R

C

RP

S

a

Pb

E

R

1

(

0d

esuscitation (2007) 74, 253—258

LINICAL PAPER

eliability testing and update of the Resuscitationredictor Scoring (RPS) Scale�

imon Coopera,∗, Fiona Duncanb,1

Faculty of Health and Social Work, C501 Portland Square, University of Plymouth,lymouth, Devon PL4 8AA, United KingdomFaculty of Health and Social Work, Veysey Building, University of Plymouth,arl Richards Road North, Exeter, Devon EX2 6AS, United Kingdom

eceived 8 November 2006; received in revised form 28 November 2006; accepted 12 December 2006

KEYWORDSAdvanced life support(ALS);Cardiopulmonaryresuscitation;Do not attemptresuscitate orders;Emergency team;Emergency medicalservices;Predicting survival;

SummaryObjectives: The purpose of this study was to test the reliability of the Resuscita-tion Predictor Scoring (RPS) Scale1 (Appendix A), a survival prediction nomogramdesigned to aid resuscitation termination decisions during a resuscitation attempt.Method: Bivariate comparisons of predictors of survival and survival rates betweenthe primary RPS Scale data set (1993—2000) and a secondary data set (2000—2003).A total of 2121 patients were included in the study.Results: Comparisons of the two sets of data showed an increase in resuscitationattempts for patients ≥80 years (p < 0.001); an increase in pulseless electrical activ-ity (PEA) (p = 0.01) and an increase in the duration of arrests (p = 0.012). However,in relation to the RPS Scale there were no statistical differences in survival betweenany of the sub groups demonstrating the reliability of the nomogram.

Resuscitation;Terminatingresuscitation

Conclusion: The final updated RPS Scale demonstrates predicted survival rates 15 mininto a resuscitation attempt. These can be poor and suggest that it is an acceptablepoint at which to first consider termination (where there has been no ROSC). TheRPS Scale has demonstrated reliability and validity, but can only be a guide for thecessation of resuscitation.© 2007 Elsevier Ireland Ltd

I

Ict

� A Spanish translated version of the summary of thisarticle appears as Appendix in the final online version at

0.1016/j.resuscitation.2006.12.010.∗ Corresponding author. Tel.: +44 1752 233864.

E-mail addresses: simon.cooper@plymouth.ac.ukS. Cooper), Fiona.duncan@plymouth.ac.uk (F. Duncan).1 Tel.: +44 1392 475169.

aasl

300-9572/$ — see front matter © 2007 Elsevier Ireland Ltd. All rightoi:10.1016/j.resuscitation.2006.12.010

. All rights reserved.

ntroduction

n 2003, we published a nomogram called the Resus-itation Predictor Scoring Scale1 designed as a guideo decisions on the termination of resuscitation

ttempts. The work arose from an interest in thepplicability of resuscitation attempts, from thetandpoint of ‘when to resuscitate’ and for ‘howong to resuscitate’. To date, there is increasing

s reserved.

ypiar

M

Trbtfp

254

comment on the applicability of resuscitation inspecific situations, for example continuing caresettings2 and for out of hospital arrests,3 viewson the termination of resuscitation4 and a num-ber of pre arrest morbidity scales,5—7 developedto aid ‘do not attempt resuscitation’ (DNAR) deci-sions. These are reasonable guides to predictingsurvival in that they designate a specific ‘cut-off’value above which patients will not survive, but lacksensitivity in predicting survival below the cut off,due to the many predictors of survival.8

The RPS Scale1 was designed to acknowledgethese sensitivity issues and the subjectivity oftermination decisions. By studying the evidencebehind the key predictors of survival (length of

resuscitation, primary arrhythmia, age and primarymode of arrest) against related survival rates, ter-mination decisions can be more objective. In thisreport, we compare the results from this first 7-

hiaa

Table 1 Patient characteristics

Characteristics Number (%), first 7 years (

SexMale 883 (59.5)Female 600 (40.5)

Agea

20—49 81 (5.5)50—59 125 (8.4)60—69 277 (18.7)70—79 572 (38.6)>80 425 (28.7)

