BJD

British Journal of Dermatology

C L I N I C A L A N D LA B O R A T O R Y I N V E S T I G A T I O N S

Which factors predict incident pressure ulcers in hospitalized patients? A prospective cohort study* T. Petzold,1,2 M. Eberlein-Gonska2 and J. Schmitt1 1

Center for Evidence-Based Healthcare and 2Department of Quality and Medical Risk Management, University Hospital Carl Gustav Carus Dresden, Technical University Dresden, Fetscherstraße 74, 01307 Dresden, Germany

Summary Correspondence Jochen Schmitt. E-mail: [email protected]

Accepted for publication 17 February 2014

Funding sources No external funding.

Conflicts of interest None declared. *Plain language summary available online. DOI 10.1111/bjd.12915

Background The prevention of pressure ulcers (PU) is an important public health issue owing to their substantial clinical and economic burden. Objectives To investigate predictors of incident PU in hospitalized patients and the performance of the Braden Scale in intensive care units (ICU) and normal care units (NCU). Methods We conducted a prospective cohort study including all inpatients treated at the University Hospital Carl Gustav Carus Dresden, Germany, between 2007 and 2011. Documentation comprised patient characteristics, Braden Scale and clinical signs of PU. The primary outcome was incident PU during inpatient treatment. Predictors of PU were explored by using univariate and multivariate logistic regression models. To evaluate the performance of the Braden Scale a receiver operating characteristics (ROC) curve analysis was applied. Results The overall incidence of PU during inpatient treatment was 0
78%. A higher rate of PU was observed at ICU vs. NCU (4
77% vs. 0
59%). Multivariate analysis identified age [odds ratio (OR) 1
04, 95% confidence interval (CI) 1
035–1
041 per year], female sex (OR 1
11, 95% CI 1
01–1
22), length of stay (OR 17
79, 95% CI 15
46–20
48 for 30 or more days vs. < 10 days) and admission from care facility compared with admission from home (OR 3
14, 95% CI 2
63–3
75) as significant predictors of incident PU. The area under the ROC curve was 84
89% at NCU and 69
0% at ICU. Conclusions The identified predictors for incident PU may inform targeted, evidence-driven preventive measures to decrease the burden of PU.

What’s already known about this topic?

•

Although the prevention of pressure ulcer (PU) during inpatient treatment is an important public health issue, predictors of incident PU in hospitalized patients and the performance of existing PU risk assessment instruments are unclear.

What does this study add?

• •

© 2014 British Association of Dermatologists

This large prospective cohort study identified higher age, female sex, length of hospitalization, treatment on intensive care unit and admission from care facility as independent predictors for incident PU in hospitalized patients. The performance of the Braden Scale to assess the individual PU risk differs between normal care and intensive care units, with better performance on normal care units.

British Journal of Dermatology (2014) 170, pp1285–1290

1285

1286 Predictors of pressure ulcers, T. Petzold et al.

Pressure ulcers (PU) are localized injury to the skin and/or underlying tissue, usually over bony prominences, as a result of pressure or pressure in combination with shear forces.1 About 14 000 new cases of PU are documented in the U.K. every year, resulting in 7
9% per 10 000 bed days in inpatient provision.2 Costs associated with PU are the third highest, after those for cancer and cardiovascular diseases, up to $70 000 for each PU.3 PU are divided into four levels of severity, ranging from persistent skin redness (grade 1) to tissue necrosis (grade 4). Intrinsic risk factors of PU include age, weight, inactivity and malnutrition. Extrinsic risk factors are friction and shearing forces, humidity, bed positioning and treatment with certain drugs (e.g. analgesics, sedatives or hypnotics).4 Monitoring and treatment of PU in clinical settings are key indicators of quality of care.5 Different risk assessment scales are established for the detection of PU (e.g. Norton Scale, Braden Scale or Waterlow Scale). However, each assessment has indistinct validity, test–retest and interobserver reliability, and performance at different levels of hospital care.6–10 Despite these limitations we decided, in 2007, to use the Braden Scale. Nevertheless, the high level of clinical and public health relevance, there is a lack of current epidemiological studies on the incidence, severity and risk factors of PU in patient care. We aimed to investigate: (i) predictors of incident PU; (ii) risk factors of prevalent PU at admission; and (iii) the performance of the Braden Scale in different levels of inpatient care.

