ORTHOPAEDICS ANZJSurg.com
Determinants of time to surgery for patients with hip fracture Justin Zeltzer,*† Rebecca J. Mitchell,* Barbara Toson,* Ian A. Harris‡ and Jacqueline Close*§ *Falls and Injury Prevention Group, Neuroscience Research Australia, University of New South Wales, Randwick, New South Wales, Australia †NSW Ministry of Health, Sydney, New South Wales, Australia ‡South Western Sydney Clinical School, Whitlam Orthopaedic Research Centre, University of New South Wales, Randwick, New South Wales, Australia and §Prince of Wales Clinical School, University of New South Wales, Randwick, New South Wales, Australia
Key words accidental falls, co-morbidity, geriatrics, hip fractures, patient care management. Correspondence Dr Jacqueline Close, Falls and Injury Prevention Group, Neuroscience Research Australia, University of New South Wales, Barker Street, Randwick, NSW 2031, Australia. Email: [email protected] J. Zeltzer BCom (Econometrics), LLB; R. J. Mitchell MOHS, PhD; B. Toson BStatEcHons; I. A. Harris MBBS, PhD; J. Close MBBS, MD. Accepted for publication 10 April 2014. doi: 10.1111/ans.12671
Abstract Background: Guidelines for hip fracture care suggest that patients with hip fracture should undergo surgery on the day of or day after admission to hospital. This study examined factors affecting time to surgery for hip fracture extracted from existing administrative datasets in New South Wales (NSW), Australia. Method: A retrospective analysis of patients with hip fracture aged 65 years and over undergoing surgical intervention in NSW public hospitals between 1 July 2000 and 30 June 2011. A multinomial logistic model was used to identify factors impacting on time to surgery from 1 July 2006 to 30 June 2011. Results: A total of 49 317 hip fracture procedures were recorded during 2000–2001 to 2010–2011. Sixty-four per cent of patients received operative treatment on the day of or day after admission. Co-morbidity, type of surgical procedure and day of presentation all impacted significantly on time to surgery. Fourteen per cent required an inter-hospital transfer prior to receiving operative intervention. Transferred patients were 2.6 (95% confidence interval (CI): 2.31–2.85) times more likely to wait 2–4 days and 3.2 times more likely to wait 5 or more days (95% CI: 2.77–3.76) for surgery compared with patients presenting to an operating hospital. Conclusion: Significant variation exists between hospitals in the time to surgery that is not solely explained by measures of case mix or geography. Opportunities exist to consider other factors contributing to this variation and to ensure timely access to surgical intervention in the future.
Introduction Hip fracture is one of the most serious and costly fall-related injuries sustained by older people in the public health system. There were an estimated 17 000 hip fractures among Australians aged 65 years and older in 2008–2009, accounting for 21.6% of all fall injury hospitalizations1 and incurring an estimated AUD$579 million in direct hospital costs.2 In New Zealand there were 3803 hip fractures in 2007, costing an estimated NZD$105 million.3 The quality of care and outcomes following hip fracture have been shown to be dependent on a number of issues relating to structures, processes and service configurations, and these have been reflected in the recent National Institute for Health and Clinical Excellence (NICE) guidelines4 and quality indicators5 for hip fracture care in the United Kingdom (UK). The NICE Guideline recommends that ‘people with hip fracture should have surgery on the day of, or the © 2014 Royal Australasian College of Surgeons
day after, admission’. Measurement of this indicator is facilitated in the UK by a National Hip Fracture Database (NHFD) that reports hospital-specific time to surgery for approximately 59 000 patients with hip fracture annually.6 In the UK, performance is also financially incentivized with hospitals rewarded for delivering on a number of quality indicators for hip fracture care.6 In the absence of an equivalent registry, Australia is dependent on limited information on hip fracture care from existing administrative datasets which are usually jurisdiction-based. To obtain an accurate estimate of time to surgery, same patient records for the one care episode need to be linked, taking into account hospital transfers and service category changes. This study examines time to surgery for patients with hip fracture aged 65 years and older in New South Wales (NSW), Australia, and explores patient-related and external factors influencing this indicator using existing administrative datasets. ANZ J Surg 84 (2014) 633–638
634
Methods A retrospective analysis was conducted of patients with hip fracture aged 65 years and older undergoing operative intervention at one of the 37 NSW public hospitals, which operate on hip fracture. Data were obtained from the Admitted Patient Data Collection (APDC) and limited to records with dates of admission ranging from 1 July 2000 to 30 June 2011. Ethical approval was obtained from the NSW Population and Health Services Research Ethics Committee (2009/ 06/165).
