Mortality after osteoporotic hip fracture: incidence, trends, and associated factors

The purpose of this study is to analyze mortality trend after having a first episode of osteoporotic hip fracture and factors associated with mortality over a long period of time (17 years). It was observed that the 1-year mortality rate increased 2% per year during the study period. Institutionalization combined with comorbidities is associated with a higher mortality.

Osteoporotic hip fracture is a major health problem due to increased mortality, morbidity, and functional impact in these patients (only 30–40% of these patients recover their previous functional status) [1, 2], as well as the economic cost for the National Health System. In addition, it is expected to increase as life expectancy increases.

Increased mortality after hip fracture has been widely reported. The cumulative mortality after 1 year of a hip fracture occurrence ranges between 20 and 40% [3‐7] with higher mortality rates in men than in women [8‐10]. After having a hip fracture, the mortality risk from any other cause increases between five and eight times, and although it decreases over the first years, excess mortality stays higher than mortality for the general population for a period of at least 10 years [10, 11].

Literature describes different factors associated with mortality, among them were age, masculine gender, dementia, heart pathology, institutionalization, comorbidity, and change of residence or rehabilitation programs carried out after hip fracture [12‐14].

One aspect of mortality associated with hip fracture, and seldom researched, is the trend of mortality rates over a long period of time.

The purpose of this study is to analyze mortality, as well as factors associated and trends in mortality over time in patients with osteoporotic hip fracture in an urban municipality of Spain.

Retrospective observational cohort study for the area covered by the Hospital Universitario Fundación Alcorcon from January 1, 1999, until December 31, 2015.

During the study period (1999–2015), there were a total of 3992 patients admitted due to osteoporotic hip fracture. Out of these 3992 patients, 3109 patients (77.9%) were women with an average age of 84.47 years (SD 8.45) and 803 (22.1%) were men with an average age of 81.64 years (SD 10.08). Moreover, 86.2% were patients of age over 75 years.

The majority of patients (73.9%) were living in their own homes and not institutionalized before the fracture occurrence. Among institutionalized patients, 81.95% were women. The average Charlson Index was 1.08 (SD 1.50), and in women, it was 0.70 (SD 1.14). Out of the total amount, 7.4% of patients had a Charlson Index above 2 (men 12.1% out of the total of men and women 6.1% out of the total of women).

Numerous studies proved that the mortality risk after hip fracture is increasing [10, 11, 16, 17]. This increase in mortality is not exclusive to hip fracture, and it is also associated with vertebral fractures [18, 19] and practically all major osteoporotic fractures, such as humerus fracture [20, 21], pelvis fracture [22, 26], distal end of femur fracture[22], and rib fracture [22, 26], and at all ages [18, 19]. Moreover, a second fracture results in an additional increase in mortality [18].

In this study, mortality in patients with hip fracture has been analyzed over a long period of time (1999–2015). This allowed for research into trends in mortality rates, which is currently a controversial topic. The main finding of this study is that the trend in 1-year mortality rate has increased 2% per year during the period of study. Other findings corroborate well-known results obtained from other studies: in the first place, mortality in patients with hip fracture increases a lot compared to the general population; secondly, mortality rate is higher in men than in women; thirdly, death risk is very high right after the fracture occurrence (it drops over the first year and remains stable afterwards); fourthly, institutionalization combined with comorbidity is associated with a higher mortality.

A few studies presented results on trends in mortality after osteoporotic hip fracture. Some of them, from different nationalities, show that mortality remained stable over the years [16, 23, 24]. Other studies described small increases in excess mortality due to hip fracture in specific population groups (elderly women) [10], and others described a variable reduction, although moderate in general, in mortality rates over time [25‐28]. Being a study that covers a long period of time, this study allows us to analyze trends in mortality in patients with hip fracture. The study observes a growing trend of mortality of 2% per year at 1 year, while short-term mortality remains stable (1 to 6 months). We believe that the observed increase in mortality rate is due to a growing proportion of men over women occurred over the years of the study period, and men, as is explained below, have a higher mortality rate. While in the first period of the study (1999–2002) men had 21.5% of all fractures, in the final period (2011–2015), this figure increased up to 25.7%. Another factor, probably having an impact on this raise, is the increase in comorbidity observed over the period of study. While in the first period (1999–2002) 5.3% of patients had a Charlson Index over 2, in the final period (2011–2015), it increased up to 11.1%.

In accordance with different studies, cumulative 1-year mortality after hip fracture varies between 20 and 40% [5, 29, 30] with higher rates of mortality observed in men than women [8, 31]. In previous studies done in Spain, 1-year mortality has been placed around 21–23% [12, 13] and varies widely depending on the regions [32]. In this study, a cumulative 1-year mortality of 30% in women and 43% in men has been obtained, which is clearly higher than data obtained from previous studies. The difference between results obtained in this study and previous studies can possibly be found in the methodology used. In this study, we have searched for patients’ deaths through the national database of death certificates, while in the rest of studies this was not done.

