JAMDA 16 (2015) 334e340
JAMDA journal homepage: www.jamda.com
Original Study
The Characteristics of Diabetic Residents in European Nursing Homes: Results from the SHELTER Study ska MD, PhD a, *, Eva Topinková MD, PhD b, Piotr Brzyski PhD a, Katarzyna Szczerbin Henriëtte G. van der Roest PhD c, Tomás Richter MD, PhD b, Harriet Finne-Soveri MD, PhD d, Michael D. Denkinger MD e, Jacob Gindin MD f, Graziano Onder MD, PhD g, Roberto Bernabei MD g a Department of Sociology of Medicine, Epidemiology and Preventive Medicine Chair, Faculty of Medicine, Jagiellonian University Medical College, Kraków, Poland b Department of Geriatrics, 1st Faculty of Medicine, Charles University, Prague, Czech Republic c EMGO Institute for Health and Care Research, Department of General Practice and Elderly Care, VU University Medical Centre, Amsterdam, The Netherlands d Unit for Ageing and Services, National Institute for Health and Welfare, Helsinki, Finland e Agaplesion Bethesda Clinic, Geriatric Centre Ulm/Alb-Donau, University of Ulm, Ulm, Germany f The Centre for Standards in Health and Disability, Research Authority, University of Haifa, Haifa, Israel g Centro Medicina dell’Invecchiamento, Università Cattolica Sacro Cuore, Rome, Italy
a b s t r a c t Keywords: Diabetes mellitus nursing home older adults
Objectives: The objectives of this study were to describe the prevalence of diabetes mellitus (DM) in European nursing homes (NHs), and the health and functional characteristics of diabetic residents (DMR) aged 60 years and older. Design: Between 2009 and 2011, the Services and Health for Elderly in Long TERm care (SHELTER) project, a 12-month prospective cohort study, was conducted to assess NH residents across different health care systems in 7 European countries and Israel. Methods: The study included 59 NHs in 8 countries with a total of 4037 residents living in or admitted to a NH during the 3-month enrollment period. The multidimensional InterRAI instrument for Long-Term Care Facilities (InterRAI-LTCF) was used to assess health and functional status among residents. Descriptive statistics and linear, ordinal, and logistic regression were used to perform the analyses. Results: We found a 21.8% prevalence of DM among NH residents. Residents with DM (DMRs) were significantly younger compared with non-DMRs (82.3, SD 7.7; 84.6, SD 8.4; P < .001). DMRs were more frequently overweight or obese, and presented more often with ischemic heart disease, congestive heart failure, hypertension, and stroke than residents without DM. DMRs also took more drugs, had pressure ulcers (PU) or other wounds more often, and more frequently had urinary incontinence (UI); they also reported worse self-perceived health. DM independently of other factors increased risk of PU occurrence (odds ratio 1.38; 95% confidence interval [CI] 1.02e1.86; P ¼ .036) and decreased probability of higher pain scores (B ¼ 0.28; 95% CI 0.41 to 0.14; P < .001). DM was not associated with ADL dependency, cognitive impairment, and depression in NH residents. Conclusion: Prevalence of DM in European NH residents is comparable to US national NH surveys, and to UK and German NH data based on glucose-level testing. DMRs compared with non-DMRs have more comorbid conditions, and a particularly higher incidence of cardiovascular diseases and obesity, PU, and severe UI. DMRs should be regarded as a specific group of residents who require an interdisciplinary approach in medical and nursing care. Ó 2015 AMDA e The Society for Post-Acute and Long-Term Care Medicine.
KS, ET, HFS, JG, and RB are members of InterRAI. The other authors declare no conflicts of interest. The SHELTER study was funded by the European Union’s Seventh Framework Programme for Research, European Commission Grant 223115. The work of ET and
http://dx.doi.org/10.1016/j.jamda.2014.11.009 1525-8610/Ó 2015 AMDA e The Society for Post-Acute and Long-Term Care Medicine.
