American Journal of Epidemiology Copyright ©1992 by The Johns Hopkins University School of Hygiene and Pubfc Health
Vol. 135, No 3 Printed in U.S.A.
AS rights reserved
The Association between the Use of Urinary Catheters and Morbidity and Mortality among Elderly Patients in Nursing Homes Calvin M. Kunln,1 Suzanne Douthitt,1 Judith Dancing,1 Judith Anderson,2 and Melvin Moeschberger2
To determine whether the use of urinary catheters in elderly patients in nursing homes has an independent effect on morbidity and mortality, the authors conducted a 1 -year prospective study among 1,540 patients in a stratified random sample of nursing homes. Patient mortality was assessed at 1 year in relation to the presence or absence of a catheter at entry to the study, acquisition of a catheter, and the proportion of nursing home days spent catheterized during the study year. The independent association of catheter use with mortality was assessed by logistic regression analysis adjusted for age, activities of daily living, mental status, skin condition (decubitus ulcers), and 20 medical diagnoses. The effect of catheterization on hospitalization, use of systemic antimicrobial drugs, and mortality was also examined by matched pairs analysis. At entry, 10.5% of patients had catheters, and they tended to remain catheterized during most of the study year. An additional 10% were catheterized during the year. The following factors were found to have a significant independent association with mortality: urinary catheters, age, mental status or activities of daily living, cancer, cardiac disease, diabetes, and skin condition. There was a stepwise increase in mortality with duration of catneterization. Patients who were catheterized for 76% or more of their days in the nursing home were three times more likely to die within a year. The number of hospttalizations, duration of hospitalization, and use of antimicrobial drugs were all three times greater among catheterized patients. Am J Epidemiol 1992;135:291-301. activities of daily living; aged; antibiotics; catheters, indwelling; costs and cost analysis; mental health; nursing homes
Indwelling urinary catheters are used commonly for the management of incontinent patients in nursing homes (1-3). The risks are well known. The patient inevitably will develop bacteriuria, often with multiple organisms; the infections are virtually impossible to eradicate as long as the catheter remains in place; and there are increased risks of pyelonephritis, bacteremia, and sep-
sis (4-14). An increase in mortality has been reported among hospitalized patients who acquire bacteriuria in association with the urinary catheter (15). In contrast, in a study reported from Sweden (16) of catheterized and noncatheterized elderly patients assigned to two different wards, no difference in mortality between was found over a 4year period.
Received for publication February 19,1991, and in final fcxm June 6, 1991. 1 Department of Internal Medicine, The Ohio State University, Columbus, OH. 2 Department of Preventive MeoWne, The Ohio State University, Cofumbus, OH. Reprint requests to Dr. Calvin M. Kunin, Department of Internal Mecflcirte, The Ohio State University, M110 Starling Loving Hal), 320 West 10th Avenue, Columbus, OH 43210. This study was supported in part by grant 12153 from
the Robert Wood Johnson Foundation. The authors gratefully acknowledge the administration and staff of the following nursing homes in the Columbus, Ohio, metropolitan area for their permission to conduct these studies and for their assistance: Bon-Ing Care Center, Capital South Care Center, Columbus West Park, Edgewood Manor, Heritage House, Mann Nursing Home, Mayfair Village, Northland Terrace, Rosegate Care Center, St. Raphael's Home for the Aged, Wecare, Whetstone Convalescent Center, and the Wortnington Nursing Home.
291
292
Kunin et al.
In a previous study, we found the crude death rate among catheterized elderly nursing home patients to be three times greater than that among those who were not catheterized (17). It was not possible, however to attribute the increased mortality to catheterization alone, since the catheterized patients differed significantly from noncatheterized patients according to mental status, activities of daily living, and the frequency of underlying diseases. All of these factors might have had an independent or combined impact on mortality. The current study was designed to determine whether the use of indwelling catheters in nursing home patients might have an independent effect on morbidity and mortality. Because of ethical issues and problems of obtaining informed consent in nursing home patients (18, 19), it was not possible to perform a randomized clinical trial. The only alternative was to conduct a large, prospective study among representative populations and to address the questions by stepwise logistic regression analysis and by comparison of pairs of catheterized and noncatheterized patients matched for major risk factors for death. METHODS Selection of nursing homes
A one-third sample of 14 nursing homes was randomly selected from a list of 42 skilled or intermediate care facilities with 50 or more beds in the Columbus, Ohio, metropolitan area. The sample was representative of all of the nursing homes in respect to location (zip code), certified level of care (skilled or intermediate), form of ownership (proprietary or philanthropic), and bed size. One nursing home with 87 beds withdrew after entry, because of local administrative problems, and was not included in the study. Characteristics of the nursing homes
Information was obtained on bed size, census, number of patients receiving skilled or intermediate care, staffing patterns, characteristics of the attending physicians, pro-
cedures for the care of incontinent and catheterized patients, and policies concerning referral to emergency care facilities and hospitals. Patient characteristics at entry
The medical and nursing records of all patients were reviewed by two nurses and a medical assistant. Information was obtained for demographic characteristics, medical diagnoses, mental status, activities of daily living, skin and voiding status, level of care, medications, and duration of use and care of an indwelling urinary catheter. Medical diagnoses