sex), admission from home or other facilities, length of stay, stay on the intensive care unit (ICU) and overall morbidity measured by means of the patient clinical complexity level (PCCL). Statistical analysis Initially, we calculated the incidence of PU during hospitalization and the prevalence of PU at admission along with corresponding 95% confidence intervals (CI). Descriptive statistics were used to characterize the study population and prevalence of hypothetical PU determinants such as age, sex, level of care [normal care unit (NCU) and ICU], morbidity (PCCL), length of stay and Braden Scale. Primary outcome was incident PU during hospital stay. Secondary outcomes were prevalent PU at admission and improvement of prevalent PU during hospital stay. Univariate and multivariate logistic regression models were used to identify determinants of incident PU during hospital stay, taking into account the above mentioned patient characteristics and characteristics of hospital stay as predefined confounders. Length of stay was not included in the analysis for prevalent PU as it is not a baseline characteristic. Receiver operating characteristics (ROC) curve analysis was applied to evaluate the performance of the Braden Scale in NCU vs. ICU. The area under the curve (AUC) indicates how well the Braden Scale separates the inpatients being tested into those with and without PU.15

Methods and materials We conducted a prospective cohort study of the University Hospital Carl Gustav Carus Dresden (UKD). The UKD is a tertiary care facility offering inpatient treatment in all medical disciplines. All patients hospitalized and discharged between 1 January 2007 and 31 December 2011 were included in the study (n = 246 162). During the entire study period, a PU screening was performed for all hospitalized patients at all clinical departments. At admission, discharge and at change in patient’s condition (subjective estimation of the clinician), trained nurses performed an examination of the entire skin with documentation of existing or newly occurring PU. A detected PU had to be verified by the attending physician or consulting physician and encoded as primary or secondary diagnosis. In the case of an existing PU, the localization and degree (grade 1–4) was documented in the hospital information system. The Braden Scale11–14 is a validated, standardized instrument for assessing PU risk using the following six patientrelated items: sensory sentience, skin moisture, activity, mobility, nutrition status, friction and shear forces. One to four points are assigned for each item of the Braden Scale. The resulting total score varies between 6 and 23 points. A higher score corresponds with a lower risk of PU. The cut-off among risk and nonrisk of PU is 19 points. Other potential risk factors for incident PU were also recorded, such as patient demographic characteristics (age and British Journal of Dermatology (2014) 170, pp1285–1290

Results No temporal trends were observed in terms of increase or decrease of incidence or prevalence. All analyses are based on the entire data of the years 2007–2011 (Table 1). A total of 1914 patients (0
78%, 95% CI 0
74–0
81%) developed an incident PU during inpatient stay, with significant differences between the departments. The incidence ranged from 0
0% to 12
7% between the clinical departments. The prevalence of PU during the entire study period was 1
21% (95% CI 1
16–1
25%; n = 2971). In 37% of the patients (n = 1089) with PU at admission no more PU was observed at discharge. During hospitalization 43% of all patients with PU (n = 1278) improved and had a lower grade at discharge compared with admission. A total of 327 patients (11%) with prevalent PU worsened during hospitalization. As expected, a higher rate of PU was observed in ICU vs. NCU (4
77% vs. 0
59%). Compared with other disciplines, patients of the dermatology department had a lower risk of developing a PU during inpatient treatment. PU incidence and prevalence in the dermatology department were 0
2% and 1
0%, respectively. The most frequent localizations of incident PU were heel (21
7%; n = 416), ischium (19
7%; n = 378), sacrum (19
5%; n = 374), buttocks (13
4%; n = 257) and ankle (3
2%; n = 63). Localizations of PU present on admission were heel (22
1%; n = 655), ischium (21
5%; n = 641), © 2014 British Association of Dermatologists

