- Click here for explanation of classification scheme.

To provide a systematic examination of specific sociodemographic characteristics and health performance variables associated with frailty in a large, well-defined sample of Mexican-American older adult males and females.

• Participants in the Hispanic Established Populations for Epidemiologic Study of the Elderly (EPESE) at Wave Three:
  • Mexican-American
  • 70 years of age or older
  • Residing in Texas, New Mexico, Colorado, Arizona and California
  • Non-institutionalized.
• Reported having Medicare coverage at EPESE Wave One or Two.

• Refusal to participate
• Proxy interviews, due to the physical nature of assessment required for the study.

Recruitment

The sample from the current study is a sub-sample from the Hispanic Established Populations for Epidemiologic Study of the Elderly (EPESE).

• The EPESE includes:
  • Longitudinal study of Mexican Americans
  • Aged 65 and over
  • Residing in Texas, New Mexico, Colorado, Arizona and California
  • Probability sampling based on counties, census tracts and households within those tracts
  • Total sample which is generalizable to 500,000 Mexican Americans in the southwest
  • Surveyed subjects in waves every two to three years beginning 1993 to 1994.
• The sub-sample:
  • Was recruited from subjects in EPESE Wave Three, 1998 to 1999
  • Included only subjects who reported having Medicare coverage at either Wave One or Two
  • Were randomly selected from this group to participate in this study.

Design

• Participants who qualified were given survey questions in addition to those provided for them in Wave Four of the EPESE study in 2000 and 2001
• Due to the physical nature of the questions, proxy interviews were not allowed
• One year later, subjects were contacted and re-assessed using the same instruments and interview questions
• Regression was used to come up with models that best predict frailty based on sociodemographic and health performance variables.

Statistical Analysis

Examined distribution of variables using descriptives and univariate statistics for all continuous variables and contingency tables (chi-square) for categorical variables

Two multi-variable linear regression models were computed to predict frailty score at baseline

• First model included sociodemographic variables:
  
  • Gender
  • Age
  • Education
  • Marital status
  • Financial strain
  • Number of persons in household.
• Second model included the first model sociodemographic variables as well as performance and health status variables:
  • Lower extremity strength (LE strength)
  • Upper extremity strength (UE strength)
  • Total number of comorbid conditions
  • Mini Mental State Examination (MMSE)
  • Disability score (ADL/IADL)
  • Body mass index (BMI).
• Criteria of P<0.20 selected to identify predictor variables
• Computed different regression models for males and females.

Timing of Measurements

The interview survey and measurements were administered initially. The same interview survey and instruments were used at the one-year follow-up.

Dependent Variables

Variable	How Measured	Scale	Comments
Frailty	Fried and Walston Frailty Index, modified	0-4	Includes sub-scales on:
			Variable	How measured	Scale
			Weight loss	Measured in pounds	1	More than 10 pounds unexplained weight loss between measurements (one-year)
			Exhaustion	Two items from Center for Epidemiological Studies - Depression (CES-D) "I felt everything I did was an effort" and "I could not get going" "How often in the last week did you feel this way"	0	"rarely or none of the time" or "some or a little of the time (one to two days)"
					1	"a moderate amount of the time (three to four days)" or "most of the time (five to seven days)
			Walking speed	Assessed over an eight-foot walk	1	≥9.0 seconds (≥75th percentile)
			Grip strength		1	Men	≤21kg (≤25th percentile)
						Women	≤14kg (≤25th percentile)

Independent Variables

Variable	How Measured		Units
Muscle Strength	Hand held device (Nicholas Manual Muscle Tester, Lafayette Instruments)		0.0 to 199.9
Disability: Activities of Daily Living (ADL) and Instrumental Activities of Daily Living (IADL)	Katz Activities of Daily Living scale, modified; OARS Instrumental Activities of Daily Living scale; Rosow-Breslau scale		0: No limitation
			1: Any IADL or mobility-related ADL limitation
			2: Basic ADL limitation
Cognitive function	The Mini Mental State Examination (MMSE)		0 to 30
Body Mass Index (BMI)	Weight in kilograms/height in meters squared
Sociodemographic Variables	Age
	Gender	Male/female
	Marital status	Married/unmarried
	Years of schooling
	Prevalent medical conditions	Physician diagnosis of heart attack, stroke, arthritis, cancer, hip fracture and diabetes
	Financial strain	How much difficulty meeting monthly payments
	Household structure	Members in household

Control Variables

None noted.