Data source The APDC is an electronic administrative data collection retained by the NSW Ministry of Health, which contains information on patient demographics, source of referral, diagnoses, separation type and clinical procedures. Each health record refers to an individual episode of care in hospital, which ends with the discharge, transfer, or death of the patient, or when the service category for the admitted patient changes. The diagnoses and external cause codes are classified using the International Statistical Classification of Diseases and Related Health Problems, 10th Revision, Australian Modification (ICD-10-AM).7 The NSW Ministry of Health’s Local Health Districts (LHDs) were used to categorize hospitals as either metropolitan or rural/regional LHDs.8
Case identification Patients with hip fracture undergoing surgical intervention were identified using the principal diagnosis of hip fracture (ICD-10-AM: S72.0, S72.1 or S72.2) and the principal procedure code of: internal fixation of fracture of trochanteric or subcapital femur; hemi arthroplasty of fracture of subcapital femur; open reduction of fracture of femur with fixation; closed reduction of fracture of femur with fixation; total arthroplasty of hip; unilateral or total arthroplasty of hip, bilateral (ICD-10-AM: 47519-00, 47522-00, 47528-01, 47531-00, 49318-00, 49319-00, respectively). The pairing of the principal hip fracture diagnosis and the surgical procedures was conducted by electronically selecting cases that fit the criteria of having a principal diagnosis of a hip fracture and a corresponding hip fracture surgical procedure. This ensured that any data inconsistencies and hip procedures for other pathology were not included (n = 17 513).
Data linkage Data linkage of the APDC was conducted by the Centre for Health Record Linkage (CHeReL) using the Choicemaker software (Choicemaker Technologies, New York, NY, USA).9 In the absence of a unique patient identifier in NSW, CHeReL uses information including name, address, date of birth and gender to probabilistically link distinct records, allowing them to be identified as records for the same patient. Upper and lower probability cut-offs started at 0.75 and 0.25 for a linkage and were adjusted for each individual linkage to ensure false links were kept to a minimum. Record groups with probabilities in between the cut-offs were subject to clerical review.
Co-morbidity identification The Charlson co-morbidity index (CCI), with a 1-year look back period, was used as a surrogate for case mix/complexity.10 The CCI
Zeltzer et al.
is a weighted score based upon the presence of 17 predetermined conditions known to be associated with mortality over 1 year. The CCI adaptation for use with ICD-10-AM11 has been used in this paper for its proven validity in an older population.12,13 Individual items of the CCI were explored to look at their independent impact on time to surgery.
Time to surgery Time to surgery was estimated using calendar days elapsing between the day of admission of the first recorded hip fracture diagnosis and the day of procedure. The proportion of patients undergoing surgery on the day of or day after surgery was identified.
Data analysis and management All data analyses were performed using SAS Enterprise Guide v5.1 (SAS Institute, Cary, NC, USA).14 Descriptive analyses were conducted using all 11 years of data while the regression analyses used only the most recent 5 years of hospital separations (i.e. 1 July 2006 to 30 June 2011) to attempt to retain temporal relevance. To examine the association between rural/regional and metropolitan hospitals, a chi-square test of independence was used. Patient-related factors (age, sex, co-morbidity status) and external factors (day of admission, procedure type, starting journey in a non-operating hospital), and the effect on time to surgery were initially considered in univariate models. A multinomial logistic model incorporated all variables with a minimum significance of 20% from the univariate analyses using a backwards elimination process at 5% significance. Included in the multinomial logistic model were patient sex and co-morbidities, weekday of admission, procedure type and transfer status of patient. The proportion of hip fracture procedures undertaken on the day of or day after admission, adjusted for age, sex and co-morbidity status was calculated.
Results There were 49 317 hip fracture procedures recorded on patients aged 65 years and older in NSW public hospitals between 1 July 2000 and 30 June 2011. Women accounted for 36 639 (74.3%) of procedures with a median age of 85.0 years (standard deviation (SD) 7.1) compared with 83.0 years (SD 7.3) for men. There were 6946 patients (14.1%) transferred between hospitals prior to undergoing operative intervention, and this percentage was higher for operating hospitals in rural/regional LHDs (24.7%) than hospitals in metropolitan LHDs (7.8%) (χ2 = 2730, degrees of freedom = 1, P < 0.0001).
Time to surgery A quarter (25.2%) of patients underwent surgery on the day of their initial admission, and a further 39.2% on the day following admission. The proportion of patients with hip fracture undergoing surgical intervention within the first 2 calendar days of admission has declined over the whole period with an estimated annual percent change (EAPC) of –0.63 (95% CI: 1.52–0.27) (P = 0.15). The period 1 July 2000 to 30 June 2004 was characterized by a significant decline in the proportion of patients undergoing surgery within 2 calendar days (EAPC = –4.76 (95% CI: –7.57 to –1.87), P = 0.007) © 2014 Royal Australasian College of Surgeons
Time to surgery for hip fracture
Fig. 1. Percentage of patients with hip fracture undergoing surgery on the day of or day after admission by year, NSW public hospitals, 1 July 2000 to 30 June 2011. Number of procedures performed (n) and 95% confidence intervals shown for estimates.
