Journal of Evaluation in Clinical Practice ISSN 1356-1294

Pressure ulcer risk screening in hospitals and nursing

homes with a general nursing assessment tool: evaluation
of the care dependency scale
Elke I. Mertens RN MEd,1 Ruud J. G. Halfens PhD (FEANS),2 Ekkehart Dietz PhD,3
Ramona Scheufele BA4 and Theo Dassen RN, PhD5
1
Assistant Professor, 5Professor, Department of Nursing Science, Centre for the Humanities and Health Sciences, Charité – Universitätsmedizin
Berlin, Charitéplatz 1, Berlin, Germany
2
Associated Professor, Department of Health Care Studies, Section of Nursing Science, Maastricht, the Netherlands
3
Associated Professor, 4Assistant Professor, Department of Biometry and Clinical Epidemiology, Charité – Universitätsmedizin Berlin Charitéplatz
1, Berlin, Germany

Keywords
care dependency, hospitals, nursing homes,
pressure ulcer risk, pressure ulcer, risk
assessment
Correspondence
Elke Mertens
Department of Nursing Science
Centre for the Humanities and Health
Sciences
Charité – Universitätsmedizin Berlin
Charitéplatz 1, 10117 Berlin
Germany
E-mail: elke.mertens@charite.de
Accepted for publication: 24 July 2007
doi:10.1111/j.1365-2753.2007.00935.x
Abstract
Rationale In view of an increasing necessity for systematic assessments, nursing practice
would benefit from a simplification of assessment procedures. These assessments should be
scientifically based.
Aims To evaluate the possibility of assessing pressure ulcer risk as well as care depen-
dency simultaneously with a standardized instrument for nursing homes and hospitals.
Methods Care dependency was measured with the Care Dependency Scale (CDS). The
quantitative analyses were accomplished with data from a cross-sectional study that was
performed in 2005 in 39 German nursing homes and 37 hospitals with a total of more than
10 000 participants. The scale’s construct validity was calculated with Pearson’s r, and
predictive validity was evaluated by computing sensitivity and specificity values and the
area under the curve (AUC). Item-level analyses included calculations of odds ratios,
relative risks and logistic regression analyses.
Results Construct validity of the CDS was r = 0.79 (P < 0.01) in nursing homes and
r = 0.89 (P < 0.01) in hospitals. AUC was 0.80 in hospitals and 0.65 in nursing homes.
Analyses on item level identified ‘mobility’ as a key item in both settings and additional
differing key items for nursing homes and hospitals.
Conclusions The CDS is a well-functioning tool for pressure ulcer risk detection in
hospitals. For this purpose, the most appropriate cut-off point is 69 while special regard is
given to the items ‘continence’, ‘mobility’ and ‘hygiene’. In nursing homes the usefulness
of the CDS for pressure ulcer risk detection is limited. Here, the most appropriate cut-off
point is 41 and attention is given to the items ‘mobility’, ‘getting (un)dressed’, ‘hygiene’
and ‘avoidance of danger’.

documentary tasks. Patients are also taken through partially redun-

Introduction
Pressure ulcers are one of the major challenges in nursing care. A
large number of individuals develop pressure ulcers [1] which
cause pain and restrictions to the life of the individuals concerned
[2], increase costs for health care systems [3] and are connected to
higher mortality rates [4].
Risk assessment is recommended as the first step in pressure
ulcer prevention [5,6]. Assessment in general is an essential part of
nursing care and standardized assessments are recommended for a
rising number of nursing care problems. Together with an ageing
population this leads to nurses being increasingly burdened with
dant assessment procedures and it would therefore be sensible to
conduct an initial screening with a general assessment tool that is
able to detect different risks simultaneously. The Care Dependency
Scale (CDS) possibly is a tool enabling such a course of action. The
CDS is a 15-item, Likert-type scale [7] with which the care depen-
dency of a person can be displayed in a short and comprehensive
way. It was developed as the starting point and framework for the
nursing care process and some of its items can also be found as
factors in various risk assessment scales, for example, ‘mobility’
and ‘continence’. It has recently been shown that a screening for fall
risk using the CDS can be accomplished with good results in