Predictors of pressure ulcers, T. Petzold et al. 1287 Table 1 Characteristics of the study population (n = 246 162) Year of admission

Sex Female Male Unknown Age (in years) 0–9 10–19 20–29 30–39 40–49 50–59 60–69 70–79 80–89 > 89 Length of stay (in days) < 10 11–20 21–30 > 30 Patient clinical complexity level 0 1 2 3 4 Unknown Admission from Home Nursing home Other hospital Other Braden Score at admission No PU risk (< 19 points) Pressure ulcer risk (≥ 19 points) Treatment in Normal care unit Intensive care unit Prevalent PU at admission Patients without prevalent PU Patients with prevalent PU Incidence PU during hospitalization Patients without incidence PU Patients with incidence PU

Total n (%)

2007, n (%)

2008, n (%)

2009, n (%)

2010, n (%)

2011, n (%)

119 247 (48
4) 126 912 (51
6) 3 (0
001)

21 936 (48
4) 23 424 (51
6) 0

23 439 (48
2) 25 204 (51
8) 0

24 504 (48
7) 25 798 (51
3) 0

24 811 (48
5) 26 308 (51
5) 0

24 557 (48
4) (51.5) 3 (0
01)

27 869 13 832 20 374 19 189 23 657 31 684 42 357 44 671 19 622 2907

(11
3) (5
6) (8
3) (7
8) (9
6) (12
9) (17
2) (18
1) (8
0) (1
2)

5401 2911 3621 3522 4494 5612 8485 7473 3336 505

(11
9) (6
4) (8
0) (7
8) (9
9) (12
4) (18
7) (16
5) (7
4) (1
1)

5687 2793 3900 3744 4829 6415 8558 8646 3632 439

(11
7) (5
7) (8
0) (7
7) (9
9) (13
2) (17
6) (17
8) (7
5) (0
9)

5528 2837 4271 3850 4860 6477 8765 9081 4096 537

(11
0) (5
6) (8
5) (7
7) (9
7) (12
9) (17
4) (18
1) (8
1) (1
1)

5601 2714 4424 4012 4865 6583 8378 9545 4311 686

(11
0) (5
3) (8
7) (7
8) (9
5) (12
9) (16
4) (18
7) (8
4) (1
3)

5652 2577 4158 4061 4609 6597 8171 9926 4247 740

(11
1) (5
1) (8
2) (8
0) (9
1) (13
0) (16
1) (19
6) (8
4) (1
5)

202 070 27 309 8164 8619

(82
1) (11
1) (3
3) (3
5)

36 627 5464 1561 1708

(80
7) (12
0) (3
4) (3
8)

39 694 5489 1623 1837

(81
6) (11
3) (3
3) (3
8)

41 319 5537 1711 1735

(82
1) (11
0) (3
4) (3
4)

42 173 5511 1640 1795

(82
5) (10
8) (3
2) (3
5)

42 257 5308 1629 1544

(83
38) (10
5) (3
2) (3
0)

140 080 3091 29 904 37 639 27 440 8008

(56
9) (1
3) (12
1) (15
3) (11
1) (3
3)

24 448 379 5790 6992 6262 1489

(53
9) (0
8) (12
8) (15
4) (13
8) (3
3)

27 559 824 6443 7109 5194 1514

(56
7) (1
7) (13
2) (14
6) (10
7) (3
1)

28 551 579 6733 7518 5263 1658

(56
8) (1
2) (13
4) (14
9) (10
5) (3
3)

29 295 711 5733 8119 5490 1771

(57
3) (1
4) (11
2) (15
9) (10
7) (3
5)

30 227 598 5205 7901 5231 1576

(59
6) (1
2) (10
3) (15
6) (10
3) (3
1)

229 901 4417 5860 5984

(93
4) (1
8) (2
4) (2
4)

45 033 141 163 23

(99
3) (0
3) (0
4) (0
1)

48 317 158 153 15

(99
3) (0
3) (0
3) (0
03)