635
75%
70%
65%
60%
55%
2000–01 2001–02 2002–03 2003–04 2004–05 2005–06 2006–07 2007–08 2008–09 2009–10 2010–11 (n = 4098) (n = 4379) (n = 4421) (n = 4353) (n = 4310) (n = 4348) (n = 4633) (n = 4595) (n = 4524) (n = 4812) (n = 4844)
while over subsequent years, there has been a non-significant increase in this figure (EAPC = 0.8 (95% CI: –0.07–1.68), P = 0.065) (Fig. 1).
Patient-related factors Women were more likely to undergo earlier surgical intervention than men (odds ratio (OR) 0.93; 95% CI: 0.87–1.00 for 2–4 days and OR 0.89; 95% CI: 0.79–1.00 for 5 or more days). An increasing number of co-morbidities was associated with longer times to surgery. A number of specific diseases were independently associated with time to surgery of which the presence of congestive cardiac failure was the most commonly reported (Table 1).
External factors Weekday of admission had a significant impact on time to surgery. Patients admitted on a weekend were more likely to be operated on within 2 days than 2–4 days (OR 0.8; 95% CI: 0.81–0.93) or 5 or more days (OR 0.77; 95% CI: 0.68–0.87) than if they had been admitted on a weekday. Interhospital transfer had a significant impact on increasing time to surgery; 2–4 days (OR 2.57; 95% CI: 2.31–2.85) or 5 or more days (OR 3.23; 95% CI: 2.77–3.76) when compared with patients admitted directly to an operating hospital. Patients undergoing fracture reduction and internal fixation were more likely to be undergo surgery earlier when compared with patients undergoing arthroplasty (Table 1). There was considerable variation among hospitals in the proportion of patients undergoing surgery on the day of or day after admission even after adjusting for patient-related factors (age, sex, individual Charlson co-morbidities) and transfer status (Fig. 2).
Discussion This is one of the first Australian studies to examine factors associated with variation in time to surgery for hip fracture using population-based, linked hospital admission data. It has shown that almost two-thirds of patients admitted to a public hospital in NSW © 2014 Royal Australasian College of Surgeons
with a hip fracture underwent surgery on the day of or day after admission. These results are consistent with the international literature where the percentage of people operated on within 48 h ranges between 50 and 87%.15–18 However, it should be remembered that this study examined calendar days as opposed to hours to surgery, and as such will underestimate the true proportion of patients undergoing surgery within 48 h. In the UK, 84.1% of patients with hip fracture undergo surgery within 48 h.6 Caution is required when comparing time to surgery in NSW, Australia to that of the UK for a number of reasons, not least of which is the geography of NSW. In NSW, a number of patients will start their hospital journey in a rural/regional hospital not resourced to provide operative intervention, and so require transport to another facility. Our results highlight the significant impact interhospital transfer has on time to surgery and that it is important to include transfer time in calculation of time to surgery to provide an accurate picture of the true time to surgery from the initial presentation to a hospital with a hip fracture. Of the available patient-related factors, co-morbidity was the most strongly associated with time to surgery and particularly so for patients with cardiac, liver, renal and respiratory disease. Optimizing patients for surgery is critical to outcomes and delaying surgery to address modifiable factors in some patients is clinically appropriate.18–20 The absence of a significant association between age and time to surgery is consistent with the findings of several other studies examining the causes of operative delay.17,18 Gender was also not associated with time to surgery and, while the absolute number and rate of hip fracture in men were less, it is recognized that men tend to have more cardiac disease, including undiagnosed cardiac disease, and do less well than women in relation to mediumand longer-term outcomes.21 Interestingly, time to surgery appears to be shorter if the hospital admission occurs during the weekend. This may reflect issues with access to theatre, and in the absence of regularly scheduled weekday trauma lists, access to theatres at weekend may be easier in some hospitals. However, this is also of concern given that most hospitals
636
Zeltzer et al.
