Predictive validity of the Braden Scale for pressure ulcer risk in hospitalized patients

Journal of Tissue Viability (2015)

--, -e-

www.elsevier.com/locate/jtv

Review

Predictive validity of the Braden Scale for pressure ulcer risk in hospitalized patients Seong-Hi Park a, Yun-Kyoung Choi b,*, Chang-Bum Kang c a

School of Nursing, Pai Chai University, South Korea Department of Nursing, Korea National Open University, South Korea c Health Promotion Fund Management Team, Korea Health Promotion Foundation, South Korea b

KEYWORDS Pressure sore; Sensitivity; Specificity; Meta-analysis

Abstract Purpose: Although the Braden Scale has been used as a basic tool to assess pressure ulcer risk, the validity of its effectiveness and accuracy was insufficient. Therefore, this study developed the groundwork for the predictive validity of the Braden Scale through a meta-analysis of prospective diagnosis assessment research. Methods: Articles published between 1966 and 2013 from periodicals indexed in the Ovid Medline, Embase, CINAHL, KoreaMed, NDSL and other databases were selected, using the keyword ‘pressure ulcer’. QUADAS-II was applied to assess the internal validity of the diagnostic studies. Selected studies were analyzed using meta-analysis with MetaDiSc 1.4. Results: Twenty-one diagnostic studies with high methodological quality, involving 6070 patients, were included. The meta-analysis revealed that the pooled sensitivity was 0.72 (95% CI 0.68, 0.75); pooled specificity was 0.81 (95% CI 0.80, 0.82), and the sROC AUC was 0.84 (SE ¼ 0.02). A detail analysis confirmed that age and reference standards were the factors that affected the diagnostic accuracy of the Braden Scale. Conclusion: The results suggest that the Braden Scale has a moderate predictive validity. This research also revealed the possibility that the predictive validity of the Braden Scale could be enhanced if it was applied differently according to the attributes of the study subjects. ª 2015 Tissue Viability Society. Published by Elsevier Ltd. All rights reserved.

* Corresponding author. Korea National Open University, Department of Nursing, 1-83, Dongsung-dong, Jongno-gu 110-809, Seoul, 110-809, South Korea. Tel.: þ82 2 3668 4745; fax: þ82 2 3673 4274. E-mail address: [email protected] (Y.-K. Choi). http://dx.doi.org/10.1016/j.jtv.2015.05.001 0965-206X/ª 2015 Tissue Viability Society. Published by Elsevier Ltd. All rights reserved.

Please cite this article in press as: Park S-H et al., Predictive validity of the Braden Scale for pressure ulcer risk in hospitalized patients, Journal of Tissue Viability (2015), http://dx.doi.org/10.1016/j.jtv.2015.05.001

2

1. Introduction Pressure ulcers can occur in any patient whose movement is limited, and once present, treatment is difficult. Therefore, it is of the utmost importance that all measures are taken to prevent it from occurring [1]. An ideal method for preventing pressure ulcers is possible by identifying the small group of patients at risk for pressure ulcers out of the entire patient group by utilizing a validated pressure ulcer risk assessment tool, and its effectiveness is maximized through systematic and focused training of patients and caregivers as well as proper nursing management [2]. The currently used pressure ulcer risk assessment tools include the Norton Scale, which was developed in 1962 in the U.K., the Gosnell Scale, which was developed in 1973, the Waterlow Scale, which was developed in 1985, the Cubbin and Jackson Scale, which was developed in 1991 for critical patients, and the Braden Scale which was developed in 1987 in the U.S.A. for seniors in longterm care facilities [3e5]. A pressure ulcer risk assessment tool needs to identify the occurrence of pressure ulcers when applied to patients; thus, the validity of the tool is very important, and it must be accompanied by a sufficient amount of research [6]. There have been some studies to assess the validity of pressure ulcer risk assessment tools; however, a solid conclusion has not been drawn that indicates which tool is the most appropriate. Additionally, the existing tools were not developed in the countries that intend to use the tools; thus, it can create issues regarding validity due to the differences in medical culture. According to the study of Zimmerman et al. [7], the predictive tools inevitably reflect the population characteristics and the medical culture of the country in which they were developed, it is essential to test the validity of the predictive assessment tools before applying to the certain patients. Therefore, the validity of the tool should be more widely reviewed in advance in order for these issues to be resolved. Predictive validity is a criterion that is estimated based on the relationship between a score from a test or an evaluation tool and a result in the future. The evaluation tool for pressure ulcers is not a diagnostic test that confirms the occurrence of pressure ulcers. Rather, it is a screening test that predicts the risk of pressure ulcers. A screening test is implemented for general people who possess potential risk factors but do not have symptoms, and an easy, simple, and non-invasive tool is normally recommended. Ideally, a screening

S.-H. Park et al. tool is good when the sensitivity, specificity, and positive and negative likelihoods are all high. However, this is not realistically feasible [8]. A screening test should be highly sensitive not to miss any potential risk from a disease, and it is necessary to prove its benefits against its harms before conducting the test in order not to burden patients with the test results [9]. The Braden Scale has been recommended as the most basic tool because it incorporates fewer questions and is easier to use than that of other tools [10,11]. However, during the process in which the Braden Scale was re-verified by other researchers, the sensitivity and specificity showed a wide range of differences from 50 to 100% depending on the research subjects or conditions [12], and the cut-off point differed as well [13]. Additionally, some studies revealed that using the Braden Scale as a standard, daily tool in clinical practices is inefficient, and there has been consistent assertion that pressure an ulcer risk assessment tool should be accurately verified based on prospectively collected materials [6,10,14e16]. Therefore, this study focused on confirming the predictive validity of the Braden Scale, the most widely used pressure ulcer risk assessment tool, by conducting a meta-analysis based on prospective diagnostic assessment research in order to confirm its ability to determine the occurrence of pressure ulcers. In addition, factors that affect it’s the predictive power were analyzed in order to propose a scientific evidence for its usefulness as a screening tool of pressure ulcer risk assessment.