The medical diagnoses were based on the International Classification of Diseases, Ninth Revision, Clinical Modification (20). Twenty of the 312 diagnoses reported at the time of entry were chosen for detailed study because of their frequency or relevance to clinically important outcomes. Ten were compiled as general groups: 1) all cancers, 2) pneumonias, 3) diseases of the digestive system, 4) skin conditions, 5) fractures, 6) arthritis, 7) hematologic disorders, 8) organic psychotic conditions, 9) ischemic heart disease, and 10) renal diseases other than infectious. Ten medical diagnoses were selected as separate groups: 1) diabetes, 2) thyroid disorders, 3) fluid and electrolyte disturbances, 4) septicemia, 5) Alzheimer's disease, 6) Parkinson's disease, 7) seizure disorders, 8) hypertension, 9) urinary tract infections, and 10) pulmonary disorders (chronic obstructive pulmonary disease, bronchitis, and emphysema/asthma). An individual diagnosis was recorded only once for each patient, regardless of the number of times it was listed in the record. Classification of patient characteristics
Mental status was classified as 1) alert, no cognitive defects; 2) alert with recent memory loss or decreased ability to concentrate; 3) decreased alertness; and 4) nonresponsive. Activities of daily living was classified as 1) independent; 2) supervised for one or more
Urinary Catheters in Nursing Homes
tasks; 3) minimal assistance for one to three tasks; 4) moderate assistance for four to six tasks; 5) maximum assistance for seven to nine tasks; and 6) total care required. Skin status was categorized in five stages by the extent of tissue involvement (from redness to deep tissue involvement) and by site(s) (head, torso, upper or lower extremities. Voiding status consisted of 1) continent with less than one episode of incontinence per day; 2) incontinent with one or more episodes per day; and 3) indwelling catheter in place. Observations made during the follow-up year
All patients were followed up for 1 year or until they died, were transferred to another nursing home, or were discharged home. Visits were conducted at approximately 3-month intervals. The charts were reviewed to determine voiding status, number of days catheterized, transfers to emergency rooms or hospitals, and number of days of administration of systemic antimicrobial drugs. Information was obtained concerning death, discharge home, transfer to another facility, and hospitalization during the preceding period. Hospital records were reviewed to obtain the diagnoses at admission and discharge, the number of hospitalizations, and the duration of hospitalization. Death certificates were reviewed for stated causes of death. Regression analysis
The data were entered into the Epi Info program, Version 5 (Stone Mountain, Georgia, 1990). Each of the 20 diseases and the patient characteristics recorded at entry and during the study year were examined individually by the chi-square test in relation to mortality by 1 year. All the variables (except those noted below) for which the level of significance was p < 0.2 were included in a stepwise logistic regression analysis to determine the association of risk factors with death. Septicemia and urinary tract infection were excluded in this analysis to avoid confounding, but were included in all other data
293
sets. Patients under 40 years of age were excluded to decrease skewing on age. The Statistical Analysis System (SAS Institute, Carey, North Carolina) program was employed using the mainframe computer at The Ohio State University. First, each of the significant (p < 0.2) disease categories were entered into the stepwise logistic procedure. The diseases that exhibited an association with death at p < 0.2 were retained in the model. Next, the patients' significant (p < 0.2) characteristics were added into the the stepwise logistic, and all independent variables with p < 0.01 were retained in the final model. Separate models were used for 1) patients who had catheters in place at entry to the study, 2) patients who received catheters at any time during the 1-year followup period, and 3) the proportion of nursing home days spent catheterized during the 1year follow-up period (1-25, 26-50, 51-75, or 76-100 percent). The strength of the association between the independent variables and mortality was determined and expressed as an adjusted odds ratio with 95 percent confidence interval. Matched pairs analysis
Matched pairs of catheterized and noncatheterized patients were selected to examine in greater detail the association of catheter risk with several variables, such as number of days hospitalized, hospital diagnoses, and receipt of antibiotics. Matching was used also to provide an independent assessment of the association between catheterization and mortality. Separate matches with those who were never catheterized during the study year were conducted for patients who had catheters at entry and patients who were catheterized at any time during the study year. To adjust for possible confounding risk factors, the pairs were matched by demographic characteristics and by the factors that had been shown to be significantly associated with mortality by the regression analyses. These included age (in decades, up to >80 years), sex, race, activities of daily living, mental status, ischemic heart disease, diabetes, cancer, and skin condition. Statis-
294
Kunin et al.
tical analysis was conducted using the Wilcoxon signed rank test, the exact sign test (21) and McNemar's test (22). RESULTS Characteristics of the study population
The population consisted of 1,540 patients 40 years or older located in 13 nursing homes (table 1). The oldest patient was 106 years old. Most patients were elderly, white females over the age of 70 years. The major sources of payment were Medicaid (64.3 percent) and private pay (32.6 percent); Medicare supported only 1.8 percent of the patients; the Veterans Administration paid for 1.3 percent. Most patients (97.7 percent) received intermediate care, and about one half (50.4 percent) had been in a nursing home for 1 or more years. The populations in the 13 facilities varied in respect to mean age (range, 71.3-85.2 years); proportion of females (range, 61.593.2 percent); mental status, classes 3 and 4 (range, 0-11.7 percent); activities of daily living, classes 5 and 6 (range, 47.9-84.6 percent); number of urinary catheters present at entry (range, 1.4-25.1 percent); and mortality at 1 year (range, 7.6-32.2 percent). One facility, which provided more frequent skilled care than the others, tended to have a younger population, a greater proportion of patients with poor mental status, a higher rate of death, and the most frequent use of urinary catheters.
