Abstract
Background and Aims:
Osteoporosis is a major health concern in elderly population. Low-trauma fractures offer one way of identifying these patients for treatment. Population-specific incidences of osteoporotic fractures are essential to validate tools for clinical decision making.
Material and Methods:
To evaluate population-based osteoporotic fracture incidences in Central Finland, we performed a manual search of low-trauma fractures 2005–2006 from the records of all the local health-care organizations treating trauma patients. We identified low-trauma fracture patients from radiology reports from five health centers serving the inhabitants of nine municipalities and from the patient records of two hospitals in Central Finland. The manually collected data were then compared against data from the Finnish Health Care Register.
Results:
The crude incidence of all osteoporotic fractures in 2005–2006 in the population aged 50 years was 1254/100,000 person years: 694/100,000 person years in men and 1718/100,000 person years in women. Fracture numbers derived from register data were similar to those manually collected for hip and humerus fractures, but clearly smaller than those for wrist and ankle fractures.
Conclusions:
Population-based low-trauma fracture incidences, reported here for Finland, constitute a basis for calibration of fracture risk evaluation tools (e.g. the World Health Organization fracture risk evaluation tool, FRAX). This study showed that register data underestimate the incidences of, in particular, distal radius and ankle fractures.
Keywords
Introduction
Osteoporotic fractures are a major source of morbidity and mortality in the aging population and of considerable economic importance for society. Osteoporosis is often recognized only at the occurrence of a low-trauma fracture. To avoid more fractures, it is important to evaluate the need for osteoporosis care for patients with fractures.
Currently, about 7000 hip fractures occur annually in Finland (population 5.3 million) (1). Recent data from the Finnish Health Care Register show that 7220 hip fractures occurred in Finland in 2009, 7330 in 2010, and about 7150 in 2011.
Earlier Finnish studies have reported an annual increase in the number of hip fractures of about 10% (2–7). The age-standardized incidence increased between 1970 and 1997 (8). The nationwide age-standardized hip fracture incidence stabilized, or even decreased, from the end of the 1990s to 2008 (1, 9 –11). However, small regional differences have been reported (10). In Central Finland in 1982–1983 versus 1992–1993, no change was observed in the age-standardized incidence of hip fractures (12). However, owing to the change in the age distribution of the population, the number of hip fractures at that time increased by 11% (12). Ten years later, between 1992–1993 and 2002–2003, the incidence of hip fracture among the at-risk population in Central Finland showed an increasing trend (13).
Data and studies on non-hip osteoporotic fracture incidences in Finland, excluding the cohort study in Lieto (14) and a recent report on distal radius fractures (15), are scarce and have mainly originated from register-based studies with their potential problems of incompleteness, as not all fractures require hospital inpatient care.
The aim of this study was to estimate the incidence of osteoporotic fractures in Finland and to compare manually checked fracture data from two hospitals and five health centers with the national register data on the same population. It is important to evaluate the burden on resources of these fractures as the population ages. Such information may also help to calibrate the incidence estimates used in the Finnish version of the World Health Organization (WHO) fracture risk evaluation tool (FRAX). In order to calculate the fracture incidences, we performed a manual collection of population-based data on all low-energy fractures that occurred during the years 2005–2006 in Central Finland (population 270,000).
Methods
We collected retrospective data on all low-trauma fractures in the Central Finland Health Care District over the period 1.1.2005 to 31.12.2006. Only fractures recorded for patients 50 years of age or older were included. A fracture was considered low-energy, if the trauma mechanism was falling, sliding, or dropping from a maximum height of 1 m.
The patient registry data of the Central Finland Central Hospital were searched to identify fractures. Fractures were classified using the 10th edition of the International Classification of Diseases (ICD-10) (16). We listed distal radius fractures (S52.5, S52.6), ankle (S82.5, S82.6, S82.8), vertebral (S22.0, S22.1, S32.0, S32.7), proximal humerus (S42.2), and hip fractures (S72.0, S72.1, S72.2) in the patient registry. We also classified the type of hip fractures: femoral neck (S72.0), pertrochanteric (S72.1), or subtrochanteric (S72.2). Hip fractures were analyzed separately.