1018 © 2008 The Authors. Journal compilation © 2008 Blackwell Publishing Ltd, Journal of Evaluation in Clinical Practice 14 (2008) 1018–1025
E.I. Mertens et al.
hospitals [8] and there are preliminary references that a screening
for pressure ulcer risk may also be possible. Stable relationships
between the occurrence of pressure ulcers and increased care
dependency were found in annual surveys [9] and first analyses
revealed that high care dependency corresponded with high pres-
sure ulcer risk measured with the Braden scale [10,11]. These clues
have to be investigated in more detail. Therefore, the aim of this
paper is to evaluate the screening of pressure ulcer risk using the
CDS. The following research questions are to be answered:
1 What is the construct validity of the CDS for measuring the
pressure ulcer risk in nursing homes and hospitals?
2 Which sum score of the CDS is related to pressure ulcers?
3 Which items of the CDS are related to pressure ulcers?
Methods and instruments
Study design
This study is an in-depth analysis of cross-sectional data collected
nationwide in German nursing homes and hospitals in 2005. Since
2001, this voluntary survey has been conducted annually to evalu-
ate the prevalence of pressure ulcers, incontinence and falls, and to
describe the care dependency of nursing home residents and
hospital patients [12].
In the run-up of the study in 2005, registered staff nurses of the
participating institutions received a standardized training and a
comprehensive study manual. During the period of data collection
the researchers were contactable on a telephone hotline to answer
current questions. The staff nurses performed the data collection
by filling in a standardized, anonymous questionnaire for each
patient/resident who had given his/her informed consent. Ethical
approval for the study was given by the General Medical Council.
Instruments
The CDS was used to measure the participants’ care dependency.
Based on the theoretical framework of Virginia Henderson [13],
the scale was developed in 1996 by Dijkstra [7,14] using a Delphi
Table 1 Care Dependency Scale items and item descriptions
Item Item description
Eating/drinking The extent to which the patient/resident is
Continence The extent to which the patient/resident is
Body posture The extent to which the patient/resident is
Mobility The extent to which the patient/resident is
Day/night pattern The extent to which the patient/resident is
Getting (un)dressed The extent to which the patient/resident is
Body temperature The extent to which the patient/resident is
Hygiene The extent to which the patient/resident is
Avoidance of danger The extent to which the patient/resident is
Communication The extent to which the patient/resident is
Contact with others The extent to which the patient/resident is
Sense of rules/values The extent to which the patient/resident is
Daily activities The extent to which the patient/resident is
Recreational activities The extent to which the patient/resident is
Learning ability The extent to which the patient/resident is
previously learnt unaided
© 2008 The Authors. Journal compilation © 2008 Blackwell Publishing Ltd
Pressure ulcer risk screening with the CDS
approach. Scientific testing led to moderate to good results in
different countries and different settings of acute and chronic care
[15–23]. The CDS contains 15 items (Table 1), for which individu-
als can be rated as completely dependent (1), almost dependent
(2), partially dependent (3), almost independent (4) or completely
independent (5) on care. Sum scores range from 15 to 75 with a
low score indicating a high care dependency.
The Braden Scale was used for measuring the participants’
pressure ulcer risk. The scale was developed in 1987 by Braden
and colleagues [24] and contains six items (sensory perception,
moisture, activity, mobility, nutrition, friction and shear) scored
from 1 to 3 or 1 to 4. Sum scores range from 6 to 23 whereby a low
score indicates a high pressure ulcer risk [25]. The scale is widely
used and is estimated the best of three analysed tools in a recent
systematic review from 2006 [26]. Nevertheless, the scale is criti-
cized owing to its insufficient predictive validity [27,28].
Pressure ulcers were ascertained by the staff nurses who exam-
ined each participant and documented the location, grade and
duration of the ulcers. According to the European Pressure Ulcer
Advisory Panel (EPUAP), a pressure ulcer was defined as ‘[. . .] an
area of localized damage to the skin and underlying tissue caused
by pressure, shear, friction and/or a combination of these.’ [6].
Additionally, photographic examples of different pressure ulcers
were shown in the study manual.
Data analysis
The occurrence of pressure ulcers in the two settings is described
in percentages; the care dependency of the participants is evaluated
using the mean CDS scores. The statistical significance of sex
differences was tested with chi-squared test and t-test was used to
evaluate the statistical significance of age differences and CDS
mean score differences.
In order to evaluate the construct validity of the CDS in relation
to pressure ulcers, Pearson’s correlation coefficient was calculated
for the sum scores of the CDS and the Braden scale.
Sensitivity and specificity values of all CDS sum scores were
calculated to analyse the predictive validity of the scale in relation
able to satisfy his/her need for food and drink
able to voluntarily control the discharge of urine and faeces
able to adopt a position appropriate to a certain activity
able to move about unaided
able to maintain an appropriate day/night cycle unaided
able to get dressed and undressed unaided
able to protect his/her body temperature against external influences unaided
able to take care of his/her personal hygiene unaided
able to assure his/her own safety unaided
able to communicate
able to appropriately make, maintain and end social contacts
able to observe rules by him/herself
able to structure daily activities within the facility unaided
able to participate in activities outside the facility unaided
able to acquire knowledge and/or skills and/or to retain anything that was
1019
Pressure ulcer risk screening with the CDS E.I. Mertens et al.