Status at 1 year of follow-up
The mean period of observation was as follows: for the total population, 300.7 days (standard deviation, 114.8 days; range, 1455 days); for those who were hospitalized, 278.0 days (standard deviation, 115.3 days; range, 1-455 days); and for those who died during the year, 148.1 days (standard deviation, 104.7 days; range, 2-394 days). The status of the population at 1 year is summarized in table 2. Slightly more than half (53.3 percent) the population remained in the nursing home and were not hospitalized or discharged during the period of observation. There were 468 patients who were hospitalized during the year. Of these, 151 (32.3 percent) died in the hospital, 26 (5.6 percent) were transferred to another nursing home, five (1.1 percent) went home, and 293 (62.6 percent) returned to the original nursing home. There were 640 admissions for the 468 hospitalized patients as well as 146 admissions to emergency rooms without subsequent hospitalization. Use of catheters, voiding status, and activities of daily living
At entry to the study 162 (10.5 percent) of the patients were being managed with indwelling urinary catheters. An additional 743 (48.2 percent) of the patients were incontinent, but not using catheters. By the end of the study year, 154 more patients (10.0 percent) had received a catheter (table 3). Patients who had catheters in place at
TABLE 1. Demographic characteristics of a stratified random sample of patients from 13 nursing homes in Columbus, Ohio Deaths within the year
Population at entry No.
%
Mean age*
Standard deviation
No.
%
Female Male
1,168 372
75.8 24.2
81.5 74.6
11.1 12.3
229 83
19.6 22.3
Caucasian Black & other
1,384 156
89.9 10.1
80.4 74.1
11.5 12.8
289 23
20.9 14.7
Total
1,540
100.0
79.8
11.8
312
20.3
• AS patients were 40 years erf age or older.
Urinary Catheters in Nursing Homes
295
TABLE 3. Use and duration of indwelling urinary catheters in a stratified random sample (n = 1,540) of patients from 13 nursing homes in Columbus, Ohio Catheter In place at entry
Catheter placed during the year
No.
%
No.
%
Catheterized patients
162
10.5
154
10.0
Duration of catheter use (% nursing home days) 1-25 26-50 51-75 76-100 Total
13 7 5 137 162
8.0 4.3 3.1 84.6 100.0
101 22 19 12 154
65.6 14.3 12.3 7.8 100.0
TABLE 2. Status after 1 year of follow-up of a stratified random sample of patients from 13 nursing homes in Columbus, Ohio Male
Died Discharged home Transferred to another nursing home Hospitalized and returned to the same nursing home Stayed the year, never hospitalized Total
Female %
83 21
22.3 5.6
229 37
19.6 3.2
312 58
20.3 3.8
26
7.0
30
2.6
56
3.6
83
22.3
210
17.9
293
19.0
159
42.7
662
56.7
821
53.3
372
100.0
1,168
100.0
1,540
100.0
entry tended to remain catheterized throughout most of the year, whereas those who received a catheter during the study year retained it for a shorter period of time (table 3). The most frequent indication for use of catheters was for incontinence in females. The median duration of catheterization before entry into the study in 58 patients for whom reliable information was available was 4 months. About one fourth of these had been catheterized for 12 months or more. The catheters were mostly latex or silicone-coated (76.5 percent), 18 French or larger (67.7 percent), and fitted with a 30ml balloon (69.7 percent). Orders for catheter change were usually at monthly intervals (49.4 percent) or as needed (40.6 percent).
No.
Total
No.
%
No.
%
Logistic regression analysis of risk factors associated with death
The risk factors, other than catheterization, found to have an independent, significant effect on mortality within the study year are shown in table 4. Six conditions were found to be significantly associated with mortality. When mental status (classes 3 and 4) was added to the regression analysis, it was found to have an odds ratio of 4.29 (95 percent confidence interval 2.27-8.11) and to reduce the odds ratio for activities of daily living to 1.43 (95 percent confidence interval 1.05-1.95). Mental status was deleted from the model to avoid confounding with activities of daily living. There were only 46 patients in mental status classes 3 or 4; all
296
Kunin et al.
TABLE 4. Risk factors, exclusive of catrteterization, that had a major impact on mortality at 1 year in a stratified random sample of patients from 13 nursing homes in Columbus, Ohio Mortality Adjusted odds ratiot
95% confidence interval
14.3 22.8
1.00 1.53
1.13-2.08
256 56
18.7 32.6
1.00 2.19
1.53-3.15
762 778
123 189
16.1 24.3
1.00 1.57
1.20-2.06
Diabetes mellltus No Yes
1,287 253
243 69
18.9 27.3
1.00 1.54
1.11-2.13
Skin condition Good Decubitus ulcers
1,162 378
191 121
16.4 32.0
1.00 2.33
1.77-3.07
Age in decades^
1,540
312
20.3
1.17
1.03-1.32
No. of patients observed (n = 1,540)
No.