45 445 1298 1784 1775

(90
3) (2
6) (3
5) (3
5)

45 912 1419 1939 1849

(89
8) (2
8) (3
8) (3
6)

45 194 1401 1821 2322

(89
1) (2
8) (3
6) (4
6)

213 584 (86
8) 32 578 (13
2)

39 443 (87
0) 5917 (13
0)

42 342 (87
0) 6301 (13
0)

43 378 (86
2) 6924 (13
8)

44 292 (86
6) 6827 (13
4)

44 129 (87
0) 6609 (13
0)

235 658 (95
7) 10 504 (4
3)

43 106 (95
0) 2254 (5
0)

46 504 (95
6) 2139 (4
4)

48 207 (95
8) 2095 (4
2)

49 018 (95
9) 2101 (4
1)

48 823 (96
2) 1915 (3
8)

243 191 (98
8) 2971 (1
2)

44 816 (98
8) 544 (1
2)

48 093 (98
9) 550 (1
1)

49 684 (98
8) 618 (1
2)

50 485 (98
8) 634 (1
2)

50 113 (98
8) 625 (1
2)

244 248 (99
2) 1914 (0
8)

44 996 (99
2) 364 (0
8)

48 227 (99
1) 416 (0
9)

49 901 (99
2) 401 (0
8)

50 728 (99
2) 391 (0
8)

50 396 (99
3) 342 (0
7)

PU, pressure ulcer.

sacrum (18
5%; n = 551), buttocks (10
6%; n = 315) and ankle (3
7%; n = 111) (Fig. 1). Determinants of incident pressure ulcers Table 2 summarizes predictors for incident PU during hospital stay. Univariate analyses indicated a significant association with all predefined factors except sex and PCCL. The incidence of PU of patients aged between 20 and 29 years was 0
22% compared with 4
51% in patients aged more than © 2014 British Association of Dermatologists

89 years (Fig. 2). The adjusted, multivariate analysis indicated an increased risk of PU with increasing patient age [odds ratio (OR) 1
037, 95% CI 1
035–1
041 per year], female sex (OR 1
11, 95% CI 1
01–1
22), length of stay (OR 17
79, 95% CI 15
46–20
48 for patients with more than 30 days length of stay compared with patients with fewer than 10 days), treatment on ICU (OR 4
03, 95% CI 3
59–4
53) and admission from care facility (OR 3
14, 95% CI 2
63–3
75). PCCL was not associated with an incident PU in the multivariate model. British Journal of Dermatology (2014) 170, pp1285–1290

1288 Predictors of pressure ulcers, T. Petzold et al.

Fig 1. Number of prevalent (n = 2971) and incident (n = 1914) pressure ulcer by localization (2007–2011).

Table 2 Predictors for incident pressure ulcer (n = 246 162) Logistic regression analysis, OR (95% CI) Determinants

Univariate analysis

Sex 1
01 (0
91–1
09) Age 1
03 (1
02–1
03) 1
19 (1
18–1
21) Braden Scale (per decrease of one point) Length of stay (reference: < 10 days) 11–20 days 8
71 (7
76–9
76) 21–30 days 14
82 (12
88–17
04) > 30 days 18
26 (16
04–20
79) Treatment on 2
88 (2
58–3
22) intensive care unit Patient clinical 0
99 (0
97–1
00) complexity level Admission from (reference: home) Care facility 6
05 (5
13–7
11) Other hospital 4
58 (3
90–5
39) Other 2
65 (2
16–3
25)

Multivariate analysisa 1
11 (1
01–1
22) 1
04 (1
035–1
041)

1
033–1
039 per year), length of stay (OR 1
023, 95% CI 1
021–1
026 per day), treatment on ICU (OR 2
46, 95% CI 2
11–2
88), and admission from care facility (OR 15
97, 95% CI 13
62–18
72) or other hospital (OR 13
32, 95% CI 11
37–15
61) compared with admission from home (Table 4). Performance of the Braden Scale

6
54 11
08 17
79 4
03

(5
81–7
36) (9
56–12
84) (15
46–20
48) (3
59–4
53)

0
987 (0
98–0
99)

3
14 (2
63–3
75) 1
10 (0
92–1
31) 1
28 (1
04–1
59)

CI, confidence interval; OR, odds ratio. aAdjusted for sex, age, grouped length of stay, treatment on intensive care unit, patient clinical complexity level and admission.