Table 1 Patient-related and external determinants of time to surgery: univariate and multivariate multinomial logistic regression odds ratios (95% confidence intervals), NSW public hospitals, 1 July 2006 to 30 June 2011 Univariate multinomial logistic regression
Patient-related factors Sex Male Female Age group (years)‡ 65–75 75–85 +85 Charlson index§ 0 1,2 +3 Co-morbidities¶ Acute myocardial infarction Congestive heart failure Peripheral vascular disease Cerebral vascular accident Dementia Pulmonary disease Connective tissue disorder Peptic ulcer Liver disease Diabetes Diabetes complications Paraplegia Renal disease Cancer Metastatic cancer External factors†† Weekday of admission Monday Tuesday Wednesday Thursday Friday Saturday Sunday Procedure type Hemi-arthroplasty Total arthroplasty Internal fixation (including open or closed reduction) Transfer status of patient Not transferred Transferred Rural versus Metro Metro Rural
Multivariate multinomial logistic regression
Total
2–4 days†
5 or more days†
2–4 days†
5 or more days†
(n = 23 400)
(n = 6576)
(n = 1826)
(n = 6576)
(n = 1826)
6259 17 141
1 0.89 (0.83–0.95)
1 0.77 (0.69–0.85)
1 0.93 (0.87–1.00)
1 0.89 (0.79–1.00)
2825 9000 11 575
1 1.00 (0.91–1.10) 1.04 (0.89–1.22)
1 0.97 (0.88–1.06) 0.95 (0.81–1.11)
— — —
— — —
13 198 4512 5690
1 1.19 (1.11–1.29) 1.45 (1.35–1.55)
1 1.72 (1.51–1.96) 2.65 (2.37–2.97)
— — —
— — —
1791 2599 242 1748 2422 1943 271 221 88 972 1091 906 2158 902 186
1.16 (1.03–1.29) 1.6 (1.46–1.75) 1.18 (0.88–1.56) 1.37 (1.23–1.53) 1.07 (0.98–1.18) 1.31 (1.18–1.45) 1.29 (0.99–1.69) 1.12 (0.83–1.52) 1.51 (0.94–2.41) 1.12 (0.97–1.30) 1.17 (1.02–1.34) 1.55 (1.35–1.80) 1.39 (1.26–1.54) 1.07 (0.92–1.25) 1.13 (0.88–1.45)
2.28 2.87 1.77 1.68 1.04 2.09 1.80 1.94 2.82 1.46 1.74 1.91 2.31 1.37 1.88
(1.97–2.63) (2.53–3.25) (1.19–2.64) (1.43–1.98) (0.89–1.22) (1.80–2.42) (1.23–2.63) (1.30–2.90) (1.56–5.07) (1.18–1.81) (1.43–2.12) (1.54–2.37) (2.01–2.65) (1.09–1.72) (1.34–2.63)
1.11 (0.84–1.45) 1.56 (1.41–1.73) — 1.19 (1.02–1.39) — 1.18 (1.05–1.33) 1.35 (1.01–1.80) — 1.54 (0.93–2.55) — — 1.27 (1.03–1.56) 1.21 (1.09–1.36) — 1.11 (0.84–1.45)
1.79 (1.24–2.59) 2.27 (1.96–2.63) — 1.29 (1.01–1.64) — 1.60 (1.36–1.89) 1.78 (1.19–2.69) — 2.39 (1.25–4.58) — — 1.35 (0.98–1.85) 1.59 (1.36–1.86) — 1.79 (1.24–2.59)
3416 3393 3471 3464 3539 3205 2909
0.88 0.93 0.93 0.82 0.82 0.77
1 (0.80–0.98) (0.84–1.04) (0.83–1.03) (0.73–0.91) (0.73–0.91) (0.69–0.87)
0.97 1.43 1.06 0.99 1.00 0.65
1 (0.81–1.17) (1.20–1.70) (0.88–1.27) (0.82–1.18) (0.83–1.21) (0.53–0.80)
0.88 0.92 0.93 0.81 0.79 0.79
1 (0.79–0.99) (0.82–1.04) (0.83–1.04) (0.72–0.90) (0.70–0.88) (0.70–0.89)
0.96 1.39 1.04 0.96 0.96 0.67
1 (0.79–1.17) (1.16–1.67) (0.86–1.26) (0.80–1.17) (0.79–1.16) (0.54–0.83)
7064 1047 15 289
1 0.85 (0.73–0.98) 0.74 (0.73–0.98)
1 0.88 (0.70–1.11) 0.66 (0.59–0.73)
1 1.01 (0.86–1.18) 0.74 (0.69–0.80)
1 1.30 (1.01–1.67) 0.66 (0.59–0.74)
20 257 3143
1 2.16 (1.99–2.34)
1 2.57 (2.27–2.90)
1 2.57 (2.31–2.85)
1 3.23 (2.77–3.76)
15 413 7987
1 1.35 (1.27–1.43)
1 1.40 (1.26–1.56)
— —
— —
†Reference category was 0–1 day (n = 14 998). ‡Severe liver disease and AIDS were not presented because of small numbers (cell number
Determinants of time to surgery for patients with hip fracture Justin Zeltzer,*† Rebecca J. Mitchell,* Barbara Toson,* Ian A. Harris‡ and Jacqueline Close*§ *Falls and Injury Prevention Group, Neuroscience Research Australia, University of New South Wales, Randwick, New South Wales, Australia †NSW Ministry of Health, Sydney, New South Wales, Australia ‡South Western Sydney Clinical School, Whitlam Orthopaedic Research Centre, University of New South Wales, Randwick, New South Wales, Australia and §Prince of Wales Clinical School, University of New South Wales, Randwick, New South Wales, Australia
Key words accidental falls, co-morbidity, geriatrics, hip fractures, patient care management. Correspondence Dr Jacqueline Close, Falls and Injury Prevention Group, Neuroscience Research Australia, University of New South Wales, Barker Street, Randwick, NSW 2031, Australia. Email: [email protected] J. Zeltzer BCom (Econometrics), LLB; R. J. Mitchell MOHS, PhD; B. Toson BStatEcHons; I. A. Harris MBBS, PhD; J. Close MBBS, MD. Accepted for publication 10 April 2014. doi: 10.1111/ans.12671
