting a good degree of
reproducibility.
Anthropometric evaluation
Body
weight assessment and anthropometric measurements were performed 5 to 20 minutes
immediately after the termination of a hemodialysis treatment. Biceps skinfold
and triceps skinfold (TSF) thickness were measured using a conventional skinfold
caliper, described elsewhere.19,20
Midarm circumference (MAC) was measured with a plastic tape. All anthropometric
measurements were performed by a single trained physician (K.K.-Z.) three times
in rapid succession on the non-access-containing arm of each dialysis patient,
and the three measurements were averaged to give the final result. Midarm muscle
circumference (MAMC) was calculated from the formula20:
MAMC
= MAC – (3.1415 × TSF)
Height was obtained from the patient's chart.
Near-infrared interactance
To
evaluate the percentage of body fat and lean body mass, near-infrared
interactance (NIR)14,20
was performed at the same time as anthropometric measurements. A commercial NIR
sensor (portable Futrex 5000; Futrex Inc, Gaithersburg, MD) was used. NIR
measurements were performed by placing a Futrex sensor on the nonaccess upper
arm for several seconds after entering the required data (sex, weight, height,
and body frame size) from each patient, stipulating that physical activity
levels were uniform for all patients. It previously was shown that NIR
measurements of body fat correlate significantly with the SGA and other
nutritional measures in MHD patients.14,20
Hospitalization
Hospitalization
data during the 12-month period after the completion of these measurements were
obtained on all 83 hemodialysis patients. Hospitalization was defined as any
hospital admission that included at least one overnight stay in the hospital.
The admission day was counted as 1 full hospitalization day, but the discharge
day was not. Therefore, the minimum duration of hospitalization per admission
was 1 day. No exclusion criterion was used. Thus, hospital admissions for a
variety of disorders were counted. However, because the vast majority of
dialysis access-related hospitalizations did not require overnight admission,
essentially only those access-related hospitalizations complicated by other
morbid events, such as infection or cardiovascular complications, were included.
For the few patients in a hospital at the start of the 6-month cohort, that
hospitalization was not counted. For patients who were still in a hospital at
the end of the 1-year cohort, all hospitalization days of the last admission
were counted. For 9 patients who died and 6 patients who left the cohort during
the prospective 12-month follow-up, hospitalization indices during the survival
time were standardized by using the factor 12/survival time (in months).
Three methods were used to assess the 12-month prospective
hospitalization as clinical outcomes. Annual hospitalization frequency
(H1) was the total number of hospital admissions during the 12-month
prospective cohort, defined previously, regardless of the length of each
admission. Annual hospitalization days (H2) were the sum of all
hospitalization days of a given patient during the same period. The number of
days at risk from study start to the first hospitalization event for each
individual per year was assessed in a survival model (H3).
Accordingly, risk time for each individual is defined as days from study entry
until the first hospitalization, a censoring event, or a study anniversary day
occurs. A patient's risk period is truncated 3 days before transplantation to
avoid attributing the transplantation-related hospitalization to observed days
to event.
Laboratory evaluation
Laboratory
values, except for postdialysis serum urea nitrogen levels used to calculate
urea reduction ratio, were measured immediately before the dialysis session at
least 16 days after the last intravenous administration of iron. Serum
C-reactive protein (CRP) was measured as an indicator of an inflammatory state
and assessed by the immunoturbidimetric method (Hitachi 747). The lower-limit
sensitivity of the CRP assay is 6.9 ng/mL. For patients with a reported CRP
level less than 6.9 ng/mL, an arbitrary average of 3.4 ng/mL was used for
statistical analyses. Laboratory values were obtained by automated methods. All
laboratory measurements were performed by Spectra Laboratories (Fremont, CA).
Statistical and epidemiological
methods
The initial cross-sectional study included 83 patients
who were subsequently followed up as a 12-month prospective cohort to evaluate
hospitalization data as continuous outcome variables and mortality as a
dichotomized outcome. We used Pearson's correlation coefficient r for
selected analyses between continuous variables. Student's t-test
(two-tailed) was used for group mean comparisons between men and women.