Determinants of prevalent pressure ulcers According to multivariate analysis, prevalent PU at admission was significantly associated with higher age (OR 1
033, 95% CI 1
031–1
035 per year), female sex (OR 1
27, 95% CI 1
18–1
38), admission on ICU (OR 2
80, 95% CI 2
50–3
13), PCCL (OR 1
002, 95% CI 1
005–1
004) and admission from care facility (OR 22
97, 95% CI 20
71–25
46) and from other hospitals (OR 13
62, 95% CI 12
16–15
25) compared with admission from home (Table 3). Determinants of pressure ulcer improvement Associated risk factors for PU improvement during stay (lower degree at discharge since admission) was female sex (OR 1
14, 95% CI 1
02–1
28), higher age (OR 1
036, 95% CI British Journal of Dermatology (2014) 170, pp1285–1290

The performance of the Braden Scale differed between NCU and ICU. On NCU, the ROC AUC was 84
89%, indicating that the risk for incident PU was correctly classified in 84
89% of all patients (Fig. 3). At the proposed cut-off of 19 points,14 the sensitivity, specificity, positive predictive value (PPV) and negative predictive value (NPV) of the Braden Scale at NCU were 44
57%, 89
95%, 3
14% and 99
69%, respectively. In contrast, on ICU, the area under the ROC curve was only 69
00%, indicating worse performance compared with the NCU. Sensitivity, specificity, PPV and NPV of the Braden Scale on ICU were 96
61%, 28
46%, 5
86% and 99
24%, respectively.

Discussion This study investigated predictors of incident and prevalent PU in a large, unselected cohort of patients in tertiary care. Incidence rates of PU vary greatly between the different clinical disciplines ranging from 0% to 12
3% on NCU and from 0% to 12
7% on ICU. Dermatological inpatients have a lower risk for both incident and prevalent PU compared with the overall average across disciplines. PU was not linked to a specific clinical department. Age, length of hospital stay and medical/nursing care of the patient before admission had an impact on the occurrence of incident PU. Inpatients who were transferred from a care facility had a threefold increased risk of PU over patients admitted from home. Despite the increasing number of cases, a constant average age of the patients and the same staffing key, PU incidence was constant during the observation period. The German Expert Standard for Pressure Ulcer Prevention recommends no specific measuring instrument for determin© 2014 British Association of Dermatologists

Predictors of pressure ulcers, T. Petzold et al. 1289

5·0

Pressure ulcer incidence (in%)*

4·5

Fig 2. Pressure ulcer incidence by age (n = 246 162; 2007–2011). *With 95% confidence interval.

4·0 3·5 3·0 2·5 2·0 1·5 1·0 0·5 0·0

0–9

10–19

Table 3 Logistic regression analyses on risk factors of prevalent pressure ulcer at admission (n = 246 162)

20–29

30–39

40–49 50–59 Age (in years)

60–69

70–79

80–89

>89

Table 4 Logistic regression analyses of risk factors of pressure ulcer improvement (n = 1278) during stay (lower degree at discharge since admission)

Logistic regression analysis, OR (95% CI) Determinants

Univariate analysis

Sex 0
95 (0
89–1
03) Age 1
04 (1
04–1
05) Braden score 1
35 (1
34–1
36) (per decrease of one point) Treatment on 5
21 (4
73–5
73) intensive care unit Patient clinical 1
003 (1
001–1
004) complexity level Admission from (reference: home) Care facility 46
11 (42
00–50
62) Other hospital 20
07 (18
09–22
27) Other 11
42 (10
09–12
92)

1
27 (1
18–1
38) 1
033 (1
031–1
035)

2
80 (2
50–3
13)