Time of arrest0001—0700 347 (23.9)0701—1500 582 (40.1)1501—2400 522 (36.0)

Mode of arrestRespiratory 286 (19.3)Cardiac 1197 (80.7)

Primary arrhythmiab

Pulseless electrical activity 591 (39.9)Ventricular tachycardia 87 (5.9)Ventricular fibrillation 290 (19.6)Asystole 307 (20.7)Unknown 208 (14.0)

Survival stageImmediate 577 (38.9)At 24 h 378 (25.9)At discharge 247 (16.7)At 12 months 154 (10.5)

Arrest durationb

Under 15 min 633 (45.6)Over 15 min 755 (54.4)a p = <0.001.b p = <0.05.

S. Cooper, F. Duncan

ear study with the following 3 years, consideringresentation and survival changes and thereby test-ng the reliability of the RPS Scale. In Cooper etl.,9 we summarise results from the full decade ofesuscitation.

ethods

he recruitment methods, patient population,esuscitation survival predictors and rates haveeen described elsewhere.1,9 In summary, a longi-udinal prospective case register study conductedrom April 1993 to March 2003, at Derriford Hos-ital, Plymouth (UK), a 1200 bed district general

ospital. As recommended in the Utstein report-ng style10 patients less than 20 years of age, falselarms, respiratory arrests only, and multiple re-rrests were excluded from the studies.

1993—2000) Number (%), last 3 years (2000—2003)

369 (57.8)269 (42.2)

29 (4.6)47 (7.4)

106 (16.6)208 (32.7)247 (38.8)

155 (24.4)253 (39.9)226 (35.6)

145 (22.7)493 (77.3)

300 (47.0)26 (4.1)

98 (15.4)135 (21.2)79 (12.4)

242 (37.9)146 (22.9)91 (14.3)86 (13.5)

238 (39.5)364 (60.5)

R

A

Cnpouwttw

R

P

DfdMMwtT

iysbsaa3

1w

(iic

P

Roraad(scuvficptvod

rffVir

eliability testing and update of the RPS scale

nalysis

rude survival proportions were calculated as theumber of patients surviving divided by the totalopulation within each specific stratum. Proportionf characteristics of participants were comparedsing Chi-square (�2) tests with Fishers Exact testhere applicable. All analyses was carried out using

he SPSS for Windows (11.5) computer package. Allests for statistical significance were two-tailed,ith the level of significance at ˛ < 0.05.

esults

hase 1

ata set comparisons: In Cooper et al.9 the resultsrom a 10 year data set are described. We took thisata and separated the first 7 years (April 1993 toarch 2000) from the last 3 years (April 2000 toarch 2003) of the study. The original RPS Scaleas derived from the first 7 years.1 Patient charac-

eristics in each of these new groups are shown inable 1.

Of the 2121 CPR calls, 1483 (69.9%) occurredn the first 7 years and 638 (30.1%) in the last 3ears. For sex, time of arrest, mode of arrest andurvival stage there are no significant differencesetween the first 7 years and last 3 years of thetudy. However in the second year band, there was

significant increase in the rate of resuscitationttempts for those ≥80 years, 28.7% compared to

8.8% (�2 20.952, p < 0.001).

There was also a significant difference (�2

3.371, p = 0.010) in the primary arrhythmia ratesith an increase in pulseless electrical activity

Piac

Table 2 RPS Scale comparisons

RPS Scale categories First 7 yea(1993—200

Alive at24 h (%)

PEA/Asystole + ≥70 years 4.7PEA/Asystole + ≤69 years or less 8.3VT/VF + ≥70 years 13.5VT/VF + ≤69 years 28.6PEA/Asystole + ≥70 years + primary cardiac 3.8PEA/Asystole + ≤69 years + primary cardiac 6PEA/Asystole + ≥70 years + primary respiratory 9.8PEA/Asystole + ≤69 years + primary respiratory 16.7VT/VF + ≥70 years + primary cardiac 12.6VT/VF + ≤69 years + primary cardiac 30VT/VF + ≥70 years + primary respiratory 33.3VT/VF + ≤69 years + primary respiratory 0

a Insufficient data.