to pressure ulcers. The results are displayed as receiver operating
characteristics (ROC) curves and the explanatory power of the
model was evaluated by calculating the area under the curve
(AUC).
The relation between single items and the occurrence of
pressure ulcers is represented with odds ratios and statistical
significance was tested with chi-squared test. As this study is a
prevalence study, these odds ratios are prevalence odds ratio esti-
mates and interpretation can be misleading, owing to a possible
prevalence–incidence bias. We could adjust for this bias using the
data about the disease duration to date of the prevalence cases and
calculated relative risks, by means of a method recently suggested
by Dietz et al. [29]. Statistical significance was assessed with the
Wald test.
As the CDS items may affect each other, logistic regression
analysis was performed in order to determine if some items have a
particularly strong relationship with the occurrence of pressure
ulcers.
Sample
Thirty-nine nursing homes and 37 hospitals took part in the study
in 2005. All patients and residents who gave their informed
consent and were 18 years or older were included in the study. The
response rate was 77.2% (87.7% in nursing homes, 73.6% in
hospitals). Most of the 10 734 participants were hospital patients
(n = 7204, 67.1%). Women were higher represented in both set-
tings; in nursing homes (hospitals) 79.0% (55.7%) of the residents
(patients) were female. Nursing home residents had a mean age of
83.0 and were almost 20 years older than hospital patients (mean
age 65.8). The number of participants, mean age and sex distribu-
tion is demonstrated in Table 2.
Results
Pressure ulcers in relation to sex and age
At the time of the study, 987 (9.2%) of all participants had at least
one pressure ulcer (Table 2). In nursing homes the rate (6.1%) was
smaller than that in hospitals (10.9%). More women than men had
pressure ulcers because of the higher proportion of women in both
settings. The percentage differences were small and clinically not
relevant, chi-squared tests for the differences between men and
women were statistically not significant in both settings.
There was no relation found between age and pressure ulcers in
nursing homes. In hospitals, patients with pressure ulcers were
9 years older than patients without pressure ulcers and this differ-
ence was statistically significant according to t-test analysis
(P < 0.001).
Care dependency
The mean CDS sum score of all participants was 55.9 (SD 20.3)
(Table 2) and all possible sum scores (15–75) were present in the
sample. Nursing home residents had a statistically significant,
lower care dependency score than hospital patients (t-test,
P < 0.001), which means that nursing home residents are more
care dependent. Patients and residents with pressure ulcers were
substantially more care dependent than those without pressure
ulcers (t-test, P < 0.001).
Validity of the CDS regarding pressure
ulcer risk
Construct validity
Pearson’s correlation coefficient for the sum scores of the CDS
and the Braden scale was r = 0.79 (P < 0.01) in nursing homes and
r = 0.89 (P < 0.01) in hospitals.
Predictive validity
Sensitivity and specificity values of different CDS sum scores are
listed in Table 3 and plotted as a graph in the ROC curves in Fig. 1.
The ROC curves give a good idea of the differentiation abilities
of the CDS sum score in the two settings. The AUC was 0.80 in
hospitals. It is similar to the values of the Braden scale (AUC
0.80), which was also calculated for comparison reasons. In
nursing homes the AUC of the CDS was 0.65 and gives a slightly
lower differentiation than the Braden scale (AUC 0.72).

Table 2 Sex distribution, frequency of pres-

Nursing homes Hospitals Total
sure ulcers, mean age and mean Care Depen-
(n = 3530) (n = 7204) (n = 10 734)
dency Scale (CDS) scores of individuals with
Female 79.0% 55.7% 63.4% and without pressure ulcers
Numbers (and percentages) of 214 (6.1%) 773 (10.9%) 987 (9.2%)
individuals with pressure ulcers
Female* 162 (5.9%) 448 (11.4%) 610 (9.0%)
Male* 52 (7.1%) 325 (10.4%) 377 (9.6%)
Mean age (SD) 83.0 (11.0%) 65.8 (17.4%) 71.5 (17.6%)
With pressure ulcer† 82.7 (10.8%) 73.8 (14.7%) 75.7 (17.8%)
Without pressure ulcer† 83.0 (11.0%) 64.8 (17.5%) 71.1 (17.8%)
Mean CDS score (SD) 41.6 (19.2%) 62.6 (17.1%) 55.9 (20.3%)
With pressure ulcer‡ 32.3 (17.1%) 44.8 (20.5%) 42.4 (19.7%)
Without pressure ulcer‡ 42.2 (19.2%) 64.8 (15.2%) 57.4 (18.7%)

*Chi-squared tests were statistically not significant, either in nursing homes or in hospitals.
†
t-test was statistically significant in hospitals and not significant in nursing homes (P < 0.05).
‡
t-test was statistically significant in both settings (P < 0.05).