Pearson's correlation r was used to determine the significance and
strength of associations. Spearman's rank correlation also was used for such
variables with nonparametric features as SGA, race, and underlying diseases, and
results were compared with Pearson's r. Multivariate regression analysis
was performed to obtain partial (adjusted) correlations controlled for sex, age,
race, and renal disease.
To calculate the relative risk for first
hospitalization and death in the prospective cohort, hazard ratios and 95%
confidence intervals (CIs) were obtained using Cox proportional hazard models to
control for the previously mentioned demographic variables. A 95% CI not
including 1.00 is considered statistically significant. The Cox proportional
hazard-calculated R2 (also known as
pseudo-R2) was used for comparison with other multivariate
correlations. Plots of log (–log [survival rate]) against log (survival time)
were performed to establish the validity of the proportionality assumption. Each
multivariate model included one outcome (dependent) variable and five predicting
(independent) variables, ie, age, sex, race, underlying kidney disease, and the
variable under study for that particular model. Therefore, the general
multivariate model for Cox regression is:
(t) = 0(t)[exp(b1age + b2sex +
b3race + b4disease + b5X)]
where is the estimated hazard, t is time to
event or censorship, b1 through b5 are coefficients of the
model terms, and X is the predicting variable, including the MIS result, its
components, or other variables (SGA, DMS, BMI, NIR body fat, Kt/V, or a
pertinent laboratory measurement). Therefore, the association between each
predicting variable and outcomes (first hospitalization or death) was studied
through separate multivariate models, but with uniform case-mix adjustment for
each model.
Descriptive and multivariate statistics were performed using
the statistical software Stata, version 5.0 (Stata Corp, College Station, TX),
and all results were verified using a second statistical software, Statistica
for Windows, release 5.1 (StatSoft Inc, Tulsa, OK). Fiducial limits are given as
mean ± SD. P less than 0.05 is considered statistically significant,
P between 0.05 and 0.10 is considered marginally significant, and
Pgreater than 0.10 is not significant.
| Results | TOP |
 |
All Patients | Men | Women | P |
| No. of patients | 83 | 44 | 39 | — |
| Race (black/Hispanic/Asian) | 40/20/19 | 22/9/10 | 18/11/9 | NS |
| Age (y) | 55.8 ± 15.3 | 50.8 ± 15.2 | 61.5 ± 13.4 | 0.002 |
| Vintage (dialysis mon) | 43.1 ± 32.9 | 42.9 ± 31.7 | 43.3 ± 34.6 | NS |
| Conventional SGA (1-3) | 2.0 ± 0.7 | 1.8 ± 0.7 | 2.1 ± 0.7 | NS |
| DMS (7-35) | 12.1 ± 3.2 | 11.5 ± 3.3 | 12.7 ± 2.9 | NS |
| MIS (0-30) | 8.3 ± 4.2 | 7.5 ± 4.6 | 9.2 ± 3.7 | NS |
| Annual hospitalization frequency (H1)* | 1.99 ± 3.14 | 2.58 ± 3.88 | 1.33 ± 1.84 | NS |