255

PEA) from 39.9 to 47%. Finally, there was anncrease in the duration of the arrests (<or>15 min)n the second year band, 60.5% lasting for >15 minompared to 54.4% (�2 6.287, p = 0.012).

hase 2

eliability of the RPS Scale (Appendix A): Theriginal ‘RPS Scale was developed from a logisticegression model, for the key predictors of survivalt 24 h (resuscitation duration, primary arrhythmia,ge and primary mode of arrest—–respiratory or car-iac) which gave a weighting as an odds ratio scorethe exponential beta) for each variable. Thesecores (rounded to whole numbers) were then allo-ated to all patients in the dataset. This enableds to calculate the relative rankings of all theariables and combine them appropriately on thenal scale (the higher the score the greater thehance of survival). Finally cross-tabulations wereerformed (for those patients who were resusci-ated for more than 15 min) based on the dependantariable ‘survival to 24 h’ against patients totaldds ratio scores, thereby producing a specific pre-iction of survival for each set of variables,1 p. 7.

To test the reliability of the RPS Scale, weepeated this process for the full 10 year data set,or example all the patients who were resuscitatedor more than 15 min and whose initial rhythm wasT or VF and were over 70 years of age were coded

nto one category and all the patients who wereesuscitated for more than 15 min who arrested in

EA/Asystole and were less than 69 were codednto another category. This was continued untilll patients were categorised into one of the RPSombinations (see Table 2). The 24 h survival rate

rs0)

Last 3 years(2000—2003)

Pearson’s�2 value

p-Value

Total Alive at24 h (%)

Total

337 4.9 185 0.004 df = 1 0.952169 9.6 83 1.953 df = 1 0.162133 22 50 0.128 df = 1 0.72063 21.4 14 0.295 df = 1 0.587

286 3.9 155 0.000 df = 1 0.990133 6.8 59 0.041 df = 1 1.00051 10 30 0.001 df = 1 1.00036 16.7 24 0.000 df = 1 1.000

127 23.8 42 3.047 df = 1 0.08160 23.1 13 0.250 df = 1 0.7456 12.5 8 0.884 df = 1 0.5383 0 1 a a

Twirr

mpTiatnul

itdorsRtgsi(it

srfcRi5tcci9sg

C

T

256

was then calculated for each category and the dataset was split into the first 7 years and last 3 years.Finally, we examined the difference in survivalbetween the year groupings (�2). The resultsin Table 2 indicate that there is no statisticaldifference between any of the sub groups in the2-year groups, demonstrating the reliability of theRPS Scale.

Discussion

Interestingly, we found a significant increase inthe rate of PEA over the last 3 years of the study.The reason is unknown; perhaps staff are failingto identify deteriorating patients or patients aresicker on admission? We also found that moreelderly patients (≥80) were being resuscitated. Isthis a failure to designate patients not for resus-citation or does it imply that elderly patients arenow considered to be more viable? Finally, againfor unknown reasons, resuscitation attempts lastedfor longer. Is this a reluctance to stop, influencedperhaps by a lack of understanding of survivalrates?

To test the reliability of the scale, we performedthe same procedures required in a split half ortwo-thirds one-third comparison but with thephilosophy that it was more applicable to comparethe whole original data set with the new to test theoriginal RPS Scale’s reliability. Despite the changeslisted above, there was no significant differencebetween survival rates for the first and second peri-ods indicating a good level of reliability. However,in order to update the scale, we have included themean survival rate from the full 10-year data basein the final scale (see Appendix A). No survivalrates are given in the last two rows of Table A2as the sample was small, with a wide survivalrange between the years. With a larger sample itis likely that survival will increase in every otherrow of these VT/VF groupings; we have thereforeincluded survival estimates of 25—35% as a guide forclinicians.

