- Email: [email protected]
Independence in Activities of Daily Living for End-Stage Renal Disease Patients: Biomedical and Demographic Correlates
Independence in Activities of Daily Living for End-Stage Renal Disease Patients: Biomedical and Demographic Correlates Mara Julius, ScD, Victor M. Hawthorne, MD, Patricia Carpentier-Alting, MA, Jill Kneisley, MS, Robert A. Wolfe, PhD, and Friedrich K. Port, MD • Factors associated with physical well-being were examined in a population-based sample of adult end-stage renal disease (ESRD) patients in Michigan (n = 459). The dependent variables were two measures of physical functioning: (1) a ten-item measure of activities of daily living (ADL), and (2) the 45-item physical dysfunction dimension of the Sickness Impact Profile (SIP). Independent variables included four modalities of treatment (incenter hemodialysis, continuous ambulatory peritoneal dialysis [CAPO], related transplant, and cadaver transplant); primary cause of ESRD (eg, diabetes, glomerulonephritis); comorbidity (other illnesses besides primary cause of ESRD); and demographic characteristics (sex, race, age, marital status, education). ADL and SIP unadjusted mean scores differed significantly by category for each of the eight study factors (analysis of variance [ANOVA], P < 0.0001), with the exception of sex for SIP means. The highest levels of dependency in ADL were reported by patients who were older, female, black, widowed, less educated, treated with in-center hemodialysis, had diabetes as the primary cause of ESRD, and/or reported more comorbidity. The partial effect of each factor on the dependent measures with adjustment for the seven other factors was assessed using analysis of covariance (ANCOVA). In the ADL analysis, sex, race, age, primary cause of kidney failure, and comorbidity were significant factors (probability values ranging from 0.05 for race to 0.0001 for sex, primary cause of ESRD, and comorbidity). The SIP physical dysfunction measure gave slightly different results. Race, age, primary cause of ESRD, comorbid status, and modality of treatment were significantly related to physical dysfunction (P < 0.05 to P < 0.0001). CAPO patients reported the most functional limitation and physical dysfunction and patients with functioning grafts at time of interview reported the least. By far the strongest explanatory factors for both high dependency in daily activities and high physical dysfunction were age (older), primary cause of ESRD (diabetes), and greater number of comorbid conditions. © 1989 by the National Kidney Foundation, Inc. INDEX WORDS: End-stage renal disease; hemodialysis; continuous ambulatory peritoneal dialysis; renal transplantation; activities of daily living; sickness impact profile.
T
HE PURPOSE OF the analyses reported here was to identify the correlates of functional dependence for patients with end-stage renal disease (ESRD), with an emphasis on the four major modalities of treatment: in-center hemodialysis, continuous ambulatory peritoneal dialysis (CAPD), related transplant, and cadaver transplant. In addition, the validation of a brief measure ofthe activities of daily living (ADL) is described, using the physical dysfunction dimension of the Sickness Impact Profile (SIP) as a reference standard. Interest in quality of life as a pertinent issue for health care studies is a fairly recent phenomenon,1.2 and the concept has many definitions. For the Michigan ESRD study, quality of life is operationalized as having three major components: psychologic, social, and physical well-being. An important aspect of physical well-being is functional independence, particularly in daily activities. Constraints on normal activities of daily living can reduce the enjoyment of close friendships, contentment with family, performance and satisfaction with work, and satisfaction with life in general. A
recent report describing research on a sample of ESRD patients in Canada found physical functioning to be of prime importance in patients' assessments of their global quality of life. 3 Previous studies investigating quality of life for ESRD patients have often been conducted with small convenience samples,4 and have frequently focused on a single modality of care. 4,5 Many have not included functional status in their consideration of patients' quality of life, putting emphasis on aspects such as work status or rehabilitation. 6,7 Most recently, however, some studies have inFrom the Departments of Epidemiology and Biostatistics. School of Public Health. University of Michigan. and the Department of Internal Medicine. Medical School. University of Michigan. Ann Arbor. Supported by Health Care Financing Administration Grant No, 14-C-98372/5-04, Address reprint requests to Mara Julius. SeD, University of Michigan, Department of Epidemiology, School of Public Health I, 109 Observatory St, Ann Arbor, MI48109-2029, © 1989 by the National Kidney Foundation, Inc, 0272-6386/89/1301-0017$3.00/0
American Journal of Kidney Diseases, Vol XIII, No 1 (January), 1989: pp 61-69
61
62
JULIUS ET AL
cluded measures of physical functioning 8' lo when assessing ESRD patients' quality of life. A variety of rating scales and other techniques for measuring functional capacity are in use today. A critical review of the merits and drawbacks of each of these measures is not the purpose of this article, but some problems associated with existing instruments for assessing physical function should be mentioned. These problems include the need for highly trained observers or interviewers l l ' 14 ; limited application to one disease, health problem, or type offunction I5 . 16 ; and/or excessive length, which makes administration time consuming and potentially irritating to respondents.1722 The overall study aims and the corresponding type of data collection (ie, structured personal interviews using patient self-reports on all measures) were determining factors in selecting a ten-item ADL index that combined the Katz index of ADL items 12 and items from the Duke University Older Americans Resources and Services (OARS) program protocol.2 1 A novel format, similar to one developed by Lawton,14 was used to minimize gender bias for studies of aged populations. The other measure of physical well-being was the SIP physical dysfunction dimension. The rationale for using this measure was to have a metrically validated scale to use as a reference standard for findings on the ADL index in the Michigan sample of ESRD patients. Use of the SIP also permits comparison between findings on the Michigan study with those of a large national study of ESRD patients conducted by the Battelle Human Affairs Research Centers. \0 METHODS AND MEASURES
Sampling, Data Collection, and Distribution of Cases The four major treatment modalities constitute the strata for sample selection. Date of diagnosis (or onset) of ESRD identified the cohort for the sample in 1985, the date of onset was defined as diagnosis of ESRD on or after November I, 1981 up to and including December 31, 1984. Patients available for sample selection were identified through the Michigan Kidney Registry (MKR), which has records for Michigan ESRD patients from 1973 to the present. The MKR data base is accumulated from an obligatory statewide health information system whereby all treatment facilities report selected data about their patients to the MKR. Data include demographic characteristics, treatment and hospitalization history, and survival information for all Michigan patients treated for ESRD.
Formulating criteria for sample selection by ESRD treatment modality presents a challenge to investigators in this area of research. Initial modality of treatment is often provided on an urgent basis and will not always be the treatment the physician and patient would select for long-term therapy. Additionally, consideration of and evaluation for renal transplantation may be delayed for some patients until after they are stabilized on dialysis. Therefore, for this study, patients were selected by the modality of care they were receiving or awaiting (transplantation) at 6 months after onset of ESRD. Analyses reported here are based on treatment at the time of interview. Further criteria for patient inclusion in the sampling frame were aged ~ 18 years and residency in Michigan, both at the time of first treatment for ESRD and at time of interview. For this report, the potential effects of treatment history before interview were not tested. Future analyses will investigate the stability and variability of treatment experience for these patients. All transplant cases of both types were selected for interviewing, as were all CAPD cases from the cohort. In-center hemodialysis patients were systematically sampled with a random start from a list provided by the MKR. Analyses are based on information collected from 500 patients interviewed during 1985. Data collection was performed by trained interviewers after consent was received from treatment facilities and from respondents who had received information about the study. The response rate for this study was 86.7%. Interviews were conducted at each patient's home or treatment facility, whichever was most convenient for the patient. Twenty-four of the 500 cases were excluded from this analysis for not meeting treatment requirements (ie, treatment other than one of the major four prescribed at time of interview) or for missing data. Another 17 from a variety of racial groups other than white or black were excluded because reliable inferences could not be made with such a small sample size. The resulting sample included 243 male and 216 female patients. Although more male patients received related donor transplants, the distribution of gender did not differ significantly by treatment categories (P > 0.05; see Table I). Other observable differences in treatments can be noted. For instance, blacks were less likely to have functioning grafts at the time of interview and also less likely to have been treated with CAPD than whites. It should be noted that the 1985 sample was drawn from a list of patients in whom onset of ESRD occurred from late 1981 through 1984, when CAPD was just being established as a treatment modality in Michigan. Although proportionately fewer black than white dialysis patients were being treated with CAPD at that time, the situation has changed in the succeeding years (ie, blacks make up 19.2% of the total 1984 to 1987 CAPD subsample v 15.2% for the 1985 subsample). Furthermore, rural/urban demographic differences may partially explain this phenomenon; the Michigan black population is largely urban and treatment facilities, including those for in-center hemodialysis, are concentrated in the cities and other urban areas. (In fact, these facilities are clustered in the southeastern section of Michigan. For instance, of the 48 centers treating chronic renal failure patients, only three are in the upper peninsula of the state and three are in the upper lower peninsula.) Transplant patients were significantly younger than dialysis patients of both types and were more likely to be better edu-
INDEPENDENCE IN LIVING FOR KIDNEY PATIENTS
63
Table 1. Demographics, Primary Cause of ESRD, and Comorbidity by Treatment Modality for a Sample of ESRD Patients (n = 459) In·center Hemodialysis
CAPD
Cadaver Transplant
Related Transplant
%
(37.3%) %
(27.2%) %
(18.1%) %
(17.4%) %
52.9
48.5
54.4
51.8
61.3
3.7
26.4
46.8
15.2
22.9
3.8
66.3*
33.1 33.8 33.1
20.5 26.9 52.6
16.0 38.4 45.6
47.0 47.0 6.0
72.5 27.5 0.0
140.5*
34.8 31.5 33.7
52.1 26.6 21.3
33.6 32.8 33.6
18.1 30.1 51.8
17.5 41.3 41.3
49.0*
15.9 57.7 16.8 9.6
14.6 48.5 19.9 17.0
8.0 67.2 14.4 10.4
22.9 55.4 20.5 1.2
23.8 65.0 10.0 1.3
41.9*
24.6 29.8 20.5 25.1
25.7 19.9 30.4 24.0
16.8 34.4 21.6 27.2
25.3 32.5 13.3 28.9
33.8 41.3 5.0 20.0
37.6*
11.5 26.5 23.4 38.7
9.8 20.1 22.6 47.6
11.7 24.2 25.0 39.2
12.3 32.1 24.7 30.9
13.8 37.5 21.3 27.5
15.8
Total Sample Variable Sex Male Race Black Age (yr) 20·40 41·60 ~61
Education (yr) 1-11 12 ~13
Marital status Never married Married Divorced/separated Widowed Primary cause of ESRD Diabetes Glomerulonephritis Hypertension Othert Illness index (no. of illnesses/conditions) 0 1 2 ~3
X2
*P < 0.0001. tSuch as interstitial pyelonephritis and polycystic kidney disease.