Initial N

• N=800 respondents selected from EPESE study
• N=621 respondents interviewed at beginning of study
• Attrition for initial interview due to:
  • Refusal to participate
  • Proxy interviews
  • Deceased
  • Lost to follow-up
  • In nursing home or too sick.

Attrition (Final N)

• N=551 respondents with both initial and follow-up interview
• Attrition for follow-up interview not noted.

Age

Variable	Female	Male
Age	78.1 (59%)	78.0 (41%)

Ethnicity

All Mexican-American.

Other Relevant Demographics

Male and female groups did not differ based on age, education or household, but did statistically differ* on:

Variable	Female	Male
Married	128 (35%)	189 (75%)

*P<0.05.

Anthropometrics

Males and female groups did not differ based on comorbid conditions, MMSE, BMI, frailty score, but did statistically differ* on:

Variable	Female	Male
LE strength (kg)	24.6 (8.5)	34.9 (11.7)
UE strength (kg)	12.6 (5.1)	19.9 (7.4)
Disability: None	166 (45%)	159 (63%)
Disability: IADL	138 (38%)	68 (27%)
Disability: ADL	64 (17%)	26 (10%)
Frailty: Non-Frail	148 (25%)	125 (50%)
Frailty: Pre-Frail	141 (38%)	83 (33%)
Frailty: Frail	80 (22%)	44 (17%)

*P<0.05.

Location

EPESE was conducted in homes in the Southwest United States (Texas, New Mexico, Colorado, Arizona and California).

• Regression Model One included sociodemographic variables
• Regression Model Two added in health status and performance variables
  • Since gender was statistically different in Model One, Model Two was separated into two separate regression models: One for males and one for females.
• There are seven variables that met the P<0.20 criterion and were considered potential predictor variables for examining frailty at the one-year follow-up:

Males

Variable	P-Value
Disability	Less than 0.01
Comorbidity	0.04
UE Strength	0.03
MMSE	0.05

• The statistically significant independent variables explained 37% of the variance in the summary frailty score in males (Nagelkerke R²=0.37)
• Age, LE strength and BMI were not significant for predicting fraility in males. 

Females

Variable	P-Value
Disability	Less than 0.01
LE Strength	0.04
BMI	0.02

• The statistically significant independent variables explained 29% of the variance in the summary frailty score in females (Nagelkerke R²=0.29)
• Age, comorbidity, UE strength and MMSE were not significant for predicting frailty in females
• Results were further classified to see in how many subjects the gender specific regression equations actually predicted frailty at the one-year follow-up.

Males

83% of males were correctly classified at follow-up.

Predicted Status	Actual Status*
	Frail	Not frail
Frail	38 (23%)	19 (12%)
Not Frail	8 (5%)	99 (60%)

*Subjects with a score of one (pre-frail) were not included.

Females

79% of females were correctly classified at follow-up.

Predicted Status	Actual Status*
	Frail	Not frail
Frail	48 (21%)	29 (12%)
Not Frail	20 (9%)	131 (58%)

*Subjects with a score of one (pre-frail) were not included.

Other Findings

It is noted that:

• Females tend to have both more comorbidity (77% versus 62%) and more disability (49% vs. 34%) than males in the sample
• Females were also more likely to have all three conditions, comorbiditiy, disability and frailty, than men in the sample (23% vs. 16%).

Using regression analysis of variables, the authors found different combinations of variables were statistically significant predictors of frailty in Mexican-American men and women over 70 years of age

• Upper extremity strength, comorbidities, MMSE and disability were associated with frailty in men
• Lower extremity strength, BMI and disability were associated with frailty in women
• Disability (ADL/IADL) was the best single predictor of frailty status at one-year follow-up for both men and women.

Government:

National Institute on Aging, National Institutes of Health

• Subject characteristics and demographics were noted for the initial sample that completed the initial interview (N=621), but not the final sample that completed the one-year follow-up (N=551)
• Accounted for significant gender differences in the sample by running separate regression models for males and females
• Attrition was accounted for in the initial sample, but not in the final one-year follow-up (N=621 at baseline interview and N=551 at one-year follow-up without citing reasons for the attrition)
• Sample had some bias by only selecting individuals with Medicare, although it is noted that this was greater than 80% of the EPESE sample in Wave Three.