1020 © 2008 The Authors. Journal compilation © 2008 Blackwell Publishing Ltd
E.I. Mertens et al. Pressure ulcer risk screening with the CDS

Table 3 Sensitivity and specificity percentage values for different Care In hospitals, sensitivity and specificity values above 60% can be
Dependency Scale (CDS) sum scores attained with CDS sum scores between 53 and 68 (Table 3). A
cut-off point of 64 would provide high sensitivity and specificity
Nursing homes Hospitals
values finding the most right positives as well as the most right
CDS sum score Sensitivity Specificity Sensitivity Specificity negatives. A cut-off point of 69 identifies 84.6% right positives
15 20.1 90.1 13.0 98.0 whereas 58.3% right negatives are identified as well.
16 24.0 87.9 14.0 97.8 In the nursing homes we found very high sensitivity values
17 26.5 86.3 15.7 97.4 above 85% for all CDS sum scores above 53, but specificity was
18 31.4 84.6 17.3 97.2 quite low then. A cut-off point of 64 would identify 94.8% right
19 34.8 83.0 19.2 96.9 positives and only 16.6% right negatives. A substantially lower
20 36.8 81.1 19.7 96.5
cut-off point of 41 would identify 70.1% right positives and at least
21 40.2 79.6 20.8 96.2
22 42.2 77.7 22.0 95.9 50.7% right negatives.
23 45.1 76.0 23.0 95.6
24 47.5 74.4 24.3 95.4
25 48.5 72.5 25.4 95.2
Odds ratios for CDS items in relation to
26 50.5 70.8 26.2 94.9 pressure ulcers
27 52.9 69.5 27.5 94.6
Figure 2 illustrates the odds ratios for all 15 CDS items as well as
28 53.9 68.1 28.1 94.2
29 54.9 66.8 29.1 93.9
for age (ⱕ66 years) and sex (male).
30 56.4 65.2 29.9 93.5 The calculations for hospitals resulted in odds ratios of almost
31 56.9 63.6 30.9 93.2 four or more for all CDS items. Eleven items had odds ratios above
32 57.8 62.1 31.7 92.9 five, and three items (‘continence’, ‘mobility’, ‘hygiene’) had odds
33 59.3 60.8 33.0 92.5 ratios above seven. Chi-squared tests (P < 0.05) were statistically
34 60.3 59.7 34.5 92.1 significant for all CDS items. An age of more than 65 years
35 60.8 58.3 36.0 91.8
resulted in an odds ratio of 2.9, which was also statistically sig-
36 61.3 57.2 37.7 91.6
37 63.7 55.7 39.5 91.2 nificant in chi-squared test (P < 0.05). The male sex had an insig-
38 65.2 54.3 40.9 90.8 nificant odds ratio of 1.1.
39 67.2 53.3 42.3 90.4 In nursing homes all CDS items showed odds ratios above 1.5,
40 69.1 51.9 43.9 90.0 except learning ability (odds ratio 1.1). Eight items had odds ratios
41 70.1 50.7 44.3 89.6 above two, and four items (‘mobility’, ‘getting (un)dressed’,
42 71.1 49.4 45.6 89.4
‘hygiene’, ‘avoidance of danger’) had odds ratios above three.
43 72.1 48.2 46.4 88.9
44 73.5 46.6 48.1 88.5 Chi-squared tests (P < 0.05) were significant for all CDS items in
45 74.5 45.5 50.2 88.1 nursing homes except for ‘learning ability’. In this setting, an age
46 75.0 44.1 51.0 87.7 above 65 (odds ratio 1.1) and the male sex (odds ratio 0.8) showed
47 76.5 43.0 51.9 87.2 no statistically significant relationships in chi-squared tests
48 78.4 41.8 52.6 86.7 (P < 0.05).
49 81.4 40.2 54.0 86.1
Mobility and hygiene belonged to the items with the highest
50 82.4 39.0 55.8 85.6
51 82.8 37.7 57.0 84.9
odds ratios in both settings.
52 84.8 36.1 58.4 84.3
53 85.3 34.1 60.0 83.5
Relative risks for CDS items in relation to
54 86.3 32.6 62.3 82.7
55 87.3 31.3 63.9 81.7 pressure ulcers
56 87.7 29.3 65.1 81.0 Figure 3 represents the relative risks of all CDS items, age and sex,
57 87.7 27.6 66.3 80.1
which were computed according to the algorithm suggested by
58 89.7 25.8 68.3 78.9
59 90.2 24.2 69.9 78.0 Dietz et al. [29].
60 91.2 22.7 70.7 76.7 The relative risks of the CDS items in hospitals were above 5.7
61 92.2 21.1 71.7 75.4 each. ‘Continence’, ‘mobility’, ‘getting (un)dressed’ and ‘hygiene’
62 93.1 19.8 73.4 74.0 were the items with the highest relative risks. Wald tests (P < 0.05)
63 93.6 18.0 74.8 72.3 were statistically significant for all CDS items in this setting.
64 94.6 16.6 76.2 70.4
Being older than 65 resulted in a relative risk of 5.0 which was
65 95.1 15.1 77.2 68.2
66 95.6 13.5 79.1 66.2 statistically significant in Wald test (P < 0.05) and being male
67 97.1 11.7 81.1 63.6 corresponded to a statistically insignificant relative risk of 1.2.
68 97.1 10.1 82.8 61.2 In nursing homes, the relative risks were higher than 1.3 for each
69 98.0 8.9 84.6 58.3 CDS item apart from ‘learning ability’ (relative risk 1.0). Eight
70 98.0 7.4 86.3 55.0 items had relative risks above 2.5. ‘Mobility’, ‘getting (un)dressed’,
71 99.0 5.9 88.3 51.1
‘hygiene’ and ‘avoidance of danger’ were the most concise items
72 99.0 4.8 90.5 46.4
73 99.5 3.5 92.6 40.4
with values above 6.9. Statistical significances were present for all
74 99.5 2.6 94.8 33.4 CDS items except for ‘contact with others’, ‘sense of rules/values’,
75 100 0 100 0 ‘recreational activities’ and ‘learning ability’. Age and sex corre-
sponded to statistically insignificant relative risks beneath 1.