All hip fractures in the Central Finland Health Care District were treated either in the Central Finland Central Hospital or in Jokilaakso Hospital, as orthopedic services are only provided in these centers. In Jokilaakso Hospital, fracture patients were found through radiograph files, as described below. Pathological fractures were excluded and the fracture mechanism checked against patient records.
A significant proportion of non-hip fracture patients do not need surgery, and can be diagnosed and treated in primary care health centers. To calculate the incidences of low-energy fractures, we also searched the records of five health centers (Jyväskylä urban municipality, Jyväskylä rural municipality, Jämsä, Laukaa, and Viitasaari). These health centers have their own X-ray units. The inhabitants of four other municipalities (Jämsänkoski, Längelmäki, Toivakka, and Uurainen) were also in the catchment areas of these X-ray units, as they did not have X-ray services of their own. All radiographs taken during the period in question were checked. The bone X-rays were evaluated by the radiologists or the specialist in internal medicine in addition to the immediate analysis of the general practitioners. In this study, we included fractures of the hip, vertebra (clinical), distal radius, proximal humerus, and ankle. The details of the fractures, for example the energy of the fracture, were confirmed from patient records. The X-rays of the patients in Jokilaakso Hospital had been taken in the radiological department shared with Jämsä health center, and hence the fracture patient data from these two health care organizations were retrieved together.
Finally, the data from all these units (two hospitals and five health centers) were combined and checked for duplicates and retreatment periods for the same fracture using the patients’ personal identity codes. Patients from municipalities other than the nine mentioned above were excluded. All together, we had complete data on (1) hip fractures in the entire population of Central Finland aged 50 years or more (population about 100,000) and (2) all osteoporotic fractures in persons aged 50 years or more resident in nine municipal corporations (population about 60,000).
We also identified fractures in the nationwide Finnish Health Care Register, using the same diagnostic criteria as described above with the Jyväskylä Patient Registry. The Finnish Health Care Register contains data on all inpatient care in public and private hospitals and health centers and outpatient care in Finnish hospitals. In order to avoid counting readmissions as new fractures, we calculated only the first admissions of each fracture type as a new fracture (11). By using the admission dates and municipality codes, we calculated the register-based numbers of fractures for the same period and population as in the manual data collection. The total counts of fractures found in the register data were divided by the counts obtained from the manual data collection and presented as the percentages.
Age differences between men and women were tested using Student’s t-test. The equivalence of the distribution of the hip fracture types was tested using the chi square goodness of fit test. Crude incidences were calculated by dividing the number of cases by the corresponding follow-up time calculated in person years (PYs), and standard formulas for rates were used to calculate 95% confidence intervals (CIs) for the incidences. The effects of age and gender on incidences were examined using a Poisson regression model. Statistical comparisons between rates derived from register and manual data were performed using standard rate ratio tests with Poisson assumption on counts. Statistical analyses were performed using software package R v2.15.1 (http://www.r-project.org) with an extension package, Muste v0.5.18 (http://www.survo.fi/muste).
The study was approved by the ethical review board of the Central Finland Health Care District, Jyväskylä, Finland. Permission to access the medical records of the study population was received from the Ministry of Social Affairs and Health.
Results
Detected Fractures
We identified 1749 low-energy fractures over the period 1 January 2005 to 31 December 2006. The numbers of the observed fractures with the corresponding follow-up times are presented by age group in Table 1. The mean age of the patients with hip fractures in the whole population of Central Finland was 80 years. The mean ages for patients with distal radius, proximal humerus, vertebral (clinical), and ankle fractures were 69, 73, 78, and 65 years, respectively. The mean ages were significantly (all p-values < 0.05) higher among women than among men (Table 1).
The observed numbers of low-trauma hip, distal radius, proximal humerus, vertebral, and ankle fractures by age group. The at-risk population is shown separately for hip fractures and other fractures. Register coverage is depicted at the end of table.
SD: standard deviation.
Hip fractures were listed according to fracture type. The proportions of identified femoral neck, pertrochanteric, and subtrochanteric hip fractures were 58%, 33%, and 9%, respectively, in men, and 63%, 28%, and 9%, respectively, in women. The distribution of hip fracture types was similar among men and women (p-value for difference 0.37).
