We report gender differences in the epidemiology and outcome after hip fracture from the Scottish Hip Fracture Audit, with data on admission and at 120 days follow-up from 22 orthopaedic units across the country between 1998 and 2005. Outcome measures included early mortality, length of hospital stay, 120-day residence and mobility. A multivariate logistic regression model compared outcomes between genders. The study comprised 25 649 patients of whom 5674 (22%) were men and 19 975 (78%) were women. The men were in poorer pre-operative health, despite being younger at presentation (mean 77 years (60 to 101) vs 81 years (50 to 106)). Pre-fracture residence and mobility were similar between genders.
Multivariate analysis indicated that the men were less likely to return to their home or mobilise independently at the 120-day follow-up. Mortality at 30 and 120 days was higher for men, even after differences in case-mix variables between genders were considered.
Approximately one-quarter of patients who sustain a hip fracture are men1 and they are reported to have higher associated morbidity and mortality than age-matched women.2–5 However, several authors have reported that when differences in pre-operative health are considered, men do not have increased early mortality after hip fracture.6–8 The aim of this study was to use the Scottish Hip Fracture Audit to establish gender differences in case-mix variables and correlate them with outcome after surgery for fracture of the hip.
Patients and Methods
The Scottish Hip Fracture Audit collects data on patients over 50 years old who are admitted to hospital with a hip fracture.1 A core dataset includes case-mix, process and outcome during the acute stay, at 120 days after this, and on any re-operations within four months of the index procedure. Review data collection at 120 days is by telephone, or using postal questionnaires when necessary.
Data collection is by locally funded, dedicated audit coordinators. The central team is funded by the Information Services Division of National Services Scotland and comprises a Clinical Coordinator, a Statistician and a Data Coordinator. Complete and accurate data are achieved by standardised collection procedures, subcontracted dual data entry and a monthly ongoing validation process.
Statistical analysis uses SPSS version 13.0 (SPSS Inc., Chicago, Illinois). The variables studied were in two groups, namely case mix (patient characteristics) and outcome measures. The former includes age, gender, pre-fracture mobility and residence, medical co-morbidity (American Society of Anesthesiologists (ASA) score) and fracture pattern. Outcome measures include 30- and 120-day mortality, duration of hospital stay, and residence and mobility at 120 days. Initial multivariate analysis using the Mann-Whitney U and chi-squared tests was performed to compare differences in case-mix variables and outcome measures. A p-value < 0.05 was considered significant. The independent effect of gender on outcome was examined by offering gender to multivariate regression models already containing the other case-mix factors.
Data were collected from 22 acute orthopaedic units between January 1998 and December 2005. Men were more likely to undergo non-operative management than women (5% vs 3%, chi-squared test p < 0.001). Overall, 25 649 patients, 5674 (22%) men and 19 975 (78%) women were treated surgically with a ratio of women-to-men of 3.5:1. Figure 1⇓ documents the number of fractures per five-year age cohort for both genders. The women-to-men ratio varies significantly with age. Between 50 and 54 years the ratio is similar, but between 85 and 89 years, women outnumber men by 5:1.
Significant differences in case-mix variables were observed between genders (Table I⇓). Men had a younger mean age at presentation of 77 years (60 to 101) compared with 81 years (50 to 106) for women. Despite this, men were more likely to have a higher ASA score, indicating greater pre-fracture co-morbidity.
Men were more likely to be living in their own homes than in institutional care at the time of injury (chi-squared test, p = 0.001). The percentage of patients who were independently mobile prior to fracture was similar between groups (men 45% (2564 of 5674) vs women 44% (8735 of 19 975), chi-squared test, p = 0.051).
Men were more likely to have sustained a fracture secondary to a neoplastic deposit (men 3% (171 of 5674) vs women 1.3% (274 of 19 975), chi squared test, p < 0.001).
Men were less likely to survive to 30 or 120 days after admission (Table II⇓). Even after multivariate analysis to allow for the effects of differences in case-mix variables between groups, mortality at 30 and 120 days remained higher for men (odds ratio (OR) 1.93 (95% confidence interval (CI) 1.73 to 2.14) and 1.98 (95% CI 1.84 to 2.14); Table III⇓). Figure 2⇓ shows the 120-day survival by five-year age cohort for each gender. For patients aged 50 to 59 years, men had improved survival, however, from 60 years and above women had significantly improved survival.