© 2008 The Authors. Journal compilation © 2008 Blackwell Publishing Ltd 1021
Pressure ulcer risk screening with the CDS E.I. Mertens et al.

Nursing homes Hospitals

1.0 1.0

0.8 0.8
sensitivity

sensitivity
0.6 0.6

0.4 0.4

0.2 0.2

0.0 0.0
0.0 0.2 0.4 0.6 0.8 1.0 0.0 0.2 0.4 0.6 0.8 1.0
Figure 1 Receiver operating characteristics
1-specificity 1-specificity curves in nursing homes and hospitals.

8
Hospitals
Nursing homes
7

1
** ** ** ** ** ** ** ** ** ** ** ** ** ** * *
Mobility

Hygene

Age 66+
Continence

being male
Body posture

Lerning ability
Eating/drinking

Daily Activities
Communication
Day/night pattern

Body temperature

Contact with others

Getting (un)dressed

Avoidance of danger

Sense of rules/values

Recreational activities

Figure 2 Odds ratios in nursing homes and

hospitals [statistically significant difference
(*P < 0.05)].

relationship to the existence of pressure ulcers, whereby being

Logistic regression analysis
dependent on help in the items ‘learning ability’ (Exp(B) 0.5) and
The analyses of odds ratios and relative risks indicated moderate to ‘recreational activities’ (Exp(B) 0.4) seemed to have a slightly
very high relationships between pressure ulcers and every CDS protecting effect. Sex and age showed no independent relationship
item and a moderate relationship between age and pressure ulcers to pressure ulcers in nursing homes.
in hospitals. However, we can assume that some CDS items and In hospitals, we found three CDS items to have statistically
age or sex may influence each other. In order to explore the significant relationships to pressure ulcers according to logistic
possible mutual influences and to reveal items with relevant regression analysis. ‘Mobility’ had an Exp(B) of 1.8, indicating
relationships to pressure ulcers, logistic regression analyses were that the chance for individuals being care dependent in their mobil-
conducted. All factors remaining statistically significant in the ity to have a pressure ulcer was almost twice as high as for those
regression analyses are separately listed in Table 4 for nursing whose mobility was intact.
homes and hospitals. ‘Body posture’ and ‘body temperature’ had an Exp(B) value of
It could be shown that for nursing home residents, ‘mobility’ 1.4, indicating that the chance to have a pressure ulcer was one and
(Exp(B) 2.5) and ‘daily activities’ (Exp(B) 4.1) had a high positive a half times higher for individuals with dependencies in those

1022 © 2008 The Authors. Journal compilation © 2008 Blackwell Publishing Ltd
E.I. Mertens et al. Pressure ulcer risk screening with the CDS

16
Hospitals
Nursing homes
14

12

10

** ** ** ** ** ** ** ** ** ** * * ** * * *

Mobility

Hygene

Age 66+
Continence

being male
Body posture

Lerning ability
Eating/drinking

Daily Activities
Communication
Day/night pattern

Body temperature

Contact with others

Getting (un)dressed

Avoidance of danger

Sense of rules/values

Recreational activities
Figure 3 Relative risks in nursing homes and
hospitals [statistically significant difference
(*P < 0.05)].