cated. Whereas married patients made up a large share of patients across modalities, because of the age association dialysis patients were more likely to have been widowed and less likely never to have been married. Patients with glomerulonephritis as the primary cause of ESRD were the least likely to be receiving in-center hemodialysis at the time of interview. Diabetic patients were the least likely to have CAPD prescribed, although this pattern has also changed in recent years. Of the total subsample of CAPD patients from 1984 to 1987,21.6% had diabetes as the primary cause, compared with 16.8% of the patients in the 1985 subsample. Many of the diabetic patients in this sample were in the youngest age group (45.1 %), and 66.7% of these had functioning transplants on the whole. In other words, more younger diabetics receive transplants and older diabetic patients are treated with dialysis. The substantial relationships seen among these independent factors make evaluation of their simultaneous association with physical well-being important.
Dependent Measures The ADL measure used to assess independence of function was our own adaptation of the index of ADL12 items and in-
cluded additional items from the OARS protocoJ.2' The format of the questions allowed for measures of both performance and ability to perform in three areas of physical functioning to minimize gender bias. There are four self-care items (eg, feeding and bathing self), four mobility items (eg, climbing stairs, transferring), and three items to measure competency in performing instrumental ADL (eg, doing housework or household chores). Respondents were asked first if they performed each of the selected activities during the preceding 2 weeks and if they needed help to do it. If they reported that they did not perform the activity, they were asked if they did not want or need to do it or if they were unable to do it. Each item was scored on a three-point scale: 0, no help needed; 1, some help needed; and 3, unable to do an activity. After combining the walking and wheeling responses for a mobility item, the ADL index was created by adding scores on all ten items. Thus, the range of possible scores was from 0 (complete independence in daily living) through 30 (complete dependence). Observed scores were highly skewed and ranged from 0 to 20, with a mean score of 2.7 for the total sample. The index was shown to have high internal reliability, with an alpha coefficient of 0.76.
64 The SIP's 45-item physical dysfunction dimension" measures a respondent's perception of sickness-related dysfunction (eg, staying in bed, dressing only witb help). During its extensive development, the SIP has been shown to be both reliable and valid. 23-25 For this study, the SIP questionnaire was self-administered by most respondents while under observation, although slightly over one fourth of the respondents had the items read to them because of personal impediments (vision problems, illiteracy, or receiving dialysis treatments at tbe time of interview). The respondents were requested to relate tbeir health status to a number of activities on the day of the interview. SIP scores of dysfunction were obtained using prescribed procedures. 26 Possible scores ranged from 0 to 100, and observed scores ranged from 0 to 76.8 with a mean of 13.2 (higher scores indicate more dysfunction). The Spearman correlation between tbe SIP and ADL scores was 0.73, showing that patients' scores were well matched across instruments. However, scores on both dependent measures were positively skewed because of the large number of respondents who reported no dependency in ADL and no daily physical dysfunction. Therefore, after one point was added to all scores on both variables, log transformations were performed on tbe indexes to give a closer approximation to tbe normal distribution. In order to allow the reader to compare scores for the two dependent variables, the ADL scores were multiplied by 3.5 before tbe log transformation. After transformation, SIP-log has a range of 0 to 1.89, a mean score of 0.86, and a standard deviation of 0.55. The ADL-Iog variable has a range of 0 to 1.85, a mean score of 0.65, and a standard deviation of 0.60.
Independent Variables Demographic information on ESRD patients from MKR records and survey information included sex, race, age, education, and marital status. Black and white were the racial categories for these analyses. Age at interview and years of education completed were used both as continuous covariates and as trichotomized strata variables. Modality of treatment at tbe time of interview was the treatment variable tested. The sample had more patients witb functioning transplants and more CAPD patients tban would be representative of the state and fewer in-center hemodialysis patients; however, unweighted results are reported in this analysis to describe the sample. Primary cause of chronic kidney failure was obtained from tbe MKR files, and these records provided three major categories of primary renal diagnosesdiabetic renal disease, hypertensive nephrosclerosis, glomerulonephritis-and a fourth category of otber diagnoses. In 1981 tbrough 1983, tbe distribution of new cases of ESRD in Michigan across primary causes was diabetes, 25.3%; hypertension, 28.6%; glomerulonephritis, 20.6%; and other, 25.4%.27 Because we selected all transplant cases meeting the criteria and sampled in-center hemodialysis patients in 1985, those with glomerulonephritis as primary cause of ESRD were represented in a slightly larger proportion and those with hypertension as primary cause in a slightly smaller proportion than the population of new cases (See Table 1). An illness index, derived from a 14-item inventory of chronic conditions and diseases, was a simple count of tbe illnesses and conditions reported by each respondent (artbritis or rheumatism, eye disease, respiratory disease, high BP, heart
JULIUS ET AL
trouble, circulation trouble, diabetes, gastrointestinal disease, liver disease, urinary tract disorders, cancer or leukemia, effects of stroke, and up to one otber category). The primary cause of ESRD was excluded from the count, as were items (headaches and infections) considered to be related to otber conditions. The range of possible scores on the illness index was o to 13. Observed scores ranged from 0 to 8, with a mean of 2.2 for the total sample. The scores were classified into four categories: 0, 1, 2, and;::: 3 conditions. Ordinary least squares (OLS) metbods were used for all estimation and hypothesis test results reported here except for tests of association for contingency tables, in which the X2 test was used. Because many important hypotheses were tested during the course of tbese analyses, only the nominal statistical significance of each result, witbout any adjustment for the overall experiment-wide error rate, is reported. Analyses witb a single categorical factor, referred to as analysis of variance (ANOYA), were used to test for differences between the unadjusted mean scores of SIP-log and ADL-Iog in the groups corresponding to the factor. Analysis of covariance (ANCOYA) was used to study the relationship of SIP-log and ADL-Iog to a categorical factor with adjustment for all otber factors. Inference related to categorical factors (witb adjustment for other covariates) was evaluated based on the mUltiple (partial) F test. If category means differed significantly according to tbe F test, the contrast for each pair of category means was tested as an a priori hypotbesis.