Among patients living in their own homes prior to fracture, univariate analysis suggested that men were more likely than women to have returned to their previous residence by 120 days, although the difference was small (men 81% (2515 of 3100) vs women 80% (9021 of 11 344), chi-squared test, p = 0.043). However, when we adjusted for differences in case-mix variables, women were more likely to return to their own home by 120 days (OR 1.19 (95% CI 1.06 to 1.32), Tables II⇑ and III⇑). Similarly, among patients who walked unaided and unaccompanied immediately prior to fracture, univariate analysis suggested that there was no significant differences in case-mix variables, women were more likely to return to their own home by 120 days (OR = 1.25, (95% CI 1.11 to 1.41), Tables II⇑ and III⇑).
Despite differences in age and pre-fracture ASA score, the median hospital stay at 120 days was 23 days for both men and women, excluding those who died within 120 days (Mann-Whitney U test p = 0.76).
In this study, significant differences were observed in case-mix variables and outcome between gender groups.
Men who sustain a hip fracture have been reported to have a higher associated mortality than age-matched women.2–5 However, other patient characteristics may act as confounding variables, and several authors have challenged the notion that gender as an isolated variable affects the rate of mortality after hip fracture.6–8 Mortality after hip fracture increases with age.6 Although men have a lower mean age at injury compared with women (77 years vs 81 years) they had reduced survival at 120 days post-fracture (27% vs 18%, Table II⇑). Patients with a high ASA score have greater post-operative mortality after surgery for hip fracture.8,9 In this study, men were more likely than women to be ASA grade 3 or above. When survival outcome is analysed, using multivariate regression models to control for the effects of case-mix variables, women continue to be more likely to survive both to 30 days (OR = 1.93) and 120 days (OR = 1.98) than men, confirming the results of the univariate analysis.
It is reported that those with a hip fracture have a higher age and gender-adjusted rate of mortality than the general population.4,5 Using the mean age for each gender group at the time of fracture, government actuarial data indicate that the annual age- and gender-adjusted mortality for men in the general population is 6.6%, compared with 6.4% for women.10 Therefore, even at four months, the mortality rate after hip fracture for men (27%) and women (18%) is significantly higher (p < 0.001) than in annual age- and gender-matched uninjured populations.
Where a patient lives after discharge is an increasingly important measure of outcome, owing to its effect on social function and the associated cost of care.11–13 Patients admitted from home represent an important group because a principal aim of surgery is to return them to independence. We have focused on this group because patients admitted from institutional care are rarely discharged to an independent residence. At the time of injury 67% of men and 65% of women lived independently at home (Table I⇑). Univariate analysis suggested that men admitted from home were more likely to return by 120 days, although the difference was small (81% vs 80%, p = 0.043). However, differences in case-mix variables, such as younger age for men, may confound such analyses. When we controlled for these differences using multivariate regression models, the opposite was found to be the case: women were more likely to return to independent living (OR 1.19).
Patients who are independently mobile before hip fracture also represent an important group, as loss of mobility may significantly affect social care and place of residence. In this analysis we looked exclusively at patients able to walk independently without aids prior to injury. At the time of fracture those who could mobilise independently were similar in both groups (45% men vs 44% women, p = 0.051). It is evident that hip fracture has a significant effect on independent mobility as only 26% of men and 27% of women who were independently mobile beforehand had regained this status at 120 days after fracture (p = 0.776). Mobility depends on a number of factors, including age and comorbidity. When this outcome was analysed using the multivariate regression model, women were more likely to return to independence than men (OR = 1.25). Furthermore, whereas univariate analysis suggests that men are more likely to return to independent mobility, more comprehensive multivariate analysis indicates that women have better outcomes.
In conclusion, we have demonstrated that compared with women, men have significant differences in case-mix variables, such as higher ASA scores, which are likely to influence outcome. However, even when adjusting for differences in patient characteristics using multivariate logistic regression models, gender as an isolated variable has a significant effect on outcome after fracture of the hip. The men have higher early post-operative mortality and are less likely to return to independent living or mobility.
The authors wish to thank the Local Audit Coordinators who collected Scottish Hip Fracture Audit (SHFA) data, and SHFA’s Steering Group, who provided access to the database. The authors would particularly like to acknowledge the contributions of Dr D. Reid, Chair of SHFA Steering Group and D. Beard, SHFA Project Manager.
No benefits in any form have been received or will be received from a commercial party related directly or indirectly to the subject of this article.
- Received September 27, 2007.
- Accepted December 21, 2007.
- © 2008 British Editorial Society of Bone and Joint Surgery