Table 4 Care Dependency Scale items with a statistically significant with pressure ulcers were considerably more care dependent than
relationship to pressure ulcers in logistic regression analysis in nursing those without pressure ulcers. Nursing home residents had a mod-
home residents (n = 3530) and hospital patients (n = 7204) erately higher care dependency when they had a pressure ulcer.
Exp(B) 95.0% confidence interval
A second indication is the high correlation between the sum
scores of the CDS and the Braden scale, a tool specifically devel-
†
Nursing homes oped for pressure ulcer risk detection. The Pearson’s coefficient
Mobility 2.5* 1.2–5.5 was r = 0.79 (P < 0.01) in nursing homes and r = 0.89 (P < 0.01)
Daily activities 4.1* 1.7–9.8 in hospitals. According to Polit and Beck [30] this is a high value
Learning ability 0.5* 0.3–1.0 and indicates a good construct validity of the CDS in relation to
Recreational activities 0.4* 0.2–0.9
pressure ulcer risks. This result confirms the clinical experience
Hospitals‡
that a high care dependency increases the risk of developing a
Mobility 1.8* 1.3–2.6
pressure ulcer.
Body posture 1.4* 1.0–1.9
For evaluating the predictive validity of the CDS, sensitivity and
Body temperature 1.4* 1.1–1.8
specificity values were calculated and plotted in ROC curve graphs.
*Statistically significant difference (P < 0.05). The AUC (0.80) indicates a good differentiation ability of the CDS
†
Total explained variance (Nagelkerke’s R2) = 0.045. in hospitals. In nursing homes the value for the AUC was a mere
‡
Total explained variance (Nagelkerke’s R2) = 0.214. 0.65, indicating a weak differentiation ability. This result is allege-
able by the distribution of sensitivity and specificity values. While
we found CDS sum scores with high sensitivity as well as high
items. Sex had no statistically significant relationship in hospitals. specificity in hospitals, a high sensitivity came along with very low
Age had a statistically significant Exp(B), but this was very small specificity in the nursing homes. For use of the scale in practice this
(1.01, confidence interval 1.00–1.02) and therefore clinically not means that the CDS has a high hit ratio in hospitals. The most
relevant. balanced value was 64, with this cut-off point it was possible to
identify 76.4% right positives and 70.4% right negatives. However,
for screening purposes a cut-off point of 69 is more appropriate,
Discussion because it correctly classifies 84.6% as being at risk whereas 58.3%
The aim of the study was to evaluate if it is possible to conduct a right negatives are identified as well. That means it identifies a high
pressure ulcer risk screening using the CDS. The presented results proportion of truly endangered individuals without assigning a risk
indicate that the CDS is a well-functioning tool for pressure ulcer to too many individuals who are not actually endangered.
risk detection in both hospitals and nursing homes. Owing to the weak balance between sensitivity and specificity
This is indicated by the statistically significant differences of the in nursing homes, it is unattainable to specify a cut-off point that
CDS sum scores in relation to pressure ulcers. Hospital patients would determine a likewise large proportion of correctly positives