RESULTS
When the unadjusted mean scores on ADL-Iog across categories of the eight explanatory factors (sex, race, age, education, marital status, treatment modality, primary cause of ESRD, and number of comorbid conditions) were examined separately, each factor was found to be significantly related (P < 0.0001) to functional dependence as measured by the ADL index (see Table 2). Tests for contrasts showed that those patients who were female, black, older, had less than a high school education, were widowed, were being treated by in-center hemodialysis or CAPD, had diabetes as the primary cause of ESRD, or had more comorbid conditions had higher levels of functionallimitation. Using ANCOVA, which adjusts for the effects of the other factors in the model simultaneously, race (P < 0.05), age (P < 0.001), sex, primary cause of ESRD, and comorbidity (all at the P < 0.0001 level of significance), were found to be associated with dependency in ADL; marital status, education, and treatment modality failed to differentiate significantly, at the P < 0.05 level, between dependency mean scores. Tests for contrasts demonstrated that ESRD patients in the sample who were female, black, older, receiving CAPD (as opposed to functioning related transplants) had diabetes as
65
INDEPENDENCE IN LIVING FOR KIDNEY PATIENTS Table 2.
Unadjusted and Adjusted Mean Scores of Functional Dependence (ADL-Iog Scores) and Physical Dysfunction (SIP-log Scores) for Categories of Each Explanatory Variable (N = 459) Slp·log
ADL·log Variable
Sex Male Female Race Black White Age (yr)
20-40 41-60 ~61
Unadjusted Scores
Adjusted Scores'
0.54 0.74 t
0.83 0.91
0.84 0.86
0.83 0.58 t
0.74:j: 0.60
1.04 0.80 t
0.94:j: 0.82
0.41 0.63 t 0.90
0.46 0.64§ 0.81
0.65 0.83 t 1.12
0.68 0.85t 1.03
0.83 0.58 t 0.51
0.69 0.62 0.60
1.04 0.76 t 0.78
0.87 0.81 0.86
0.47 0.60 0.74 t 1.01
0.64 0.63 0.66 0.60
0.76 0.81 0.95 t 1.20
0.93 0.82 0.87 0.86
0.80 0.73 0.49 t 0.34
0.64 0.71 0.59 0.55
1.01 0.96 0.73 t 0.54
0.86 0.94:j: 0.82 0.73
0.93 0.45 0.72 t 0.54
0.88 0.52 0.61 t 0.56
1.14 0.69 0.96 t 0.73
1.09 0.76 0.84 t 0.74
0.36 0.41 0.62 t 0.89
0.45 0.51 t 0.61 0.79
0.56 0.56 0.90 t 1.11
0.64 0.65 0.90 t 1.02
Unadjusted Scores
Adjusted Scores'
0.53 0.78 t
Education (yr)
1-11 12 ~13
Marital status Never married Married Divorced/separated Widowed Treatment modality In-center hemodialysis CAPD Cadaver transplant Related transplant Primary cause of ESRD Diabetes Glomerulonephritis Hypertension Otherll Illness index (no. of illnesses/conditions)
0 1 2 ~3
Mean SD
0.65 0.60
0.86 0.55
Higher scores indicate more impairment. • Adjusted to the average for all other factors by covariance analyses. tP < 0.0001. :j:P < 0.05. §P < 0.001. IISuch as interstitial pyelonephritis and polycystic kidney disease.
primary cause of ESRD, or had three or more comorbid conditions had higher levels of dependency. When adjusting for other factors in the ADL-Iog ANCOVA analysis, the rank order for modality of treatment changed. In this new order, CAPD had the highest adjusted mean score, replacing in-center hemodialysis. With or without adjustment, the two dialysis treatments were associated with
higher ADL-Iog mean scores than the two transplant categories. However, the differences were not statistically significant across treatments, except between CAPD and related transplant patients, at P < 0.05, as indicated by tests for contrasts. Differences in unadjusted SIP-log mean scores among categories of explanatory factors also showed the factors to be significantly associated
66
JULIUS ET AL
Table 3.
Comparison of Census Data and Michigan ESRD Sample Data for Those Unable to Perform Basic Activities Without Help by Type of Activity and Sex (% With Functional limitation) Michigan ESRD Data
ADL Items
Mobility Walking/wheelingt Transferri ng Climbing stairs Self-care Feeding self Dressing self Bathing self Using toilet Instrumental Using phone Doing chores Getting placesll
(n
Men = 235)
Women (n = 208)
Census Data' Men
Women
18.1 2.1 17.7
24.1 5.1 34.7:1:
1.5 0.4
1.8 0.6
4.1 6.2 7.8 2.5
1.4 9.7 14.8§ 6.0
0.2 0.6 0.8 0.5
0.2 0.8 1.0 0.6
6.2 42.0 17.3
10.2 62.0:1: 33.8:1:
1.0
1.7
'Census percentages are calculated from rates per 1,000 for data from the 1979 health survey (household interviews with the civilian noninstitutionalized population aged ~ 18 years). tMichigan study combined ability to use wheelchair independently if necessary and ability to walk to produce one score for ability to convey oneself. US census assessed walking only. :l:P:$; 0.001. §P:$; 0.05. liThe US census used needs help in going outside; Michigan ESRD study used needs help in getting places beyond walking distance.
with daily physical dysfunction except for sex. The five factors retaining statistically significant associations with SIP-log physical dysfunction adjusted scores were race and treatment modality at P < 0.05 and age, primary cause of ESRD, and number of comorbid conditions at P ~ 0.0001. ESRD patients in the sample who were black, older, receiving CAPD (as opposed to functioning related transplants), diabetic, or had more comorbid conditions, were shown by tests for contrasts to have reported greater levels of daily physical dysfunction. The pattern of association between modality of treatment and SIP-log scores was similar to that for ADL-log and modality of treatment (see Table 2). When tested separately, the highest SIP-log mean score of impairment was associated with incenter hemodialysis (unadjusted scores). When adjusted for all the other factors, the rank order changed and the highest impairment scores were associated with CAPD treatment. In the ANCOVA analysis for the SIP-log measure, the treatment factor remained a statistically significant indicator of physical well-being at the P < 0.05 level. However, as in the ADL-log analyses, the tests for
contrasts between pairs of categories of the factor showed that only the difference in mean scores between CAPD patients and related transplant patients was significant (P < 0.005). The significant sex difference found for the ADL-log scores was analyzed for each item in the three areas of activity. For every item except feeding self, women reported higher levels of dependency (see Table 3). However, for the mobility items, only scores for climbs stairs showed statistically significant differences between sexes (P ~ 0.001). On the self-care item of bathing self, women's dependency levels were significantly higher than men's (P ~ 0.05), and two of the instrumental items (doing chores and getting places) produced higher impairment rates for women than for men (P ~ 0.001). Although the majority of Michigan ESRD patients were functionally independent, a comparison with findings from the 1979 National Health Interview survey shows that, on complementary items, the patients reported considerably more impairment (see Table 3). It is interesting to note that women in the general population also reported higher rates of functional limitation than men, al-
INDEPENDENCE IN LIVING FOR KIDNEY PATIENTS
though the national study determined that the differences between the rates per 1,000 people generally were not statistically significant by sex.28 DISCUSSION
The analyses presented here illustrate the associations between a number of coexisting factors and current functional limitation for a sample of ESRD patients in Michigan. ANCOVA was used from the regression perspective to examine the effect or contribution of each independent variable while statistically adjusting for the effects of all other independent variables or covariates. In these analyses, the differences examined among the sample ESRD patients include those in sex, race, age, education, marital status, primary cause of ESRD, and health status. Treatment modality was the eighth independent variable of interest. It is known that ESRD patients are not assigned randomly to various treatment modalities, but rather are prescribed a treatment after expert deliberation of patient characteristics and health status, among other considerations. Therefore, treatment effects for functional impairment are substantially associated with those of patient assignment and must be addressed correspondingly. It must be kept in mind that this was a crosssectional or correlational study and the data cannot be interpreted as indicating either cause or effect. However, correlational analyses are useful in confirming the associations among the variables of interest and the strengths ofthese associations. The relationships of all eight independent variables to functional limitation were found to be significant by ANOVA with the ADL measure as the dependent variable (all at P < 0.0001). Thus, all of the factors were included in the covariance analyses to further test relationships. The results of these tests show that education, marital status, and treatment modality do not make statistically significant contributions to the explanation of functional impairment on the ADL measure for this sample. However, sex, age, primary cause of ESRD, and health status as measured by the illness index did prove to be statistically significant indicators of functional impairment (P < 0.001 for age and P < 0.0001 for all others). The explanatory power of race for functional impairment on the ADL was reduced to borderline significance (P < 0.05).