2. Methods 2.1. Criteria for considering studies for this review (1) Type of participants: Studies involving adult patients over 18 years of age without pressure ulcers at the time of hospitalization were eligible for inclusion. (2) Types of Index test: Studies that were reported prospective diagnostic test accuracy using the Braden Scale for pressure ulcer risk assessment were eligible for inclusion in this review. (3) Reference standards: Evaluation standards of skin condition for defining pressure ulcers and stages of pressure ulcers were extracted from each study and listed as the National Pressure Ulcer Advisory Panel (NPUAP), Agency for Health Care Policy and Research (AHCPR), and others.

Please cite this article in press as: Park S-H et al., Predictive validity of the Braden Scale for pressure ulcer risk in hospitalized patients, Journal of Tissue Viability (2015), http://dx.doi.org/10.1016/j.jtv.2015.05.001

Predictive validity of Braden Scale

3

(4) Types of outcome measures: Studies involving in a 2  2 contingency table with true positive and negative, false positive and negative results defined by the reference standards were eligible for inclusion in this review. The cut-off point was extracted using the researcher’s judgment.

2.2. Search methods for identification of studies A search using electronic databases was conducted on October 1, 2013. Seven electronic databases including KoreaMed, National Digital Science Library (NDSL), Korea Education and Research Information Service (KERIS), Ovid-Medline, Embase and Cochrane Library as well as CINAHL Plus with Full Text were searched. In order to include all pressure ulcer-related academic journals, eight Korean medicine and nursing journals were added. A search strategy was devised by extracting pressure ulcer-related Medical Subject Headings (MeSH) and concept words that consisted of the core question from the study target and diagnostic

Fig. 1

method. Search terms that were used include ‘pressure ulcer’, ‘decubitus ulcer’, ‘skin ulcer’, ‘bed sore’ and ‘risk assessment’. Regarding the search filter for diagnostic method evaluation, a search strategy proposed by the Scottish Intercollegiate Guidelines Network (SIGN) was utilized. After eliminating duplicated studies from the first search, we selected studies based on the selection and elimination standards after searching the title and abstract of each study. We found full texts of the studies that were selected in the first search, and applied the selection and elimination standards again. The study selection process was conducted by two researchers independently, and in the case of disagreement, they discussed and reached a consensus.

2.3. Risk of bias in included studies and data extraction The quality of the study was evaluated independently by two authors using the Quality Assessment of Diagnostic Accuracy Studies-II (QUADAS-II) [17].

Flow diagram of article selection.

Please cite this article in press as: Park S-H et al., Predictive validity of the Braden Scale for pressure ulcer risk in hospitalized patients, Journal of Tissue Viability (2015), http://dx.doi.org/10.1016/j.jtv.2015.05.001

4

S.-H. Park et al. Table 1

Characteristics of selected studies.

Year of Authors publication

Location

2011 2009 2009 2007 2006 2005 2003 2002

Brazil Hong Kong Korea Korea Indonesia China Korea USA

2000 1998

Serpa et al. Chan et al. Kim et al. Oh et al. Suriadi et al. Kwong et al. Lee Bergstrom & Braden Halfens et al. Baldwin & Ziegler

1998 1998

Lyder et al. Olson et al.

1998 1997 1996

Pang & Wong Watkinson Capobiano & McDonald Harrison et al. VandenBosch et al. Barnes & Payton Salvadalena et al. Choi & Song Bergstrom et al.

1996 1996 1993 1992 1991 1987

Participants Setting

Sampling method

Age (yrs)

M:F (n) Total (n)

Reference standards

Consecutive 60.9  16.5 48:24 72 NPUAP, (2007) 79.4  10.9 30:167 197 NPUAP (2007) SICU 58.1  1.2 145:74 219 AHCPR (1994) Ward & ICU 51.1 997:885 1882 e ICU 47.5e50.9 72:33 105 NPUAP (1989) Hospitals 54.1  16.9 253:176 429 NPUAP (1989) NSICU 54.1 48:18 66 NPUAP (1989) Randomly 58.1e63.2 825 NPUAP (1998)

Cut-off point <13 <16 <14 <18 <11 <15 <16 <18

Netherlands USA Trauma center USA Canada Nursing units Hong Kong UK USA

Consecutive 60.9  18.3 167:153 Consecutive 31.7  10.9 16:10

320 e 36 NPUAP (1992)

<20 <10

Consecutive 71.0  6.5 Consecutive 54.8e62.4

15:21

36 NPUAP 128 NPUAP

16 <16

45e92 Consecutive 82.7 66.9  19.3

52:54 24:68 18:32

106 TDCPS (1983) 92 Lowthian (1987) 50 NPUAP (1989)

<18 <16 18

Canada USA USA USA Korea USA

60.0  19.0 376:362 62.4e67.0 49:54 50e90 183:178 72.0  13.0 34:63 54.1 89:57 Consecutive 58.5  14.5 28:32

738 103 361 99 146 60

<19 <17 <16 <16 <16 <16

Randomly Randomly

NS wards ICU

AHCPR (1992) Bergstrom (1994) Lyder (1991) Bergstrom (1987) Bergstrom (1987) Bergstrom (1987)

Notes: TP ¼ True positive; FP ¼ False positive; FN ¼ False negative; TN ¼ True negative; SN ¼ Sensitivity; SP ¼ Specificity; PLR ¼ Positive likelihood ratio; NLR ¼ Negative likelihood ratio; DOR ¼ Diagnosis odds ratio; NPUAP ¼ National Pressure Ulcer Advisory Panel; AHCPR ¼ Agency for Health Care Policy and Research; TDCPS ¼ Torrence Developmental Classification of Pressure Sore.