© 2008 The Authors. Journal compilation © 2008 Blackwell Publishing Ltd 1023
Pressure ulcer risk screening with the CDS
as well as correctly negatives. The most appropriate cut-off point
is 41, with which 70.1% right positives and 50.7% right negatives
can be identified. As the nursing home population is very homog-
enous and care dependency is generally high, it is difficult to detect
a pressure ulcer risk using solely the CDS sum score, especially
when the assessment is not conducted frequently. Therefore, all
nursing home residents should be assumed to have a (potential)
pressure ulcer risk and an individual and close-meshed assessment
of their care dependency, with special respect to the items named
in the following, and main pressure ulcer risk factors should be
conducted.
A further aim of the study was to evaluate if there are special
items that can be used as alarm signals for pressure ulcer risks.
Analyses on item level confirmed that the CDS is sensitive for
pressure ulcers, because calculations of odds ratios indicated that
all CDS items (apart from ‘learning ability’ in nursing homes) had
a moderate to very high relation to pressure ulcers. Calculations of
relative risks indicated moderate to very high relation to pressure
ulcers for all CDS items in hospitals and for 11 items in nursing
homes. ‘Mobility’ and ‘hygiene’ were outstanding because they
belonged to the items with the highest odds ratios in hospitals and
nursing homes.
In each setting, odds ratio and relative risk calculations revealed
very high values for the same items: in hospitals they were ‘con-
tinence’, ‘mobility’ and ‘hygiene’; in nursing homes they were
‘mobility’, ‘getting (un)dressed’, ‘hygiene’ and ‘avoidance of
danger’. As the results were stable in both analyses it can be
assumed that individuals, who are care dependent in these items,
have an increased pressure ulcer risk.
When adjusting for mutual influences by using logistic regres-
sion analyses, ‘mobility’ and ‘daily activities’ were significant as
items with positive relation to pressure ulcers in nursing homes. In
hospitals this applied to ‘mobility’, ‘body posture’ and ‘body tem-
perature’. The substance of the logistic regression analyses is
limited owing to the low explained variances of 4.5% in nursing
homes and 21.4% in hospitals. However, when considering all
analyses on item level, it is remarkable that mobility is repeatedly
identified as one of the items with the strongest relation to pressure
ulcers in all performed analyses. Therefore it can be assumed that
mobility is a key item for the detection of pressure ulcer risks.
The bivariate analyses showed a statistical significance for mean
age differences in relation to pressure ulcers in hospitals but not in
nursing homes. Hospital patients become more care dependent
with an increasing age and our further analyses revealed strong
relationships between care dependency/CDS items and pressure
ulcers in both settings. Considering this, we can assume that care
dependency is the effective variable for pressure ulcers in spite
of age.
In a total of 987 individuals with pressure ulcers, none of our
analyses gave statistically significant clues if women suffered
more often from pressure ulcers than men did, or vice versa.
Therefore, we can conclude that sex has no relationship to pressure
ulcers, at least in the presented large and heterogeneous sample.
The results of the research are limited by the cross-sectional
design, which does not allow conclusions as to the direction of
causalities. For example, it may well be that a pressure ulcer
causes a restricted mobility and not the other way round. Consid-
ering this, expanded analyses were accomplished in which only
ulcers grade 2 or worse and only institution-acquired ulcers were
1024
E.I. Mertens et al.
included. These analyses confirmed the represented results.
However, in order to obtain sound evidence, the findings should be
validated in a prospective design.
Acknowledgements
We thank all participating institutions for their cooperation and
especially the patients and residents, as well as the nurses who
performed the data collecting, for their valuable contribution to
this research.
References
1. Vanderwee, K., Clark, M., Dealey, C., Gunningberg, L. & Defloor, T.
(2007) Pressure ulcer prevalence in Europe: a pilot study. Journal of
Evaluation in Clinical Practice, 13 (2), 227–235.
2. Hopkins, A., Dealey, C., Bale, S., Defloor, T. & Worboys, F. (2006)
Patient stories of living with a pressure ulcer. Journal of Advanced
Nursing, 56 (4), 345–353.
3. Bennett, G., Dealey, C. & Posnett, J. (2004) The cost of pressure
ulcers in the UK. Age and Ageing, 33 (3), 230–235.
4. Landi, F., Onder, G., Russo, A. & Bernabei, R. (2007) Pressure ulcer
and mortality in frail elderly people living in community. Archives of
Gerontology and Geriatrics, 44 (Suppl. 1), 217–223.
5. DNQP. (2004) Expertenstandard Dekubitusprophylaxe in der Pflege.
Osnabrück: Deutsches Netzwerk für Qualitätsentwicklung in der
Pflege.
6. EPUAP. (2007) Pressure Ulcer Prevention Guidelines. In: European
Pressure Ulcer Advisory Panel. Available at: http://www.epuap.