67
The effect of multicollinearity between some important factors may cause those variables to lose statistical significance. For example, age and education are negatively correlated (older patients have less education) at r = - 0.33. However, factors which are not significant statistically may still be important in explaining functional impairment in ESRD. Although age was modestly correlated with health status (r = 0.20), both factors retained significance in the covariance analyses. Age was expected to have a significant linear relationship with functional impairment. Nothing in the literature indicates that research on ESRD patients or other chronically ill patients has ever found age unassociated with functional limitation. 6,7,26,29-31 As expected, the illness index also retained a significant relationship with functional status after adjustments were made for the other factors in the analyses. The primary cause of ESRD was found to be one of the strongest determinants of functional independence. Diabetics reported the highest dependency and patients with glomerulonephritis reported the least. Of course, diabetic complications such as blindness and vascular disease with amputation are expected to cause severe limitations in function. Covariance analyses also revealed that female ESRD patients in the Michigan sample reported significantly more functional dependence in daily activities as measured by the ADL index than male patients. As previously reported, the results of the National Health Interview survey also exhibited this ordering. The sex differences are also supported by previous research by Verbrugge; more women in middle and older ages report trouble performing secondary activities because of chronic health problems than men of the same age. 32 ,33 Haug and Folmar also found that older women were more disadvantaged in ADL than men. 34 Therefore, the ADL index may correctly differentiate functional status for men and women, especially in older age groups. However, there may be some gender bias in the items studied, especially those for instrumental or task-oriented activities. Lawton and Brody have noted the difficulties related to selecting bias-free measures to assess general competence in ADL for studies conducted with elderly men and women. 35
68
It appears that the average ESRD patient in the Michigan sample, with the exception of women who needed help with bathing, was able to carry out self-care functions without assistance. (Overall, < 10% reported needing help in this area of function.) These findings demonstrate once again the resiliency of the human organism. One consistent finding in most studies of high-risk populations (eg, cancer patients, ESRD patients, the aged), regardless of what impairment measures were used, demonstrates that the majority of people are able to perform the simple tasks needed for maintenance of life to the very end. However, more help was needed in the mobility area, with about 18 % of men needing help with walking (or using a wheelchair) and climbing stairs. Women in this sample reported needing help with these activities at 24 % and 35 %, respectively. Both sexes reported the greatest functional limitation in the instrumental area. Forty-two percent of men and 62 % of women reported needing a great deal of assistance in performing routine household tasks, while 17 % of the men and 34 % of women had problems with getting to places outside of walking distance. Previous research with ESRD patients is supported by our findings, which show that treatment for chronic renal failure by transplantation, if successful, has a positive association with physical well-being. In addition, treatment by CAPD is shown to have a weak adverse association with physical well-being. The rank order of functional limitations across modalities of treatment for the Michigan ESRD sample is similar to those found by the Battelle study29.30 and by Simmons and associates. 31 Patients with functioning transplants consistently report less functional limitation and physical dysfunction than those receiving either type of dialysis. The rates of impairment for our sample were higher than those in the two above-mentioned studies because of difference in measures of function, questionnaire format, administration procedures, and sample composition (eg, the Michigan sample included more severe cases, such as patients with diabetes as primary cause of ESRD, older patients). Adjusting for the other factors in the covariance analyses reduced the significance of current treatment and changed the rank order of the type of treatment with highest impairment scores (from in-center hemodialysis to CAPD) for both depend-
JULIUS ET AL
ent measures. However, the differences in mean ADL scores were not statistically significant (except between CAPD and related transplant scores). Our finding that the mean SIP physical dysfunction scores have the same pattern of high and low scores found for the ADL index across categories of the explanatory variables supports the usefulness of the shorter instrument. Some differences in associations with other factors were expected because the two scales measure different but closely related concepts. The strong association (Spearman's Rho = 0.73) between the two variables for the raw scores (before log transformation) provides concurrent validity for the ADL index because the SIP instrument has been extensively tested and may be considered acceptable for use as a reference standard. The new ADL index developed for this research meets two additional key criteria: it is very practical (brief and acceptable to respondents) and very simple to administer. Because of its brevity the ADL index may allow more research into the physical aspects of quality of life issues with ESRD patients and is expected to raise pertinent questions regarding optimal care for patients with ESRD.
REFERENCES 1. Edlund M, Tancredi LR: Quality of life: An ideological critique. Perspect Bioi Med 28:591-607, 1985 2. Chubon RA: Quality of life and persons with end-stage renal disease. Dial Transpl 15:450-452, 1986 3. Churchill DN, Torrance Gw, Taylor DW, et al: Measurement of quality of life in end-stage renal disease: The time trade-off approach. Clin Invest Med 10:14-20, 1987 4. Levy NB, Wynbrandt GD: The quality of life on maintenance haemodialysis. Lancet 1:1328-1330, 1975 5. Richmond JM, Lindsay RM, Burton HJ, et al: Psychological and physiological factors predicting the outcome on home hemodialysis. Clin Nephrol 17: 109-113, 1982 6. Kutner NG, Cardenas DD: Rehabilitation status of chronic renal disease patients undergoing dialysis: Variations by age category. Arch Phys Med Rehab 62:626-630, 1981 7. Gutman RA, Stead Ww, Robinson RR: Physical activity and employment status of patients on maintenance dialysis. N Engl J Med 304:309-313, 1981 8. Rozenbaum EA, Pliskin JS, Barnoon S, et al: Comparative study of costs and quality of life of chronic ambulatory peritoneal dialysis and hemodialysis patients in Israel. Isr J Med Sci 21:335-339, 1985 9. Karnofsky DA, Burchenal JH: The clinical evaluation of chemotherapeutic agents in cancer, in McCleod CM (ed): Evaluation of Chemotherapeutic Agents. Symposium held at New York Academy of Medicine, New York, 1948. New York, Columbia, 1949, pp 191-205
INDEPENDENCE IN LIVING FOR KIDNEY PATIENTS
10. Evans RW, Manninen DL, Garrison LP Jr, et al: The quality of life of patients with end-stage renal disease. N Engl J Med 312:553-559, 1985 11. Hutchinson TA, Boyd NF, Feinstein AR, et al: Scientific problems in clinical scales, as demonstrated in the Karnofsky Index of Performance Status. J Chron Dis 32:661-666, 1979 12. Katz S, Ford AB, Downs TD, et al: Effects of Continued Care: A Study of Chronic Illness in the Home. Washington, DC, US Department of Health, Education, and Welfare, National Center for Health Services Research and Development, 1972 13. Katz S, Akpom CA: A measure of primary sociobiological functions. Int J Health Serv 6:493-507, 1976 14. Lawton MP, Moss M, Fulcomer M, et al: A research and service oriented multilevel assessment instrument. J Gerontol 37:91-99, 1982 15. Mahoney FI, Barthel DW: Functional evaluation: The Barthel Index. Md State Med J 14:61-65, 1965 16. Deniston OL, Jette AM: A functional status assessment instrument: Validation in an elderly population. Health Serv Res 15:21-34, 1980 17. Marsh SK, Konar V, Langton MS, et al: The functional assessment profile: A rehabilitation model. J Appl Rehab Counsel 11: 140-144, 1980 18. Sarno JE, Sarno MT, Levita E: The functional life scale. Arch Phys Med Rehab 54:214-220, 1973 19. Stewart AL, Ware JE Jr, Brook RH: Construction and Scoring of Aggregate Functional Status Indexes, vol I. Santa Monica, CA, Rand, 1982 20. Bergner M, Bobbitt RA, Carter WB, et al: The Sickness Impact Profile: Development and final revision of a health status measure. Med Care 19:787-805, 1981 21. Pfeiffer E: OARS Multidimensional Functional Assessment Questionnaire. Durham, NC, Duke University Center for the Study of Aging and Human Development, 1975 22. Hunt SM, McKenna Sp, McEwen J, et al: The Nottingham health profile: Subjective health status and medical consultations. Soc Sci Med 15A:221-229, 1981
69 23. Bergner M, Bobbitt RA, Kressel S, et al: The Sickness Impact Profile: Conceptual formation and methodology for the development of a health status measure. Int J Health Serv 6:393-415, 1976 24. Carter WB, Bobbitt RA, Bergner M: Validation of an internal scaling: The Sickness Impact Profile. Health Serv Res 74:516-528, 1976 25. Pollard WE, Bobbitt RA, Bergner M: The sickness impact profile: Reliability of a health status measure. Med Care 14: 146-155, 1976 26. Gilson BS, Bergner M, Bobbitt RA, et al: The Sickness Impact Profile: Final development and testing, 1975-1978. Unpublished final report to the National Center for Health Services Research, USDHEW, 1978 27. Michigan Kidney Registry: 1983 Annual Report (unpublished material) 28. National Center for Health Statistics, Feller BA: Americans Needing Help to Function at Home. Vital and Health Statistics, No. 92, 1983 29. Health Care Financing Administration: Special Report: Findings from the National Kidney Dialysis and Kidney Transplantation Study. HCFA Publ No. 03230. Washington, DC, US Government Printing Office, 1987 30. Hart LG, Evans RW: The functional status of ESRD patients as measured by the Sickness Impact Profile. J Chron Dis 40: 117S-130S, 1987 31. Simmons RG, Anderson C, Kamstra L: Comparison of quality of life of patients on continuous ambulatory peritoneal dialysis, hemodialysis, and after transplantation. Am J Kidney Dis 4:253-255, 1984 32. Verbrugge LM: A health profile of older women with comparisons to older men. Res Aging 6:291-322, 1984 33. Verbrugge LM: Gender and health: An update on hypotheses and evidence. J Health Soc Behav 26:156-182, 1985 34. Haug MR, Folmar SJ: Longevity, gender, and life quality. J Health Soc Behav 27:332-345, 1986 35. Lawton MP, Brody EM: Assessment of older people: Self-maintaining and instrumental activities of daily living. Gerontology 9: 179-186, 1986