With respect to data extraction, a basic format for the evidence table was first determined and the research type, location, the number of institutions, gender distribution of the research subjects, age, length of hospital stay, and cut-off point for the Braden Scale as well as the diagnostic result of the pressure ulcer risk evaluation tool such as the values for a true positive reaction, false positive reaction, false negative reaction and true negative reaction were described. Based on this, a 2  2 table of specification was devised, and sensitivity, specificity, positive likelihood, negative likelihood, diagnostic odds ratio, and 95% confidence interval (95% CI) of each category were calculated using 2-way Contingency Table Analysis. This process was conducted independently by two researchers, and in the case of disagreement, a third party mediation was planned; however, it proceeded without any disagreements.

2.4. Statistical analysis A meta-analysis was conducted with the MetaDiSc 1.4 program. The basic principle of the statistical

model was to analyze with the random effect model in order to include the heterogeneity among studies [18], and it was analyzed with pooled sensitivity, specificity, positive and negative likelihoods, diagnostic odds ratio and sROC curve (symmetric summary receiver-operating characteristic [sROC] curve). In terms of sROC statistics, the test accuracy was calculated with the Area under the curve (AUC) and the index Q* value. With respect to the AUC value, a value of 0.5 was deemed non-informative; 0.5 < AUC  0.7 was considered less accurate; 0.7 < AUC  0.9 was thought to be moderate; 0.9 < AUC < 1 was deemed very accurate, and AUC ¼ 1 was considered a perfect test [19]. The existence of heterogeneity among studies was confirmed by checking the forest plot to visually identify any common areas in the confidence interval and effect estimate. Additionally, with a significance level of below 5%, the I2 homogeneity test (I2 test) by Higgins was used. If the value was I2 25%, the heterogeneity was considered low; 25% < I2  75% was deemed to indicate a moderate amount of heterogeneity, and higher than I2 >75% was

Please cite this article in press as: Park S-H et al., Predictive validity of the Braden Scale for pressure ulcer risk in hospitalized patients, Journal of Tissue Viability (2015), http://dx.doi.org/10.1016/j.jtv.2015.05.001

Predictive validity of Braden Scale

5

2  2 table

Value (95% Confidence interval)

TP

SN

FP

FN TN

6 12 37 4 23 8 26 76

11 2 64 6 54 3 114 0 15 12 118 1 14 4 166 32

34 10

82 13 1 1

53 115 125 1764 55 302 22 551

0.75 0.67 0.93 1.00 0.66 0.89 0.87 0.70

SP (0.38e0.95) (0.42e0.85) (0.80e0.98) (0.40e1.00) (0.51e0.78) (0.52e0.99) (0.73e0.95) (0.62e0.78)

0.83 0.64 0.70 0.94 0.79 0.72 0.61 0.77

PLR

NLR

(0.78e0.85) 4.36 (1.73e6.52) (0.62e0.66) 1.87 (1.10e2.52) (0.67e0.71) 3.07 (2.41e3.39) (0.94e0.94) 16.47 (6.41e16.47) (0.71e0.85) 3.07 (1.79e5.06) (0.71e0.72) 3.16 (1.77e3.57) (0.49e0.68) 2.23 (1.44e3.01) (0.76e0.78) 3.04 (2.51e3.55)

0.30 0.52 0.11 0.00 0.44 0.16 0.22 0.39

DOR (0.05e0.80) 14.46 (2.17e121.64) (0.22e0.94) 3.59 (1.18e11.37) (0.03e0.30) 28.55 (7.96e121.67) (0.00e6.42) e (0.26e0.68) 7.03 (2.62e19.30) (0.01e0.69) 20.48 (2.57e441.31) (0.07e0.55) 10.21 (2.60e43.90) (0.28e0.51) 7.88 (4.93e12.66)

191 0.72 (0.58e0.84) 0.70 (0.68e0.72) 2.41 (1.80e2.97) 0.40 (0.23e0.62) 6.09 (2.92e12.89) 24 0.91 (0.65e0.99) 0.96 (0.85e1.00) 22.73 (4.26e279.82) 0.10 (0.01e0.41) 240.00 (10.37e24,609.34)

5 9

0 19

9 2

22 98

0.36 (0.18e0.36) 1.00 (0.86e1.00) e 0.82 (0.50e0.97) 0.84 (0.81e0.85) 5.04 (2.58e6.52)

0.64 (0.64e0.93) e 0.22 (0.04e0.62) 23.21 (4.15e170.28)

19 14 10

32 18 6

2 1 4

53 59 30

0.91 (0.70e0.98) 0.62 (0.57e0.64) 0.93 (0.69e1.00) 0.77 (0.72e0.78) 0.71 (0.47e0.89) 0.83 (0.74e0.90)

0.15 (0.03e0.52) 0.09 (0.00e0.44) 0.34 (0.12e0.73)

15.73 (3.19e105.01) 45.89 (5.57e1000.60) 12.50 (2.43e72.06)

176 72 30 12 32 6 24 12 8 3 13 4

343 44 307 55 122 23

0.50 0.70 0.30 0.86 0.20 0.26

3.98 2.08 25.58 1.53 66.08 8.85

147 17 16 8 13 20

0.67 0.59 0.73 0.40 0.81 0.83

(0.62e0.72) (0.42e0.74) (0.51e0.88) (0.21e0.61) (0.57e0.95) (0.66e0.94)

0.66 0.60 0.91 0.70 0.94 0.64

2.40 (1.65e2.75) 3.99 (2.44e4.50) 4.29 (1.77e8.96)

(0.64e0.68) 1.98 (1.70e2.28) (0.53e0.66) 1.44 (0.88e2.15) (0.89e0.92) 7.71 (4.72e10.42) (0.65e0.75) 1.32 (0.60e2.44) (0.91e0.96) 13.20 (6.31e21.03) (0.53e0.71) 2.31 (1.40e3.25)

determined to indicate the existence of heterogeneity [20].

