Health outcomes of adults 3 months after injury

Introduction 

Injury represents one of the major contributors to the public health burden in Australia. According to the Australian Institute of Health and Welfare, 6.8% of all separations from public, private and psychiatric hospitals in Australia in the 2001/2002 financial year were due to injury (including poisonings).2 This burden remains similar to that reported several years earlier in relation to the 1999/2000 financial year.14 In terms of expenditure, health care costs related to injury and poisoning accounted for 8.2% of the total allocated health expenditure in Australia in 2001/2002.2 Despite this burden, there is limited literature on the long-term outcomes of injury, including health related quality of life (HRQoL).

HRQoL has a variety of descriptions, but generally it includes ‘the dimensions of physical functioning, social functioning, role functioning, mental health and general health perceptions’.40 These dimensions are effectively measured using the SF-36 instrument.8, 38 HRQoL represents one aspect of the broader concept of quality of life. The measurement of HRQoL provides a useful method of assessing the outcomes of health interventions. It can also assist health workers in the identification of key areas to target in planning and monitoring different treatment options.

Given that injury is a leading cause of preventable mortality and morbidity in Australia and the world and a major cost to the health system,3, 5, 25, 31 the lack of research in this field is surprising. There has been minimal research in Australia examining the HRQoL of adults following injury, however there has been two series of work from the USA, one in general trauma patients and one in specific subgroups of trauma patients. Holbrook et al. identified reduced function up to 18 months post-injury, with function being worse in women when compared to men.15, 16, 17, 18 Similarly, MacKenzie et al. identified poorer health across most dimensions measured by the SF-36 1 year post-head injury and poor functional outcomes up to 7 years post-severe lower limb trauma.22, 24

This study was consequently developed to determine the HRQoL of adults 3 months post-injury. These HRQoL scores were then compared to non-injured Australian adults of the same age to determine whether the reported HRQoL was similar to the general population, suggesting the impact of injury was experienced for less than 3 months, or whether HRQoL was lower than the general population, suggesting the impact of injury persisted beyond 3 months. This study is making the assumption that the injured population did not differ from the general population prior to their injury.

Materials and methods 

A prospective cohort design was used to study people who were admitted to 1 of the 12 adult Queensland Trauma Registry (QTR) hospitals for the treatment of injury and who granted consent for follow-up post-discharge. The sample reported in this manuscript was limited to those people who met the following criteria:

Study participants were then categorised into one of two groups:

•admitted to hospital for ≥24h for definitive treatment of specified injury: this group met the criteria for inclusion on the QTR and therefore had information available regarding their injury and the treatment of that injury, these participants formed the Primary Study Group;

•discharged directly from the emergency department or admitted to hospital for <24h: this group of participants were those who were initially expected to be admitted to hospital for ≥24h, and so were enrolled in the study, but subsequently were discharged early. As a result they were not entered on the QTR, only had SF-36 data collected and formed the Secondary Study Group.

To determine eligibility for inclusion in the study, patients’ health records were reviewed and cases were excluded if, in the opinion of the Trauma Registry Nurse Coordinator, the primary reason for the admission of a patient exceeding 24h was for medical, social or psychiatric reasons, rather than primarily for treatment of the injury. For example, if the reason for admission was to treat the patient's chronic heart failure, even if a minor injury had caused the patient to present to hospital, then the patient was excluded from the study. If the reason for admission was a combination of trauma and other reasons the patient was included in the study. Cases were also excluded if the patient was admitted for more than 24h for elective surgical treatment of an injury or for medical complications or pathological injuries (without an associated reasonable force resulting in the injury).

Patients who met the criteria for the study were approached for consent during their hospitalisation for injury between March and June 2003. Those patients who had consented to involvement in the study received a package in early August 2003. This package included a questionnaire containing the SF-36 and an additional 13 demographic questions and a cover letter explaining the purpose of the study. Demographic questions related to age, sex, indigenous and marital status, employment, education and health insurance status. In order to improve the response rate a four-step mail out process was used. This process consisted of the original postal questionnaire, followed by a reminder letter sent to all non-respondents approximately 3 weeks after posting the survey. After 6 weeks a copy of the original questionnaire was resent, and a final reminder was sent 2 weeks after the second questionnaire was posted.