This latest RPS scale retains its high contentand face validity in that survival rates are simi-lar to other studies 10—19 and the key predictors ofsurvival are supported in other work, for exampleage,10,11,20—24 primary mode of arrest,11,25 primaryarrhythmia 9—11,13,17,20,22,25,27—30 and duration ofresuscitation11,20,23,31,32. In fact, even though alimited number of variables were selected for use in

the scale the original regression analysis predictedan overall survival accuracy of 90%.1

The survival figures also demonstrate an appro-priate progressive order of survival. For example, in

A

MM

S. Cooper, F. Duncan

able A1 younger patients are more likely to surviveithin and between the primary arrhythmia band-

ngs and in Table A2 there is always a lower survivalate for the primary cardiac mode of arrest (thanespiratory) in each arrhythmia grouping.

As previously1 the final scale (Appendix A) isade up of three tables, Table A1 is based on therimary arrhythmia and the patient’s age whilstable A2 is for use where the primary mode of arrests known. Finally, Table A3 lists additional vari-bles, from other studies5—7,11,20,23,26,31,33,34—36

hat should be taken into account before termi-ating a resuscitation attempt. The scale should besed 15 min into a resuscitation attempt; if it is usedater survival rates are likely to be reduced.

The figures reported in this paper match thosen our 10 year study9 but have minor differences tohose reported in our first paper1 due to an acci-ental random deletion of 150 patients from theriginal data base of 1633 patients. However, ine-running this smaller data sample, we found noignificant difference in proportions or the originalPS survival figures. Secondly, the results listed areaken from a single hospital site which may limiteneralisabilty. However, we would argue that theample is large (n = 2121), outcomes are replicatedn other studies, standardised Resuscitation CouncilUK) guidelines were adhered to, and the hospitals reasonably representative of in-hospital resusci-ation in the United Kingdom.

The final RPS Scale (Appendix A) clearly demon-trates that predicted survival rates 15 min into aesuscitation attempt can be very poor and there-ore it is an acceptable point at which to firstonsider termination (where there has been noOSC). ‘Clearly however there are ethical issues

nvolved in using such a tool, in some situations a% chance of survival would be deemed sufficiento continue resuscitation. In another situation thelinician may have substantial evidence to supportessation of resuscitation even where the RPS Scalendicates that there is a 29% chance of survival’,1 p.. In summary therefore the RPS Scale has demon-trated validity and reliability but can only be auide for the cessation of resuscitation.

onflict of interest

here is no conflict of interest with this work.

cknowledgement

any thanks to Dr. Hilary Sanders, Department ofathematics and Statistics, University of Plymouth,

R 257

Ui

AS

Ttru

Table A3 Additional key predictors of survival(reduced survival rates)

Resuscitation factorsDelay (>3 min) to commencement of basic life

supportNo return of spontaneous circulation during

eliability testing and update of the RPS scale

K, for her advice on the statistical methods usedn this study.

ppendix A. Resuscitation Predictorcoring Scale (RPS Scale)

his scale has been designed to aid resuscitationermination decisions by indicating 24 h survivalates (Tables A1 and A2).The RPS Scale should besed:

Table A1 Use this table 15 min into a resuscitationwhere the following are known primary arrhythmiaa

and patient’s age

Variable Alive (24 h)

PEA or Asystole + 70 years or older (%) 5PEA or Asystole + 69 years or less (%) 9VT or VF + 70 years or older (%) 16VT or VF + 69 years or less (%) 27

a Refers to initial cardiac arrest rhythm, ventricular tachy-cardia (VT), ventricular fibrillation (VF), Asystole or pulselesselectrical activity (PEA).

Table A2 Use this table 15 min into a resuscitationwhere the following are known: primary arrhythmia,patient’s age and primary mode of arrest (respiratoryor cardiac)a

Variable Alive (24 h)

PEA or Asystole + 70 years orolder + primary cardiac (%)

4

PEA or Asystole + 69 years orless + primary cardiac (%)

6

PEA or Asystole + 70 years orolder + primary respiratory (%)

10

PEA or Asystole + 69 years orless + primary respiratory (%)

17

VT or VF + 70 years or older + primarycardiac (%)

15

VT or VF + 69 years or less + primarycardiac (%)

29

VT or VF + 70 years or older + primaryrespiratory (%)

Estimate25—35b

VT or VF + 69 years or less + primaryrespiratory (%)