The register-based counts of osteoporotic fractures were similar to the manually collected ones for hip and proximal humerus fractures (p-values for difference > 0.20), while a large proportion of distal radius (p < 0.001) and ankle fractures had clearly been missed by the register (p < 0.02) (Table 1). The register-based number of all vertebral fractures was also smaller than the manually collected number, but this difference was not statistically significant (p = 0.136) nor consistent across all age groups (data not shown).
The Incidence of Osteoporotic Fractures
Follow-up time for the at-risk population was 203,280 patient years for hip fractures and 119,870 patient years for the remaining fractures.
The crude incidence of all low-trauma fractures during the years 2005–2006 in individuals aged 50 years or older was 1254/100,000 PYs with 95% CI from 1190 to 1318. The incidence was 694 (95% CI 626–768) in men and 1718 (95% CI 1617–1820) in women (p-value for difference < 0.001). The incidence rates of the different categories of fractures by age group are shown in Fig. 1 together with the crude incidence rates per 100,000 PYs.
The age-specific incidence rates of low-trauma fractures in Central Finland during the years 2005–2006 separately for men and women with crude incidences per 100,000 person years.
The total incidence of low-trauma fractures increased steadily by age in both sexes: the incidence density ratio in comparison to the age group 50–64 years was 1.6 (95% CI 1.4–1.8) in the age group 65–74 years, 3.1 (95% CI 2.7–3.5) in the age group 75–84 years, and 5.3 (95% CI 4.5–6.2) in the highest age group. The most common fracture type in the population aged 50–80 years was a distal radius fracture and in the population over 80 years of age a hip fracture. The observed numbers of humerus and vertebral fractures increased in women markedly only after the age of 70 (Table 1). All fractures showed a higher incidence among females (all age-adjusted p-values < 0.02), with vertebral fractures showing the most equal distribution between the sexes (p = 0.014). The contribution of age to fracture incidence was lowest for ankle fractures (p = 0.44 vs p < 0.001 for all other fractures).
Discussion
In this population-based study, we investigated the incidences of osteoporotic fractures in Central Finland in men and women aged 50 years or more in 2005–2006. To find low-energy fractures, we screened both hospital records and radiology reports from five health centers serving the inhabitants of nine municipalities. The manually collected data were compared to the data from the Finnish Health Care Register. In addition to providing a description of the incidences of different osteoporotic fractures in Finland, this study demonstrates that the incidence of wrist and ankle fractures is underestimated by the national register.
In Central Finland, the crude incidences of hip fractures in the population aged 50 years or older in 1992–1993 was 156/100,000 PYs in men and 323/100,000 PYs in women (12). In 2002–2003, the same incidence rates were 210 in men and 450 in women (13). In our study, the corresponding hip fracture incidences were 171 in men and 364 in women, that is, closer to the figures published in 1999 than 2006, indicating that the increase in hip fracture incidence was temporary and has at least been stabilizing during recent years. At the end of the year 2004, a new osteoporosis treatment strategy, establishing the post of osteoporosis nurse and laying down detailed instructions for general practitioners regarding treatment for osteoporosis, was launched in Central Finland. This intervention could have had an impact on the fracture incidences. In addition, a few recent studies have reported a decrease (17, 18), stabilization (19, 20), or leveling of the increase (21) in the incidence of hip fractures in the Nordic and other European countries.
Register-based studies have reported the age-standardized incidence of ankle fractures in the Finnish population. In 2006, the incidence in women at the age of 65 years or older was 173/100,000 and 100/100,000 PYs in men (22). The incidences in the present population-based study were clearly higher. In fact, the actual register-based number of ankle fractures (higher than in previous register studies owing to inclusion of the fractures found in connection with hospital outpatient visits) was only 82% of the number found with the manual data collection (Table 1).
For the proximal humerus fractures found in the register data, the incidence in women aged 80 years or older was 298/100,000 PYs in 2007 (23), which is two- to threefold lower than the corresponding incidence found in this study. The comparison with register-based numbers in the current study revealed that when hospital outpatient visits were also screened, virtually all proximal humerus fractures were present in the register. This may, however, be related to local treatment practices; the hospital admission rate for non-operated fractures might be lower in some other parts of Finland. According to the Epos study, the age-adjusted incidence of humerus fractures standardized to the European population, for persons aged 50 years or older was 70/100,000 PYs in men and 200/100,000 PYs in women (24). These numbers including all humerus fractures are not directly comparable to the numbers for proximal humerus fractures found in the current study. Based on the data from the Finnish Health Care Register for the years 1998–2010, the ratio of proximal humerus fractures (S42.2) to all humerus fractures (S42.2, S42.3, S42.4) among patients aged 50 years or more was about 73%. By taking this information into account, it seems that incidences in the current study are slightly, but not statistically significantly, higher than the ones in the Epos study.
