- Email: [email protected]
The outcome following major trauma in the elderly. Predictors of survival
Injury, Int. J. Care Injured 30 (1999) 703±706
www.elsevier.com/locate/injury
The outcome following major trauma in the elderly. Predictors of survival S.A.W. Pickering, D. Esberger, C.G. Moran* Department of Trauma and Orthopaedic Surgery, Queen's Medical Centre, University Hospital Trust, Nottingham NG7 2UH, UK Accepted 19 July 1999
Abstract Objectives: To assess the reliability of the predicted probability of survival calculated using TRISS methodology by the UK Trauma Network for elderly patients. Method: Analysis of 100 consecutive trauma patients 65 years and over, prospectively entered into the UK Trauma Network database from a single centre. The probability of survival (Ps) was calculated from the UK Trauma database and retrospectively related to survival, premorbid medical condition and mobility. Results: Of 100 patients, 16 died and 84 survived. Eleven of the 16 who died and 12 of the survivors had pre-existing medical disease (ASA grade III±V) and social dependency suggesting a poor outcome, these factors being signi®cantly associated with mortality (P < 0.005). The mean Ps for the 11 with severe medical disease who died was 0.85 (20.07) with a mean age 85 (23.5). The remaining ®ve patients who died suered high energy injuries, had a mean age of 70 (24.8) and a low probability of survival (Ps 0.402 0.24). The median pre-injury mobility score was 8 in patients who survived and 4.5 in those who died. Mobility score <5 was associated with an increased mortality following admission from Trauma (P < 0.05) Conclusions: There is a signi®cant association between severe preexisting medical disease (ASA III±V) and death during admission for trauma. The Ps score is unrealistically high in this group of patients. A simple mobility score correlates well with outcome in this group. # 1999 Elsevier Science Ltd. All rights reserved.
1. Introduction There has been considerable interest in assessing the outcome of trauma care in the United Kingdom. The UK Trauma Network collects data from a large number of hospitals within the National Health Service. The audit is based upon the American Major Trauma Outcome Survey. This method attempts to correct for case mix and provides data for the expected outcome and mortality for a range of Trauma patients [1]. It is essential that such a system is reliable as both the public and the politicians increasingly demand the publication of headline ®gures and `league tables'. In the early stages of development, the UK Trauma Network included elderly patients (65 years and above) with isolated hip fractures. However, the large
* Corresponding author.
numbers caused problems with data collection and the injury severity score was not a good predictor of outcome in this group. In view of this, elderly patients with proximal femoral fractures are now excluded from the UK Trauma Network. However, elderly trauma patients still constitute 22% of patients in the database. A potential limitation of the methodology currently used to predict outcome following trauma is that it takes no account of medical or social conditions. Many elderly patients have their accident as a direct result of poor health and mobility. It is possible that pre-existing problems will have a greater in¯uence on outcome than the injury itself. These factors could be used as a general predictor of outcome. The outcome for elderly patients with a proximal femoral fracture has been extensively studied. The mental test score and pre-injury mobility correlate well with long term outcome and survival [2].
0020-1383/99/$ - see front matter # 1999 Elsevier Science Ltd. All rights reserved. PII: S 0 0 2 0 - 1 3 8 3 ( 9 9 ) 0 0 1 8 8 - 6
704
S.A.W. Pickering et al. / Injury, Int. J. Care Injured 30 (1999) 703±706
Table 1 Assessment of mobility before the fracture. Score is the total 0±9 Mobility
No diculty
With an aid
With help from other person
Not at all
Able to get about the house Able to get out of house Able to go shopping
3 3 3
2 2 2
1 1 1
0 0 0
The aim of this study is to compare the predictive value of the injury severity score with functional mobility and general health in the elderly trauma patient.