| Annual hospitalization days (H2)* | 12.7 ± 24.3 | 15.5 ± 25.9 | 9.5 ± 9.0 | NS |
| Triceps skinfold (mm) | 14.7 ± 10.7 | 10.7 ± 8.2 | 19.2 ± 11.5 | 0.001 |
| Biceps skinfold (mm) | 10.5 ± 7.6 | 8.7 ± 6.7 | 12.6 ± 8.0 | 0.02 |
| MAC (cm) | 26.6 ± 4.6 | 27.0 ± 4.1 | 26.1 ± 5.1 | NS |
| MAMC (cm) | 21.9 ± 3.4 | 23.6 ± 2.9 | 20.0 ± 2.9 | 0.001 |
| NIR-measured body fat (%) | 28.0 ± 8.1 | 24.2 ± 6.9 | 32.2 ± 7.2 | 0.001 |
| Weight (lb) | 151.0 ± 45.6 | 157.8 ± 45.9 | 143.3 ± 44.6 | NS |
| Lean body mass by NIR (lb) | 106.9 ± 26.2 | 117.9 ± 25.0 | 94.4 ± 21.8 | 0.001 |
| BMI (kg/m2) | 24.7 ± 6.4 | 24.3 ± 6.1 | 25.2 ± 6.8 | NS |
| Serum | |
|
|
|
| Albumin (g/dL) | 3.8 ± 0.5 | 3.8 ± 0.5 | 3.8 ± 0.4 | NS |
| Creatinine (mg/dL) | 10.5 ± 3.1 | 11.5 ± 3.3 | 9.3 ± 2.4 | 0.002 |
| Cholesterol (mg/dL) | 165.2 ± 33.7 | 159.6 ± 31.7 | 171.4 ± 35.2 | NS |
| TIBC (mg/dL) | 180.0 ± 36.8 | 187.1 ± 42.8 | 171.9 ± 26.9 | NS |
| Transferrin (mg/dL) | 159.1 ± 36.5 | 165.6 ± 44.2 | 151.6 ± 23.7 | NS |
| Ferritin (ng/mL) | 826 ± 472 | 747 ± 442 | 916 ± 494 | NS |
| Iron (mg/dL) | 63.3 ± 28.8 | 67.2 ± 27.3 | 58.8 ± 30.2 | NS |
| Transferrin saturation (%) | 34.9 ± 13.8 | 36.1 ± 12.6 | 33.7 ± 15.1 | NS |
| Hematocrit (%) | 33.7 ± 4.4 | 34.4 ± 4.5 | 32.9 ± 4.3 | NS |
| CRP (ng/mL) | 18.2 ± 42.0 | 22.2 ± 56.1 | 13.6 ± 14.4 | NS |
| Kt/V | 1.37 ± 0.27 | 1.31 ± 0.25 | 1.45 ± 0.29 | 0.02 |
| rHuEPO dose (U/wk) | 7,928 ± 7,730 | 9,159 ± 9,368 | 6,538 ± 5,088 | NS |
Abbreviations: NS, not significant; rHuEPO, recombinant human erythropoietin. *Hospitalization days and frequency of hospitalization are 12-month prospective data. | ||||
| Correlation Coefficient (r) for | SGA | DMS | MIS | Change in r From DMS to MIS (%) |
| Hospitalization frequency (H1) | 0.35* | 0.30* (0.36*) | 0.39* (0.46*) | +30 (+28) |
| Hospitalization days (H2) | 0.34* | 0.30* (0.34*) | 0.41* (0.45*) | +37 (+32) |
| Triceps skinfold | –0.40* | –0.30* (–0.48*) | –0.29* (–0.47*) | –3 (–2) |
| Biceps skinfold | –0.44* | –0.35* (–0.47*) | –0.34* (–0.48*) | +3 (+2) |
| MAC | –0.50* | –0.40* (–0.42*) | –0.46* (–0.49*) | +15 (+16) |
| MAMC | –0.27† | –0.25† (–0.20) | –0.34* (–0.33*) | +56 (+65) |
| NIR body fat (%) | –0.35* | –0.27† (–0.58*) | –0.24† (–0.56*) | –11 (–3) |
| Lean body weight | –0.39* | –0.35* (–0.34*) | –0.40* (–0.43*) | +25 (+26) |
| BMI | –0.49* | –0.42* (–0.44*) | –0.45* (–0.49*) | +7 (+11) |
| Serum | |
|
|
|
| Albumin | –0.31* | –0.27† (–0.27) | –0.50* (–0.51*) | +92 (+88) |
| Creatinine | –0.14 | –0.13 (–0.04) | –0.33* (–0.24†) | +154 (+500) |
| Cholesterol | –0.03 | –0.02 (–0.06) | –0.02 (–0.11) | 0 (+83) |
| TIBC | –0.26† | –0.27† (–0.21) | –0.47* (–0.42*) | +74 (+100) |
| Transferrin | –0.26 | –0.29* (–0.23†) | –0.46* (–0.42*) | +59 (+83) |
| Ferritin | 0.27† | 0.34* (0.28†) | 0.30* (0.23†) | –12 (–17) |
| Hematocrit | –0.08 | –0.16 (–0.16) | –0.25† (–0.24†) | +56 (+50) |
| Iron | –0.06 | –0.12 (0.05) | –0.23† (–0.12) | +92 (+120) |