%
Activities of daily living Classes 1-4 Classes 5-6
462 1,078
66 246
Cancer No Yes
1,368 172
Risk factor*
Cardiac disease No Yes
Deaths
• Risk factors present at entry to the study. f Adjusted against aM other variables isted in the table. All the differences were significant (p a 0 01). % The mean age was 77.7 years for those who lived and 81.8 years for those who died (standard deviation, 15.2 and 10.6 years, respectively).
of these fit into activity of daily living classes 5 and 6. The association between mortality and the use of urinary catheters at entry versus during the study year and by duration of catheterization is shown in table 5. The data are presented as the mortality rate and as adjusted for six independent risk factors. The odds ratios for mortality were significantly (p < 0.01) increased for catheterized patients in each of the comparisons. The largest odds ratio was observed for patients who spent 76-100 percent of their nursing home days catheterized, compared with those who were never catheterized (odds ratio = 2.99). This was similar to the odds ratio of 2.38 found for patients with catheters in place at entry who were compared with those who were never catheterized during the study year. These results are not surprising, since 84.6 percent of patients who had catheters at entry continued to be catheterized for 76 percent or more nursing home days. Smaller odds ratios were observed when patients who
had catheters in place at entry were compared with those who did not, but who may have been catheterized later on, and for patients who received a catheter at any time during the year, regardless of the duration of catheterization, compared with those who were never catheterized. There was an incremental increase in the association between the use of catheters and mortality by duration of catheterization. Analysis of matched pairs
Catheterized and noncatheterized patients were matched according to the sixriskfactors shown in table 4. Several comparisons were performed, including analysis of 1) 146 pairs of patients with catheters in place at entry and patients who were never catheterized during the study year, 2) 279 pairs of patients who received a catheter at any time during the study year and patients who were never catheterized; and 3) 262 patients, matched as in the second analysis, but for
Urinary Catheters in Nursing Homes
297
TABLE 5. Effect of duration of catheterlzation ori mortality at 1 year in a stratified random sample of patients from 13 nursing homes in Columbus, Ohio MortaBty Variable
No. of patients
A/illKtfld
Died
odds ratio*
95% confidence
observed ( n - 1,540)
No.
%
1,224
201
16.4
1.00
162
63
38.9
2.38
1.62-3.50
Catheter in place at entry Not Yes
1,378 162
249 63
18.1 38.9
1.00 2.16
1.48-3.23
Catheter placed during year No Yes
1,224 316
201 111
16.4 35.1
1.00 1.47
1.29-1.66
% nursing home days catheterized None 1-25 26-75 76-100 Total
1,224 114 53 149 1,540
201 27 19 65 312
16.4 23.7 35.8 43.6 20.3
1.00 1.35 2.09 2.99
0.84-2.16 1.14-3.84 2.01-4.42
Never catheterized Catheter in place at entry
Interval
* Adjusted for activities of dally Bving, cancer, cardiac disease, diabetes meWus, skin condition, and age. t Of these 1,378 patients 154 (11.2%) received a catheter at some time during the foDow-up year.
whom mental status, rather than activities of daily living, was used as one of the risk factors. The outcomes were very similar for the three different matched pairs analyses. Data are presented in tables 6 and 7 for group 2, since it contained the largest set of matched pairs and included patients who were catheterized during the study year. The catheterized patients were hospitalized about three times more often; stayed in the hospital about three times as many days; and received antibiotics while in the nursing home about three times longer than the matched, noncatheterized patients, table 6. These findings were consistent even after adjustment for days spent in the nursing home. The discharge diagnoses were reviewed for the 156 catheterized and 37 noncatheterized patients who were admitted to the hospital during the study period. Septicemia was diagnosed in 14 (9.0 percent) of the catheterized patients and in only two (5.4 percent) of the noncatheterized group. Similarly 29, (18.6 percent) of the catheterized patients were admitted for urinary tract
infections, compared with only five (13.5 percent) of the noncatheterized group. The risk of death for pairs of patients with catheters at entry or placed during the study year and noncatheterized patients is shown in table 7. Among the discordant pairs, there were about twice as many pairs in which the catheterized patient died and the noncatheterized patient remained alive as there were pairs in which the catheterized patient lived and the noncatheterized patient died. The most frequent death certificate diagnoses, for catheterized and noncatheterized patients respectively, were as follows: cardiopulmonary arrest, 54.7 versus 45.8 percent; congestive heart failure, 6.7 versus 10.4 percent; pneumonia, 5.3 versus 10.4 percent; respiratory failure, 4.0 versus 4.2 percent; and sepsis, 6.7 versus 2.1 percent. DISCUSSION
This study was designed to determine the independent effect on morbidity and mortality associated with use of indwelling uri-
298
Kunin et al.
TABLE 6. Comparison of hospital admissions and use of antibiotics during the study year among 279 matched pairs of patients in a stratified random sample from 13 nursing homes in Columbus Ohio, according to use of indwelling urinary catheters Catheterized patients
Variable No. of days spent in the nursing home
Noncatheterized patients p vaiu©
Mean
SD*
Medtan
Mean
SD
Median
277.10
125.40
343
306.60
119.60
356
0.90
1.07
0.28
0.52
Vol. 135, No 3 Printed in U.S.A.