3. Research result 3.1. Results of the search A total of 492 studies were searched using electronic databases, and a total of 19 studies were found in relevant academic journals. There were 165 duplicated studies. Thus, based on a total of 346 studies, diagnostic method evaluation studies were selected that reported true positive results, false positive results, true negative results and false negative results using the Braden Scale on hospitalized patients. Non-original work, diagnostic method evaluation studies that were not compared against reference standards test, and studies for which the diagnostic accuracy cannot be calculated were excluded. At the end, a total of 325 studies (93.9%) were eliminated and 21 studies were selected. A flow chart is provided to show the detailed study selection process (Fig. 1).

(0.40e0.60) (0.39e1.11) (0.13e0.55) (0.52e1.22) (0.06e0.47) (0.08e0.64)

(2.81e5.65) (0.80e5.46) (8.60e79.55) (0.49e4.70) (13.46e376.46) (2.18e39.15)

3.2. Risk of bias in the included studies Based on the quality evaluation of the 21 studies that were finally selected, none of the studies were evaluated to have a high risk of bias in each area. The reasons are as: 1) they only included prospective studies, 2) test diagnosis measure and reference standards test were noninvasive and evaluated by regular observation by nurses, which means that almost none of the patients were excluded. Thus, factors for risk bias and application issues were very few. Therefore, all selected studies were confirmed to be of high quality and meet all areas of the quality evaluation.

3.3. General properties of the selected studies The twenty-one studies focused on the prospective study of the predictive accuracy of the Braden Scale for patients who had not developed pressure ulcers at the time of hospitalization, and the total number of study subjects were 6070. The United States of America published 8 studies, which is the highest

Please cite this article in press as: Park S-H et al., Predictive validity of the Braden Scale for pressure ulcer risk in hospitalized patients, Journal of Tissue Viability (2015), http://dx.doi.org/10.1016/j.jtv.2015.05.001

0.0228 0.0368 0.1538 0.0499 0.0313 0.0190 0.1360 0.9185 0.7292 0.6197 0.8604 0.8200 0.8089 0.7672 (10.31e46.72) (2.87e8.01) (1.03e9.04) (5.42e40.75) (6.45e18.77) (6.10e14.77) (1.35e27.30) 0.88 0.68 0.48 0.72 0.82 0.72 0.65 50 6 60 5 70 3 Setting Ward 6 ICU 5 Reference standard NPUAP 10 Bergstrom 4

Note: ICU ¼ Intensive care unit; NPUAP ¼ National Pressure Ulcer Advisory Panel; AUC ¼ Area under the curve; SE ¼ Standard error.

21.95 4.79 3.06 14.87 11.00 9.49 6.07 (0.12e0.30) (0.40e0.59) (0.54e0.92) (0.22e0.62) (0.16e0.46) (0.30e0.52) (0.24e0.94) 0.19 0.49 0.71 0.37 0.27 0.39 0.48 (2.37e8.68) (1.74e3.07) (1.01e3.10) (0.41e6.79) (2.40e3.42) (2.53e3.96) (1.13e6.36) 4.53 2.31 1.77 4.04 2.87 3.17 2.68 (0.87e0.89) (0.65e0.71) (0.63e0.74) (0.73e0.79) (0.67e0.77) (0.73e0.78) (0.71e0.81) 0.88 0.68 0.69 0.76 0.72 0.75 0.76 (0.81e0.93) (0.62e0.73) (0.34e0.62) (0.63e0.80) (0.74e0.88) (0.66e0.77) (0.54e0.75)

0.8353 0.0246 10.30 (6.65e15.96) 0.38 (0.30e0.48) 21

Total Sub-group analysis Mean age

0.72 (0.68e0.75) 0.81 (0.80e0.82) 3.43 (2.66e4.44)

AUC Positive likelihood Negative likelihood Diagnostic odds ratio (95% CI) ratio (95% CI) ratio (95% CI) Specificity (95% CI) Study No. Sensitivity (95% CI)

3.4.1. Overall result The meta-analysis of the Braden Scale in a total of 21 selected studies [5,11,16,21e38] revealed that the pooled sensitivity was 0.72 (95% CI 0.68, 0.75), and the heterogeneity among the studies was found to be moderate at 64.0% (X2 ¼ 55.48, p < .001). The pooled specificity was 0.81 (95% CI 0.80, 0.82) while the heterogeneity among studies was found to be high at 96.2% (X2 ¼ 525.38,

Scale

The results of the predictive validity of the Braden Scale and sub-group analysis were presented on Table 2.

Summary results of meta-analysis. A. The results of diagnostic test accuracy.