Abstract Background: Guidelines for hip fracture care suggest that patients with hip fracture should undergo surgery on the day of or day after admission to hospital. This study examined factors affecting time to surgery for hip fracture extracted from existing administrative datasets in New South Wales (NSW), Australia. Method: A retrospective analysis of patients with hip fracture aged 65 years and over undergoing surgical intervention in NSW public hospitals between 1 July 2000 and 30 June 2011. A multinomial logistic model was used to identify factors impacting on time to surgery from 1 July 2006 to 30 June 2011. Results: A total of 49 317 hip fracture procedures were recorded during 2000–2001 to 2010–2011. Sixty-four per cent of patients received operative treatment on the day of or day after admission. Co-morbidity, type of surgical procedure and day of presentation all impacted significantly on time to surgery. Fourteen per cent required an inter-hospital transfer prior to receiving operative intervention. Transferred patients were 2.6 (95% confidence interval (CI): 2.31–2.85) times more likely to wait 2–4 days and 3.2 times more likely to wait 5 or more days (95% CI: 2.77–3.76) for surgery compared with patients presenting to an operating hospital. Conclusion: Significant variation exists between hospitals in the time to surgery that is not solely explained by measures of case mix or geography. Opportunities exist to consider other factors contributing to this variation and to ensure timely access to surgical intervention in the future.
Introduction Hip fracture is one of the most serious and costly fall-related injuries sustained by older people in the public health system. There were an estimated 17 000 hip fractures among Australians aged 65 years and older in 2008–2009, accounting for 21.6% of all fall injury hospitalizations1 and incurring an estimated AUD$579 million in direct hospital costs.2 In New Zealand there were 3803 hip fractures in 2007, costing an estimated NZD$105 million.3 The quality of care and outcomes following hip fracture have been shown to be dependent on a number of issues relating to structures, processes and service configurations, and these have been reflected in the recent National Institute for Health and Clinical Excellence (NICE) guidelines4 and quality indicators5 for hip fracture care in the United Kingdom (UK). The NICE Guideline recommends that ‘people with hip fracture should have surgery on the day of, or the © 2014 Royal Australasian College of Surgeons
day after, admission’. Measurement of this indicator is facilitated in the UK by a National Hip Fracture Database (NHFD) that reports hospital-specific time to surgery for approximately 59 000 patients with hip fracture annually.6 In the UK, performance is also financially incentivized with hospitals rewarded for delivering on a number of quality indicators for hip fracture care.6 In the absence of an equivalent registry, Australia is dependent on limited information on hip fracture care from existing administrative datasets which are usually jurisdiction-based. To obtain an accurate estimate of time to surgery, same patient records for the one care episode need to be linked, taking into account hospital transfers and service category changes. This study examines time to surgery for patients with hip fracture aged 65 years and older in New South Wales (NSW), Australia, and explores patient-related and external factors influencing this indicator using existing administrative datasets. ANZ J Surg 84 (2014) 633–638
634
Methods A retrospective analysis was conducted of patients with hip fracture aged 65 years and older undergoing operative intervention at one of the 37 NSW public hospitals, which operate on hip fracture. Data were obtained from the Admitted Patient Data Collection (APDC) and limited to records with dates of admission ranging from 1 July 2000 to 30 June 2011. Ethical approval was obtained from the NSW Population and Health Services Research Ethics Committee (2009/ 06/165).