T
HE PURPOSE OF the analyses reported here was to identify the correlates of functional dependence for patients with end-stage renal disease (ESRD), with an emphasis on the four major modalities of treatment: in-center hemodialysis, continuous ambulatory peritoneal dialysis (CAPD), related transplant, and cadaver transplant. In addition, the validation of a brief measure ofthe activities of daily living (ADL) is described, using the physical dysfunction dimension of the Sickness Impact Profile (SIP) as a reference standard. Interest in quality of life as a pertinent issue for health care studies is a fairly recent phenomenon,1.2 and the concept has many definitions. For the Michigan ESRD study, quality of life is operationalized as having three major components: psychologic, social, and physical well-being. An important aspect of physical well-being is functional independence, particularly in daily activities. Constraints on normal activities of daily living can reduce the enjoyment of close friendships, contentment with family, performance and satisfaction with work, and satisfaction with life in general. A
recent report describing research on a sample of ESRD patients in Canada found physical functioning to be of prime importance in patients' assessments of their global quality of life. 3 Previous studies investigating quality of life for ESRD patients have often been conducted with small convenience samples,4 and have frequently focused on a single modality of care. 4,5 Many have not included functional status in their consideration of patients' quality of life, putting emphasis on aspects such as work status or rehabilitation. 6,7 Most recently, however, some studies have inFrom the Departments of Epidemiology and Biostatistics. School of Public Health. University of Michigan. and the Department of Internal Medicine. Medical School. University of Michigan. Ann Arbor. Supported by Health Care Financing Administration Grant No, 14-C-98372/5-04, Address reprint requests to Mara Julius. SeD, University of Michigan, Department of Epidemiology, School of Public Health I, 109 Observatory St, Ann Arbor, MI48109-2029, © 1989 by the National Kidney Foundation, Inc, 0272-6386/89/1301-0017$3.00/0
American Journal of Kidney Diseases, Vol XIII, No 1 (January), 1989: pp 61-69
61
62
JULIUS ET AL
cluded measures of physical functioning 8' lo when assessing ESRD patients' quality of life. A variety of rating scales and other techniques for measuring functional capacity are in use today. A critical review of the merits and drawbacks of each of these measures is not the purpose of this article, but some problems associated with existing instruments for assessing physical function should be mentioned. These problems include the need for highly trained observers or interviewers l l ' 14 ; limited application to one disease, health problem, or type offunction I5 . 16 ; and/or excessive length, which makes administration time consuming and potentially irritating to respondents.1722 The overall study aims and the corresponding type of data collection (ie, structured personal interviews using patient self-reports on all measures) were determining factors in selecting a ten-item ADL index that combined the Katz index of ADL items 12 and items from the Duke University Older Americans Resources and Services (OARS) program protocol.2 1 A novel format, similar to one developed by Lawton,14 was used to minimize gender bias for studies of aged populations. The other measure of physical well-being was the SIP physical dysfunction dimension. The rationale for using this measure was to have a metrically validated scale to use as a reference standard for findings on the ADL index in the Michigan sample of ESRD patients. Use of the SIP also permits comparison between findings on the Michigan study with those of a large national study of ESRD patients conducted by the Battelle Human Affairs Research Centers. \0 METHODS AND MEASURES
Sampling, Data Collection, and Distribution of Cases The four major treatment modalities constitute the strata for sample selection. Date of diagnosis (or onset) of ESRD identified the cohort for the sample in 1985, the date of onset was defined as diagnosis of ESRD on or after November I, 1981 up to and including December 31, 1984. Patients available for sample selection were identified through the Michigan Kidney Registry (MKR), which has records for Michigan ESRD patients from 1973 to the present. The MKR data base is accumulated from an obligatory statewide health information system whereby all treatment facilities report selected data about their patients to the MKR. Data include demographic characteristics, treatment and hospitalization history, and survival information for all Michigan patients treated for ESRD.
Formulating criteria for sample selection by ESRD treatment modality presents a challenge to investigators in this area of research. Initial modality of treatment is often provided on an urgent basis and will not always be the treatment the physician and patient would select for long-term therapy. Additionally, consideration of and evaluation for renal transplantation may be delayed for some patients until after they are stabilized on dialysis. Therefore, for this study, patients were selected by the modality of care they were receiving or awaiting (transplantation) at 6 months after onset of ESRD. Analyses reported here are based on treatment at the time of interview. Further criteria for patient inclusion in the sampling frame were aged ~ 18 years and residency in Michigan, both at the time of first treatment for ESRD and at time of interview. For this report, the potential effects of treatment history before interview were not tested. Future analyses will investigate the stability and variability of treatment experience for these patients. All transplant cases of both types were selected for interviewing, as were all CAPD cases from the cohort. In-center hemodialysis patients were systematically sampled with a random start from a list provided by the MKR. Analyses are based on information collected from 500 patients interviewed during 1985. Data collection was performed by trained interviewers after consent was received from treatment facilities and from respondents who had received information about the study. The response rate for this study was 86.7%. Interviews were conducted at each patient's home or treatment facility, whichever was most convenient for the patient. Twenty-four of the 500 cases were excluded from this analysis for not meeting treatment requirements (ie, treatment other than one of the major four prescribed at time of interview) or for missing data. Another 17 from a variety of racial groups other than white or black were excluded because reliable inferences could not be made with such a small sample size. The resulting sample included 243 male and 216 female patients. Although more male patients received related donor transplants, the distribution of gender did not differ significantly by treatment categories (P > 0.05; see Table I). Other observable differences in treatments can be noted. For instance, blacks were less likely to have functioning grafts at the time of interview and also less likely to have been treated with CAPD than whites. It should be noted that the 1985 sample was drawn from a list of patients in whom onset of ESRD occurred from late 1981 through 1984, when CAPD was just being established as a treatment modality in Michigan. Although proportionately fewer black than white dialysis patients were being treated with CAPD at that time, the situation has changed in the succeeding years (ie, blacks make up 19.2% of the total 1984 to 1987 CAPD subsample v 15.2% for the 1985 subsample). Furthermore, rural/urban demographic differences may partially explain this phenomenon; the Michigan black population is largely urban and treatment facilities, including those for in-center hemodialysis, are concentrated in the cities and other urban areas. (In fact, these facilities are clustered in the southeastern section of Michigan. For instance, of the 48 centers treating chronic renal failure patients, only three are in the upper peninsula of the state and three are in the upper lower peninsula.) Transplant patients were significantly younger than dialysis patients of both types and were more likely to be better edu-
INDEPENDENCE IN LIVING FOR KIDNEY PATIENTS
63
Table 1. Demographics, Primary Cause of ESRD, and Comorbidity by Treatment Modality for a Sample of ESRD Patients (n = 459) In·center Hemodialysis
CAPD
Cadaver Transplant
Related Transplant
%
(37.3%) %
(27.2%) %
(18.1%) %
(17.4%) %
52.9
48.5
54.4
51.8
61.3
3.7
26.4
46.8
15.2
22.9
3.8
66.3*
33.1 33.8 33.1
20.5 26.9 52.6
16.0 38.4 45.6
47.0 47.0 6.0
72.5 27.5 0.0
140.5*
34.8 31.5 33.7
52.1 26.6 21.3
33.6 32.8 33.6
18.1 30.1 51.8
17.5 41.3 41.3
49.0*
15.9 57.7 16.8 9.6
14.6 48.5 19.9 17.0
8.0 67.2 14.4 10.4
22.9 55.4 20.5 1.2
23.8 65.0 10.0 1.3
41.9*
24.6 29.8 20.5 25.1
25.7 19.9 30.4 24.0
16.8 34.4 21.6 27.2
25.3 32.5 13.3 28.9
33.8 41.3 5.0 20.0
37.6*
11.5 26.5 23.4 38.7
9.8 20.1 22.6 47.6
11.7 24.2 25.0 39.2
12.3 32.1 24.7 30.9
13.8 37.5 21.3 27.5
15.8
Total Sample Variable Sex Male Race Black Age (yr) 20·40 41·60 ~61
Education (yr) 1-11 12 ~13
Marital status Never married Married Divorced/separated Widowed Primary cause of ESRD Diabetes Glomerulonephritis Hypertension Othert Illness index (no. of illnesses/conditions) 0 1 2 ~3
X2
*P < 0.0001. tSuch as interstitial pyelonephritis and polycystic kidney disease.