org (accessed 25 June 2007).
7. Dijkstra, A., Buist, G. & Dassen, T. (1996) Nursing-care dependency.
Development of an assessment scale for demented and mentally handi-
capped patients. Scandinavian Journal of Caring Sciences, 10 (3),
137–143.
8. Mertens, E. I., Halfens, R. J. & Dassen, T. (2007) Using the Care
Dependency Scale for fall risk screening. Journal of Advanced
Nursing, 58 (6), 594–601.
9. Dassen, T., Heinze, C., Lahmann, N. A., Mertens, E. I. & Tannen, A.
(eds) (2006) Prävalenzerhebung 2006. Pflegeabhängigkeit, Sturzereig-
nisse, Inkontinenz, Dekubitus, Berlin: Institut für Medizin-/
Pflegepädagogik und Pflegewissenschaft.
10. Dassen T., Heinze, C., Lahmann, N. A. & Lohrmann, C. (eds) (2004)
Prävalenzerhebung 2004. Pflegeabhängigkeit, Sturzereignisse, Inkon-
tinenz, Dekubitus, Berlin: Institut für Medizin-/Pflegepädagogik und
Pflegewissenschaft.
11. Balzer, K., Pohl, C., Dassen, T. & Halfens, R. (2007) The Norton,
Waterlow, Braden and Care Dependency Scales: comparing their
validity when identifying patients’ pressure sore risk. Journal of
Wound, Ostomy and Continence Nursing, 34(4), 389–398.
12. Dassen T., Heinze, C., Lahmann, N. A., Lohrmann, C., Mertens, E. I.
& Tannen, A. (eds) (2005) Prävalenzerhebung. Pflegeabhängigkeit,
Sturzereignisse, Inkontinenz, Dekubitus, Berlin: Institut für Medizin-/
Pflegepädagogik und Pflegewissenschaft.
13. Henderson, V. (1966) The Nature of Nursing: A Definition and Its
Implications for Practice, Research and Education. New York: Mac-
Millan Press.
14. Dijkstra, A., Buist, G. & Dassen, T. (1998) Operationalization of the
concept of ‘nursing care dependency’ for use in long-term care facili-
ties. The Australian and New Zealand Journal of Mental Health
Nursing, 7 (4), 142–151.
15. Dijkstra, A., Brown, L., Havens, B., Romoren, T., Zanotti, R., Dassen,
T. & van den Heuvel, W. (2000) An international psychometric testing
of the care dependency scale. Journal of Advanced Nursing, 31 (4),
944–952.
© 2008 The Authors. Journal compilation © 2008 Blackwell Publishing Ltd
E.I. Mertens et al.
16. Dijkstra, A., Buist, G. & Dassen, T. (1998) A criterion-related validity
study of the Nursing-Care Dependency (NCD) scale. International
Journal of Nursing Studies, 35 (3), 163–170.
17. Dijkstra, A., Buist, G., Moorer, P. & Dassen, T. (1999) Construct
validity of the Nursing Care Dependency Scale. Journal of Clinical
Nursing, 8 (4), 380–388.
18. Dijkstra, A., Buist, G., Moorer, P. & Dassen, T. (2000) A reliability
and utility study of the care dependency scale. Scandinavian Journal
of Caring Science, 14 (3), 155–161.
19. Dijkstra, A., Coleman, M., Tomas, C., Valimaki, M. & Dassen, T.
(2003) Cross-cultural psychometric testing of the Care Dependency
Scale with data. Journal of Advanced Nursing, 43 (2), 181–187.
20. Dijkstra, A., Tiesinga, L. J., Goossen, W. T. & Dassen, T. W. (2002)
Further psychometric testing of the Dutch Care Dependency Scale on
two different patient Groups. International Journal of Nursing
Practice, 8 (6), 305–314.
21. Dijkstra, A., Tiesinga, L. J., Plantinga, L., Veltman, G. & Dassen, T.
W. (2005) Diagnostic accuracy of the care dependency scale. Journal
of Advanced Nursing, 50 (4), 410–416.
22. Lohrmann, C., Balzer, K., Dijkstra, A. & Dassen, T. (2003) Care
dependency in nursing homes – a psychometric study. Zeitschrift für
Gerontologie und Geriatrie, 36 (4), 255–259.
23. Lohrmann, C., Dijkstra, A. & Dassen, T. (2003) Care dependency:
testing the German version of the Care Dependency Scale in nursing
homes and on geriatric wards. Scandinavian Journal of Caring
Sciences, 17 (1), 51–56.
© 2008 The Authors. Journal compilation © 2008 Blackwell Publishing Ltd
Pressure ulcer risk screening with the CDS
24. Bergstrom, N., Braden, B. J., Laguzza, A. & Holman, V. (1987) The
Braden Scale for Predicting Pressure Sore Risk. Nursing Research, 36
(4), 205–210.
25. Braden, B. J. (2007) Braden Scale for Predicting Pressure Sore Risk.
Available at: http://www.Bradenscale.Com/Braden.PDF (accessed 25
June 2007).
26. Pancorbo-Hidalgo, P. L., Garcia-Fernandez, F. P., Lopez-Medina, I.
M. & Alvarez-Nieto, C. (2006) Risk assessment scales for pressure
ulcer prevention: a systematic review. Journal of Advanced Nursing,
54 (1), 94–110.
27. Papanikolaou, P., Lyne, P. & Anthony, D. (2007) Risk assessment
scales for pressure ulcers: a methodological review. International
Journal of Nursing Studies, 44 (2), 285–296.
28. Schlömer, G. (2003) Dekubitusrisikoskalen als Screeninginstrumente
– ein systematischer Überblick externer Evidenz. Zeitschrift für
Ärztliche Fortbildung und Qualitätssicherung, 97 (1), 33–46.
29. Dietz, E., Böhning, D., Englert, H. & Willich, S. (2005) Eine Methode
zur Schätzung von Relativen Risiken aus Querschnittsstudien mit
Krankheitsdauerdaten. In 50 Jahrestagung der GMDS und 12. Jahr-
estagung der DAE (eds R. Klar, W. Köpcke & K. Kuhn), pp. 588–593.
Köln, Düsseldorf: Deutsche Gesellschaft für Medizinische Informatik,
Biometrie und Epidemiologie.
30. Polit, D. F. & Beck, C. T. (2003) Nursing research: Principles and
Methods. Philadelphia, PA: Lippincott Co.
1025