In this study, the standardized mortality index for the whole period of study (17 years) has been analyzed in comparison with the global mortality in Madrid based on data obtained from the National Institute of Statistics (INE, by its Spanish acronym). We obtained an 8 SMR. This finding is similar to those of other authors working with series such as the Forsen study [33], the Johnell study [34], or the Fransen study [35]. Forsen et al. [33] found a 1-year mortality risk of 9 in subjects aged between 50 and 74 years, 5.1 for subjects aged between 75 and 85 years, and 5.7 for subjects older than 85. Johnel et al. [34] found a 1-year mortality risk of 10.2 in men and 9.1 in women. Fransen et al. [35, 40] found a 2-year mortality risk of 7.18 in women.

Epidemiological studies have consistently proved that men present a higher risk of mortality [8, 9]. This study, as other studies did, found a higher risk of mortality in men than in women. Masculine sex presents a 30% increase in death risk compared to women. In contrast to other studies that do not find that excess mortality in men compared to women is due to comorbidities [36], this study observed that men present higher comorbidity, measured by the Charlson Index which would justify this increase of mortality.

Likewise, the study observes, as other studies do [7], the fact that death risk is very high immediately after hip fracture and that this risk rapidly drops over the first year, stabilizing with a slight trend downwards over the following years (Fig. 2).

Lastly, the role of comorbidities as a factor increasing risk mortality is controversial [37]. In our study, the risk factors most strongly associated with mortality (HR > 2) were moderate-severe liver disease, to be older than 75 years, and heart failure. Other risk factors also associated with mortality (HR between 1 and 2) were male, Charlson index greater than 2, diabetes mellitus, living in a nursing home, COPD, myocardial infarction, dementia, and renal failure. It has been hypothesized that comorbidities play a major role as a cause of mortality after fracture. In fact, several studies [12, 38], as this study did, found an association between the severity of comorbidities and mortality risk. In a study based on Medicare USA data, after adjustment by comorbidities, the risk of mortality dropped significantly over the early post-fracture period (first 6 months) and disappeared completely over the later post fracture period of follow-up (over 6 months) [37]. Due to this dramatic drop of excess mortality with adjustment of state of health, the authors concluded that the majority of early deaths and all late deaths were not attributable to hip fracture. In a different Spanish study, the severity of comorbidities was also found to be associated with an increase of mortality after vertebral fracture [37]. Nevertheless, not all studies found an association between comorbidity and post-fracture mortality. In the Study Osteoporotic Fractures (SOF), the adjustment by comorbidity did not affect the relationship between mortality risk and any of the examined fractures [39]. Similar findings were obtained in other studies [40]. This study is groundbreaking as we had the chance to analyze the data collected from a Nursing Assessment Form on admission for a subgroup of 810 patients. This form collects specific data on the patient’s quality of life before admission, such as vision and hearing capacity, presence of urinary incontinence, fecal incontinence, Downton scale (measuring risk of falling), and Norton scale (measuring risk of developing bedsores). In this study, it was found that, with the exception of Norton scale total score, all these factors are associated with risk of death. It seems that we can conclude that a previous bad quality of life is associated with a higher mortality after hip fracture. Unfortunately, the absence of data on established treatments does not allow us to assess with certainty whether different treatments have been taken or avoided in these groups with higher risk. It is striking that among the associated factors are institutionalized and older people, that is, the more fragile patients who sometimes have to stabilize before proceeding with surgical treatment, delaying one of the known factors that increase mortality. In relation to the increase in mortality in males, the associated comorbidities that sometimes require prior stabilization is also one of the main causes of surgical delays. In daily practice, knowing the most vulnerable subgroups forces us to maximize therapeutic care in these subgroups with a higher risk of mortality.

This study also presents some limitations, among them were those typical of a retrospective study that uses administrative databases as main source, and these limitations may cause a loss of data. Additionally, since it is a large-scale population and a long period of study, some variables could not be obtained from all patients (for example, date of surgery or data collected on the Nursing Assessment Form on admission which was only implemented in our center for the last few years of study). Time delay between the fracture and surgery could not be observed. Important variables for this profile of patient, i.e., the use of different treatments such as bisphosphonates and other therapeutic methods, could not be examined since our database lacks this type of data. Another limitation of our work, it is that which refers to the use of HR in retrospective cohort studies. The use of the HR for causal inference is not straightforward even in the absence of unmeasured confounding, measurement error, and model misspecification. Endowing a HR with a causal interpretation is risky for 2 key reasons: the HR may change over time, and the HR has a built-in selection bias. Nevertheless, this study presents some strength. It is a series with a large number of patients (3992) that includes all osteoporotic hip fractures treated in a tertiary hospital. For this reason, we think that this study is very representative of the urban population of the center of Spain, and it was carried out over a long period of time (17 years). Additionally, numerous variables were collected, and some of them, related to mortality after hip fracture, have not been studied until now, such as Downton or Norton scales.

In conclusion, in this study, we found that 1-year mortality has increased over the last 17 years, and we corroborate specific aspects of mortality that were established in other previous epidemiologic studies, i.e., that mortality in patients with hip fracture increased highly compared to the general population, that mortality risk is higher in men than in women, that death risk is very high immediately after fracture and drops over the first year remaining stable afterwards, and that institutionalization combined with comorbidity are associated with higher mortality. Additionally, some factors related to quality of life before hip fracture are identified, i.e., vision disorders, hearing impairment, incontinence, or Downton scale, which are associated with an increase in mortality for these patients.