3.4. Evaluation of the predictive validity of the Braden Scale

Table 2

(38.1%) [21e28], followed by the Republic of Korea (19.0%) [5,29e31], which published 4 studies. Hong Kong [32,33] and Canada [34,35] each published 2 studies, and the Netherlands [36], Brazil [11], the United Kingdom [37], Indonesia [16], China [38] each published 1 study, which shows that many countries around the world are conducting research in this area. In terms of the average age of the research subjects, 2 studies targeted subjects in their 40s or below [16,22]; 6 studies targeted subjects in their 50s [5,28e31,38]; 5 studies targeted subjects in their 60s [11,24,25,35,36]; 4 studies targeted subjects in their 70s or older [23,27,32,37], and 4 studies targeted subjects with an unclear average age [21,26,33,34]. Large studies that targeted 100 or more subjects were 13 (61.9%) [5,16,21,25,26,29,30,32e36,38], and 6 studies targeting more than 300 subjects [21,26,30,35,36,38]. Reference standards used for risk assessment were mostly NPUAP in 10 studies (47.6%) [11,16,21e24,31,32,34,38], and they tended to be used in published studies after 1995. Two skin evaluation tools were developed by AHCPR [29,35], and the studies from the early 1900s used tools that were developed by individual researcher [5,25e28,37]. The pressure ulcer evaluation of research subjects was conducted by nurses at the time of or within 24e72 h of hospitalization. The cut-off point for the Braden Scale was defined at the most effective level, which was determined by the researchers. Five studies showed a cut-off point of 15 and lower [11,16,22,29,38], and a cut-off point of 16 was used the most in nine studies (42.9%) [5,23,26e28,31,32,34,37]. One study had a cut-off point of 17 [25]; four studies [21,24,30,33] had a cut-off point of 18, and two studies [35,36] had a cut-off point of 19 and higher (Table 1).

S.-H. Park et al. SE (AUC)

6

Please cite this article in press as: Park S-H et al., Predictive validity of the Braden Scale for pressure ulcer risk in hospitalized patients, Journal of Tissue Viability (2015), http://dx.doi.org/10.1016/j.jtv.2015.05.001

Fig. 2 Diagnosis test accuracy of Braden Scale in total selected studies. Note: sROC ¼ Summary receiver operating characteristic; AUC ¼ area under curve; SE ¼ standard error.

Please cite this article in press as: Park S-H et al., Predictive validity of the Braden Scale for pressure ulcer risk in hospitalized patients, Journal of Tissue Viability (2015), http://dx.doi.org/10.1016/j.jtv.2015.05.001

8 p < .001). The pooled positive likelihood was 3.44 (95% CI 2.66, 4.44) and the pooled negative likelihood was 0.37 (95% CI 0.30, 0.48). The pooled diagnostic odds ratio was 10.30 (95% CI 6.65, 15.95); the sROC AUC was 0.84 (SE ¼ 0.02) (Fig. 2). 3.4.2. Detailed analysis A detailed analysis was implemented when three or more studies reported on each attribute.

S.-H. Park et al. (1) By average age group The pooled sensitivity of six studies, which reported the average age group as in the 50s [5,28e31,38], was 0.88 (95% CI 0.81, 0.93) (Fig. 3A), and no heterogeneity among the studies was found at 0.0% (X2 ¼ 3.00, p ¼ .70). The pooled specificity was 0.88 (95% CI 0.87, 0.89) (Fig. 3-B), and the heterogeneity among the studies was high

Fig. 3 Subgroup analysis for diagnosis test accuracy of Braden Scale. Notes: NPUAP ¼ National Pressure Ulcer Advisory Panel. Please cite this article in press as: Park S-H et al., Predictive validity of the Braden Scale for pressure ulcer risk in hospitalized patients, Journal of Tissue Viability (2015), http://dx.doi.org/10.1016/j.jtv.2015.05.001

Predictive validity of Braden Scale

9

Fig. 3

at 97.8% (X2 ¼ 230.99, p < .001). The sROC AUC was 0.92 (SE ¼ 0.02). In five studies, in which the average age group was in the 60s [11,24,25,35,36], the pooled sensitivity was 0.68 (95% CI 0.62, 0.73) (Fig. 3-C), and the pooled specificity was 0.68 (95% CI 0.65, 0.71) (Fig. 3-D). The heterogeneities among the studies were 0.0% (X2 ¼ 1.85, p ¼ .76) and 73.8% (X2 ¼ 15.24, p ¼ .004), respectively, and the sROC AUC was 0.73 (SE ¼ 0.04). The pooled sensitivity of three studies in which the average age group was in the 70s [23,27,32] was 0.48 (95% CI 0.34, 0.62) (Fig. 3-E), and the pooled specificity was 0.69 (95% CI 0.63, 0.74) (Fig. 3-F). The heterogeneities among the studies were high at 49.1% (X2 ¼ 3.93, p ¼ .14), and 89.0% (X2 ¼ 18.16, p < .001), respectively. The sROC AUC was 0.62 (SE ¼ 0.15). (2) By the location of hospitalization In six studies, which targeted inpatients in the inpatient ward [5,23,32,34,36,37], the pooled sensitivity was 0.72 (95% CI 0.63, 0.80) (Fig. 3-G) while the pooled specificity was 0.76 (95% CI 0.73, 0.79) (Fig. 3-H). The heterogeneities among the studies were 64.0% (X2 ¼ 13.90, p ¼ .02) and 92.1% (X2 ¼ 63.55, p < .001), respectively. The sROC AUC was 0.86 (SE ¼ 0.05). In five studies that targeted inpatients in the intensive care unit [11,16,28,29,31], the pooled sensitivity was 0.82 (95% CI 0.74, 0.88) (Fig. 3-I); the pooled specificity was 0.72 (95% CI 0.67, 0.77) (Fig. 3-J), and the heterogeneities among the studies were moderate

(Continued).