Data source The APDC is an electronic administrative data collection retained by the NSW Ministry of Health, which contains information on patient demographics, source of referral, diagnoses, separation type and clinical procedures. Each health record refers to an individual episode of care in hospital, which ends with the discharge, transfer, or death of the patient, or when the service category for the admitted patient changes. The diagnoses and external cause codes are classified using the International Statistical Classification of Diseases and Related Health Problems, 10th Revision, Australian Modification (ICD-10-AM).7 The NSW Ministry of Health’s Local Health Districts (LHDs) were used to categorize hospitals as either metropolitan or rural/regional LHDs.8
Case identification Patients with hip fracture undergoing surgical intervention were identified using the principal diagnosis of hip fracture (ICD-10-AM: S72.0, S72.1 or S72.2) and the principal procedure code of: internal fixation of fracture of trochanteric or subcapital femur; hemi arthroplasty of fracture of subcapital femur; open reduction of fracture of femur with fixation; closed reduction of fracture of femur with fixation; total arthroplasty of hip; unilateral or total arthroplasty of hip, bilateral (ICD-10-AM: 47519-00, 47522-00, 47528-01, 47531-00, 49318-00, 49319-00, respectively). The pairing of the principal hip fracture diagnosis and the surgical procedures was conducted by electronically selecting cases that fit the criteria of having a principal diagnosis of a hip fracture and a corresponding hip fracture surgical procedure. This ensured that any data inconsistencies and hip procedures for other pathology were not included (n = 17 513).
Data linkage Data linkage of the APDC was conducted by the Centre for Health Record Linkage (CHeReL) using the Choicemaker software (Choicemaker Technologies, New York, NY, USA).9 In the absence of a unique patient identifier in NSW, CHeReL uses information including name, address, date of birth and gender to probabilistically link distinct records, allowing them to be identified as records for the same patient. Upper and lower probability cut-offs started at 0.75 and 0.25 for a linkage and were adjusted for each individual linkage to ensure false links were kept to a minimum. Record groups with probabilities in between the cut-offs were subject to clerical review.
Co-morbidity identification The Charlson co-morbidity index (CCI), with a 1-year look back period, was used as a surrogate for case mix/complexity.10 The CCI
Zeltzer et al.
is a weighted score based upon the presence of 17 predetermined conditions known to be associated with mortality over 1 year. The CCI adaptation for use with ICD-10-AM11 has been used in this paper for its proven validity in an older population.12,13 Individual items of the CCI were explored to look at their independent impact on time to surgery.
Time to surgery Time to surgery was estimated using calendar days elapsing between the day of admission of the first recorded hip fracture diagnosis and the day of procedure. The proportion of patients undergoing surgery on the day of or day after surgery was identified.
Data analysis and management All data analyses were performed using SAS Enterprise Guide v5.1 (SAS Institute, Cary, NC, USA).14 Descriptive analyses were conducted using all 11 years of data while the regression analyses used only the most recent 5 years of hospital separations (i.e. 1 July 2006 to 30 June 2011) to attempt to retain temporal relevance. To examine the association between rural/regional and metropolitan hospitals, a chi-square test of independence was used. Patient-related factors (age, sex, co-morbidity status) and external factors (day of admission, procedure type, starting journey in a non-operating hospital), and the effect on time to surgery were initially considered in univariate models. A multinomial logistic model incorporated all variables with a minimum significance of 20% from the univariate analyses using a backwards elimination process at 5% significance. Included in the multinomial logistic model were patient sex and co-morbidities, weekday of admission, procedure type and transfer status of patient. The proportion of hip fracture procedures undertaken on the day of or day after admission, adjusted for age, sex and co-morbidity status was calculated.
Results There were 49 317 hip fracture procedures recorded on patients aged 65 years and older in NSW public hospitals between 1 July 2000 and 30 June 2011. Women accounted for 36 639 (74.3%) of procedures with a median age of 85.0 years (standard deviation (SD) 7.1) compared with 83.0 years (SD 7.3) for men. There were 6946 patients (14.1%) transferred between hospitals prior to undergoing operative intervention, and this percentage was higher for operating hospitals in rural/regional LHDs (24.7%) than hospitals in metropolitan LHDs (7.8%) (χ2 = 2730, degrees of freedom = 1, P < 0.0001).
Time to surgery A quarter (25.2%) of patients underwent surgery on the day of their initial admission, and a further 39.2% on the day following admission. The proportion of patients with hip fracture undergoing surgical intervention within the first 2 calendar days of admission has declined over the whole period with an estimated annual percent change (EAPC) of –0.63 (95% CI: 1.52–0.27) (P = 0.15). The period 1 July 2000 to 30 June 2004 was characterized by a significant decline in the proportion of patients undergoing surgery within 2 calendar days (EAPC = –4.76 (95% CI: –7.57 to –1.87), P = 0.007) © 2014 Royal Australasian College of Surgeons
Time to surgery for hip fracture
Fig. 1. Percentage of patients with hip fracture undergoing surgery on the day of or day after admission by year, NSW public hospitals, 1 July 2000 to 30 June 2011. Number of procedures performed (n) and 95% confidence intervals shown for estimates.