| CRP | 0.24† | 0.20 (0.19) | 0.40* (0.41*) | +100 (+116) |
| Kt/V | –0.06 | –0.04 (–0.08) | –0.04 (–0.08) | 0 (0) |
| Vintage (dialysis mon) | 0.28 | 0.28† (0.33*) | 0.19 (0.21) | –32 (–36) |
| Age | 0.31* | 0.33* | 0.34* | +3 |
| Race | 0.01 | 0.01 | 0.01 | 0 |
| Sex | 0.16 | 0.18 | 0.20 | +11 |
| Renal disease | 0.00 | 0.04 | 0.04 | 0 |
NOTE. Adjusted correlation coefficients (for age, race, sex, and underlying renal disease) in parentheses. Last column shows percentage of change in r from DMS to MIS. A positive percentage denotes change toward one (±1), or increase in correlation, whereas a negative percentage indicates change toward null; or decrease in correlation. Prospective hospitalization data (days and frequency) are annual. *P < 0.01. †P between 0.05 and 0.01. | ||||
| Correlation Coefficient (r) for | Hospitalization Frequency (H1) | Hospitalization Days (H2) |
| MIS | 0.39* (0.46*) | 0.41* (0.45*) |
| DMS | 0.30* (0.36*) | 0.30* (0.34*) |
| SGA | 0.35* (0.41*) | 0.34* (0.38*) |
| NIR body fat (%) | –0.27† (–0.23†) | –0.17 (–0.14) |
| BMI | –0.22† (–0.24†) | –0.12 (–0.11) |
| Serum | |
|
| Albumin | –0.53* (–0.54*) | –0.54* (–0.54*) |
| Creatinine | –0.14 (–0.28†) | –0.25† (–0.34*) |
| Cholesterol | –0.22† (–0.20) | –0.30* (–0.28†) |
| Ferritin | 0.23† (0.24†) | 0.24† (0.24†) |
| Hematocrit | –0.17 (–0.23†) | –0.17 (–0.20) |
| Iron | –0.12 (–0.14) | –0.24† (0.26†) |
| CRP | 0.41* (0.39*) | 0.28† (0.27†) |
NOTE. Case-mix-adjusted correlation coefficients (controlled for age, race, sex, and underlying renal disease) in parentheses. *P < 0.01. †P between 0.05 and 0.01. | ||
| Variable | Relative Risk for First Hospitalization (95% CI) | P |
| MIS (per 10-U increase) | 3.83 (1.85-7.94) | <0.001 |
| DMS (per 10-U increase) | 3.68 (1.42-9.53) | 0.007 |
| SGA (per 1-U increase) | 1.80 (1.17-2.78) | 0.007 |
| Serum albumin (per 1-g/dL decrease) | 4.48 (2.16-9.28) | <0.001 |
| Creatinine (per 1-mg/dL decrease) | 1.13 (1.00-1.29) | 0.052 |
| CRP (per 10-ng/mL increase) | 1.12 (1.05-1.194) | <0.001 |
NOTE. Magnitude of increments and direction of change in parentheses. Note that serum creatinine level also is mentioned here for comparison despite a P of 0.052. | ||
. Among the three scoring systems, the MIS showed the
strongest association with H3. The relative risk for first hospital
admission for each 10-unit increase in MIS was 3.83 (95% CI, 1.85 to 7.94;
P < 0.001). The DMS had a weaker association. Among other variables,
serum albumin level was a strong predictor of H3 (hazard ratio, 4.48;
95% CI, 2.16 to 9.28; P = 0.001). Serum CRP level also showed a
statistically significant association with H3, and serum creatinine
level had a marginal association. All other clinical, laboratory, and
demographic variables did not show a statistically significant association with
H3 based on Cox proportional hazard modeling.