TR was partly supported by FP7 European Commission Grant 278803 “The MIDFRAIL Study.” ska, MD, PhD, Jagiellonian * Address correspondence to Katarzyna Szczerbin University Medical College, ul. Kopernika 7a, 31e034 Kraków, Poland. ska). E-mail address: [email protected] (K. Szczerbin
ska et al. / JAMDA 16 (2015) 334e340 K. Szczerbin
According to the International Diabetes Federation (IDF), the prevalence of diabetes mellitus (DM) in European adults aged 20 to 79 years is 8.5% (11.0% in North America).1 It rises with age, reaching the peak in older adults aged 75 to 80.2 The highest prevalence of diabetes was found in nursing home (NH) populations. Studies conducted in the United States among large samples of NH residents showed a prevalence of 33.3% in a general sample of 1,361,406 NH residents,3 with 26% of patients found to be men and 23% women older than 65.4 Another cross-sectional study conducted in 1995 and 2004 found that the increase in prevalence of DM in US NH residents older than 55 years was steeper among men, from 16.9% to 26.4%, than among women, from 16.1% to 22.2%.5 DM thus seems to be an emerging challenge in the NH population. In comparison with the United States, there is definitely less information in published literature about prevalence of DM in European NHs.6 The existing data are based on significantly smaller convenience samples. Strikingly, a very low prevalence of DM was found in one study from France (8.7% in 4896 residents)7 and in 2 British studies (5.8% and 13.0%), depending on the type of long-term care facility (residential homes or NHs; for older people or for mentally ill older adults)8 and study method used (eg, postal survey or measurement of fasting glucose).8,9 Yet, Sinclair et al10 indicated that many cases of DM may remain unrecognized, and studies may underestimate the true prevalence rate in NHs when laboratory measurements of blood glucose are not included in the study protocol. Based on patient file information Sinclair et al10 reported a 12% prevalence of previously diagnosed DM in NH residents. However, as a result of blood glucose measurements taken during the study, another 14.7% of the study population was newly diagnosed with DM. Thus, the overall DM prevalence in Birmingham NHs in the United Kingdom was found to be 26.7%. These findings are supported by another study from Germany reporting 26.2% prevalence of DM in NHs based on glucose-level testing.11 The lack of standard blood glucose measurement in NHs was seen as an indicator of substandard quality and incomplete diagnostics.10 The latest European guidelines therefore recommend routine laboratory screening for DM in NH residents at admission and at 2-year intervals.12 Timely recognition of DM may prevent or slow down potential complications and diabetes-related conditions. Known diagnosis of DM should draw attention to symptoms that are often underreported, undetected, and undertreated in diabetic patients. NH professionals’ awareness of the high prevalence rate of DM in NHs is therefore a key factor that should drive them to perform regular screening for DM in NH residents so as to provide appropriate care, bearing in mind the clinical complexity of diabetic patients. The aim of our study thus was to detect the prevalence of DM in a large sample of European NHs. Further, to describe clinical characteristics of diabetic residents (DMRs) aged 60 years and older based on data collected in 59 NHs in 7 European countries and Israel from the SHELTER study (the Services and Health for Elderly in Long TERm care study) database. The third aim was to compare health and functional status of DMRs and nondiabetic residents (nonDMRs), while focusing on health problems that are typical for patients with DM. Methods The SHELTER study, a project funded by the Seventh Framework Program of the EU, was conducted from 2009 to 2011.13 The sample consisted of 4156 residents in 59 NHs located in 7 European countries (Czech Republic: 10 NHs, England: 9, Finland: 6, France: 4, Germany: 9, Italy: 10, The Netherlands: 4), and Israel: 7 NHs. NHs were selected based on their willingness to participate in the SHELTER study and