at 58.7% (X2 ¼ 9.69, p ¼ .05) and 56.4% (X2 ¼ 9.18, p ¼ .06), respectively. The sROC AUC was 0.82 (SE ¼ 0.03). (3) Reference standards In 10 studies, which utilized NPUAP as the reference standards [11,16,21e24,31,32,34,38], the pooled sensitivity was 0.72 (95% CI 0.66, 0.77) (Fig. 3-K), and the pooled specificity was 0.75 (95% CI 0.73, 0.77) (Fig. 3-L). The heterogeneities among the studies were 47.8% (X2 ¼ 17.24, p ¼ .05) and 80.9% (X2 ¼ 47.11, p < .001), respectively. The sROC AUC was 0.81 (SE ¼ 0.02). Four studies [5,25,27,28] utilized the skin evaluation standards developed by Bergstrom as the reference standards, and they had a pooled sensitivity of 0.65 (95% CI 0.54, 0.75) (Fig. 3-M) and a pooled specificity of 0.76 (95% CI 0.71, 0.81) (Fig. 3-N). The heterogeneities among the studies were 74.4% (X2 ¼ 11.73, p ¼ .008) and 93.2% (X2 ¼ 43.81, p < .001), respectively. The sROC AUC was 0.77 (SE ¼ 0.14).

4. Discussion This study conducted a meta-analysis after thoroughly studying a total of 21 studies and 6070 diagnostic method evaluation results in order to identify the predictive validity of the Braden Scale. The prospective study result of the pooled sensitivity of the Braden Scale was 0.72, and the

Please cite this article in press as: Park S-H et al., Predictive validity of the Braden Scale for pressure ulcer risk in hospitalized patients, Journal of Tissue Viability (2015), http://dx.doi.org/10.1016/j.jtv.2015.05.001

10 pooled specificity was 0.81. The sROC AUC was 0.84. Sensitivity and specificity are generally measured values of test performance, which can be readily interpreted; however, it is not ideal to determine the appropriateness of the test performance with only these numbers. Thus, it is advisable to utilize the likelihood and AUC values [9]. This study determined the predictive validity of a pressure ulcer risk evaluation tool based on the sROC AUC value. As it was mentioned in the research methods, the sROC AUC of the Braden Scale was 0.7 < AUC  0.9; thus, the Braden Scale can be interpreted to have a moderate level of predictive validity. Predictive validity indicates the extent to which a score on a scale predicts an individuals’ future level such as likelihood to develop pressure sores on some criterion measure. Screening tools for pressure ulcer risks are not a diagnostic test for the incidence of pressure ulcers but instead are tests assessing the level of risk for a development of pressure ulcer. Pressure ulcers cause physical pain in patients, the extension of hospital stays, and additional burden from hospital costs, and they also cause feelings of guilt, failure, and negligence in healthcare providers [39]. Therefore, the issue of pressure ulcers is considered important in public health as a social health issue, and it is widely studied around the world. In several studies reported that the risk factors such as sex, age group, skin condition can enhance the sensitivity and specificity of the Braden Scale [36]. To understand the nature of variability in studies using screening tools, subsets by age, location of hospitalization and reference standards were grouped and meta-analyzed. The detailed analysis by average age group revealed that the pooled sensitivity of the patients in their 50s was 0.88 while that of patients in their 60s was 0.68, and that of the patients in their 70s was 0.48, which indicate that the pooled sensitivity decreases as the age increases. The same pattern was seen in the specificity. In particular, the sROC AUC of the patients in their 50s was 0.92, which revealed high test accuracy. Additionally, the results showed that the age of the study target is a factor that affects the diagnostic accuracy of the Braden Scale as well as the cause of the heterogeneity among the studies based on the fact that there were no heterogeneities at 0.0% in the pooled sensitivities for the patients in their 50s and 60s. In the analysis by location of hospitalization, the pooled sensitivity of the inpatients in the inpatient ward was 0.72 while that of the inpatients in the intensive care unit was 0.82. Based

S.-H. Park et al. on this result, it appears that the Braden Scale has a higher diagnostic accuracy in intensive care units than in inpatient wards; however, the sROC AUC did not have a significant difference which was 0.86 and 0.82, respectively. Such a result appears to be caused by the specificity. Therefore, it is difficult to interpret that the location of hospitalization is a factor in the diagnostic accuracy of the Braden Scale. However, the Braden Scale showed more stable predictive values in intensive care units than in inpatient wards because the heterogeneities among the studies in the intensive care units were moderate in both sensitivity (I2 ¼ 58.7%) and specificity (I2 ¼ 56.4%). Although Braden Scale is not a tool developed for ICU patients, Bergstrom and Braden who developed the Braden Scale applied the Braden Scale in ICU [15,28,32,36]. Thus the Braden Scale is interpreted to be the pressure ulcer assessment tool that can be used regardless of both ICU and general wards. A detailed analysis for reference standards was implemented by comparing between NPUAP, which was developed by an expert group, and the Bergstrom skin evaluation standards, which was developed by an individual researcher. The results revealed that the NPUAP had a pooled sensitivity of 0.72 and a sROC AUC of 0.81 while the Bergstrom standards had a pooled sensitivity of 0.65 and a sROC AUC of 0.77, which showed a difference in diagnostic accuracy. Thus, it can be deduced that the objective skin evaluation standards developed by experts yield a higher predictive validity for the pressure ulcer evaluation tool. In addition, it showed that the reference standards influence the diagnostic accuracy of the Braden Scale, and they are also a factor that causes heterogeneity among the studies. This study used a cut-off point that each researcher selected as the best, and a corresponding detailed analysis did not take place. In the 21 studies, the cut-off point of the Braden Scale did not show any consistency which was based on attributes such as age and location of hospitalization. The cut-off points using a wide range from 11 to 20 points can be a primary cause of heterogeneity, because different cut-off points used in different studies influence the ability to define a positive or negative test result, which is called the threshold effect. When the threshold effect exists, there is a negative correlation between sensitivities and specificities [40]. Although, in this study, a sub-group analysis was not performed, this result confirms that the Braden Scale is not appropriate when a cut-off point is applied uniformly. Application of a cut-off point in a category, rather than as a specific threshold, is