635
75%
70%
65%
60%
55%
2000–01 2001–02 2002–03 2003–04 2004–05 2005–06 2006–07 2007–08 2008–09 2009–10 2010–11 (n = 4098) (n = 4379) (n = 4421) (n = 4353) (n = 4310) (n = 4348) (n = 4633) (n = 4595) (n = 4524) (n = 4812) (n = 4844)
while over subsequent years, there has been a non-significant increase in this figure (EAPC = 0.8 (95% CI: –0.07–1.68), P = 0.065) (Fig. 1).
Patient-related factors Women were more likely to undergo earlier surgical intervention than men (odds ratio (OR) 0.93; 95% CI: 0.87–1.00 for 2–4 days and OR 0.89; 95% CI: 0.79–1.00 for 5 or more days). An increasing number of co-morbidities was associated with longer times to surgery. A number of specific diseases were independently associated with time to surgery of which the presence of congestive cardiac failure was the most commonly reported (Table 1).
External factors Weekday of admission had a significant impact on time to surgery. Patients admitted on a weekend were more likely to be operated on within 2 days than 2–4 days (OR 0.8; 95% CI: 0.81–0.93) or 5 or more days (OR 0.77; 95% CI: 0.68–0.87) than if they had been admitted on a weekday. Interhospital transfer had a significant impact on increasing time to surgery; 2–4 days (OR 2.57; 95% CI: 2.31–2.85) or 5 or more days (OR 3.23; 95% CI: 2.77–3.76) when compared with patients admitted directly to an operating hospital. Patients undergoing fracture reduction and internal fixation were more likely to be undergo surgery earlier when compared with patients undergoing arthroplasty (Table 1). There was considerable variation among hospitals in the proportion of patients undergoing surgery on the day of or day after admission even after adjusting for patient-related factors (age, sex, individual Charlson co-morbidities) and transfer status (Fig. 2).
Discussion This is one of the first Australian studies to examine factors associated with variation in time to surgery for hip fracture using population-based, linked hospital admission data. It has shown that almost two-thirds of patients admitted to a public hospital in NSW © 2014 Royal Australasian College of Surgeons
with a hip fracture underwent surgery on the day of or day after admission. These results are consistent with the international literature where the percentage of people operated on within 48 h ranges between 50 and 87%.15–18 However, it should be remembered that this study examined calendar days as opposed to hours to surgery, and as such will underestimate the true proportion of patients undergoing surgery within 48 h. In the UK, 84.1% of patients with hip fracture undergo surgery within 48 h.6 Caution is required when comparing time to surgery in NSW, Australia to that of the UK for a number of reasons, not least of which is the geography of NSW. In NSW, a number of patients will start their hospital journey in a rural/regional hospital not resourced to provide operative intervention, and so require transport to another facility. Our results highlight the significant impact interhospital transfer has on time to surgery and that it is important to include transfer time in calculation of time to surgery to provide an accurate picture of the true time to surgery from the initial presentation to a hospital with a hip fracture. Of the available patient-related factors, co-morbidity was the most strongly associated with time to surgery and particularly so for patients with cardiac, liver, renal and respiratory disease. Optimizing patients for surgery is critical to outcomes and delaying surgery to address modifiable factors in some patients is clinically appropriate.18–20 The absence of a significant association between age and time to surgery is consistent with the findings of several other studies examining the causes of operative delay.17,18 Gender was also not associated with time to surgery and, while the absolute number and rate of hip fracture in men were less, it is recognized that men tend to have more cardiac disease, including undiagnosed cardiac disease, and do less well than women in relation to mediumand longer-term outcomes.21 Interestingly, time to surgery appears to be shorter if the hospital admission occurs during the weekend. This may reflect issues with access to theatre, and in the absence of regularly scheduled weekday trauma lists, access to theatres at weekend may be easier in some hospitals. However, this is also of concern given that most hospitals
636
Zeltzer et al.