335
were not intended to be representative of all NHs in each participating country. Adults residing in participating NHs at the beginning of the study and those admitted in the 3-month enrollment period were assessed. To meet the goal of our current analysis, we excluded NH residents younger than 60 years (n ¼ 96) because of possible differences of characteristics in younger adults with DM. We also excluded residents (n ¼ 23) with missing data about their concomitant diseases (empty checklist of diseases), because it was not clear whether they had DM or not. The final sample was composed of 4037 persons. Residents were invited to take part in the study and were free to decline participation. Ethical approval for the study was obtained in all countries according to local regulations. The interRAI instrument for Long-Term Care Facilities (interRAILTCF) was used. This is a comprehensive standardized instrument to assess care needs, health, and functional status of NH residents. The interRAI-LTCF has been validated in several European countries and has proved to be a reliable instrument.13,14 It contains more than 350 variables, including sociodemographic items, numerous clinical diagnoses, symptoms, geriatric syndromes, care programs, and treatments. A diagnosis of DM and other clinical diagnoses were derived from the interRAI diagnoses section and a list of International Classification of Diseases, Ninth Revision (ICD-9) coded diseases. Several validated scales can be generated from the interRAI-LTCF items. For this study, the 7-point Cognitive Performance Scale (CPS) was used to assess cognitive status.15,16 Intact or nearly intact cognitive function was represented by a CPS score of 0 to 1, moderate impairment by a CPS score of 2 to 3, and severe impairment by a CPS score of 4 to 6. Functional status was represented by the 7-point scale of Activities of Daily Living Hierarchy (ADLh)17 categorizing physical functioning as independent (ADLh ¼ 0e1), moderately dependent (ADLh ¼ 2e3), and severely dependent (ADLh ¼ 4e6). The 7-point Depression Rating Scale (DRS)18 was used to indicate the presence of symptoms of depression (DRS ¼ 3 or more). Pressure ulcer (PU) stage was coded as follows: 0, no PU; 1, any area of persistent skin redness; 2, partial loss of skin layers; 3, deep craters in the skin; 4, breaks in skin exposing muscle or bone; 5, not codeable (eg, necrotic eschar predominant). Pain was coded based on frequency (pain experienced daily vs less than daily or no pain), on intensity (moderate, severe, or unbearable pain vs mild or no pain), and on occurrence of breakthrough pain within the past 3 days (yes vs no). A 5-point Pain Scale presenting level of pain starting from 0 (no pain) through 2 (daily pain) up to 4 (daily excruciating pain) was applied in regression models. The World Health Organization classification based on body mass index (BMI) was used to stratify for nutritional status. Four answers to a question concerning self-perceived health were divided into 3 categories: “excellent” (due to very low number of answers) and “good” were put in one category, and “fair” and “poor” were treated as 2 other separate categories. In line with previous publications, polypharmacy was categorized into 3 groups: nonpolypharmacy (concurrent use of 0e4 drugs), polypharmacy (concurrent use of 5e9 drugs), and excessive polypharmacy (concurrent use of 10 drugs or more).19 Urinary incontinence (UI) was defined as “severe” if the patient was frequently (daily but with some control present) or permanently incontinent (not controlled) or needed bladder catheterization versus “occasionally or infrequently incontinent” (incontinence episodes less than daily or did not occur over past 3 days), or “continent” (no incontinence problem). Bunions, hammertoes, overlapping toes or other structural foot problems, infections, and ulcers were all defined as “any foot problems.” Coding of other variables is described in Table 2 presenting results. Distributions of qualitative variables were described as frequencies (n) and percentages (%), whereas quantitative variables were described as means with SDs if they had normal distribution or as medians and quartiles otherwise. The distributions of categorical