Please cite this article in press as: Park S-H et al., Predictive validity of the Braden Scale for pressure ulcer risk in hospitalized patients, Journal of Tissue Viability (2015), http://dx.doi.org/10.1016/j.jtv.2015.05.001

Predictive validity of Braden Scale more appropriate, and standards for a cut-off point need to be established according to the attributes of the research subjects such as age and reference standards.

11

[5]

[6]

5. Conclusion The predictive validity of the Braden Scale was confirmed by conducting a meta-analysis based on twenty-one well-designed diagnostic evaluation studies. The sROC AUC for the Braden Scale was 0.7, which was analyzed to have a moderate predictive validity. The findings conclude that those screening tools have a limitation to predict the accuracy in assessing pressure ulcer risks in adults, because high heterogeneity is existed between studies. Although the Braden Scale was found to be inappropriate for routinely use across all patients under a standardized criterion, it confirmed that the predictive validity can be enhanced if the Braden Scale is applied differently based on the attributes of the study subjects. Pressure ulcer care is one of the basic cares that should be provided for patients whose mobility is limited. Therefore, the diagnostic accuracy of other pressure ulcer risk evaluation tools other than the Braden Scale should be studied in order to provide more effective nursing service, and more efforts should be made to develop a new pressure ulcer risk evaluation tool with higher predictive validity by improving the strengths and weaknesses of existing tools.

[7]

[8]

[9]

[10]

[11]

[12]

[13]

[14]

[15]

Conflict of interest [16]

We wish to confirm that there are no known conflicts of interest associated with this publication and there has been no significant financial support for this work that could have influenced its outcome.

References [1] Perneger TV, Ra ¨e AC, Gaspoz JM, Bort F, Vitek O, He ´iot C. Screen for pressure ulcer risk in the acute care hospital: development of a brief bedside scale. J Clin Epidemiol 2002;55(5):498e504. [2] Rosenburg CJ. New checklist for pressure ulcer prevention. J Gerontol Nurs 2002;28(8):7e12. [3] Bergstrom N, Braden BJ, Laguzza A, Holman V. The Braden Scale for predicting pressure sore risk. Nurs Res 1987; 36(4):205e10. [4] Cho MS, Park IS, Kim KH, Woo KS, Joo YH, Jeong EH, et al. Evaluation of predictive validity for the pressure ulcer risk

[17]

[18]

[19]

[20] [21]

assessment tool in a medical ward inpatient: using Braden Scale. Korean Nurses 2004;43(2):68e82. Choi KS, Song M. Test of predictive validity for the pressure risk assessment scale. J Korean Acad Soc Adult Nurs 1991; 3(1):19e28. Goodridge DM, Sloan JA, LeDoyen YM, McKenzie JA, Knight WE, Gayari M, et al. Risk-assessment scores, prevention strategies, and the incidence of pressure ulcers among the elderly in four Canadian health-care facilities. Can J Nurs Res 1998;30(2):23e44. Zimmerman JE, Wagner DP, Draper EA, Wright L, Alzola C. Evaluation of acute physiology and chronic health evaluation III predictions of hospital mortality in an independent database. Crit Care Med 1998;26(8):1317e26. Park DA, Hwang JS, Lee SH, Sul AR, Choi WJ, Oh SGH, et al. Systematic review of diagnostic tests. 1st ed. Seoul: National Evidence-Based Healthcare Collaborating Agency; 2014 January. Knottnerus JA. The evidence base of clinical diagnosis. In: Park SH, Kang CB, translator, editors. Seoul: E-Public; 2008. Bergquist S. Subscales, subscores, or summative score: evaluating the contribution of Braden Scale items for predicting pressure ulcer risk in older adults receiving home health care. J Wound Ostomy Continence Nurs 2001; 28(6):279e89. Serpa LF, Santos VL, Campanili TC, Queiroz M. Predictive validity of the Braden scale for pressure ulcer risk in critical care patients. Rev Lat Am Enfermagem 2011;19(1): 50e7. Chou R, Dana T, Bougatsos C, Blazina I, Starmer A, Reitel K, et al. Pressure ulcer risk assessment and prevention: a systematic comparative effective review. Ann Intern Med 2013;159(1):28e39. http://dx.doi.org/10.7326/00034819-159-1-201307020-00006. Cowan LJ, Stechmiller JK, Rowe M, Kairalla JA. Enhancing Braden pressure ulcer risk assessment in acutely ill adult veterans. Wound Repair Regen 2012;20(2):137e48. Defloor T, Grypdonck MFH. Pressure ulcers: validation of two risk assessment scales. J Clin Nurs 2005;14(3): 373e82. Jalali R, Rezaie M. Predicting pressure ulcer risk: comparing the predictive validity of 4 scales. Adv Skin Wound Care 2005;18(2):92e7. Suriadi, Sanada H, Sugama J, Thigpen B, Kitagawa A, Kinosita S, et al. A new instrument for predicting pressure ulcer risk in an intensive care unit. J Tissue Viability 2006; 16(3):21e6. Whiting PF, Rutjes AW, Westwood ME, Mallett S, Deeks JJ, Reitsma JB, et al. QUADAS-2: a revised tool for the quality assessment of diagnostic accuracy studies. Ann Intern Med 2011;155(8):529e36. http://dx.doi.org/10.7326/00034819-155-8-201110180-00009. Macaskill P, Gatsonis C, Deeks J, Harbord R, Takwoingi T. Cochrane handbook for systematic reviews of diagnostic test accuracy version 1.0 [Internet]. The Cochrane Collaboration; 2010 [cited 2013 May 30]. Available from:, http://www.cochrane-handbook.org. Greiner M, Pfeiffer D, Smith RD. Principles and practical application of the receiver-operating characteristic analysis for diagnostic tests. Prev Vet Med 2000;45(1e2): 23e41. Higgins JP, Thompson SG. Quantifying heterogeneity in a meta-analysis. Stat Med 2002;21(11):1539e58. Bergstrom N, Braden BJ. Predictive validity of the Braden Scale among Black and White subjects. Nurs Res 2002; 51(6):398e403.