Estimate25—35b

(Note) The RPS Scale is based on the following premise: Theshorter the resuscitation the greater the chance of survival(p ≤ 0.001), primary arrhythmias of VT/VF have a signifi-cantly higher survival rate than PEA or Asystole (p ≤ 0.001),patients of 69 years or less are significantly more likely tosurvive (p = <0.001), where the primary mode of arrest isrespiratory (as opposed to cardiac) the survival rate is sig-nificantly improved (p = 0.001).

a Refers to the initial mode of arrest. Was it a primaryrespiratory arrest or a primary cardiac arrest?

b Insufficient data.

resuscitation

Pre arrest morbidityMalignancy especially metastaticHomebound lifestyleSepsis (on admission)Pneumonia (on admission)

•••

scwrscoa

R

t

Creatinine >220 mmol/lHypotension (systolic ≤90 mmHg)Glasgow Coma Scale = 3—5

for in-hospital resuscitation;15 min into a resuscitation attempta;where there has been no return of sponta-neous circulation (ROSC) during the resuscitationattempt.

The survival rates listed below are based on aample of 2121 cases of whom 1119 received resus-itation for more than 15 min. The variables listedill predict survival with a minimum of 90% accu-

acy. The RPS scale is therefore a good predictor ofurvival; however, the decision to terminate resus-itation should only be made after considerationf the predictors listed in Table A3 and all otherpplicable factors.

eferences

1. Cooper S, Evans C. Resuscitation Predictor Scoring Scale forin hospital cardiac arrests. Emerg Med J 2003;20:6—9.

2. Conroy SP, Luxton T, Dingwall R, Harwood RH, Gladman JRF.Cardiopulmonary resuscitation in continuing care settings:time for a rethink? Br Med J 2006;332:479—82.

3. Timmermans S. Sudden death and the myth of CPR. USA:Temple University Press; 1999.

4. Mohr M, Bahr J, Schmid J, Panzer W, Kettler D. The decisionto terminate resuscitation efforts: results of a questionnaire.Resuscitation 1997;34:51—5.

5. Bowker L, Stewart K. Predicting insuccessful cardiopul-monary resuscitation (CPR): a comparison of three morbidityscores. Resuscitation 1999;40:89—95.

6. O’Keeffe S, Ebell M. Prediction of failure to survive followingin-hospital cardiopulmonary resusciation: comparison of twopredictive instruments. Resuscitation 1994;28:21—5.

7. George A, et al. Pre-arrest morbidity and other correlated of

survival after in-hospital cardiopulmonary arrest. Am J Med1989;87:28—34.

8. de Vos R, Oosterom L, Koster RW, de Haan RJ. Decisions toterminate resuscitation. Resuscitation 1998;39:7—13.

a It can be used after 15 min but the survival rates are likelyo be lower than indicated.

1

1

1

1

1

1

1

1

1

1

2

2

2

2

2

2

2

2

2

2

3

3

3

3

3

3

258

9. Cooper S, Janghorbani M, Cooper G. A decade of in-hospitalresuscitation: outcomes and prediction of survival? Resusci-tation 2006;68:231—7.

0. A statement for health care professionals from the Ameri-can Heart Association, the European Resuscitation Council,the Heart and Stroke Foundation of Canada, the Aus-tralian Resuscitation Council and the Resuscitation Councilsof Southern Africa. Recommended Guidelines for review-ing, reporting, and conducting research on in-hospitalresuscitation: the in-hospital ‘Utstein Style’. Resuscitation1997;34:151—83.

1. Tunstall-Pedoe H, Bailey L, Chamberlain DA, MarsdenAK, Ward ME, Zideman DA. Survey of 3765 cardiopul-monary resuscitations in British hospitals (the BRESUSstudy): methods and overall results. Br Med J 1992;304:1347—51.

2. Taffet GI, Teasdale TA, Luchi RJ. In-hospital cardiopulmonaryresuscitation. JAMA 1988;260:2069—72.

3. Murphy DI, Murray AM, Robinson BE. Outcomes of car-diopulmonary resuscitation in the elderly. Ann Intern Med1989;111:199—205.