AS rights reserved
The Association between the Use of Urinary Catheters and Morbidity and Mortality among Elderly Patients in Nursing Homes Calvin M. Kunln,1 Suzanne Douthitt,1 Judith Dancing,1 Judith Anderson,2 and Melvin Moeschberger2
To determine whether the use of urinary catheters in elderly patients in nursing homes has an independent effect on morbidity and mortality, the authors conducted a 1 -year prospective study among 1,540 patients in a stratified random sample of nursing homes. Patient mortality was assessed at 1 year in relation to the presence or absence of a catheter at entry to the study, acquisition of a catheter, and the proportion of nursing home days spent catheterized during the study year. The independent association of catheter use with mortality was assessed by logistic regression analysis adjusted for age, activities of daily living, mental status, skin condition (decubitus ulcers), and 20 medical diagnoses. The effect of catheterization on hospitalization, use of systemic antimicrobial drugs, and mortality was also examined by matched pairs analysis. At entry, 10.5% of patients had catheters, and they tended to remain catheterized during most of the study year. An additional 10% were catheterized during the year. The following factors were found to have a significant independent association with mortality: urinary catheters, age, mental status or activities of daily living, cancer, cardiac disease, diabetes, and skin condition. There was a stepwise increase in mortality with duration of catneterization. Patients who were catheterized for 76% or more of their days in the nursing home were three times more likely to die within a year. The number of hospttalizations, duration of hospitalization, and use of antimicrobial drugs were all three times greater among catheterized patients. Am J Epidemiol 1992;135:291-301. activities of daily living; aged; antibiotics; catheters, indwelling; costs and cost analysis; mental health; nursing homes
Indwelling urinary catheters are used commonly for the management of incontinent patients in nursing homes (1-3). The risks are well known. The patient inevitably will develop bacteriuria, often with multiple organisms; the infections are virtually impossible to eradicate as long as the catheter remains in place; and there are increased risks of pyelonephritis, bacteremia, and sep-
sis (4-14). An increase in mortality has been reported among hospitalized patients who acquire bacteriuria in association with the urinary catheter (15). In contrast, in a study reported from Sweden (16) of catheterized and noncatheterized elderly patients assigned to two different wards, no difference in mortality between was found over a 4year period.
Received for publication February 19,1991, and in final fcxm June 6, 1991. 1 Department of Internal Medicine, The Ohio State University, Columbus, OH. 2 Department of Preventive MeoWne, The Ohio State University, Cofumbus, OH. Reprint requests to Dr. Calvin M. Kunin, Department of Internal Mecflcirte, The Ohio State University, M110 Starling Loving Hal), 320 West 10th Avenue, Columbus, OH 43210. This study was supported in part by grant 12153 from
the Robert Wood Johnson Foundation. The authors gratefully acknowledge the administration and staff of the following nursing homes in the Columbus, Ohio, metropolitan area for their permission to conduct these studies and for their assistance: Bon-Ing Care Center, Capital South Care Center, Columbus West Park, Edgewood Manor, Heritage House, Mann Nursing Home, Mayfair Village, Northland Terrace, Rosegate Care Center, St. Raphael's Home for the Aged, Wecare, Whetstone Convalescent Center, and the Wortnington Nursing Home.
291
292
Kunin et al.
In a previous study, we found the crude death rate among catheterized elderly nursing home patients to be three times greater than that among those who were not catheterized (17). It was not possible, however to attribute the increased mortality to catheterization alone, since the catheterized patients differed significantly from noncatheterized patients according to mental status, activities of daily living, and the frequency of underlying diseases. All of these factors might have had an independent or combined impact on mortality. The current study was designed to determine whether the use of indwelling catheters in nursing home patients might have an independent effect on morbidity and mortality. Because of ethical issues and problems of obtaining informed consent in nursing home patients (18, 19), it was not possible to perform a randomized clinical trial. The only alternative was to conduct a large, prospective study among representative populations and to address the questions by stepwise logistic regression analysis and by comparison of pairs of catheterized and noncatheterized patients matched for major risk factors for death. METHODS Selection of nursing homes
A one-third sample of 14 nursing homes was randomly selected from a list of 42 skilled or intermediate care facilities with 50 or more beds in the Columbus, Ohio, metropolitan area. The sample was representative of all of the nursing homes in respect to location (zip code), certified level of care (skilled or intermediate), form of ownership (proprietary or philanthropic), and bed size. One nursing home with 87 beds withdrew after entry, because of local administrative problems, and was not included in the study. Characteristics of the nursing homes
Information was obtained on bed size, census, number of patients receiving skilled or intermediate care, staffing patterns, characteristics of the attending physicians, pro-
cedures for the care of incontinent and catheterized patients, and policies concerning referral to emergency care facilities and hospitals. Patient characteristics at entry
The medical and nursing records of all patients were reviewed by two nurses and a medical assistant. Information was obtained for demographic characteristics, medical diagnoses, mental status, activities of daily living, skin and voiding status, level of care, medications, and duration of use and care of an indwelling urinary catheter. Medical diagnoses