ska et al. / JAMDA 16 (2015) 334e340 K. Szczerbin
336
Table 1 Clinical Characteristics of DMRs and non-DMRs in 59 European Nursing Homes: The SHELTER Study Characteristics of NH Residents
Age, mean (SD) Age, y, % (n) 60e75 76e85 >85 Female gender Self-perceived health, % (n) Excellent and Good Fair Poor Selected diseases, % (n) Coronary heart disease Congestive heart failure Hypertension Stroke/transient ischemic attack Hemiplegia Alzheimer disease Other dementia Urinary tract infection Chronic kidney disease Cancer Chronic obstructive pulmonary disease No. of diseases, mean (SD) No. of drugs, mean (SD) No. of drugs, % (n) 0e4 5e9 10
Diabetic
Nondiabetic
All
21.8% (n ¼ 879)
78.2% (n ¼ 3158)
100.0% (n ¼ 4037)
82.3 ( 7.7)
84.6 ( 8.4)
84.1 ( 8.3)
17.7 40.8 41.4 73.8
13.6 33.6 52.8 73.7
14.5 35.2 50.3 73.7
(156) (359) (364) (649)
(428) (1062) (1668) (2327)
(584) (1421) (2032) (2976)
34.4 (224) 45.1 (294) 20.6 (134)
41.7 (917) 40.1 (882) 18.1 (398)
40.0 (1141) 41.3 (1176) 18.7 (134)
37.0 21.2 40.5 29.4 13.2 17.0 37.8 7.9 4.6 12.5 11.7 4.1 8.2
24.0 17.2 22.7 21.8 10.5 21.2 36.4 6.0 2.8 11.1 9.6 2.5 6.8
26.8 18.1 26.6 23.5 11.1 20.3 36.7 6.4 3.2 11.4 10.1 2.9 7.1
(325) (186) (356) (258) (116) (149) (332) (69) (40) (110) (103) (1.8)* (3.5)
17.0 (145) 49.9 (452) 33.0 (281)
(756) (543) (716) (689) (330) (669) (1145) (188) (90) (352) (304) (1.6) (3.3)
27.9 (855) 50.3 (1540) 21.8 (668)
(1081) (729) (1072) (947) (446) (818) (1477) (257) (130) (462) (40.7) (1.8) (3.6)
n
P
4037 4037
JAMDA journal homepage: www.jamda.com
Original Study
The Characteristics of Diabetic Residents in European Nursing Homes: Results from the SHELTER Study ska MD, PhD a, *, Eva Topinková MD, PhD b, Piotr Brzyski PhD a, Katarzyna Szczerbin Henriëtte G. van der Roest PhD c, Tomás Richter MD, PhD b, Harriet Finne-Soveri MD, PhD d, Michael D. Denkinger MD e, Jacob Gindin MD f, Graziano Onder MD, PhD g, Roberto Bernabei MD g a Department of Sociology of Medicine, Epidemiology and Preventive Medicine Chair, Faculty of Medicine, Jagiellonian University Medical College, Kraków, Poland b Department of Geriatrics, 1st Faculty of Medicine, Charles University, Prague, Czech Republic c EMGO Institute for Health and Care Research, Department of General Practice and Elderly Care, VU University Medical Centre, Amsterdam, The Netherlands d Unit for Ageing and Services, National Institute for Health and Welfare, Helsinki, Finland e Agaplesion Bethesda Clinic, Geriatric Centre Ulm/Alb-Donau, University of Ulm, Ulm, Germany f The Centre for Standards in Health and Disability, Research Authority, University of Haifa, Haifa, Israel g Centro Medicina dell’Invecchiamento, Università Cattolica Sacro Cuore, Rome, Italy
a b s t r a c t Keywords: Diabetes mellitus nursing home older adults
Objectives: The objectives of this study were to describe the prevalence of diabetes mellitus (DM) in European nursing homes (NHs), and the health and functional characteristics of diabetic residents (DMR) aged 60 years and older. Design: Between 2009 and 2011, the Services and Health for Elderly in Long TERm care (SHELTER) project, a 12-month prospective cohort study, was conducted to assess NH residents across different health care systems in 7 European countries and Israel. Methods: The study included 59 NHs in 8 countries with a total of 4037 residents living in or admitted to a NH during the 3-month enrollment period. The multidimensional InterRAI instrument