Table 1 Patient-related and external determinants of time to surgery: univariate and multivariate multinomial logistic regression odds ratios (95% confidence intervals), NSW public hospitals, 1 July 2006 to 30 June 2011 Univariate multinomial logistic regression
Patient-related factors Sex Male Female Age group (years)‡ 65–75 75–85 +85 Charlson index§ 0 1,2 +3 Co-morbidities¶ Acute myocardial infarction Congestive heart failure Peripheral vascular disease Cerebral vascular accident Dementia Pulmonary disease Connective tissue disorder Peptic ulcer Liver disease Diabetes Diabetes complications Paraplegia Renal disease Cancer Metastatic cancer External factors†† Weekday of admission Monday Tuesday Wednesday Thursday Friday Saturday Sunday Procedure type Hemi-arthroplasty Total arthroplasty Internal fixation (including open or closed reduction) Transfer status of patient Not transferred Transferred Rural versus Metro Metro Rural
Multivariate multinomial logistic regression
Total
2–4 days†
5 or more days†
2–4 days†
5 or more days†
(n = 23 400)
(n = 6576)
(n = 1826)
(n = 6576)
(n = 1826)
6259 17 141
1 0.89 (0.83–0.95)
1 0.77 (0.69–0.85)
1 0.93 (0.87–1.00)
1 0.89 (0.79–1.00)
2825 9000 11 575
1 1.00 (0.91–1.10) 1.04 (0.89–1.22)
1 0.97 (0.88–1.06) 0.95 (0.81–1.11)
— — —
— — —
13 198 4512 5690
1 1.19 (1.11–1.29) 1.45 (1.35–1.55)
1 1.72 (1.51–1.96) 2.65 (2.37–2.97)
— — —
— — —
1791 2599 242 1748 2422 1943 271 221 88 972 1091 906 2158 902 186
1.16 (1.03–1.29) 1.6 (1.46–1.75) 1.18 (0.88–1.56) 1.37 (1.23–1.53) 1.07 (0.98–1.18) 1.31 (1.18–1.45) 1.29 (0.99–1.69) 1.12 (0.83–1.52) 1.51 (0.94–2.41) 1.12 (0.97–1.30) 1.17 (1.02–1.34) 1.55 (1.35–1.80) 1.39 (1.26–1.54) 1.07 (0.92–1.25) 1.13 (0.88–1.45)
2.28 2.87 1.77 1.68 1.04 2.09 1.80 1.94 2.82 1.46 1.74 1.91 2.31 1.37 1.88
(1.97–2.63) (2.53–3.25) (1.19–2.64) (1.43–1.98) (0.89–1.22) (1.80–2.42) (1.23–2.63) (1.30–2.90) (1.56–5.07) (1.18–1.81) (1.43–2.12) (1.54–2.37) (2.01–2.65) (1.09–1.72) (1.34–2.63)
1.11 (0.84–1.45) 1.56 (1.41–1.73) — 1.19 (1.02–1.39) — 1.18 (1.05–1.33) 1.35 (1.01–1.80) — 1.54 (0.93–2.55) — — 1.27 (1.03–1.56) 1.21 (1.09–1.36) — 1.11 (0.84–1.45)
1.79 (1.24–2.59) 2.27 (1.96–2.63) — 1.29 (1.01–1.64) — 1.60 (1.36–1.89) 1.78 (1.19–2.69) — 2.39 (1.25–4.58) — — 1.35 (0.98–1.85) 1.59 (1.36–1.86) — 1.79 (1.24–2.59)
3416 3393 3471 3464 3539 3205 2909
0.88 0.93 0.93 0.82 0.82 0.77
1 (0.80–0.98) (0.84–1.04) (0.83–1.03) (0.73–0.91) (0.73–0.91) (0.69–0.87)
0.97 1.43 1.06 0.99 1.00 0.65
1 (0.81–1.17) (1.20–1.70) (0.88–1.27) (0.82–1.18) (0.83–1.21) (0.53–0.80)
0.88 0.92 0.93 0.81 0.79 0.79
1 (0.79–0.99) (0.82–1.04) (0.83–1.04) (0.72–0.90) (0.70–0.88) (0.70–0.89)
0.96 1.39 1.04 0.96 0.96 0.67
1 (0.79–1.17) (1.16–1.67) (0.86–1.26) (0.80–1.17) (0.79–1.16) (0.54–0.83)
7064 1047 15 289
1 0.85 (0.73–0.98) 0.74 (0.73–0.98)
1 0.88 (0.70–1.11) 0.66 (0.59–0.73)
1 1.01 (0.86–1.18) 0.74 (0.69–0.80)
1 1.30 (1.01–1.67) 0.66 (0.59–0.74)
20 257 3143
1 2.16 (1.99–2.34)
1 2.57 (2.27–2.90)
1 2.57 (2.31–2.85)
1 3.23 (2.77–3.76)
15 413 7987
1 1.35 (1.27–1.43)
1 1.40 (1.26–1.56)
— —
— —
†Reference category was 0–1 day (n = 14 998). ‡Severe liver disease and AIDS were not presented because of small numbers (cell number