The medical diagnoses were based on the International Classification of Diseases, Ninth Revision, Clinical Modification (20). Twenty of the 312 diagnoses reported at the time of entry were chosen for detailed study because of their frequency or relevance to clinically important outcomes. Ten were compiled as general groups: 1) all cancers, 2) pneumonias, 3) diseases of the digestive system, 4) skin conditions, 5) fractures, 6) arthritis, 7) hematologic disorders, 8) organic psychotic conditions, 9) ischemic heart disease, and 10) renal diseases other than infectious. Ten medical diagnoses were selected as separate groups: 1) diabetes, 2) thyroid disorders, 3) fluid and electrolyte disturbances, 4) septicemia, 5) Alzheimer's disease, 6) Parkinson's disease, 7) seizure disorders, 8) hypertension, 9) urinary tract infections, and 10) pulmonary disorders (chronic obstructive pulmonary disease, bronchitis, and emphysema/asthma). An individual diagnosis was recorded only once for each patient, regardless of the number of times it was listed in the record. Classification of patient characteristics
Mental status was classified as 1) alert, no cognitive defects; 2) alert with recent memory loss or decreased ability to concentrate; 3) decreased alertness; and 4) nonresponsive. Activities of daily living was classified as 1) independent; 2) supervised for one or more
Urinary Catheters in Nursing Homes
tasks; 3) minimal assistance for one to three tasks; 4) moderate assistance for four to six tasks; 5) maximum assistance for seven to nine tasks; and 6) total care required. Skin status was categorized in five stages by the extent of tissue involvement (from redness to deep tissue involvement) and by site(s) (head, torso, upper or lower extremities. Voiding status consisted of 1) continent with less than one episode of incontinence per day; 2) incontinent with one or more episodes per day; and 3) indwelling catheter in place. Observations made during the follow-up year
All patients were followed up for 1 year or until they died, were transferred to another nursing home, or were discharged home. Visits were conducted at approximately 3-month intervals. The charts were reviewed to determine voiding status, number of days catheterized, transfers to emergency rooms or hospitals, and number of days of administration of systemic antimicrobial drugs. Information was obtained concerning death, discharge home, transfer to another facility, and hospitalization during the preceding period. Hospital records were reviewed to obtain the diagnoses at admission and discharge, the number of hospitalizations, and the duration of hospitalization. Death certificates were reviewed for stated causes of death. Regression analysis
The data were entered into the Epi Info program, Version 5 (Stone Mountain, Georgia, 1990). Each of the 20 diseases and the patient characteristics recorded at entry and during the study year were examined individually by the chi-square test in relation to mortality by 1 year. All the variables (except those noted below) for which the level of significance was p < 0.2 were included in a stepwise logistic regression analysis to determine the association of risk factors with death. Septicemia and urinary tract infection were excluded in this analysis to avoid confounding, but were included in all other data
293
sets. Patients under 40 years of age were excluded to decrease skewing on age. The Statistical Analysis System (SAS Institute, Carey, North Carolina) program was employed using the mainframe computer at The Ohio State University. First, each of the significant (p < 0.2) disease categories were entered into the stepwise logistic procedure. The diseases that exhibited an association with death at p < 0.2 were retained in the model. Next, the patients' significant (p < 0.2) characteristics were added into the the stepwise logistic, and all independent variables with p < 0.01 were retained in the final model. Separate models were used for 1) patients who had catheters in place at entry to the study, 2) patients who received catheters at any time during the 1-year followup period, and 3) the proportion of nursing home days spent catheterized during the 1year follow-up period (1-25, 26-50, 51-75, or 76-100 percent). The strength of the association between the independent variables and mortality was determined and expressed as an adjusted odds ratio with 95 percent confidence interval. Matched pairs analysis
Matched pairs of catheterized and noncatheterized patients were selected to examine in greater detail the association of catheter risk with several variables, such as number of days hospitalized, hospital diagnoses, and receipt of antibiotics. Matching was used also to provide an independent assessment of the association between catheterization and mortality. Separate matches with those who were never catheterized during the study year were conducted for patients who had catheters at entry and patients who were catheterized at any time during the study year. To adjust for possible confounding risk factors, the pairs were matched by demographic characteristics and by the factors that had been shown to be significantly associated with mortality by the regression analyses. These included age (in decades, up to >80 years), sex, race, activities of daily living, mental status, ischemic heart disease, diabetes, cancer, and skin condition. Statis-
294
Kunin et al.
tical analysis was conducted using the Wilcoxon signed rank test, the exact sign test (21) and McNemar's test (22). RESULTS Characteristics of the study population
The population consisted of 1,540 patients 40 years or older located in 13 nursing homes (table 1). The oldest patient was 106 years old. Most patients were elderly, white females over the age of 70 years. The major sources of payment were Medicaid (64.3 percent) and private pay (32.6 percent); Medicare supported only 1.8 percent of the patients; the Veterans Administration paid for 1.3 percent. Most patients (97.7 percent) received intermediate care, and about one half (50.4 percent) had been in a nursing home for 1 or more years. The populations in the 13 facilities varied in respect to mean age (range, 71.3-85.2 years); proportion of females (range, 61.593.2 percent); mental status, classes 3 and 4 (range, 0-11.7 percent); activities of daily living, classes 5 and 6 (range, 47.9-84.6 percent); number of urinary catheters present at entry (range, 1.4-25.1 percent); and mortality at 1 year (range, 7.6-32.2 percent). One facility, which provided more frequent skilled care than the others, tended to have a younger population, a greater proportion of patients with poor mental status, a higher rate of death, and the most frequent use of urinary catheters.