for Long-Term Care Facilities (InterRAI-LTCF) was used to assess health and functional status among residents. Descriptive statistics and linear, ordinal, and logistic regression were used to perform the analyses. Results: We found a 21.8% prevalence of DM among NH residents. Residents with DM (DMRs) were significantly younger compared with non-DMRs (82.3, SD 7.7; 84.6, SD 8.4; P < .001). DMRs were more frequently overweight or obese, and presented more often with ischemic heart disease, congestive heart failure, hypertension, and stroke than residents without DM. DMRs also took more drugs, had pressure ulcers (PU) or other wounds more often, and more frequently had urinary incontinence (UI); they also reported worse self-perceived health. DM independently of other factors increased risk of PU occurrence (odds ratio 1.38; 95% confidence interval [CI] 1.02e1.86; P ¼ .036) and decreased probability of higher pain scores (B ¼ 0.28; 95% CI 0.41 to 0.14; P < .001). DM was not associated with ADL dependency, cognitive impairment, and depression in NH residents. Conclusion: Prevalence of DM in European NH residents is comparable to US national NH surveys, and to UK and German NH data based on glucose-level testing. DMRs compared with non-DMRs have more comorbid conditions, and a particularly higher incidence of cardiovascular diseases and obesity, PU, and severe UI. DMRs should be regarded as a specific group of residents who require an interdisciplinary approach in medical and nursing care. Ó 2015 AMDA e The Society for Post-Acute and Long-Term Care Medicine.
KS, ET, HFS, JG, and RB are members of InterRAI. The other authors declare no conflicts of interest. The SHELTER study was funded by the European Union’s Seventh Framework Programme for Research, European Commission Grant 223115. The work of ET and
http://dx.doi.org/10.1016/j.jamda.2014.11.009 1525-8610/Ó 2015 AMDA e The Society for Post-Acute and Long-Term Care Medicine.
TR was partly supported by FP7 European Commission Grant 278803 “The MIDFRAIL Study.” ska, MD, PhD, Jagiellonian * Address correspondence to Katarzyna Szczerbin University Medical College, ul. Kopernika 7a, 31e034 Kraków, Poland. ska). E-mail address: [email protected] (K. Szczerbin
ska et al. / JAMDA 16 (2015) 334e340 K. Szczerbin
According to the International Diabetes Federation (IDF), the prevalence of diabetes mellitus (DM) in European adults aged 20 to 79 years is 8.5% (11.0% in North America).1 It rises with age, reaching the peak in older adults aged 75 to 80.2 The highest prevalence of diabetes was found in nursing home (NH) populations. Studies conducted in the United States among large samples of NH residents showed a prevalence of 33.3% in a general sample of 1,361,406 NH residents,3 with 26% of patients found to be men and 23% women older than 65.4 Another cross-sectional study conducted in 1995 and 2004 found that the increase in prevalence of DM in US NH residents older than 55 years was steeper among men, from 16.9% to 26.4%, than among women, from 16.1% to 22.2%.5 DM thus seems to be an emerging challenge in the NH population. In comparison with the United States, there is definitely less information in published literature about prevalence of DM in European NHs.6 The existing data are based on significantly smaller convenience samples. Strikingly, a very low prevalence of DM was found in one study from France (8.7% in 4896 residents)7 and in 2 British studies (5.8% and 13.0%), depending on the type of long-term care facility (residential homes or NHs; for older people or for mentally ill older adults)8 and study method used (eg, postal survey or measurement of fasting glucose).8,9 Yet, Sinclair et al10 indicated that many cases of DM may remain unrecognized, and studies may underestimate the true prevalence rate in NHs when laboratory measurements of blood glucose are not included in the study protocol. Based on patient file information Sinclair et al10 