1
002 (1
0005–1
004)

22
97 (20
71–25
46) 13
62 (12
16–15
25) 10
34 (9
08–11
78)

CI, confidence interval; OR, odds ratio. aAdjusted for sex, age, treatment on intensive care unit, patient clinical complexity level and admission.

ing the risk of PU.16 Validation studies using the Braden Scale disclose different results based on the setting.8,10,11 Our study extends previous research as it indicates that the performance of the Braden Scale is very dependent on the setting. While its performance was good on NCU, its usefulness for the ICU setting may be questioned in accordance with our results. Developing a customized instrument for PU risk on ICU is required. Modifications in PU screening should be evidencebased and prospectively evaluated. To comply with the principles of evidence-based healthcare, quality management measures must provide scientifically approved methods. Finally, consequences and adjustments of preventive measures can be derived and implemented solely on the basis of valid data.17 This study included an unselected cohort of all patients treated at a German university hospital during the period 2007–2011. The integrity of the data and the large number of cases are advantages of the study. With the Braden Scale, a validated risk assessment11–14 was used by trained personnel. © 2014 British Association of Dermatologists

Logistic regression analysis, OR (95% CI)

Multivariate analysisa Determinants

Univariate analysis

Sex 0
93 (0
83–1
03) Age 1
04 (1
04–1
05) Braden Score 1
33 (1
32–1
34) (per decrease of one point) Length of stay 1
025 (1
023–1
027) (for each additional day) Treatment on 6
08 (5
30–6
96) intensive care unit Patient clinical 1
004 (1
001–1
007) complexity level Admission from (reference: home) Care facility 33
64 (29
11–38
88) Other hospital 24
56 (21
25–28
37) Other 11
93 (9
95–14
31)

Multivariate analysisa 1
14 (1
02–1
28) 1
036 (1
033–1
039)

1
023 (1
021–1
026)

2
46 (2
11–2
88)

0
999 (0
996–1
002)

15
97 (13
62–18
72) 13
32 (11
37–15
61) 9
19 (7
59–11
13)

CI, confidence interval; OR, odds ratio. aAdjusted for sex, age, length of stay, treatment on intensive care unit, patient clinical complexity level and admission.

The use of routine data has not been sufficiently validated. So far, there are hospital-based internal, random audits to verify the documented data. The University Hospital Dresden is a hospital of tertiary care. In the present study, all clinical departments were included and analysed. We believe our findings are generalizable to tertiary care hospitals in Germany. The finding of an increased PU risk and prevalence of patients transferred from other care facilities may not be generalizable. The risk factors for incident PU in inpatient care, such as age, length of stay and the medical/nursing care of the patient before admission are most likely widely generalizable. One limitation of this study is that it was not possible for us to evaluate the individual items of the Braden Scale. We adjusted for PCCL as an indicator of overall morbidity. HowBritish Journal of Dermatology (2014) 170, pp1285–1290

0·75

0·75 Sensitivity 0·5 0·25

0·25

Sensitivity 0·5

1

1

1290 Predictors of pressure ulcers, T. Petzold et al.

AUC*: 69·00% 0

0

AUC*: 84·89% 0 0·25 0·5 0·75 1 - Specificity Area under curve = 0·8498 se(area) = 0·0044

1

0 0·25 0·5 0·75 1 - Specificity Area under curve = 0·6900 se(area) = 0·0103

ever, residual confounding cannot be excluded as valid data on risk factors of PU such as obesity, smoking, history of diabetes and cardiovascular disease were not available for analysis. The performance of risk assessments for the collection of PU risk should be further investigated for NCU and ICU. A varying cut-off point for the identification of patients at risk or adding further items should be considered. Also, the comparison between the risk assessments is necessary. There may be other assessments that provide better performance. Our study has important implications on healthcare management as it identified patient characteristics that may be used for targeted preventive measures. With increasing age, the risk for PU incidence is rising. Older patients require more attention concerning their PU risk. The risk of PU incidence increases with the length of stay. The highest risk exists for patients with more than 30 days of inpatient treatment. Patients referred from a care facility appear to be at particularly high risk for incident PU, as well as patients admitted to ICU.