Status at 1 year of follow-up
The mean period of observation was as follows: for the total population, 300.7 days (standard deviation, 114.8 days; range, 1455 days); for those who were hospitalized, 278.0 days (standard deviation, 115.3 days; range, 1-455 days); and for those who died during the year, 148.1 days (standard deviation, 104.7 days; range, 2-394 days). The status of the population at 1 year is summarized in table 2. Slightly more than half (53.3 percent) the population remained in the nursing home and were not hospitalized or discharged during the period of observation. There were 468 patients who were hospitalized during the year. Of these, 151 (32.3 percent) died in the hospital, 26 (5.6 percent) were transferred to another nursing home, five (1.1 percent) went home, and 293 (62.6 percent) returned to the original nursing home. There were 640 admissions for the 468 hospitalized patients as well as 146 admissions to emergency rooms without subsequent hospitalization. Use of catheters, voiding status, and activities of daily living
At entry to the study 162 (10.5 percent) of the patients were being managed with indwelling urinary catheters. An additional 743 (48.2 percent) of the patients were incontinent, but not using catheters. By the end of the study year, 154 more patients (10.0 percent) had received a catheter (table 3). Patients who had catheters in place at
TABLE 1. Demographic characteristics of a stratified random sample of patients from 13 nursing homes in Columbus, Ohio Deaths within the year
Population at entry No.
%
Mean age*
Standard deviation
No.
%
Female Male
1,168 372
75.8 24.2
81.5 74.6
11.1 12.3
229 83
19.6 22.3
Caucasian Black & other
1,384 156
89.9 10.1
80.4 74.1
11.5 12.8
289 23
20.9 14.7
Total
1,540
100.0
79.8
11.8
312
20.3
• AS patients were 40 years erf age or older.
Urinary Catheters in Nursing Homes
295
TABLE 3. Use and duration of indwelling urinary catheters in a stratified random sample (n = 1,540) of patients from 13 nursing homes in Columbus, Ohio Catheter In place at entry
Catheter placed during the year
No.
%
No.
%
Catheterized patients
162
10.5
154
10.0
Duration of catheter use (% nursing home days) 1-25 26-50 51-75 76-100 Total
13 7 5 137 162
8.0 4.3 3.1 84.6 100.0
101 22 19 12 154
65.6 14.3 12.3 7.8 100.0
TABLE 2. Status after 1 year of follow-up of a stratified random sample of patients from 13 nursing homes in Columbus, Ohio Male
Died Discharged home Transferred to another nursing home Hospitalized and returned to the same nursing home Stayed the year, never hospitalized Total
Female %
83 21
22.3 5.6
229 37
19.6 3.2
312 58
20.3 3.8
26
7.0
30
2.6
56
3.6
83
22.3
210
17.9
293
19.0
159
42.7
662
56.7
821
53.3
372
100.0
1,168
100.0
1,540
100.0
entry tended to remain catheterized throughout most of the year, whereas those who received a catheter during the study year retained it for a shorter period of time (table 3). The most frequent indication for use of catheters was for incontinence in females. The median duration of catheterization before entry into the study in 58 patients for whom reliable information was available was 4 months. About one fourth of these had been catheterized for 12 months or more. The catheters were mostly latex or silicone-coated (76.5 percent), 18 French or larger (67.7 percent), and fitted with a 30ml balloon (69.7 percent). Orders for catheter change were usually at monthly intervals (49.4 percent) or as needed (40.6 percent).
No.
Total
No.
%
No.
%
Logistic regression analysis of risk factors associated with death
The risk factors, other than catheterization, found to have an independent, significant effect on mortality within the study year are shown in table 4. Six conditions were found to be significantly associated with mortality. When mental status (classes 3 and 4) was added to the regression analysis, it was found to have an odds ratio of 4.29 (95 percent confidence interval 2.27-8.11) and to reduce the odds ratio for activities of daily living to 1.43 (95 percent confidence interval 1.05-1.95). Mental status was deleted from the model to avoid confounding with activities of daily living. There were only 46 patients in mental status classes 3 or 4; all
296
Kunin et al.
TABLE 4. Risk factors, exclusive of catrteterization, that had a major impact on mortality at 1 year in a stratified random sample of patients from 13 nursing homes in Columbus, Ohio Mortality Adjusted odds ratiot
95% confidence interval
14.3 22.8
1.00 1.53
1.13-2.08
256 56
18.7 32.6
1.00 2.19
1.53-3.15
762 778
123 189
16.1 24.3
1.00 1.57
1.20-2.06
Diabetes mellltus No Yes
1,287 253
243 69
18.9 27.3
1.00 1.54
1.11-2.13
Skin condition Good Decubitus ulcers
1,162 378
191 121
16.4 32.0
1.00 2.33
1.77-3.07
Age in decades^
1,540
312
20.3
1.17
1.03-1.32
No. of patients observed (n = 1,540)
No.