2. Method
3. Results One hundred patients (mean age 78; range 65±93) were included in the study. There were 66 females and 34 males. Sixteen patients died during their ®rst admission to hospital following trauma. 3.1. Deaths
One hundred consecutive patients, 65 years and over, admitted to the Queen's Medical Centre between August 1997 and August 1998 were reviewed with respect to outcome. All patients had been prospectively entered into the UK Trauma Audit Network according to strict entry criteria at the time of their admission. Details of age, previous health, mobility and dependence on social services were documented. The mobility score was calculated using the method of Parker and Palmer [2] (see Table 1). The anatomic injuries were used to give an injury severity score using the UK Trauma Network database from which a probability of survival (Ps) could be calculated [3]. Coexisting medical conditions and their severity were recorded and used to estimate the ASA grade [4], all patients having ASA grade III or above being considered as high risk (see Table 2). It must be accepted that the accuracy of this type of retrospective assessment is limited. This information was related to mortality during hospital admission. A Ps score >0.75 was considered to predict an avoidable death [3] and a mobility score of <5 was considered to predict high chance of a poor outcome [2]. The results were analysed in the form of contingency tables by the Chi squared test with Yates correction and the Student's t-test for independent samples where equal variances were not assumed (see Tables 4 and 5).
There were 12 females and 4 males in the group who died. The mean age was 80 years (range 66±90). The mean Ps was 0.71 (SD=0.26) and the median mobility score was 4.5 (range 2±9). Review of the group identi®ed 11 patients (69%) with multiple coexisting medical problems (ASA III-V), poor mobility and high dependence on social services. The mean age for this subgroup was 85 (SD=3.5), but the Ps suggested a high probability of survival (mean 0.85, SD=0.07). Their mobility was poor (median 4, range 2±9). The remaining ®ve patients who died (31%) had an entirely dierent clinical picture (see Table 3). They were signi®cantly younger (P < 0.005) with a mean age of 70 (SD=4.8) and they were all independently mobile (mobility score 9 in all cases). This subgroup suered more severe injuries and had a signi®cantly lower (P < 0.01) probability of survival (mean Ps 0.42, SD=0.25). 3.2. Survivors Eighty-four patients survived and in this group the mean age was 76 years. There were 54 females and 30 males. The mean Ps was 0.90 (SD=0.14) and the median mobility score was 8 (range 1±9). Within this group 12 patients (15%) were identi®ed who had mul-
Table 2 ASA classi®cation I II III IV V
Normal healthy patient Patient with mild systemic disease Severe systemic disease that limits activity but is not incapacitating Patient with an incapacitating disease that is constant threat to life Moribund patient not expected to survive more than 24 h with or without operation
S.A.W. Pickering et al. / Injury, Int. J. Care Injured 30 (1999) 703±706
705
Table 3 Elderly trauma deaths, collected data Patient
Age
Injury
Ps
Mobility
Coexisting disease
ASA
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16
90 77 89 84 78 66 87 84 84 88 86 84 66 69 73 81
Pubic ramus # Fall down stairs, severe head injury Simple fall, suprcond. # femur Slipped on ¯oor, multiple rib # Slipped on stairs, sternal #, scalp cut RTA, pelvic #, rib #s, cord transection Fall on stairs, T2 # Fall on stairs, head injury, rib #s Simple fall, multiple rib#s Colles #, head injury NOF, Colles # Ankle # Severe head injury Hit by car, head injury, tibia #, wrist # Severe head injury Fell in garden, rib #s
0.91 0.16 0.86 0.79 0.82 0.39 0.67 0.91 0.91 0.84 0.86 0.9 0.17 0.7 0.59 0.91
5 9 2 4 4 9 4 4 4 4 3 5 9 9 9 9
CCF, BP, Dementia Nil TIAs, CCF COAD, Angina, Dementia Severe COAD Nil IHD, CCF, Thrombocytopaenia CVA, AF, MI, BP Dementia, Hypertension CCF, Anaemia Dementia, AF, BP, NIDDM, Hypothyroid Asthma, Anaemia, Endometrial Ca. Nil Ca Breast Nil MI, CCF, Asthma
3 1 3 3 4 1 3 3 3 4 3 3 1 2 1 3