| Variable | Relative Risk for Death (95% CI) | P |
| MIS (per 10-U increase) | 10.43 (2.28-47.64) | 0.002 |
| DMS (per 10-U increase) | 7.74 (0.94-64.02) | 0.06 |
| SGA (per 1-U increase) | 3.90 (1.29-11.74) | 0.02 |
| Serum albumin (per 1-g/dL decrease) | 7.21 (2.47-20.99) | 0.001 |
| Creatinine (per 1-mg/dL decrease) | 1.33 (1.06-1.65) | 0.01 |
| Cholesterol (per 10-mg/dL decrease) | 1.51 (1.10-2.09) | 0.01 |
| CRP (per 10-ng/mL increase) | 1.13 (1.05-1.22) | 0.001 |
NOTE. Magnitude of increments and direction of change in parentheses. Note that the DMS is mentioned here for comparison despite P of 0.06. | ||
, except for the DMS (P = 0.06), mentioned here for
comparison with other scoring systems. Among the three scoring systems, the MIS
showed the strongest association with prospective mortality. The relative risk
for death for each 10-unit increase in MIS was 10.43 (95% CI, 2.28 to 47.64;
P = 0.002). The SGA had a weaker association, and the DMS did not have a
statistically significant hazard ratio. Among other variables, serum albumin
level also was a strong predictor of mortality (hazard ratio, 7.21; 95% CI, 2.47
to 11.74; P = 0.001). Serum CRP, cholesterol, and creatinine levels also
showed statistically significant associations with annual mortality. All other
clinical, laboratory, and demographic variables did not show a statistically
significant association with mortality based on Cox proportional hazard
modeling.
|
First Hospitalization | Death | ||
| MIS Components | Hazard Ratio (95% CI) | P | Hazard Ratio (95% CI) | P |
| 1. Weight change | 1.78 (1.14-2.79) | 0.012 | 0.63 (0.16-2.40) | 0.49 |
| 2. Dietary intake | 1.05 (0.69-1.59) | 0.82 | 1.88 (0.77-4.56) | 0.17 |
| 3. GI symptoms | 1.28 (0.92-1.762) | 0.14 | 1.93 (0.96-3.89) | 0.06 |
| 4. Functional capacity | 1.52 (1.01-2.29) | 0.049 | 2.97 (1.29-6.86) | 0.01 |
| 5. Vintage and morbidity | 1.01 (0.66-1.54) | 0.96 | 2.14 (1.02-4.48) | 0.05 |
| 6. Subcutaneous fat | 1.77 (1.22-2.56) | 0.003 | 3.45 (1.62-7.34) | 0.005 |
| 7. Muscle wasting | 1.86 (1.22-2.84) | 0.004 | 2.67 (0.73-9.77) | 0.14 |
| 8. BMI | 1.52 (1.02-2.25) | 0.04 | 0.71 (0.28-1.82) | 0.48 |
| 9. Albumin | 1.95 (1.33-2.88) | 0.01 | 2.29 (1.06-4.94) | 0.03 |
| 10. TIBC | 1.60 (0.92-2.78) | 0.09 | 2.39 (1.01-5.70) | 0.05 |
| MIS | 3.83 (1.85-7.94) | <0.001 | 10.43 (2.28-47.64) | 0.002 |
NOTE. Each component consists of a number between 0 and 3 (see text). Note that the MIS, a number between 0 and 30, also is divided into three equal increments (0 to 10, 11 to 20, and 21 to 30); thus, the hazard ratio of the MIS can be compared with those of its three-incremental components. | ||||
).