reported a 12% prevalence of previously diagnosed DM in NH residents. However, as a result of blood glucose measurements taken during the study, another 14.7% of the study population was newly diagnosed with DM. Thus, the overall DM prevalence in Birmingham NHs in the United Kingdom was found to be 26.7%. These findings are supported by another study from Germany reporting 26.2% prevalence of DM in NHs based on glucose-level testing.11 The lack of standard blood glucose measurement in NHs was seen as an indicator of substandard quality and incomplete diagnostics.10 The latest European guidelines therefore recommend routine laboratory screening for DM in NH residents at admission and at 2-year intervals.12 Timely recognition of DM may prevent or slow down potential complications and diabetes-related conditions. Known diagnosis of DM should draw attention to symptoms that are often underreported, undetected, and undertreated in diabetic patients. NH professionals’ awareness of the high prevalence rate of DM in NHs is therefore a key factor that should drive them to perform regular screening for DM in NH residents so as to provide appropriate care, bearing in mind the clinical complexity of diabetic patients. The aim of our study thus was to detect the prevalence of DM in a large sample of European NHs. Further, to describe clinical characteristics of diabetic residents (DMRs) aged 60 years and older based on data collected in 59 NHs in 7 European countries and Israel from the SHELTER study (the Services and Health for Elderly in Long TERm care study) database. The third aim was to compare health and functional status of DMRs and nondiabetic residents (nonDMRs), while focusing on health problems that are typical for patients with DM. Methods The SHELTER study, a project funded by the Seventh Framework Program of the EU, was conducted from 2009 to 2011.13 The sample consisted of 4156 residents in 59 NHs located in 7 European countries (Czech Republic: 10 NHs, England: 9, Finland: 6, France: 4, Germany: 9, Italy: 10, The Netherlands: 4), and Israel: 7 NHs. NHs were selected based on their willingness to participate in the SHELTER study and
335
were not intended to be representative of all NHs in each participating country. Adults residing in participating NHs at the beginning of the study and those admitted in the 3-month enrollment period were assessed. To meet the goal of our current analysis, we excluded NH residents younger than 60 years (n ¼ 96) because of possible differences of characteristics in younger adults with DM. We also excluded residents (n ¼ 23) with missing data about their concomitant diseases (empty checklist of diseases), because it was not clear whether they had DM or not. The final sample was composed of 4037 persons. Residents were invited to take part in the study and were free to decline participation. Ethical approval for the study was obtained in all countries according to local regulations. The interRAI instrument for Long-Term Care Facilities (interRAILTCF) was used. This is a comprehensive standardized instrument to assess care needs, health, and functional status of NH residents. The interRAI-LTCF has been validated in several European countries and has proved to be a reliable instrument.13,14 It contains more than 350 variables, including sociodemographic items, numerous clinical diagnoses, symptoms, geriatric syndromes, care programs, and treatments. A diagnosis of DM and other clinical diagnoses were derived from the interRAI diagnoses section and a list of International Classification of Diseases, Ninth Revision (ICD-9) coded diseases. Several validated scales can be generated from the interRAI-LTCF items. For this study, the 7-point Cognitive Performance Scale (CPS) was used to assess cognitive status.15,16 Intact or nearly intact cognitive function was represented by a CPS score of 0 to 1, moderate impairment by a CPS score of 2 to 3, and severe impairment by a