4. Johnson AL, Tanser PH, Ulan RA, Wood TE. Results of cardiacresuscitation in 552 patients. Am J Cardiol 1967;20:831—5.

5. Lemire JG, Johnson AL. Is cardiac resuscitation worthwhile?A decade of experience. N Engl J Med 1972;286:970—2.

6. Tweed WA, Bristow G, Donen N, Kirk BW. Evaluationof hospital-based cardiac resuscitation 1973—1977. CMAJ1980;122:301—4.

7. Gulati RS, Bhan GL, Horan MA. Cardiopulmonary resuscita-tion of the elderly. Lancet 1983;2:267—9.

8. Bedell SE, Delbanco TL, Cook EF. Survival after car-diopulmonary resuscitation in the hospital. N Engl J Med1983;309:569—75.

9. Berger R, Kelley M. Survival after in-hospital cardiopul-monary arrest of non critically ill patients—–a prospectivestudy. Chest 1994;106:872—9.

0. Lazzam C, McCans JL. Predictors of survival of in-hospitalcardiac arrest. Can J Cardiol 1991;7:113—6.

1. de Vos R, Koster RW, de Haan RJ, Oosting H, van der WouwPA, Lampe-Schoenmaeckers AJ. In-hospital cardiopulmonaryresuscitation. Arch Intern Med 1999;159:845—50.

2. Parish DC, Dane FC, Montgomery M, Wynn LJ, DurhamMD. Resuscitation in the hospital: differential relationships

3

S. Cooper, F. Duncan

between age and survival across rhythms. Crit Care Med1999;27:2137—41.

3. Cooper S, Cade J. Predicting survival, in-hospital cardiacarrests: resuscitation survival variables and training effec-tiveness. Resuscitation 1997;35:17—22.

4. Roberts D, Landolfo K, Light RB, Dobson K. Early predictorsof mortality for hospitalized patients suffering cardiopul-monary arrest. Chest 1990;97:413—9.

5. Brindley PG, Markland DM, Mayers I, Kutsogiannis DJ. Predic-tors of survival following in-hospital adult cardiopulmonaryresuscitation. CMAJ 2002;167:343—8.

6. Larsen MP, Eisenberg MS, Cummins RO, Hallstrom AP. Pre-dicting survival from out of hospital cardiac arrest: a graphicmodel. Ann Emerg Med 1993;22(11):1652—8.

7. Camp BN, Parish DC, Andrews RH. The impact of advancedcardiac life support training on resuscitation efforts and sur-vival in a rural hospital. Ann Emerg Med 1997;29:529—33.

8. Robinson G.R., Hess D. Post discharge survival and functionalstatus following in-hospital cardiopulmonary resuscitation.Chest 1994;105:991—6.

9. Rosendberg M, Wang C, Hoffman-Wilde S. Results ofcardiopulmonary resuscitation. Arch Intern Med 1993;153:1370—5.

0. Tresch DD, Heudebert G, Kutty K. Cardiopulmonary resus-citation in elderly patients hospitalized in the 1990s: afavorable outcome. J Am Geriatr Soc 1994;42:137—41.

1. Gwinnutt C, Columb M, Harris R. Outcome after cardiacarrest in adults in UK hospitals: effect of the 1997 guidelines.Resuscitation 2000;47:125—35.

2. Schultz SC, Cullinane DC, Pasquale MD, Magnant C, EvansSRT. Predicting in-hospital mortality during cardiopulmonaryresuscitation. Resuscitation 1996;33:13—7.

3. Henry JA. Resuscitation from poisoning. In: Baskett PJF,editor. Cardiopulmonary resuscitation. Amsterdam: Elsevier;1989. p. 231—57.

4. Wik L, Steen PA, Bircher NG. Quality of bystander cardiopul-monary resuscitation influences outcome after pre-hospitalcardiac arrest. Resuscitation 1994;28:195—203.

5. Bulut S, et al. Successful out-of-hospital cardiopulmonary

resuscitation: what is the optimum in-hospital strategy.Resuscitation 2000;47:155—61.

6. Wyatt JP, Tomlinson GS, Busuttil A. Resuscitation ofdrowning victims in South-East Scotland. Resuscitation1999;41:101—4.