References 1 Sipponen A, Jokinen JJ, Sipponen P et al. Beneficial effect of resin salve in treatment of severe pressure ulcers: a prospective, randomized and controlled multicentre trial. Br J Dermatol 2008; 158:1055–62. 2 National Health Service. Quarterly Data Summary Issue 11: Learning from Reporting – Pressure Ulcers [internet document]. National Health Service, 2009. Available from: http://www.nrls.npsa.nhs.uk/ resources/?EntryId45=59862 (last accessed 5 April 2014). 3 Kuhn BA, Coulter SJ. Balancing the pressure ulcer cost and quality equation. Nurs Econ 1992; 10:353–9. 4 Lubatsch H. Dekubitusmanagement auf der Basis des Nationalen Expertenstandards: Ein Qualit€at entwickelndes Pflegemanagement. Hannover: Schl€ utersche Verlagsgesellschaft mbH & Co. KG, 2004.

British Journal of Dermatology (2014) 170, pp1285–1290

1

Fig 3. Performance of the Braden Scale at normal care unit (NCU) and intensive care unit (ICU). *The area under curve (AUC) is equal to the probability that a classifier will rank a randomly chosen positive instance higher than a randomly chosen negative one.

5 Kristensen S, Mainz J, Bartels P. Catalogue of Patient Safety Indicators. Safety Improvement for Patients in Europe. Aarhus: European Society for Quality in Healthcare, 2007. 6 Cho I, Noh M. Braden Scale: evaluation of clinical usefulness in an intensive care unit. J Adv Nurs 2010; 66:293–302. 7 Pancorbo-Hidalgo PL, Garcia-Fernandez FP, Lopez-Medina IM et al. Risk assessment scales for pressure ulcer prevention: a systematic review. J Adv Nurs 2006; 54:94–110. 8 Schoonhoven L, Haalboom JR, Bousema MT et al. Prospective cohort study of routine use of risk assessment scales for prediction of pressure ulcers. BMJ 2002; 12:325. 9 Tescher AN, Branda ME, Byrne TJ, Naessens JM. All at-risk patients are not created equal: analysis of Braden pressure ulcer risk scores to identify specific risks. J Wound Ostomy Continence Nurs 2012; 39:282–91. 10 Jun Seongsook RN, Jeong Ihnsook RN, Lee Younghee RN. Validity of pressure ulcer risk assessment scales; Cubbin and Jackson, Braden, and Douglas scale. Int J Nurs Stud 2004; 41:199–204. 11 Braden BJ, Bergstrom N. Clinical utility of the Braden Scale for predicting pressure sore risk. Decubitus 1989; 2:44–6, 50–1. 12 Bergstrom N, Braden BJ, Laguzza A, Holman V. The Braden Scale for predicting pressure sore risk. Nurs Res 1987; 36:205–10. 13 Bergstrom N, Demuth PJ, Braden BJ. A clinical trial of the Braden Scale for predicting pressure sore risk. Nurs Clin North Am 1987; 22:417–28. 14 Bergstrom N, Braden B, Kemp M et al. Predicting pressure ulcer risk: a multisite study of the predictive validity of the Braden Scale. Nurs Res 1998; 47:261–9. 15 Zhou XH, Obuchwoski NA, McClish DK. Statistical Methods in Diagnostic Medicine. New York: John Wiley & Sons, 2011; 27–9. 16 Expertenstandard Dekubitusprophylaxe in der Pflege. Deutsches Netzwerk f€ur Qualit€atsentwicklung in der Pflege. Osnabr€ uck: Expertenstandard Dekubitusprophylaxe in der Pflege, 2010. 17 Eberlein-Gonska M. [What is evidence-based concerning quality management? Thoughts about an apparently simple question.] Bundesgesundheitsblatt Gesundheitsforschung Gesundheitsschutz 2011; 54:148–53 [Article in German].

© 2014 British Association of Dermatologists