cated. Whereas married patients made up a large share of patients across modalities, because of the age association dialysis patients were more likely to have been widowed and less likely never to have been married. Patients with glomerulonephritis as the primary cause of ESRD were the least likely to be receiving in-center hemodialysis at the time of interview. Diabetic patients were the least likely to have CAPD prescribed, although this pattern has also changed in recent years. Of the total subsample of CAPD patients from 1984 to 1987,21.6% had diabetes as the primary cause, compared with 16.8% of the patients in the 1985 subsample. Many of the diabetic patients in this sample were in the youngest age group (45.1 %), and 66.7% of these had functioning transplants on the whole. In other words, more younger diabetics receive transplants and older diabetic patients are treated with dialysis. The substantial relationships seen among these independent factors make evaluation of their simultaneous association with physical well-being important.
Dependent Measures The ADL measure used to assess independence of function was our own adaptation of the index of ADL12 items and in-
cluded additional items from the OARS protocoJ.2' The format of the questions allowed for measures of both performance and ability to perform in three areas of physical functioning to minimize gender bias. There are four self-care items (eg, feeding and bathing self), four mobility items (eg, climbing stairs, transferring), and three items to measure competency in performing instrumental ADL (eg, doing housework or household chores). Respondents were asked first if they performed each of the selected activities during the preceding 2 weeks and if they needed help to do it. If they reported that they did not perform the activity, they were asked if they did not want or need to do it or if they were unable to do it. Each item was scored on a three-point scale: 0, no help needed; 1, some help needed; and 3, unable to do an activity. After combining the walking and wheeling responses for a mobility item, the ADL index was created by adding scores on all ten items. Thus, the range of possible scores was from 0 (complete independence in daily living) through 30 (complete dependence). Observed scores were highly skewed and ranged from 0 to 20, with a mean score of 2.7 for the total sample. The index was shown to have high internal reliability, with an alpha coefficient of 0.76.
64 The SIP's 45-item physical dysfunction dimension" measures a respondent's perception of sickness-related dysfunction (eg, staying in bed, dressing only witb help). During its extensive development, the SIP has been shown to be both reliable and valid. 23-25 For this study, the SIP questionnaire was self-administered by most respondents while under observation, although slightly over one fourth of the respondents had the items read to them because of personal impediments (vision problems, illiteracy, or receiving dialysis treatments at tbe time of interview). The respondents were requested to relate tbeir health status to a number of activities on the day of the interview. SIP scores of dysfunction were obtained using prescribed procedures. 26 Possible scores ranged from 0 to 100, and observed scores ranged from 0 to 76.8 with a mean of 13.2 (higher scores indicate more dysfunction). The Spearman correlation between tbe SIP and ADL scores was 0.73, showing that patients' scores were well matched across instruments. However, scores on both dependent measures were positively skewed because of the large number of respondents who reported no dependency in ADL and no daily physical dysfunction. Therefore, after one point was added to all scores on both variables, log transformations were performed on tbe indexes to give a closer approximation to tbe normal distribution. In order to allow the reader to compare scores for the two dependent variables, the ADL scores were multiplied by 3.5 before tbe log transformation. After transformation, SIP-log has a range of 0 to 1.89, a mean score of 0.86, and a standard deviation of 0.55. The ADL-Iog variable has a range of 0 to 1.85, a mean score of 0.65, and a standard deviation of 0.60.
Independent Variables Demographic information on ESRD patients from MKR records and survey information included sex, race, age, education, and marital status. Black and white were the racial categories for these analyses. Age at interview and years of education completed were used both as continuous covariates and as trichotomized strata variables. Modality of treatment at tbe time of interview was the treatment variable tested. The sample had more patients witb functioning transplants and more CAPD patients tban would be representative of the state and fewer in-center hemodialysis patients; however, unweighted results are reported in this analysis to describe the sample. Primary cause of chronic kidney failure was obtained from tbe MKR files, and these records provided three major categories of primary renal diagnosesdiabetic renal disease, hypertensive nephrosclerosis, glomerulonephritis-and a fourth category of otber diagnoses. In 1981 tbrough 1983, tbe distribution of new cases of ESRD in Michigan across primary causes was diabetes, 25.3%; hypertension, 28.6%; glomerulonephritis, 20.6%; and other, 25.4%.27 Because we selected all transplant cases meeting the criteria and sampled in-center hemodialysis patients in 1985, those with glomerulonephritis as primary cause of ESRD were represented in a slightly larger proportion and those with hypertension as primary cause in a slightly smaller proportion than the population of new cases (See Table 1). An illness index, derived from a 14-item inventory of chronic conditions and diseases, was a simple count of tbe illnesses and conditions reported by each respondent (artbritis or rheumatism, eye disease, respiratory disease, high BP, heart
JULIUS ET AL
trouble, circulation trouble, diabetes, gastrointestinal disease, liver disease, urinary tract disorders, cancer or leukemia, effects of stroke, and up to one otber category). The primary cause of ESRD was excluded from the count, as were items (headaches and infections) considered to be related to otber conditions. The range of possible scores on the illness index was o to 13. Observed scores ranged from 0 to 8, with a mean of 2.2 for the total sample. The scores were classified into four categories: 0, 1, 2, and;::: 3 conditions. Ordinary least squares (OLS) metbods were used for all estimation and hypothesis test results reported here except for tests of association for contingency tables, in which the X2 test was used. Because many important hypotheses were tested during the course of tbese analyses, only the nominal statistical significance of each result, witbout any adjustment for the overall experiment-wide error rate, is reported. Analyses witb a single categorical factor, referred to as analysis of variance (ANOYA), were used to test for differences between the unadjusted mean scores of SIP-log and ADL-Iog in the groups corresponding to the factor. Analysis of covariance (ANCOYA) was used to study the relationship of SIP-log and ADL-Iog to a categorical factor with adjustment for all otber factors. Inference related to categorical factors (witb adjustment for other covariates) was evaluated based on the mUltiple (partial) F test. If category means differed significantly according to tbe F test, the contrast for each pair of category means was tested as an a priori hypotbesis.
RESULTS
When the unadjusted mean scores on ADL-Iog across categories of the eight explanatory factors (sex, race, age, education, marital status, treatment modality, primary cause of ESRD, and number of comorbid conditions) were examined separately, each factor was found to be significantly related (P < 0.0001) to functional dependence as measured by the ADL index (see Table 2). Tests for contrasts showed that those patients who were female, black, older, had less than a high school education, were widowed, were being treated by in-center hemodialysis or CAPD, had diabetes as the primary cause of ESRD, or had more comorbid conditions had higher levels of functionallimitation. Using ANCOVA, which adjusts for the effects of the other factors in the model simultaneously, race (P < 0.05), age (P < 0.001), sex, primary cause of ESRD, and comorbidity (all at the P < 0.0001 level of significance), were found to be associated with dependency in ADL; marital status, education, and treatment modality failed to differentiate significantly, at the P < 0.05 level, between dependency mean scores. Tests for contrasts demonstrated that ESRD patients in the sample who were female, black, older, receiving CAPD (as opposed to functioning related transplants) had diabetes as
65
INDEPENDENCE IN LIVING FOR KIDNEY PATIENTS Table 2.