tiple coexisting medical problems, poor mobility and social dependence. Preexisting severe medical disease (ASA r 3) and poor mobility were signi®cantly associated with mortality following trauma in the elderly (P < 0.005 and P < 0.05 respectively). 4. Discussion The UK Trauma Audit and Research Network has been set up to audit trauma care. Patient survival probabilities are worked out by reference to a national database and use of TRISS methodology [1]. The criteria for entry include trauma admissions longer than 72 hours, any patient admitted to a high dependency area and all deaths of injured patients occurring in the hospital. It is then possible to compare expected with observed outcome as part of ongoing audit. A number of specialists in trauma care have expressed reservations with regard to the accuracy of this process. Our concern is that a subgroup of elderly patients with signi®cant health and mobility problems, who suered low energy trauma, are being entered into the UK Trauma Audit and Research Network. This audit methodology was developed in North Table 4 Contingency table relating survival to severe coexisting medical disease (ASA > 2)
ASA > 2 ASA < 3 Total
Died
Survived
Total
11 5 16
12 72 84
23 77 100
American Trauma Centres which generally deal with a young population Thus, no account is taken of a patient's state of health at the time of injury. This is not of major importance in the younger age groups, but becomes increasingly signi®cant beyond 65 years. When a survival score is calculated, elderly patients may be given unrealistically high survival probabilities. As a result, any department that deals with a large amount of elderly trauma would appear to perform less well than expected. In the elderly patients who survived injury, only 15% had major health problems. In contrast, 69% of the patients who died had major coexisting medical problems (P < 0.001). The study identi®ed two dierent groups of elderly trauma patients who die. Those suering high energy trauma who are younger (mean age 70 years), active and without an extensive medical history, and those suering relatively low energy trauma who are older (mean age 85 years) and have several major medical conditions with severe mobility restriction. Age is linked to signi®cant medical disease in the elderly and our ®ndings con®rm this. The mean Ps of 0.85 in the older group of patients who died, suggests that these deaths could have been avoided. However, clinical assessment indicated they would tolerate poorly a period of bed rest in hospital. A typical example was an 82-year-old patient with Table 5 Contingency table relating survival to mobility score < 5
Mobility < 5 Mobility > 4 Total
Died
Survived
Total
8 8 16
16 68 84
24 76 100
706
S.A.W. Pickering et al. / Injury, Int. J. Care Injured 30 (1999) 703±706
severe congestive cardiac failure, who fell at home sustaining multiple rib fractures. She was quickly overwhelmed by a chest infection yet had a Ps of 0.91. Parker and Palmer [2] assessed a large number of patients presenting with proximal femoral fractures. They looked at survival at one year and found there to be a signi®cant relationship between death and a mobility score of less than 5 or a reduced mental test score (less than 8/11). We could only reliably calculate the mobility score as the mental test score was not routinely documented. We found the median mobility score in the frail subgroup of those who died to be 4.5 compared with 9 in the ®t elderly patients who died (P < 0.05). Use of a simple mobility score, proven to be of bene®t in long-term prediction, may be useful in predicting mortality in the elderly trauma patient. This is a relatively small study and interpretation must be cautious. However, we have demonstrated the important in¯uence of coexisting medical disease in outcome in elderly trauma patients. A factor that is not currently taken into account when survival probabilities are calculated in the UK Trauma audit, although they acknowledge the importance of ®nding ways of routinely collecting and using such information to give more reliable survival probabilities. 5. Conclusions This study has identi®ed two groups of elderly trauma patients, included in the UK Trauma Network Audit, who have a poor outcome. The ®rst group are relatively young (mean age 70 years), medically ®t and mobile. They suer high energy trauma and the mor-
tality is high. The second larger group are more frail patients, often house-bound, who suer low energy trauma. Use of the Ps is unreliable in this latter group. The present study suggests that pre-existing medical conditions and mobility are important factors in outcome in this group of patients. The UK Trauma Network Audit is continuing to evolve and we suggest that further re®nements could take account of these factors. An alternative is a further age band (over 75 years) or excluding all patients over 75 years from TRISS analysis.
Acknowledgements The authors would like to thank Prof D. Yates and A. Whyte for their helpful discussions in the preparation of this study.