This study used the Australian and New Zealand version of the SF-36 (version 1) to measure the functional health status and well being of adults following their injury. The short form 36 (SF-36) is a 36 item self-administered instrument that was designed to measure HRQoL in persons 14 years of age and older.38 The SF-36 examines concepts of limitation due to health problems in the following areas: (1) physical functioning; (2) social functioning; (3) physical role (4) bodily pain; (5) mental health; (6) emotional role; (7) vitality; (8) general health. The SF-36 has been shown to have good reliability and validity in a variety of contexts both within Australia and elsewhere.6, 10, 12, 21, 22, 25, 26, 30, 34, 35, 36, 37, 38, 39 In addition to the individual subscales two summary scores, the physical component summary (PCS) and mental component summary (MCS) scores were calculated using data from each of the relevant scales within the SF-36. The aim of this process was to produce meaningful summary scores without significant loss of information.

Data that were collected as part of the routine operation of the QTR for the cases hospitalised for ≥24h included date of birth, sex, date of injury, external cause of injury, hospital length of stay, ICU length of stay, injury description coded using the abbreviated injury scale (AIS) 1990, and injury severity score (ISS), calculated from the AIS. ISS is an interval scale which extends from 1 to 75 and incorporates the three highest scoring regions from the AIS in an attempt to quantify injuries.

This study was approved by each of the Human Research Ethics Committees of the relevant hospitals in accordance with the National Health and Medical Research Council's Guidelines. All patients who participated in the study provided an informed consent to be contacted after discharge from hospital for the purposes of research. Further consent was implied by their return of the questionnaire.

Descriptive data is presented to describe the study sample. The SF-36 scores for study respondents were compared with those of the Australian norms using z-statistics. Australian normative data for SF-36 scores were obtained from a subsample of the National Health Survey collected by the Australian Bureau of Statistics and included 18,800 adult residents of private dwellings.1

Multivariate analysis was performed to determine the effect of factors on the outcome of physical component summary (PCS). One-way ANOVA for categorical variables or simple linear regression for continuous variables, was used in univariate analyses of PCS score to identify potential effects for inclusion in a multivariate model. A multivariate ANCOVA was then applied to examine the partial effects of the covariates and factors that were significant with p<0.05 in the univariate analyses. The model building process involved a step-by-step omission of non-significant effects (p>0.05).

Unless otherwise specified, data was considered statistically significant at the five percent level. SAS version 8© (1999–2001, SAS Institute Inc., Cary, NC, USA) was used to conduct descriptive and comparative analyses, while Stata/SE 9.1© (2005, StatCorp LP, Texas, USA) and SPSS Release 13.0© (2004, SPSS Inc., Chicago, Il, USA) were used to conduct the multivariate analyses.

Results 

Initially 843 potential participants consented to be contacted regarding participation in the study (Fig. 1). Of the 843 consenting patients, 286 (33.9%) patients did not respond, while a further 58 (6.9%) questionnaires were returned as ‘unknown at that address’. Eleven potential participants were ineligible including seven patients who had died since hospital admission, one patient who had left the country, four patients who declined to participate and three patients who were unable to participate. Of the remaining 484 participants that returned their questionnaires, 145 (30.0%) participants were hospitalised for less than 24h and are not considered in the main cohort for this study, but do form a Secondary Study Group for comparison of SF-36 data to the Australian norms. The remaining 339 participants met the criteria for inclusion in the Primary Study Group reported here.

View Large Image Download to PowerPoint

Figure 1. Summary of participant response.

Comparison of the Primary Study Group to the 3367 patients entered onto the QTR during the enrolment period who met the study inclusion criteria identified few differences. Patients who participated were slightly younger than the other eligible QTR patients (48.8±19.6 years versus 48.1±24.0 years; p=0.003), although this age difference is unlikely to hold any clinical significance. Patients who participated had a similar ISS to the other eligible QTR patients (7.8±6.2 years versus 8.9±8.6 years; p=0.102).

With regard to the Primary Study Group, completed questionnaires were received from 210 (62.0%) males and 127 (37.5%) females (data were not provided by two participants). The mean age of participants was 48.8 (S.D.=19.6) years. Participants were relatively evenly spread across the age groups corresponding to those used in the Australian National Health Survey from which the Australian norms for the SF-36 have been obtained. There were 18% of participants in the 25–34-year age group, 23% in the 45–54-year age group and 11–12% in all other age groups.