|
Hazard Ratio of Death | Hospitalization Frequency (H1) | Hospitalization Days (H2) | Hazard Ratio of First Hospitalization (H3) | CRP | Hemoglobin | MAMC | NIR Body Fat (%) |
| 1) * | 0.0747 (0.031) | 0.1337 (0.001) | 0.1486 (<0.001) | 0.0366 (0.006) | 0.0509 (0.045) | 0.0332 (0.108) | 0.0429 (0.067) | 0.2752 (<0.001) |
| 2) * + ALB | 0.1130 (0.005) | 0.2111 (<0.001) | 0.2305 (<0.001) | 0.0466 (0.001) | 0.0995 (0.005) | 0.0680 (0.020) | 0.0489 (0.050) | 0.2186 (<0.001) |
| 3) * + TIBC | 0.0807 (0.022) | 0.1284 (0.001) | 0.1533 (<0.001) | 0.0383 (0.004) | 0.0759 (0.014) | 0.0377 (0.087) | 0.0493 (0.049) | 0.2636 (<0.001) |
| 4) * + BMI | 0.0877 (0.016) | 0.1537 (<0.001) | 0.1340 (0.001) | 0.0384 (0.003) | 0.1011 (0.004) | 0.0391 (0.081) | 0.0691 (0.019) | 0.3213 (<0.001) |
| 5) * + ALB + TIBC | 0.1149 (0.004) | 0.1972 (<0.001) | 0.2275 (<0.001) | 0.0474 (0.001) | 0.1251 (0.001) | 0.0703 (0.018) | 0.0542 (0.039) | 0.2111 (<0.001) |
| 6) * + ALB + BMI | 0.1267 (0.003) | 0.2271 (<0.001) | 0.2066 (<0.001) | 0.0479 (<0.001) | 0.1550 (<0.001) | 0.0724 (0.017) | 0.0738 (0.015) | 0.2615 (<0.001) |
| 7) * + TIBC + BMI | 0.0932 (0.012) | 0.1467 (<0.001) | 0.1388 (0.001) | 0.0399 (0.002) | 0.1272 (0.001) | 0.0430 (0.067) | 0.0740 (0.015) | 0.3057 (<0.001) |
| 8) * + ALB + TIBC + BMI (=MIS) | 0.1280 (0.002) | 0.2123 (<0.001) | 0.2053 (<0.001) | 0.0487 (<0.001) | 0.1792 (<0.001) | 0.0741 (0.015) | 0.0777 (0.013) | 0.2505 (<0.001) |
| 9) * + TIBC + BMI + CRP | 0.1134 (0.005) | 0.1724 (<0.001) | 0.1572 (<0.001) | 0.0448 (0.001) | 0.2000 (<0.001) | 0.0740 (0.015) | 0.1032 (0.004) | 0.2709 (<0.001) |
NOTE. R2 and P for hazard ratios of death, three measures of hospitalization indices (H1 through H3), serum CRP, hemoglobin, MAMC, and NIR body fat percentage to compare the DMS (shown by *), which consists of the first seven components of the MIS, and its modified versions through the stepwise evolution toward the full version of the MIS by adding three additional components to the DMS (*), ie, TIBC, albumin, and BMI. The ninth row is based on replacing albumin with CRP in the MIS. The case-mix-adjusted correlation coefficients R2 are controlled for age, race, sex, and underlying renal disease. For each multivariate R2, P is listed in parentheses. Abbreviation: ALB, albumin. | ||||||||
, the ninth model, created by replacing albumin level with
CRP level, had somewhat stronger correlations with MAMC, NIR body fat, and, of
course, CRP level itself, but its correlations with death hazard ratio, all
three hospitalization indices, and hemoglobin level were less strong. Therefore,
the original MIS with albumin level, not CRP level, appears to be a more useful
scoring system. Other models also were constructed and evaluated, eg, one model
based on the addition of CRP level to the 10 components of the MIS, thus
including 11 components, and although most correlations improved as expected,
the gain was minimal (data not shown). Moreover, CRP is not a routine laboratory
value in many clinical institutions and dialysis centers. Therefore, in general,
the MIS appears to be the best model with overall good correlation with outcome,
as well as indices of inflammation, anthropometric values, and anemia. | Discussion | TOP |