The incidence rate for distal radius fractures in Finland was calculated in the Oulu district in 2008. In the population aged 50 years or older, the incidence was 710/100,000 PYs in women and 223/100,000 PYs in men (15). In our study, which concerned only low-trauma fractures, the incidence of distal radius fractures was 658/100,000 PYs for women and 168/100,000 PYs for men. Only 60% of these fractures were found in the register despite the inclusion of hospital outpatient visits (Table 1).
The age-standardized incidence of morphometric vertebral fractures in the Epos study was 1070/100,000 PYs in women and 570/100,000 PYs in men aged 50 years and older (25). Compression fractures were identified from lateral spinal radiographs and were not clinical. Härmä et al. (26) found in a Finnish study conducted in 1966–1972 that at age 65 years or older, the incidence of thoracic vertebral compression fractures analyzed from thorax radiographs was 183/100,000 PYs in women and 44/100,000 PYs in men. In the younger age groups, men had a higher fracture rate than women (26). The limitation of the study was that the participation rate in the population aged 80 years or older was only 35%. Clinical vertebral fracture incidence in the present study was 183/100,000 PYs in women, indicating that less than 20% of compression fractures are identified by health-care organizations and radiologically verified. This also explains the inconsistent numbers found in the register in this study.
In the Finnish Lieto cohort study, all the fractures in people aged 65 years or older living in one municipality were reported from 1991 to 2002 (14). The total fracture incidences were about twofold greater than in our survey, but all fracture types were counted and low-energy fractures comprised only 79% of total fractures. In our study, the number of fractures is three to four times greater than that reported for the Lieto cohort. In women, the incidences of distal radius fracture were higher in Lieto in all the age categories, which is at least partly due to the use of wider diagnostic criteria: all S52 ICD-10 codes were applied in Lieto. In men, the corresponding incidences were similar to ours, except for the age groups in the Lieto study in which very few fractures were observed. The incidences of hip fracture were also higher in Lieto in both men and women, especially in the age group 70–84 years.
The low-trauma fracture incidence data collected directly from patient care units in this epidemiological, population-based study can be considered the most reliable obtained thus far in Finland. All the fracture data were checked from patient records. This minimizes the number of wrong positive findings, which is the major strength of this study. It is possible that some cases are missed when using manual search techniques, but the careful checking of the radiological records, fracture by fracture, was implemented to overcome this. Indeed, comparison of the manually collected data with the corresponding fracture data derived from the national register showed that the register underestimates the incidence of distal radius and ankle fractures, as a significant portion of these fractures are treated in outpatient primary care. The weakness of this study is that the detailed fracture data were obtained only from one region of the country, namely Central Finland. The question remains, whether fracture incidences are similar across the whole country.
The numbers of hip and proximal humerus fractures found in this manual data collection were similar to those in the national register. This finding might suggest that fractures of these kinds can be comprehensively identified from the current nationwide hospital registers recording inpatient and outpatient care. However, while for hip fractures, this is a reasonable assumption, for fractures of the proximal humerus, local primary care treatment practices may prefer to avoid hospital admissions if there is no need for surgery. Other studies validating the coverage of the nationwide register are needed to clarify this issue. Especially useful would be studies that conduct linkages at the individual level and also incorporate data from the new outpatient primary care data bank of the Health Care Register launched in 2011.
The incidences of population-specific low-trauma fractures, reported here for Finland, are essential for validation of the fracture risk evaluation tools (e.g. FRAX (27)), which have growing implications for clinical decision making. This study shows that register data alone are not reliable to produce required information.
Footnotes
Declaration of Conflicting Interests
Anna-Mari Koski: Consulting fees (Duodecim, Servier), lecture fees (Amgen, Astra, Bristol-Myers Squibb, GlaxoSmithKline, Eli Lilly, MSD, Novo Nordisk, Servier) and travel support for a congress (Amgen, Eli Lilly, MSD, Novo Nordisk, Sanofi, Servier).