Unadjusted and Adjusted Mean Scores of Functional Dependence (ADL-Iog Scores) and Physical Dysfunction (SIP-log Scores) for Categories of Each Explanatory Variable (N = 459) Slp·log
ADL·log Variable
Sex Male Female Race Black White Age (yr)
20-40 41-60 ~61
Unadjusted Scores
Adjusted Scores'
0.54 0.74 t
0.83 0.91
0.84 0.86
0.83 0.58 t
0.74:j: 0.60
1.04 0.80 t
0.94:j: 0.82
0.41 0.63 t 0.90
0.46 0.64§ 0.81
0.65 0.83 t 1.12
0.68 0.85t 1.03
0.83 0.58 t 0.51
0.69 0.62 0.60
1.04 0.76 t 0.78
0.87 0.81 0.86
0.47 0.60 0.74 t 1.01
0.64 0.63 0.66 0.60
0.76 0.81 0.95 t 1.20
0.93 0.82 0.87 0.86
0.80 0.73 0.49 t 0.34
0.64 0.71 0.59 0.55
1.01 0.96 0.73 t 0.54
0.86 0.94:j: 0.82 0.73
0.93 0.45 0.72 t 0.54
0.88 0.52 0.61 t 0.56
1.14 0.69 0.96 t 0.73
1.09 0.76 0.84 t 0.74
0.36 0.41 0.62 t 0.89
0.45 0.51 t 0.61 0.79
0.56 0.56 0.90 t 1.11
0.64 0.65 0.90 t 1.02
Unadjusted Scores
Adjusted Scores'
0.53 0.78 t
Education (yr)
1-11 12 ~13
Marital status Never married Married Divorced/separated Widowed Treatment modality In-center hemodialysis CAPD Cadaver transplant Related transplant Primary cause of ESRD Diabetes Glomerulonephritis Hypertension Otherll Illness index (no. of illnesses/conditions)
0 1 2 ~3
Mean SD
0.65 0.60
0.86 0.55
Higher scores indicate more impairment. • Adjusted to the average for all other factors by covariance analyses. tP < 0.0001. :j:P < 0.05. §P < 0.001. IISuch as interstitial pyelonephritis and polycystic kidney disease.
primary cause of ESRD, or had three or more comorbid conditions had higher levels of dependency. When adjusting for other factors in the ADL-Iog ANCOVA analysis, the rank order for modality of treatment changed. In this new order, CAPD had the highest adjusted mean score, replacing in-center hemodialysis. With or without adjustment, the two dialysis treatments were associated with
higher ADL-Iog mean scores than the two transplant categories. However, the differences were not statistically significant across treatments, except between CAPD and related transplant patients, at P < 0.05, as indicated by tests for contrasts. Differences in unadjusted SIP-log mean scores among categories of explanatory factors also showed the factors to be significantly associated
66
JULIUS ET AL
Table 3.
Comparison of Census Data and Michigan ESRD Sample Data for Those Unable to Perform Basic Activities Without Help by Type of Activity and Sex (% With Functional limitation) Michigan ESRD Data
ADL Items
Mobility Walking/wheelingt Transferri ng Climbing stairs Self-care Feeding self Dressing self Bathing self Using toilet Instrumental Using phone Doing chores Getting placesll
(n
Men = 235)
Women (n = 208)
Census Data' Men
Women
18.1 2.1 17.7
24.1 5.1 34.7:1:
1.5 0.4
1.8 0.6
4.1 6.2 7.8 2.5
1.4 9.7 14.8§ 6.0
0.2 0.6 0.8 0.5
0.2 0.8 1.0 0.6
6.2 42.0 17.3
10.2 62.0:1: 33.8:1:
1.0
1.7
'Census percentages are calculated from rates per 1,000 for data from the 1979 health survey (household interviews with the civilian noninstitutionalized population aged ~ 18 years). tMichigan study combined ability to use wheelchair independently if necessary and ability to walk to produce one score for ability to convey oneself. US census assessed walking only. :l:P:$; 0.001. §P:$; 0.05. liThe US census used needs help in going outside; Michigan ESRD study used needs help in getting places beyond walking distance.
with daily physical dysfunction except for sex. The five factors retaining statistically significant associations with SIP-log physical dysfunction adjusted scores were race and treatment modality at P < 0.05 and age, primary cause of ESRD, and number of comorbid conditions at P ~ 0.0001. ESRD patients in the sample who were black, older, receiving CAPD (as opposed to functioning related transplants), diabetic, or had more comorbid conditions, were shown by tests for contrasts to have reported greater levels of daily physical dysfunction. The pattern of association between modality of treatment and SIP-log scores was similar to that for ADL-log and modality of treatment (see Table 2). When tested separately, the highest SIP-log mean score of impairment was associated with incenter hemodialysis (unadjusted scores). When adjusted for all the other factors, the rank order changed and the highest impairment scores were associated with CAPD treatment. In the ANCOVA analysis for the SIP-log measure, the treatment factor remained a statistically significant indicator of physical well-being at the P < 0.05 level. However, as in the ADL-log analyses, the tests for
contrasts between pairs of categories of the factor showed that only the difference in mean scores between CAPD patients and related transplant patients was significant (P < 0.005). The significant sex difference found for the ADL-log scores was analyzed for each item in the three areas of activity. For every item except feeding self, women reported higher levels of dependency (see Table 3). However, for the mobility items, only scores for climbs stairs showed statistically significant differences between sexes (P ~ 0.001). On the self-care item of bathing self, women's dependency levels were significantly higher than men's (P ~ 0.05), and two of the instrumental items (doing chores and getting places) produced higher impairment rates for women than for men (P ~ 0.001). Although the majority of Michigan ESRD patients were functionally independent, a comparison with findings from the 1979 National Health Interview survey shows that, on complementary items, the patients reported considerably more impairment (see Table 3). It is interesting to note that women in the general population also reported higher rates of functional limitation than men, al-
INDEPENDENCE IN LIVING FOR KIDNEY PATIENTS
though the national study determined that the differences between the rates per 1,000 people generally were not statistically significant by sex.28 DISCUSSION
The analyses presented here illustrate the associations between a number of coexisting factors and current functional limitation for a sample of ESRD patients in Michigan. ANCOVA was used from the regression perspective to examine the effect or contribution of each independent variable while statistically adjusting for the effects of all other independent variables or covariates. In these analyses, the differences examined among the sample ESRD patients include those in sex, race, age, education, marital status, primary cause of ESRD, and health status. Treatment modality was the eighth independent variable of interest. It is known that ESRD patients are not assigned randomly to various treatment modalities, but rather are prescribed a treatment after expert deliberation of patient characteristics and health status, among other considerations. Therefore, treatment effects for functional impairment are substantially associated with those of patient assignment and must be addressed correspondingly. It must be kept in mind that this was a crosssectional or correlational study and the data cannot be interpreted as indicating either cause or effect. However, correlational analyses are useful in confirming the associations among the variables of interest and the strengths ofthese associations. The relationships of all eight independent variables to functional limitation were found to be significant by ANOVA with the ADL measure as the dependent variable (all at P < 0.0001). Thus, all of the factors were included in the covariance analyses to further test relationships. The results of these tests show that education, marital status, and treatment modality do not make statistically significant contributions to the explanation of functional impairment on the ADL measure for this sample. However, sex, age, primary cause of ESRD, and health status as measured by the illness index did prove to be statistically significant indicators of functional impairment (P < 0.001 for age and P < 0.0001 for all others). The explanatory power of race for functional impairment on the ADL was reduced to borderline significance (P < 0.05).