%
Activities of daily living Classes 1-4 Classes 5-6
462 1,078
66 246
Cancer No Yes
1,368 172
Risk factor*
Cardiac disease No Yes
Deaths
• Risk factors present at entry to the study. f Adjusted against aM other variables isted in the table. All the differences were significant (p a 0 01). % The mean age was 77.7 years for those who lived and 81.8 years for those who died (standard deviation, 15.2 and 10.6 years, respectively).
of these fit into activity of daily living classes 5 and 6. The association between mortality and the use of urinary catheters at entry versus during the study year and by duration of catheterization is shown in table 5. The data are presented as the mortality rate and as adjusted for six independent risk factors. The odds ratios for mortality were significantly (p < 0.01) increased for catheterized patients in each of the comparisons. The largest odds ratio was observed for patients who spent 76-100 percent of their nursing home days catheterized, compared with those who were never catheterized (odds ratio = 2.99). This was similar to the odds ratio of 2.38 found for patients with catheters in place at entry who were compared with those who were never catheterized during the study year. These results are not surprising, since 84.6 percent of patients who had catheters at entry continued to be catheterized for 76 percent or more nursing home days. Smaller odds ratios were observed when patients who
had catheters in place at entry were compared with those who did not, but who may have been catheterized later on, and for patients who received a catheter at any time during the year, regardless of the duration of catheterization, compared with those who were never catheterized. There was an incremental increase in the association between the use of catheters and mortality by duration of catheterization. Analysis of matched pairs
Catheterized and noncatheterized patients were matched according to the sixriskfactors shown in table 4. Several comparisons were performed, including analysis of 1) 146 pairs of patients with catheters in place at entry and patients who were never catheterized during the study year, 2) 279 pairs of patients who received a catheter at any time during the study year and patients who were never catheterized; and 3) 262 patients, matched as in the second analysis, but for
Urinary Catheters in Nursing Homes
297
TABLE 5. Effect of duration of catheterlzation ori mortality at 1 year in a stratified random sample of patients from 13 nursing homes in Columbus, Ohio MortaBty Variable
No. of patients
A/illKtfld
Died
odds ratio*
95% confidence
observed ( n - 1,540)
No.
%
1,224
201
16.4
1.00
162
63
38.9
2.38
1.62-3.50
Catheter in place at entry Not Yes
1,378 162
249 63
18.1 38.9
1.00 2.16
1.48-3.23
Catheter placed during year No Yes
1,224 316
201 111
16.4 35.1
1.00 1.47
1.29-1.66
% nursing home days catheterized None 1-25 26-75 76-100 Total
1,224 114 53 149 1,540
201 27 19 65 312
16.4 23.7 35.8 43.6 20.3
1.00 1.35 2.09 2.99
0.84-2.16 1.14-3.84 2.01-4.42
Never catheterized Catheter in place at entry
Interval
* Adjusted for activities of dally Bving, cancer, cardiac disease, diabetes meWus, skin condition, and age. t Of these 1,378 patients 154 (11.2%) received a catheter at some time during the foDow-up year.
whom mental status, rather than activities of daily living, was used as one of the risk factors. The outcomes were very similar for the three different matched pairs analyses. Data are presented in tables 6 and 7 for group 2, since it contained the largest set of matched pairs and included patients who were catheterized during the study year. The catheterized patients were hospitalized about three times more often; stayed in the hospital about three times as many days; and received antibiotics while in the nursing home about three times longer than the matched, noncatheterized patients, table 6. These findings were consistent even after adjustment for days spent in the nursing home. The discharge diagnoses were reviewed for the 156 catheterized and 37 noncatheterized patients who were admitted to the hospital during the study period. Septicemia was diagnosed in 14 (9.0 percent) of the catheterized patients and in only two (5.4 percent) of the noncatheterized group. Similarly 29, (18.6 percent) of the catheterized patients were admitted for urinary tract
infections, compared with only five (13.5 percent) of the noncatheterized group. The risk of death for pairs of patients with catheters at entry or placed during the study year and noncatheterized patients is shown in table 7. Among the discordant pairs, there were about twice as many pairs in which the catheterized patient died and the noncatheterized patient remained alive as there were pairs in which the catheterized patient lived and the noncatheterized patient died. The most frequent death certificate diagnoses, for catheterized and noncatheterized patients respectively, were as follows: cardiopulmonary arrest, 54.7 versus 45.8 percent; congestive heart failure, 6.7 versus 10.4 percent; pneumonia, 5.3 versus 10.4 percent; respiratory failure, 4.0 versus 4.2 percent; and sepsis, 6.7 versus 2.1 percent. DISCUSSION
This study was designed to determine the independent effect on morbidity and mortality associated with use of indwelling uri-
298
Kunin et al.
TABLE 6. Comparison of hospital admissions and use of antibiotics during the study year among 279 matched pairs of patients in a stratified random sample from 13 nursing homes in Columbus Ohio, according to use of indwelling urinary catheters Catheterized patients
Variable No. of days spent in the nursing home
Noncatheterized patients p vaiu©
Mean
SD*
Medtan
Mean
SD
Median
277.10
125.40
343
306.60
119.60
356
0.90
1.07
0.28
0.52