References [1] Champion HR, Copes WS, Sacco WJ, Lawnick MM, Keast SL, Bain LW, Flanagan ME, Frey CF. The Major Trauma Outcome Study: establishing norms for trauma care. Journal of Trauma 1990;30(11):1356±65. [2] Parker MJ, Palmer CR. A new mobility score for predicting mortality after hip fracture. Journal of Bone and Joint Surgery (Br) 1993;75-B:797±8. [3] Yates DW, Woodford M, Hollis S. Trauma audit: clinical judgement or statistical analysis? Annals of the Royal College of Surgeons of England 1993;75:321±4. [4] American Society of Anesthesiologists. New classi®cation of physical status. Anesthesiology 1963;24:111.
www.elsevier.com/locate/injury
The outcome following major trauma in the elderly. Predictors of survival S.A.W. Pickering, D. Esberger, C.G. Moran* Department of Trauma and Orthopaedic Surgery, Queen's Medical Centre, University Hospital Trust, Nottingham NG7 2UH, UK Accepted 19 July 1999
Abstract Objectives: To assess the reliability of the predicted probability of survival calculated using TRISS methodology by the UK Trauma Network for elderly patients. Method: Analysis of 100 consecutive trauma patients 65 years and over, prospectively entered into the UK Trauma Network database from a single centre. The probability of survival (Ps) was calculated from the UK Trauma database and retrospectively related to survival, premorbid medical condition and mobility. Results: Of 100 patients, 16 died and 84 survived. Eleven of the 16 who died and 12 of the survivors had pre-existing medical disease (ASA grade III±V) and social dependency suggesting a poor outcome, these factors being signi®cantly associated with mortality (P < 0.005). The mean Ps for the 11 with severe medical disease who died was 0.85 (20.07) with a mean age 85 (23.5). The remaining ®ve patients who died suered high energy injuries, had a mean age of 70 (24.8) and a low probability of survival (Ps 0.402 0.24). The median pre-injury mobility score was 8 in patients who survived and 4.5 in those who died. Mobility score <5 was associated with an increased mortality following admission from Trauma (P < 0.05) Conclusions: There is a signi®cant association between severe preexisting medical disease (ASA III±V) and death during admission for trauma. The Ps score is unrealistically high in this group of patients. A simple mobility score correlates well with outcome in this group. # 1999 Elsevier Science Ltd. All rights reserved.
1. Introduction There has been considerable interest in assessing the outcome of trauma care in the United Kingdom. The UK Trauma Network collects data from a large number of hospitals within the National Health Service. The audit is based upon the American Major Trauma Outcome Survey. This method attempts to correct for case mix and provides data for the expected outcome and mortality for a range of Trauma patients [1]. It is essential that such a system is reliable as both the public and the politicians increasingly demand the publication of headline ®gures and `league tables'. In the early stages of development, the UK Trauma Network included elderly patients (65 years and above) with isolated hip fractures. However, the large
* Corresponding author.
numbers caused problems with data collection and the injury severity score was not a good predictor of outcome in this group. In view of this, elderly patients with proximal femoral fractures are now excluded from the UK Trauma Network. However, elderly trauma patients still constitute 22% of patients in the database. A potential limitation of the methodology currently used to predict outcome following trauma is that it takes no account of medical or social conditions. Many elderly patients have their accident as a direct result of poor health and mobility. It is possible that pre-existing problems will have a greater in¯uence on outcome than the injury itself. These factors could be used as a general predictor of outcome. The outcome for elderly patients with a proximal femoral fracture has been extensively studied. The mental test score and pre-injury mobility correlate well with long term outcome and survival [2].
0020-1383/99/$ - see front matter # 1999 Elsevier Science Ltd. All rights reserved. PII: S 0 0 2 0 - 1 3 8 3 ( 9 9 ) 0 0 1 8 8 - 6
704
S.A.W. Pickering et al. / Injury, Int. J. Care Injured 30 (1999) 703±706
Table 1 Assessment of mobility before the fracture. Score is the total 0±9 Mobility
No diculty
With an aid
With help from other person
Not at all
Able to get about the house Able to get out of house Able to go shopping
3 3 3
2 2 2
1 1 1
0 0 0
The aim of this study is to compare the predictive value of the injury severity score with functional mobility and general health in the elderly trauma patient.