67
The effect of multicollinearity between some important factors may cause those variables to lose statistical significance. For example, age and education are negatively correlated (older patients have less education) at r = - 0.33. However, factors which are not significant statistically may still be important in explaining functional impairment in ESRD. Although age was modestly correlated with health status (r = 0.20), both factors retained significance in the covariance analyses. Age was expected to have a significant linear relationship with functional impairment. Nothing in the literature indicates that research on ESRD patients or other chronically ill patients has ever found age unassociated with functional limitation. 6,7,26,29-31 As expected, the illness index also retained a significant relationship with functional status after adjustments were made for the other factors in the analyses. The primary cause of ESRD was found to be one of the strongest determinants of functional independence. Diabetics reported the highest dependency and patients with glomerulonephritis reported the least. Of course, diabetic complications such as blindness and vascular disease with amputation are expected to cause severe limitations in function. Covariance analyses also revealed that female ESRD patients in the Michigan sample reported significantly more functional dependence in daily activities as measured by the ADL index than male patients. As previously reported, the results of the National Health Interview survey also exhibited this ordering. The sex differences are also supported by previous research by Verbrugge; more women in middle and older ages report trouble performing secondary activities because of chronic health problems than men of the same age. 32 ,33 Haug and Folmar also found that older women were more disadvantaged in ADL than men. 34 Therefore, the ADL index may correctly differentiate functional status for men and women, especially in older age groups. However, there may be some gender bias in the items studied, especially those for instrumental or task-oriented activities. Lawton and Brody have noted the difficulties related to selecting bias-free measures to assess general competence in ADL for studies conducted with elderly men and women. 35
68
It appears that the average ESRD patient in the Michigan sample, with the exception of women who needed help with bathing, was able to carry out self-care functions without assistance. (Overall, < 10% reported needing help in this area of function.) These findings demonstrate once again the resiliency of the human organism. One consistent finding in most studies of high-risk populations (eg, cancer patients, ESRD patients, the aged), regardless of what impairment measures were used, demonstrates that the majority of people are able to perform the simple tasks needed for maintenance of life to the very end. However, more help was needed in the mobility area, with about 18 % of men needing help with walking (or using a wheelchair) and climbing stairs. Women in this sample reported needing help with these activities at 24 % and 35 %, respectively. Both sexes reported the greatest functional limitation in the instrumental area. Forty-two percent of men and 62 % of women reported needing a great deal of assistance in performing routine household tasks, while 17 % of the men and 34 % of women had problems with getting to places outside of walking distance. Previous research with ESRD patients is supported by our findings, which show that treatment for chronic renal failure by transplantation, if successful, has a positive association with physical well-being. In addition, treatment by CAPD is shown to have a weak adverse association with physical well-being. The rank order of functional limitations across modalities of treatment for the Michigan ESRD sample is similar to those found by the Battelle study29.30 and by Simmons and associates. 31 Patients with functioning transplants consistently report less functional limitation and physical dysfunction than those receiving either type of dialysis. The rates of impairment for our sample were higher than those in the two above-mentioned studies because of difference in measures of function, questionnaire format, administration procedures, and sample composition (eg, the Michigan sample included more severe cases, such as patients with diabetes as primary cause of ESRD, older patients). Adjusting for the other factors in the covariance analyses reduced the significance of current treatment and changed the rank order of the type of treatment with highest impairment scores (from in-center hemodialysis to CAPD) for both depend-
JULIUS ET AL
ent measures. However, the differences in mean ADL scores were not statistically significant (except between CAPD and related transplant scores). Our finding that the mean SIP physical dysfunction scores have the same pattern of high and low scores found for the ADL index across categories of the explanatory variables supports the usefulness of the shorter instrument. Some differences in associations with other factors were expected because the two scales measure different but closely related concepts. The strong association (Spearman's Rho = 0.73) between the two variables for the raw scores (before log transformation) provides concurrent validity for the ADL index because the SIP instrument has been extensively tested and may be considered acceptable for use as a reference standard. The new ADL index developed for this research meets two additional key criteria: it is very practical (brief and acceptable to respondents) and very simple to administer. Because of its brevity the ADL index may allow more research into the physical aspects of quality of life issues with ESRD patients and is expected to raise pertinent questions regarding optimal care for patients with ESRD.
REFERENCES 1. Edlund M, Tancredi LR: Quality of life: An ideological critique. Perspect Bioi Med 28:591-607, 1985 2. Chubon RA: Quality of life and persons with end-stage renal disease. Dial Transpl 15:450-452, 1986 3. Churchill DN, Torrance Gw, Taylor DW, et al: Measurement of quality of life in end-stage renal disease: The time trade-off approach. Clin Invest Med 10:14-20, 1987 4. Levy NB, Wynbrandt GD: The quality of life on maintenance haemodialysis. Lancet 1:1328-1330, 1975 5. Richmond JM, Lindsay RM, Burton HJ, et al: Psychological and physiological factors predicting the outcome on home hemodialysis. Clin Nephrol 17: 109-113, 1982 6. Kutner NG, Cardenas DD: Rehabilitation status of chronic renal disease patients undergoing dialysis: Variations by age category. Arch Phys Med Rehab 62:626-630, 1981 7. Gutman RA, Stead Ww, Robinson RR: Physical activity and employment status of patients on maintenance dialysis. N Engl J Med 304:309-313, 1981 8. Rozenbaum EA, Pliskin JS, Barnoon S, et al: Comparative study of costs and quality of life of chronic ambulatory peritoneal dialysis and hemodialysis patients in Israel. Isr J Med Sci 21:335-339, 1985 9. Karnofsky DA, Burchenal JH: The clinical evaluation of chemotherapeutic agents in cancer, in McCleod CM (ed): Evaluation of Chemotherapeutic Agents. Symposium held at New York Academy of Medicine, New York, 1948. New York, Columbia, 1949, pp 191-205
INDEPENDENCE IN LIVING FOR KIDNEY PATIENTS
10. Evans RW, Manninen DL, Garrison LP Jr, et al: The quality of life of patients with end-stage renal disease. N Engl J Med 312:553-559, 1985 11. Hutchinson TA, Boyd NF, Feinstein AR, et al: Scientific problems in clinical scales, as demonstrated in the Karnofsky Index of Performance Status. J Chron Dis 32:661-666, 1979 12. Katz S, Ford AB, Downs TD, et al: Effects of Continued Care: A Study of Chronic Illness in the Home. Washington, DC, US Department of Health, Education, and Welfare, National Center for Health Services Research and Development, 1972 13. Katz S, Akpom CA: A measure of primary sociobiological functions. Int J Health Serv 6:493-507, 1976 14. Lawton MP, Moss M, Fulcomer M, et al: A research and service oriented multilevel assessment instrument. J Gerontol 37:91-99, 1982 15. Mahoney FI, Barthel DW: Functional evaluation: The Barthel Index. Md State Med J 14:61-65, 1965 16. Deniston OL, Jette AM: A functional status assessment instrument: Validation in an elderly population. Health Serv Res 15:21-34, 1980 17. Marsh SK, Konar V, Langton MS, et al: The functional assessment profile: A rehabilitation model. J Appl Rehab Counsel 11: 140-144, 1980 18. Sarno JE, Sarno MT, Levita E: The functional life scale. Arch Phys Med Rehab 54:214-220, 1973 19. Stewart AL, Ware JE Jr, Brook RH: Construction and Scoring of Aggregate Functional Status Indexes, vol I. Santa Monica, CA, Rand, 1982 20. Bergner M, Bobbitt RA, Carter WB, et al: The Sickness Impact Profile: Development and final revision of a health status measure. Med Care 19:787-805, 1981 21. Pfeiffer E: OARS Multidimensional Functional Assessment Questionnaire. Durham, NC, Duke University Center for the Study of Aging and Human Development, 1975 22. Hunt SM, McKenna Sp, McEwen J, et al: The Nottingham health profile: Subjective health status and medical consultations. Soc Sci Med 15A:221-229, 1981
69 23. Bergner M, Bobbitt RA, Kressel S, et al: The Sickness Impact Profile: Conceptual formation and methodology for the development of a health status measure. Int J Health Serv 6:393-415, 1976 24. Carter WB, Bobbitt RA, Bergner M: Validation of an internal scaling: The Sickness Impact Profile. Health Serv Res 74:516-528, 1976 25. Pollard WE, Bobbitt RA, Bergner M: The sickness impact profile: Reliability of a health status measure. Med Care 14: 146-155, 1976 26. Gilson BS, Bergner M, Bobbitt RA, et al: The Sickness Impact Profile: Final development and testing, 1975-1978. Unpublished final report to the National Center for Health Services Research, USDHEW, 1978 27. Michigan Kidney Registry: 1983 Annual Report (unpublished material) 28. National Center for Health Statistics, Feller BA: Americans Needing Help to Function at Home. Vital and Health Statistics, No. 92, 1983 29. Health Care Financing Administration: Special Report: Findings from the National Kidney Dialysis and Kidney Transplantation Study. HCFA Publ No. 03230. Washington, DC, US Government Printing Office, 1987 30. Hart LG, Evans RW: The functional status of ESRD patients as measured by the Sickness Impact Profile. J Chron Dis 40: 117S-130S, 1987 31. Simmons RG, Anderson C, Kamstra L: Comparison of quality of life of patients on continuous ambulatory peritoneal dialysis, hemodialysis, and after transplantation. Am J Kidney Dis 4:253-255, 1984 32. Verbrugge LM: A health profile of older women with comparisons to older men. Res Aging 6:291-322, 1984 33. Verbrugge LM: Gender and health: An update on hypotheses and evidence. J Health Soc Behav 26:156-182, 1985 34. Haug MR, Folmar SJ: Longevity, gender, and life quality. J Health Soc Behav 27:332-345, 1986 35. Lawton MP, Brody EM: Assessment of older people: Self-maintaining and instrumental activities of daily living. Gerontology 9: 179-186, 1986