Please cite this article in press as: Park S-H et al., Predictive validity of the Braden Scale for pressure ulcer risk in hospitalized patients, Journal of Tissue Viability (2015), http://dx.doi.org/10.1016/j.jtv.2015.05.001

12 [22] Baldwin KM, Ziegler SM. Pressure ulcer risk following critical traumatic injury. Adv Wound Care 1998;11(4): 168e73. [23] Lyder CH, Yu C, Stevenson D, Mangat R, Empleo-Frazier O, Emerling J, et al. Validating the Braden Scale for the prediction of pressure ulcer risk in blacks and Latino/Hispanic elders: a pilot study. Ostomy Wound Manage 1998; 44(Suppl. 3A):42Se9S. [24] Capobianco ML, McDonald DD, Capobianco ML, McDonald DD. Factors affecting the predictive validity of the Braden Scale. Adv Wound Care 1996;9(6):32e6. [25] VandenBosch T, Montoye C, Satwicz M, Durkee-Leonard K, Boylan-Lewis B. Predictive validity of the Braden Scale and nurse perception in identifying pressure ulcer risk. Appl Nurs Res 1996;9(2):80e6. [26] Barnes D, Payton RG. Clinical application of the Braden Scale in the acute-care setting. Dermatol Nurs 1993;5(5): 386e8. [27] Salvadalena GD, Snyder ML, Brogdon KE. Clinical trial of the Braden Scale on an acute care medical unit. J ET Nurs 1992;19(5):160e5. [28] Bergstrom N, Demuth PJ, Braden BJ. A clinical trial of the Braden Scale for predicting pressure sore risk. Nurs Clin North Am 1987;22(2):417e28. [29] Kim EK, Lee S, Lee E, Eom M. Comparison of the predictive validity among pressure ulcer risk assessment scales for surgical ICU patients. Aust J Adv Nurs 2009;26(4):87e94. [30] Oh DY, Kim JH, Lee PK, Ahn ST, Lee JW. Prevention of pressure ulcer using the pressure ulcer risk assessment based on Braden Scale. J Korean Soc Plast Reconstr Surg 2007;34(4):456e69. [31] Lee JK. The relationship of risk assessment using Braden scale and development of pressure sore in neurologic

S.-H. Park et al.

[32]

[33]

[34] [35]

[36]

[37] [38]

[39]

[40]

intensive care unit. J Korean Acad Adult Nurs 2003;15(2): 267e77. Chan WS, Pang SM, Kwong EWY. Assessing predictive validity of the modified Braden scale for prediction of pressure ulcer risk of orthopaedic patients in an acute care setting. J Clin Nurs 2009;18(11):1565e73. http://dx.doi. org/10.1111/j.1365-2702.2008.02757.x. Pang SW, Wong TK. Predicting pressure sore risk with the Norton, Braden and Waterlow Scales in a Hong Kong rehabilitation hospital. Nurs Res 1998;47(3):147e53. Olson K, Tkachuk L, Hanson J. Preventing pressure sores in oncology patients. Clin Nurs Res 1998;7(2):207e24. Harrison MB, Wells G, Fisher A, Prince M, Harrison MB, Wells G, et al. Practice guidelines for the prediction and prevention of pressure ulcers: evaluating the evidence. Appl Nurs Res 1996;9(1):9e17. Halfens RJ, Van Achterberg T, Bal RM. Validity and reliability of the Braden scale and the influence of other risk factors: a multi-centre prospective study. Int J Nurs Stud 2000;37(4):313e9. Watkinson C. Developing a pressure sore risk assessment scale. Prof Nurse 1997;12(5). 341e342, 344e6, 348. Kwong E, Pang S, Wong T, Ho J, Shao-ling X, Li-jun T, et al. Predicting pressure ulcer risk with the modified Braden, Braden, and Norton scales in acute care hospitals in Mainland China. Appl Nurs Res 2005;18(2):122e8. Lee YH, Jeong IS, Jeon SS. A comparative study predictive validity among pressure ulcer risk assessment scales. J Korean Acad Nurs 2003;33(2):162e9. Zamora J, Abraira V, Muriel A, Khan KS, Coomarasamy A. Meta-DiSc: a software for meta-analysis of test accuracy data. Med Res Methodol 2006;6:31. http://dx.doi.org/10. 1186/1471-2288-6-31.

Please cite this article in press as: Park S-H et al., Predictive validity of the Braden Scale for pressure ulcer risk in hospitalized patients, Journal of Tissue Viability (2015), http://dx.doi.org/10.1016/j.jtv.2015.05.001