2. Method
3. Results One hundred patients (mean age 78; range 65±93) were included in the study. There were 66 females and 34 males. Sixteen patients died during their ®rst admission to hospital following trauma. 3.1. Deaths
One hundred consecutive patients, 65 years and over, admitted to the Queen's Medical Centre between August 1997 and August 1998 were reviewed with respect to outcome. All patients had been prospectively entered into the UK Trauma Audit Network according to strict entry criteria at the time of their admission. Details of age, previous health, mobility and dependence on social services were documented. The mobility score was calculated using the method of Parker and Palmer [2] (see Table 1). The anatomic injuries were used to give an injury severity score using the UK Trauma Network database from which a probability of survival (Ps) could be calculated [3]. Coexisting medical conditions and their severity were recorded and used to estimate the ASA grade [4], all patients having ASA grade III or above being considered as high risk (see Table 2). It must be accepted that the accuracy of this type of retrospective assessment is limited. This information was related to mortality during hospital admission. A Ps score >0.75 was considered to predict an avoidable death [3] and a mobility score of <5 was considered to predict high chance of a poor outcome [2]. The results were analysed in the form of contingency tables by the Chi squared test with Yates correction and the Student's t-test for independent samples where equal variances were not assumed (see Tables 4 and 5).
There were 12 females and 4 males in the group who died. The mean age was 80 years (range 66±90). The mean Ps was 0.71 (SD=0.26) and the median mobility score was 4.5 (range 2±9). Review of the group identi®ed 11 patients (69%) with multiple coexisting medical problems (ASA III-V), poor mobility and high dependence on social services. The mean age for this subgroup was 85 (SD=3.5), but the Ps suggested a high probability of survival (mean 0.85, SD=0.07). Their mobility was poor (median 4, range 2±9). The remaining ®ve patients who died (31%) had an entirely dierent clinical picture (see Table 3). They were signi®cantly younger (P < 0.005) with a mean age of 70 (SD=4.8) and they were all independently mobile (mobility score 9 in all cases). This subgroup suered more severe injuries and had a signi®cantly lower (P < 0.01) probability of survival (mean Ps 0.42, SD=0.25). 3.2. Survivors Eighty-four patients survived and in this group the mean age was 76 years. There were 54 females and 30 males. The mean Ps was 0.90 (SD=0.14) and the median mobility score was 8 (range 1±9). Within this group 12 patients (15%) were identi®ed who had mul-
Table 2 ASA classi®cation I II III IV V
Normal healthy patient Patient with mild systemic disease Severe systemic disease that limits activity but is not incapacitating Patient with an incapacitating disease that is constant threat to life Moribund patient not expected to survive more than 24 h with or without operation
S.A.W. Pickering et al. / Injury, Int. J. Care Injured 30 (1999) 703±706
705
Table 3 Elderly trauma deaths, collected data Patient
Age
Injury
Ps
Mobility
Coexisting disease
ASA
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16
90 77 89 84 78 66 87 84 84 88 86 84 66 69 73 81
Pubic ramus # Fall down stairs, severe head injury Simple fall, suprcond. # femur Slipped on ¯oor, multiple rib # Slipped on stairs, sternal #, scalp cut RTA, pelvic #, rib #s, cord transection Fall on stairs, T2 # Fall on stairs, head injury, rib #s Simple fall, multiple rib#s Colles #, head injury NOF, Colles # Ankle # Severe head injury Hit by car, head injury, tibia #, wrist # Severe head injury Fell in garden, rib #s
0.91 0.16 0.86 0.79 0.82 0.39 0.67 0.91 0.91 0.84 0.86 0.9 0.17 0.7 0.59 0.91
5 9 2 4 4 9 4 4 4 4 3 5 9 9 9 9
CCF, BP, Dementia Nil TIAs, CCF COAD, Angina, Dementia Severe COAD Nil IHD, CCF, Thrombocytopaenia CVA, AF, MI, BP Dementia, Hypertension CCF, Anaemia Dementia, AF, BP, NIDDM, Hypothyroid Asthma, Anaemia, Endometrial Ca. Nil Ca Breast Nil MI, CCF, Asthma
3 1 3 3 4 1 3 3 3 4 3 3 1 2 1 3
tiple coexisting medical problems, poor mobility and social dependence. Preexisting severe medical disease (ASA r 3) and poor mobility were signi®cantly associated with mortality following trauma in the elderly (P < 0.005 and P < 0.05 respectively). 4. Discussion The UK Trauma Audit and Research Network has been set up to audit trauma care. Patient survival probabilities are worked out by reference to a national database and use of TRISS methodology [1]. The criteria for entry include trauma admissions longer than 72 hours, any patient admitted to a high dependency area and all deaths of injured patients occurring in the hospital. It is then possible to compare expected with observed outcome as part of ongoing audit. A number of specialists in trauma care have expressed reservations with regard to the accuracy of this process. Our concern is that a subgroup of elderly patients with signi®cant health and mobility problems, who suered low energy trauma, are being entered into the UK Trauma Audit and Research Network. This audit methodology was developed in North Table 4 Contingency table relating survival to severe coexisting medical disease (ASA > 2)