The injury severity score for participants in the Primary Study Group ranged from 1 to 50, with a mean of 7.8 (S.D.=6.2). Twenty-two participants were considered to have experienced major injury (ISS≥16), while 312 experienced minor injury (ISS<16). The external cause of these injuries was predominantly falls (39%) and road traffic accidents (17%), while fractures (67%) and open wounds (6%) of the limbs constituted the greatest number of injuries.

Comparisons have been made between the study participants in the Primary Study Group and Australian norms for all subscales of the SF-36 except the reported health transition subscale for which there are no Australian norms published. Mean subscale scores for participants in this study were statistically significantly lower than the Australian norm mean subscale scores for every subscale as well as the two summary scores (Fig. 2). This confirms that adults who have sustained an injury that required hospitalisation for more than 24h have a reduced HRQoL, when compared to the Australian norms, 3 months after hospitalisation. The same pattern, with participants’ scores being less than Australian norms, continued when the mean scores for each subscale for male and female participants were considered.

View Large Image Download to PowerPoint

Figure 2. Sample means for both the Primary (hospitalised for ≥24h, n=337) and Secondary Study Groups (hospitalised for <24h, n=130) compared with Australian norms.

The potential existed for Primary Study Group participants’ scores to vary from Australian norms as a result of different age distributions between the study sample and the Australian population. Therefore, the mean scores for study participants versus Australian norms broken down by age group were analysed for each subscale. In general, younger study participants’ scores were statistically lower than Australian norms for the majority of subscales, as well as the two summary scores, in the majority of age groups up to 64 years of age (Table 1). However from 65 years of age onwards the SF-36 scores for the participants in the Primary Study Group were frequently not statistically lower than the Australian norms. This pattern was particularly evident in the general health, mental health and vitality subscales.

Table 1. Differences in SF-36 scores between Primary Study group and Australian norms for each age group

	SF-36 subscale	Overall study mean	Mean for 18–24 years		Mean for 25–34 years		Mean for 35–44 years		Mean for 45–54 years		Mean for 55–64 years		Mean for 65–74 years		Mean for 75+ years
			M	F	M	F	M	F	M	F	M	F	M	F	M	F
	Physical functioning	55.7*	71.2#	50*	63.4#	68.2#	70.4#	49#	59.3#	44.7#	43.7#	52.1#	66.5 NS	35.9#	40.3*	32.5#
	Role-physical	34.7*	46.9#	35*	39.2#	53.5#	48.9#	25#	31.9#	30.7#	22.5#	26.2#	45.3 NS	15.2#	20.3#	28.4#
	Role-emotional	54.4*	66.6#	60.0 NS	60.6#	61.9 NS	58.3#	37.7#	50.3#	50.6#	43.3#	46.6#	62.5 NS	57.8*	28.8#	55.3*
	Social functioning	58.7*	60.3#	50#	63.0#	55.7#	65.6#	37.5#	58.6#	56.0#	53.7#	54.3#	64.0*	66.2#	55.8*	59.6*
	Mental health	65.4*	64.6#	68.0 NS	66.0#	64.5*	68.1*	53.6#	67.1#	61.4#	65.9#	57.4#	72.2 NS	70.8 NS	72.9 NS	66.6*
	Vitality	51.8*	57.3#	41.0#	56.5#	46.4#	55.4#	37.3#	54.3#	44.8#	50.5#	48.1#	54.3 NS	53.1 NS	53.2 NS	46.3*
	Bodily pain	53.5*	59.7#	50.8*	52.2#	60.5#	55.1#	46.9#	55.1#	45.1#	47.8#	52.7*	54.1*	55.8*	49.3*	62.5 NS
	General health	64.8*	68.6*	66.9 NS	70.9*	70.7 NS	71.6 NS	59.8#	64.6*	61.1#	56.1*	61.2 NS	67.5 NS	64.3 NS	58.2 NS	56.0 NS
	Physical component	38.7*	44.9#	36.6*	40.3#	43.6#	43.6#	37.6#	38.5#	35.2#	33.9#	37.6#	40.7 NS	31.7#	33.0*	33.4*
	Mental component	44.6*	43.8#	44.1 NS	45.8*	41.6 NS	45.2*	36.0#	45.2#	42.9#	45.0#	41.6*	46.8*	48.2 NS	45.2*	47.3*

Where significance is indicated in this table it refers only to the case of the sample mean being statistically significantly lower than the Australian norm. NS, not significant at the 5% level.

* Significant at the 5% level. # Significant at the 1% level.