CPS score of 4 to 6. Functional status was represented by the 7-point scale of Activities of Daily Living Hierarchy (ADLh)17 categorizing physical functioning as independent (ADLh ¼ 0e1), moderately dependent (ADLh ¼ 2e3), and severely dependent (ADLh ¼ 4e6). The 7-point Depression Rating Scale (DRS)18 was used to indicate the presence of symptoms of depression (DRS ¼ 3 or more). Pressure ulcer (PU) stage was coded as follows: 0, no PU; 1, any area of persistent skin redness; 2, partial loss of skin layers; 3, deep craters in the skin; 4, breaks in skin exposing muscle or bone; 5, not codeable (eg, necrotic eschar predominant). Pain was coded based on frequency (pain experienced daily vs less than daily or no pain), on intensity (moderate, severe, or unbearable pain vs mild or no pain), and on occurrence of breakthrough pain within the past 3 days (yes vs no). A 5-point Pain Scale presenting level of pain starting from 0 (no pain) through 2 (daily pain) up to 4 (daily excruciating pain) was applied in regression models. The World Health Organization classification based on body mass index (BMI) was used to stratify for nutritional status. Four answers to a question concerning self-perceived health were divided into 3 categories: “excellent” (due to very low number of answers) and “good” were put in one category, and “fair” and “poor” were treated as 2 other separate categories. In line with previous publications, polypharmacy was categorized into 3 groups: nonpolypharmacy (concurrent use of 0e4 drugs), polypharmacy (concurrent use of 5e9 drugs), and excessive polypharmacy (concurrent use of 10 drugs or more).19 Urinary incontinence (UI) was defined as “severe” if the patient was frequently (daily but with some control present) or permanently incontinent (not controlled) or needed bladder catheterization versus “occasionally or infrequently incontinent” (incontinence episodes less than daily or did not occur over past 3 days), or “continent” (no incontinence problem). Bunions, hammertoes, overlapping toes or other structural foot problems, infections, and ulcers were all defined as “any foot problems.” Coding of other variables is described in Table 2 presenting results. Distributions of qualitative variables were described as frequencies (n) and percentages (%), whereas quantitative variables were described as means with SDs if they had normal distribution or as medians and quartiles otherwise. The distributions of categorical
ska et al. / JAMDA 16 (2015) 334e340 K. Szczerbin
336
Table 1 Clinical Characteristics of DMRs and non-DMRs in 59 European Nursing Homes: The SHELTER Study Characteristics of NH Residents
Age, mean (SD) Age, y, % (n) 60e75 76e85 >85 Female gender Self-perceived health, % (n) Excellent and Good Fair Poor Selected diseases, % (n) Coronary heart disease Congestive heart failure Hypertension Stroke/transient ischemic attack Hemiplegia Alzheimer disease Other dementia Urinary tract infection Chronic kidney disease Cancer Chronic obstructive pulmonary disease No. of diseases, mean (SD) No. of drugs, mean (SD) No. of drugs, % (n) 0e4 5e9 10
Diabetic
Nondiabetic
All
21.8% (n ¼ 879)
78.2% (n ¼ 3158)
100.0% (n ¼ 4037)
82.3 ( 7.7)
84.6 ( 8.4)
84.1 ( 8.3)
17.7 40.8 41.4 73.8
13.6 33.6 52.8 73.7
14.5 35.2 50.3 73.7
(156) (359) (364) (649)
(428) (1062) (1668) (2327)
(584) (1421) (2032) (2976)
34.4 (224) 45.1 (294) 20.6 (134)
41.7 (917) 40.1 (882) 18.1 (398)
40.0 (1141) 41.3 (1176) 18.7 (134)
37.0 21.2 40.5 29.4 13.2 17.0 37.8 7.9 4.6 12.5 11.7 4.1 8.2
24.0 17.2 22.7 21.8 10.5 21.2 36.4 6.0 2.8 11.1 9.6 2.5 6.8
26.8 18.1 26.6 23.5 11.1 20.3 36.7 6.4 3.2 11.4 10.1 2.9 7.1
(325) (186) (356) (258) (116) (149) (332) (69) (40) (110) (103) (1.8)* (3.5)
17.0 (145) 49.9 (452) 33.0 (281)
(756) (543) (716) (689) (330) (669) (1145) (188) (90) (352) (304) (1.6) (3.3)
27.9 (855) 50.3 (1540) 21.8 (668)
(1081) (729) (1072) (947) (446) (818) (1477) (257) (130) (462) (40.7) (1.8) (3.6)
n
P
4037 4037