ASA > 2 ASA < 3 Total
Died
Survived
Total
11 5 16
12 72 84
23 77 100
American Trauma Centres which generally deal with a young population Thus, no account is taken of a patient's state of health at the time of injury. This is not of major importance in the younger age groups, but becomes increasingly signi®cant beyond 65 years. When a survival score is calculated, elderly patients may be given unrealistically high survival probabilities. As a result, any department that deals with a large amount of elderly trauma would appear to perform less well than expected. In the elderly patients who survived injury, only 15% had major health problems. In contrast, 69% of the patients who died had major coexisting medical problems (P < 0.001). The study identi®ed two dierent groups of elderly trauma patients who die. Those suering high energy trauma who are younger (mean age 70 years), active and without an extensive medical history, and those suering relatively low energy trauma who are older (mean age 85 years) and have several major medical conditions with severe mobility restriction. Age is linked to signi®cant medical disease in the elderly and our ®ndings con®rm this. The mean Ps of 0.85 in the older group of patients who died, suggests that these deaths could have been avoided. However, clinical assessment indicated they would tolerate poorly a period of bed rest in hospital. A typical example was an 82-year-old patient with Table 5 Contingency table relating survival to mobility score < 5
Mobility < 5 Mobility > 4 Total
Died
Survived
Total
8 8 16
16 68 84
24 76 100
706
S.A.W. Pickering et al. / Injury, Int. J. Care Injured 30 (1999) 703±706
severe congestive cardiac failure, who fell at home sustaining multiple rib fractures. She was quickly overwhelmed by a chest infection yet had a Ps of 0.91. Parker and Palmer [2] assessed a large number of patients presenting with proximal femoral fractures. They looked at survival at one year and found there to be a signi®cant relationship between death and a mobility score of less than 5 or a reduced mental test score (less than 8/11). We could only reliably calculate the mobility score as the mental test score was not routinely documented. We found the median mobility score in the frail subgroup of those who died to be 4.5 compared with 9 in the ®t elderly patients who died (P < 0.05). Use of a simple mobility score, proven to be of bene®t in long-term prediction, may be useful in predicting mortality in the elderly trauma patient. This is a relatively small study and interpretation must be cautious. However, we have demonstrated the important in¯uence of coexisting medical disease in outcome in elderly trauma patients. A factor that is not currently taken into account when survival probabilities are calculated in the UK Trauma audit, although they acknowledge the importance of ®nding ways of routinely collecting and using such information to give more reliable survival probabilities. 5. Conclusions This study has identi®ed two groups of elderly trauma patients, included in the UK Trauma Network Audit, who have a poor outcome. The ®rst group are relatively young (mean age 70 years), medically ®t and mobile. They suer high energy trauma and the mor-
tality is high. The second larger group are more frail patients, often house-bound, who suer low energy trauma. Use of the Ps is unreliable in this latter group. The present study suggests that pre-existing medical conditions and mobility are important factors in outcome in this group of patients. The UK Trauma Network Audit is continuing to evolve and we suggest that further re®nements could take account of these factors. An alternative is a further age band (over 75 years) or excluding all patients over 75 years from TRISS analysis.
Acknowledgements The authors would like to thank Prof D. Yates and A. Whyte for their helpful discussions in the preparation of this study.
References [1] Champion HR, Copes WS, Sacco WJ, Lawnick MM, Keast SL, Bain LW, Flanagan ME, Frey CF. The Major Trauma Outcome Study: establishing norms for trauma care. Journal of Trauma 1990;30(11):1356±65. [2] Parker MJ, Palmer CR. A new mobility score for predicting mortality after hip fracture. Journal of Bone and Joint Surgery (Br) 1993;75-B:797±8. [3] Yates DW, Woodford M, Hollis S. Trauma audit: clinical judgement or statistical analysis? Annals of the Royal College of Surgeons of England 1993;75:321±4. [4] American Society of Anesthesiologists. New classi®cation of physical status. Anesthesiology 1963;24:111.