Further analysis investigated the impact of relevant factors on the PCS score of the SF-36. Factors considered for their possible effect on PCS were: ISS, LOS, age, sex, whether or not the patient was in a tertiary hospital, private health insurance, external cause of injury, the location of the injury on the body and the mental component summary score (MCS). Factors significant in the univariate analysis or one-way ANOVA included ISS (p<0.001), LOS (p<0.001), age (p<0.001), sex (p=0.008), external cause (p<0.001) and body location (p<0.001). Investigative analysis indicated that LOS and ISS should be log-transformed. Sex and external cause of injury were omitted during the model building process as they failed to reach significance in the multivariate model; further examination of these two factors indicated confounding of sex with age and external cause with both age and ln_ISS. In the final model, ln_ISS, ln_LOS and age were all found to have a significantly negative impact on PCS (Table 2). The global test for body location also indicated a significant effect on PCS, and further investigation identified injuries to the head/neck/face, as well as abdomen, had significantly better adjusted mean PCS compared with upper and lower limb injuries.

Table 2. Multivariate model for effect on physical component summary score (n=302)

	Factor	Coefficient (95% CI)	p
	Constant	52.5 (48.38–56.62)	<0.001
	ln_ISS	−2.10 (−3.88 to −0.32)	0.021
	ln_LOS	−3.75 (−5.29 to −2.21)	<0.001
	Age	−0.09 (−0.16 to −0.03)	0.005
	Body location—global test		0.001
	Upper and lower limbs (n=211)	Reference
	Head/neck/face (n=27)	4.94 (0.78–9.10)	0.020
	Thorax (n=19)	4.12 (−0.86 to 9.10)	0.105
	Abdomen (n=10)	11.45 (4.86–18.05)	0.001
	Lower back (n=11)	−5.76 (−12.01 to 0.49)	0.071
	Pelvis (n=14)	−3.14 (−8.82 to 2.54)	0.278
	Multiple injuries (n=7)	−1.252 (−9.09 to 6.58)	0.753
	Body location not required (n=3)	−2.988 (−14.74 to 8.78)	0.618
	Adjusted R2	26.9

ln_ISS: natural log of ISS; ln_LOS: natural log of length of stay.

Discussion 

This study is one of the first in Australia using the short form 36 to assess the HRQoL of individuals hospitalised as a result of injury. The short form 36, a measure of adult health and well being that has been tested and standardised for Australian use, was chosen to obtain follow-up data from a cohort of adults admitted to 1 of the 12 Queensland Trauma Registry hospitals throughout the state. Results for the present study of HRQoL of adults following an injury, show support for the hypothesis that adults who have sustained an injury differ in health and well being compared with Australian norms.

The sample of injured people who participated in this study were spread across all adult age groups, with slightly higher numbers in the 25–34 and 45–54-year age groups and almost two-thirds of the sample were males. The most common causes of injury were falls and road traffic accidents. This demographic representation is not unlike the full sample recorded by the QTR, although the 45–54-year age group is slightly over-represented.

The participants in the present study were found to have a markedly lower HRQoL overall compared with the Australian normative population. This finding is consistent with the findings from other studies in Australia using the SF-36 to determine the HRQoL in patients who have sustained hip fractures.13, 33 It is also consistent with findings in international samples of injured people who have suffered from general injury,15, 16, 20 traumatic brain injury,10, 22, 30 amputation23, 24, 32 and orthopaedic injuries.4, 19, 27, 28, 29

Worthy of note is the fact that both the small group of participants who did not spend more than 24h in hospital (the Secondary Study Group), as well as the Primary Study Group of participants who were in hospital for 24h or more, demonstrated markedly lower HRQoL compared with the Australian population. This finding suggests that hospitalisation for either less than 24h or 24h and more results in a similar reduction in HRQoL when compared to the Australian norms.

The prevailing pattern of outcomes in patients following a traumatic injury, as represented by the SF-36, is one of reduction or impairment in every subscale measured by the instrument when compared with the Australian norms. This pattern generally holds for each sex, across age groups, and for age groups within each sex. The differences that occurred between sexes and age groups are generally differences of magnitude, rather than kind. Where there were exceptions to this, the scores of injured participants were generally lower than the Australian norms, but were not statistically significant. These exceptions tended to occur in psychosocial, rather than physical, subscales and generally only in the older age groups. This lack of difference in the older age groups may be a reflection of the fact that older people in the normative population also experience a reduced HRQoL. It may also reflect a difference in the type of injuries sustained by older trauma patients. As identified, age and external cause of injury were significantly associated. Generally, older patients experience less high impact injuries (such as road traffic crashes) and more low impact injuries (such as falls). This pattern of results deserves further investigation.

The largest impact of injury on HRQoL appeared to be on role reductions, both physical and social, in those aged between 35 and 64 years of age, while health perceptions and mental health scores, particularly in those aged 65 years or more, appeared to be closer to Australian norms. Further, females aged less than 35 years did not report statistically lower scores (when compared to Australian norms) in the areas of role—emotional, mental health, general health and the mental component summary score. The selective impact of injury on reduced HRQoL based on a relationship between age and sex, has not been noted by previous authors, although the importance of sex by itself has been noted.17 The impact of sex on outcome following major trauma requires further investigation.

A greater influence of injury on physical scores, rather than psychosocial scores, is also consistent with previous work. A small group of patients with open tibial fractures who had amputation or successful reconstruction of lower limbs reported scores lower than a healthy reference group in the physical functioning, role—physical and bodily pain subscales, but no other subscales.19 This pattern was also found in a previous study of 78 traumatic amputee patients.32 Similarly, patients with pelvic fractures reported a 14% lower PCS when compared to US norms, but only a 6% lower MCS,29 while female patients with pelvic or lower extremity injury reported significantly lower scores on all subscales except mental health.27 In a group of 20 adults with unstable ankle fractures, while five subscale scores were lower than US population scores approximately 4 months after the injury, the only subscale scores that remained significantly below the normative scores by 18 months post-injury were the physical function scores.28 Although most of the published literature has concentrated on orthopaedic patients the results are comparable to those found in this study as two-thirds of the sample experienced a fracture injury.

Limitations of this study include the assumption that participants were functioning well prior to injury, the low response rate and low proportion of participants with a major injury, the high proportion of participants with a fracture injury and the limited follow-up to only 3 months post injury.

Participants were assumed to be functioning well prior to injury and therefore that the reduction in scores at 3 months can be attributed to the injury. Pre-injury HRQoL data has not been collected retrospectively due to the recognised limited reliability of this method of data collection resulting from the potential for recall bias.11 It is recognised that data exists to indicate that injured people have a higher incidence of pre-existing morbidity than their non-injured counterparts,7 although no method currently exists to adapt normative HRQoL data to accommodate this difference.

The low response rate and small proportion of study participants (<10%) who were categorised as having experienced major injury limit the generalisability of results. It is possible that patients who did participate in the study were those who were experiencing poor HRQoL and were consequently motivated to participate in the study as a strategy to document this problem. Despite these concerns it should be noted that study participants had a similar age and ISS to the other eligible QTR patients during the enrolment period.

It is possible that any differences between participants and Australian norms would be greater if more seriously injured patients were included. The potential existed for the preferential impact of injury on physical, rather than psychosocial, scores to be different for those who experienced major injury, although there were insufficient major injury patients in this study to allow further analysis of this feature. A further limitation is the high proportion of study participants who experienced a fracture injury. While this represents the greatest group of injured people, it cannot be assumed that people with differing injuries, such as injuries to internal organs, experience similar reductions in HRQoL.

Finally, this study only extended to an average of 3 months post-hospitalisation for injury. It would be appropriate to extend the time frame to determine the length of compromise to HRQoL. In addition, no assessment has been made of the impact of any intervention, such as rehabilitation, on HRQoL and this should be incorporated into a larger study that allows analysis of multiple interrelated factors such as age, ongoing health care interventions and pre-injury HRQoL.

Conclusions 

In the majority of age groups the participants in this study reported lower HRQoL in the majority of subscales measured on the SF-36 approximately 3 months post-injury. The lowest scores, when compared to Australian norms, were reported in the physical, rather than psychosocial subscales. People who were injured and were admitted to hospital for less than 24h reported similarly reduced scores on the SF-36 when compared with Australian norms.

It is evident that significant reductions in HRQoL are apparent up to 3 months after hospitalisation for injury. This suggests that people who are injured experienced considerable burden for at least 3 months post-injury. Consequently, support through this period is necessary to assist return to the optimal HRQoL possible. Such HRQoL may be less than that experienced prior to injury, but should be optimised for each individual.

Conflict of interest 

There are no conflicts of interest by any authors in relation to this article.